Haematologica, Volume 103, Issue 8 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 Omar I. Abdel-Wahab (New York), Hélène Cavé (Paris), Simon Mendez-Ferrer (Cambridge), Pavan Reddy (Ann Arbor), Andreas Rosenwald (Wuerzburg), Monika Engelhardt (Freiburg), Davide Rossi (Bellinzona), Jacob Rowe (Haifa, Jerusalem), 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 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); 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 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 Warsaw, Poland

Highlights of Past EHA - Cairo 2018 Chairs: P Sonneveld, J Gribben, M Qari, M Mattar, A El-Beshlawy, A Kamel, Dates: September 27 - 29, 2018 Location: Cairo, Egypt

EHA-Baltic Hematology Tutorial on Lymphoid malignancies, including Waldenstrรถm's Macroglobulinemia EHA in close collaboration with the Estonian Society of Haematology, the Lithuanian Society of Hematology and the Latvian Hematology Society Chairs: J Gribben, E Laane, S Lejniece, V Peceliunas Date: October 18-19, 2018 Tallinn, Estonia

14th Educational Course of the Lymphoma Working Party European Society for Blood and Marrow Transplantations (EBMT) Lymphoma Working Party Chairs: S Montoto, A Sureda, L Bento September 27-28, 2018 Palma, Spain

Calendar of Events updated on July 10, 2018

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

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

aÂma [haima] = blood a·matow [haimatos] = of blood lÒgow [logos]= reasoning

Scientific Latin

haematologicus (adjective) = related to blood

Scientific Latin

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects

Modern English

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.

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

Table of Contents Volume 103, Issue 8: August 2018 Cover Figure Bone marrow smear showing atypical cells with abundant cytoplasm and indistinct outlines in a patient with bone marrow metastasis of amelanotic melanoma. Courtesy of Prof. Rosangela Invernizzi.

Editorials 1251

Is a matched sibling the ideal donor for hematopoietic cell transplant? Mary Eapen

1252

G-protein coupled receptor (GPCR) mutations in lymphoid malignancies: linking immune signaling activation and genetic abnormalities Jose Angel Martinez-Climent

Review Article 1256

Hide or defend, the two strategies of lymphoma immune evasion: potential implications for immunotherapy Marie de Charette and Roch Houot

Articles Hematopoiesis

1269

CARD10, a CEBPE target involved in granulocytic differentiation Pavithra Shyamsunder et al.

Bone Marrow Failure

1278

Natural history of GATA2 deficiency in a survey of 79 French and Belgian patients Jean Donadieu et al.

Myelodysplastic Syndrome

1288

Chronic Myeloid Leukemia

1298

Safety and efficacy of second-line bosutinib for chronic phase chronic myeloid leukemia over a five-year period: final results of a phase I/II study Carlo Gambacorti-Passerini et al.

Acute Myeloid Leukemia

1308

Phase I trial of plerixafor combined with decitabine in newly diagnosed older patients with acute myeloid leukemia Gail J. Roboz et al.

1317

Outcomes of hematopoietic stem cell transplantation from unmanipulated haploidentical versus matched sibling donor in patients with acute myeloid leukemia in first complete remission with intermediate or high-risk cytogenetics: a study from the Acute Leukemia Working Party of the European Society for Blood and Marrow Transplantation Dalila Salvator et al.

Non-Hodgkin Lymphoma

1329

Novel GPR34 and CCR6 mutation and distinct genetic profiles in MALT lymphomas of different sites Sarah Moody et al.

Haematologica 2018; vol. 103 no. 8 - August 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation 1337

End-of-treatment and serial PET imaging in primary mediastinal B-cell lymphoma following dose-adjusted EPOCH-R: a paradigm shift in clinical decision making Christopher Melani et al.

1345

Rituximab plus bendamustine as front-line treatment in frail elderly (>70 years) patients with diffuse large B-cell non-Hodgkin lymphoma: a phase II multicenter study of the Fondazione Italiana Linfomi Sergio Storti et al.

1351

A phase II multicenter study of the anti-CD19 antibody drug conjugate coltuximab ravtansine (SAR3419) in patients with relapsed or refractory diffuse large B-cell lymphoma previously treated with rituximab-based immunotherapy Marek Trnĕný et al.

Plasma Cell Disorders

1359

Maternal embryonic leucine zipper kinase inhibitor OTSSP167 has preclinical activity in multiple myeloma bone disease Joséphine Muller et al.

1369

Therapeutic effects of the novel subtype-selective histone deacetylase inhibitor chidamide on myeloma-associated bone disease Jingsong He et al.

Quality of Life

1380

Meaningful changes in end-of-life care among patients with myeloma Oreofe O. Odejide et al.

Cell Therapy & Immunotherapy

1390

The early expansion of anergic NKG2Apos/CD56dim/CD16neg natural killer represents a therapeutic target in haploidentical hematopoietic stem cell transplantation Alessandra Roberto et al.

Coagulation & Its Disorders

1403

Dexamethasone promotes durable factor VIII-specific tolerance in hemophilia A mice via thymic mechanisms Maria T. Georgescu et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/103/8.toc

e331

C/EBPg is dispensable for steady-state and emergency granulopoiesis Miroslava Kardosova et al. http://www.haematologica.org/content/103/8/e331

e336

GATA1s exerts developmental stage-specific effects in human hematopoiesis Sofia Gialesaki et al. http://www.haematologica.org/content/103/8/e336

e341

miR-181a regulates erythroid enucleation via the regulation of Xpo7 expression Amalia Avila Figueroa et al. http://www.haematologica.org/content/103/8/e341

e345

Concurrent treatment of aplastic anemia/paroxysmal nocturnal hemoglobinuria syndrome with immunosuppressive therapy and eculizumab: a UK experience Morag Griffin et al. http://www.haematologica.org/content/103/8/e345

Haematologica 2018; vol. 103 no. 8 - August 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation e348

Molecular genetic characterization of myeloid/lymphoid neoplasms associated with eosinophilia and rearrangement of PDGFRA, PDGFRB, FGFR1 or PCM1-JAK2 Constance Baer et al. http://www.haematologica.org/content/103/8/e348

e351

Automated decision tree to evaluate genetic abnormalities when determining prognostic risk in acute myeloid leukemia Kevin Watanabe-Smith et al. http://www.haematologica.org/content/103/8/e351

e356

Increased rituximab exposure does not improve response and outcome of patients with chronic lymphocytic leukemia after fludarabine, cyclophosphamide, rituximab. A French Innovative Leukemia Organization (FILO) study Guillaume Cartron et al. http://www.haematologica.org/content/103/8/e356

e360

RNA fusions involving CD28 are rare in peripheral T-cell lymphomas and concentrate mainly in those derived from follicular helper T cells David Vallois et al. http://www.haematologica.org/content/103/8/e360

e364

Impact of age on genetics and treatment efficacy in follicular lymphoma Stefan Alig et al. http://www.haematologica.org/content/103/8/e364

e368

Cereblon loss and up-regulation of c-Myc are associated with lenalidomide resistance in multiple myeloma patients Laurens E. Franssen et al. http://www.haematologica.org/content/103/8/e368

Case Reports Case Reports are available online only at www.haematologica.org/content/103/8.toc

e372

Germline JAK2 L611S mutation in a child with thrombocytosis Bernard Aral et al. http://www.haematologica.org/content/103/8/e372

e374

True, true unrelated? Coexistence of Waldenstrรถm macroglobulinemia and cardiac transthyretin amyloidosis Avinainder Singh et al. http://www.haematologica.org/content/103/8/e374

e377

BRAF V600E mutation detected in a case of Rosai-Dorfman disease Giancarlo Fatobene et al. http://www.haematologica.org/content/103/8/e377

Comments Comments are available online only at www.haematologica.org/content/103/8.toc

e380

Unproven value of end-of-treatment and serial follow-up FDG-PET in primary mediastinal B-cell lymphoma Hugo J.A. Adams et al. http://www.haematologica.org/content/103/8/e380

e382

End-of-treatment and serial PET imaging has prognostic value and clinical utility in primary mediastinal B-cell lymphoma following dose-adjusted EPOCH-R - Response to Adams et al. Christopher Melani et al. http://www.haematologica.org/content/103/8/e382

Haematologica 2018; vol. 103 no. 8 - August 2018 http://www.haematologica.org/

EDITORIALS Is a matched sibling the ideal donor for hematopoietic cell transplant? Mary Eapen Department of Medicine, Medical College of Wisconsin, Milwaukee, WI, USA E-mail: meapen@mcw.edu doi:10.3324/haematol.2018.196980

n this issue of Haematologica, Salvatore and colleagues compare outcomes after transplantation of grafts from a haploidentical relative or a HLA-matched sibling for adults with acute myeloid leukemia in first complete remission.1 They conclude survival was inferior after transplantation of peripheral blood or bone marrow from a haploidentical relative compared to an HLA-matched sibling for patients with intermediate-risk cytogenetics. On the other hand, for patients with high-risk cytogenetics, survival did not differ by donor type despite a non-significant reduction in relapse risk after haploidentical transplant. Non-relapse mortality risks were higher after haploidentical transplantation negated any survival advantage to be expected with the modest reduction in relapse after haploidentical transplantation for AML with high-risk cytogenetics. The observation of a reduction in relapse risk, albeit non-significant, after haploidentical transplantation is intriguing. While it is tempting to attribute this to an enhanced graft-versus-host leukemia effect in the setting of an HLA-mismatched transplant, the reduction in relapse risk was only seen for patients with high-risk cytogenetics. Can this be explained by differences in transplant conditioning regimen intensity? The study population received both myeloablative and reduced intensity transplant conditioning regimens. Among patients with intermediate risk cytogenetics, reduced intensity conditioning was associated with higher relapse. Yet, in the group of patients with high-risk cytogenetics, relapse risks did not differ by transplant conditioning regimen intensity leading us to conclude this merits further investigation. These data raise a fundamental question: when should we select an HLA-mismatched relative instead of an HLAmatched sibling? If an HLA-matched sibling is medically unfit or unwilling to donate, an HLA-mismatched relative could be the obvious choice for a number of reasons including, but not limited to, the ease of availability of the donor and timing of transplantation. Yet, when an HLA-matched sibling is medically fit and willing to donate are there circumstances that warrant selection of a haploidentical relative? A recent joint report from the European Society for Blood and Marrow Transplant and the Center for International Blood and Marrow Transplant explored whether post-transplant cyclophosphamide can nullify the detrimental effect of HLA mismatch for acute myeloid and lymphoblastic leukemia.2 The report showed haploidentical siblings donated to adult patients younger than 55 years and offspring donated to those 55 years and older. After adjusting for risk factors associated with survival the study concluded an HLA-matched sibling was a better choice than an offspring in patients 55 years and older. In the group with patients aged 18-54 years, a comparison of haploidentical to HLA-matched sibling transplant did not reveal differences in survival. The characteristics of the patients studied in two reports and their numbers differ1,2 and this is the most likely explanation for the differences between the two reports. As haematologica | 2018; 103(8)

the report by Salvatore and colleagues did not consider donor-recipient relationship, we do not know whether the effect of cytogenetic risk on survival may be explained by donor-recipient relationship and patient age on survival. However, both these reports present more questions in regards to donor selection. Donor age is associated with survival after unrelated donor transplantation.3 Survival is better after transplantation of grafts from younger donors after adjustment for donor-recipient HLA-match. Donor age is challenging to study in the setting of HLA-matched sibling transplants as generally the age of siblings falls within the same decade. Others have compared transplantation of grafts from a young unrelated donor and an HLA-matched sibling in older adults with hematologic malignancy and confirm there is no survival advantage when a young unrelated donor is chosen in favor of an older HLA-matched sibling.4 So, is there a potential advantage to selecting an offspring who is likely to be about 2-3 decades younger than the parent? The effects of donor age on adults with hematologic malignancy undergoing haploidentical transplantation has been studied by others.5 In that report, the age of the patient (â&#x2030;Ľ55 years) rather than the age of the donor was associated with higher mortality.5 The study did not identify any donor factors that were associated with mortality.5 It is worth noting that the numbers of haploidentical transplants available for study are modest when compared to the numbers of HLA-matched sibling and unrelated donor transplants. Therefore, with the increasing numbers of haploidentical transplants performed, it is incumbent upon the community of transplant physicians to carefully evaluate the effects of characteristics of haploidentical donors on transplant outcomes. Lastly, how can we best study donor selection for hematopoietic cell transplant? There is general agreement that when treatment options are being studied, a randomized trial is the gold standard. Planning and executing randomized trials is more easily said than done. In the context of related donor transplantation, subjects must have an HLAmatched sibling and a haploidentical relative for randomization. This in itself is limiting, as several more subjects will have a suitable haploidentical relative rather than an HLAmatched sibling. Secondly, we do not know whether there are differences amongst the haploidentical relatives and should randomization consider donor-recipient relationship. Thirdly, are physicians willing to randomize patients with an HLA-matched sibling to receive a haploidentical relative? While some may not, others may find this unacceptable. Regardless of the complexities of conducting randomized trials there is no denial in the lengthy duration of these trials and the expense incurred. Hence, there is reliance on data collected by large transplant registries to better understand the effect of donor types on transplant outcomes. In the meantime, the report by Salvatore and colleagues compels us to select an HLA-matched sibling when such a donor is 1251

Editorials

available. A haploidentical relative is a suitable alternative when an HLA-matched sibling is not available.

References 1. Salvatore D, Labopin M, Ruggeri A, et al. Outcomes of hematopoietic stem cell transplantation from unmanipulated haploidentical versus matched sibling donor in patients with acute myeloid leukemia in first complete remission with intermediate or high-risk cytogenetics: a study from the Acute Leukemia Working Party of the European Society for Blood and Marrow Transplantation. Haematologica. 2018;103(8): 1317-1328.

2. Robinson TM, Fuchs EJ, Zhang MJ, et al. Related donor transplants: has posttransplant cyclophosphamide nullified the detrimental effect of HLA mismatch? Blood Adv. 2018;2(11):1180-1186. 3. Kollman C, Spellman SR, Zhang MJ, et al. The effect of donor characteristics on survival after unrelated donor transplantation for hematologic malignancy. Blood. 2016;127(2):260-267. 4. Alousi AM, Le-Rademacher J, Saliba RM, et al. Who is the better donor older hematopoietic transplant recipients: an older-aged sibling or a young, matched unrelated volunteer? Blood. 2013;121(13): 2567-2573. 5. McCurdy SR, Zhang MJ, St. Martin A, et al. Effect of donor characteristics on haploidentical transplantation with posttransplant cyclophosphamide. Blood Adv. 2018;2(3):299-307.

G-protein coupled receptor (GPCR) mutations in lymphoid malignancies: linking immune signaling activation and genetic abnormalities Jose Angel Martinez-Climent Division of Hematological Oncology, Center for Applied Medical Research, University of Navarra, IDISNA, CIBERONC, Pamplona, Spain E-mail: jamcliment@unav.es doi:10.3324/haematol.2018.196998

arginal-zone B-cell lymphomas of mucosa-associated lymphoid tissue (MALT) arise from a background of chronic microbial infections or autoimmune disorders at diverse extranodal sites.1,2 The best characterized examples are gastric MALT lymphoma following Helicobacter pylori infection, and salivary gland or thyroid MALT lymphomas developing in patients with Sjögren syndrome or Hashimoto thyroiditis, respectively.3,4 It is now accepted that such chronic microenvironmental inflammation stimulates surface BCR, TLR and CD40 receptors in B lymphocytes that converge to activate downstream NF-κB signaling, which leads to the local expansion of autoreactive B cells eventually suffering malignant transformation through the acquisition of genetic changes.5 Among them, three hallmark chromosomal translocations, t(11;18)(q21;q21), t(14;18)(q32;q21) and t(1;14)(p22;q32), play a major part in MALT lymphoma origination through dysregulating MALT1 enzymatic activity that constitutively triggers the NF-κB pathway independently of antigenic stimuli.6-9 Other recurrent mutations in the MYD88, TBL1XR1, KLF2 and TNFAIP3 genes are similarly a consequence of chronic receptor stimulation and further promote NF-κB signaling, contributing to lymphoma transformation.10 A second signaling pathway recurrently found to be involved in marginalzone lymphoma (MZL) pathogenesis is NOTCH, primarily including mutations in the C-terminal PEST domain of NOTCH2 and NOTCH1 genes that enhance the stability of intracellular protein domains after being triggered by microenvironmental interactions.5 Thus, both the active chronic immunological stimuli and the acquired genetic abnormalities have critical roles during the development of MALT lymphoma through dysregulating similar molecular mechanisms. In this issue of the Journal, Moody et al. expand this intriguing oncogenic co-operation between immune 1252

receptor signaling and genetic abnormalities in MALT lymphoma. They report the discovery of somatic mutations in the G-protein coupled receptors (GPCRs) GPR34 and CCR6 not previously reported in human malignancies.11 The Authors performed whole exome sequencing of 21 salivary gland and thyroid tumors, and also carried out sequencing analysis of 249 MALT lymphomas, to define distinct mutation profiles in tumors of various sites. Those of the salivary gland were characterized by frequent TBL1XR1 and GPR34 mutations, whereas CCR6 changes were found in MALT lymphomas at different locations. The majority of GPR34 and CCR6 mutations clustered in the cytoplasmic tail, potentially leading to truncated gain-of-function proteins enabling constitutive ligand-dependent receptor activation.12 Thus, a novel synergistic mechanism between constitutively active NF-κB and GPCR signaling pathways is proposed to participate in the development of MALT lymphoma (Figure 1A). G-protein coupled receptors are made up of a large superfamily of cell surface ligands that regulate and transmit extracellular signals across the plasma membrane to induce a range of cellular and physiological responses. Despite this diversity, however, their structure, activation, signaling and regulatory mechanisms are remarkably conserved. GPCRs contain seven transmembrane spanning ahelices linked by three intracellular and three extracellular loop regions, an extracellular amino-terminal domain, and an intracellular carboxyl tail. In response to ligand binding, the receptor undergoes conformational changes to couple and activate heterotrimeric G proteins (Ga, Gb and Gγ) at the plasma membrane that regulate downstream signaling effectors. To turn off the response, GPCR kinases are recruited to phosphorylate the receptor and prepare them for b-arrestin binding, which compete with G protein coupling and desensitize the G-protein-mediated signaling response.13 Aberrant receptor activity has been shown in haematologica | 2018; 103(8)

Editorials

numerous disorders including cancer, ranging from deregulated patterns of expression in particular tumor entities to pathogenic gene mutations potentially contributing to malignant transformation.14,15 Consequently, GPCR pharmacological targeting is under development for a variety of indications such as inflammation, neurobiological and metabolic disorders, and cancer.14,16 The discovery of G-protein coupled receptor 34 (GPR34) mutations in MALT lymphoma is not totally unexpected, as deregulation of the GPR34 gene through juxtaposition to IGVH gene sequences has already been reported after the molecular cloning of a recurrent t(X;14)(p11;q32) chromosomal translocation.17 Elevated GPR34 expression was detected independently of the translocation in most other MALT lymphoma cases, leading to increased proliferation through constitutive activa-

tion of the ERK and NF-ÎşB pathways.17 In line with this gain-of-function oncogenic potential, the majority of GPR34 mutations identified by Moody et al. are nonsense or frameshift changes clustered in the C-terminal region, resulting in truncated proteins that would eliminate or impair a key phosphorylation motif and thus deregulate the receptor desensitization process (Figure 1B).12 The remaining GPR34 mutations are missense changes including Y327N, which locates in between the transmembrane domain and the cytoplasmic tail, and R84H and D151A at the intracellular loops, which also seem to induce constitutive receptor signaling activation.12,18 Lysophosphatidylserine, an endogenous lipid mediator generated by the hydrolysis of the membrane phospholipid phosphatidylserine, has been proposed as one of the ligands of GPR34.18 Because most MALT lymphoma cases

haematologica | 2018; 103(8)

Figure 1. G-protein coupled receptor (GPCR) mutations co-operate with constitutively active NF-ÎşB and NOTCH signaling pathways in mucosa-associated lymphoid tissue (MALT) lymphoma. (A) Representation of major signaling pathways affected by somatic genetic changes in patients with MALT lymphoma. (B) Schematic representation of GPR34 and CCR6 C-terminal mutant isoforms in MALT lymphoma (according to Moody et al.11). (C) GPR98 mutations in the extracellular domain in nodal MZL (as reported by Spina et al.27).

1253

Editorials

with GPR34 translocations or mutations were observed in the setting of autoimmune disorders,11,17 one can hypothesize that increased amounts of lysophosphatidylserine generated in salivary gland and thyroid tissues affected by chronic inflammation could have stimulated GPR34 in surrounding B lymphocytes to progressively induce malignant transformation through the acquired GPR34 genetic lesions. Chemokine receptor 6 (CCR6) is a chemokine receptor expressed on a variety of immune cells with a well-established role as a modulator of inflammation.19 Most CCR6 mutations reported by Moody et al. are also clustered within the C-terminal cytoplasmic tail, potentially resulting in constitutive receptor triggering (Figure 1C). However, despite sharing the mutation pattern with GPR34, CCR6 is stimulated by a different ligand, chemokine CCL20, since the CCR6/CCL20 axis is involved in the function of several cell types, including memory B lymphocytes, helper and regulatory T cells, and dendritic cells.20 Of note, a number of CCR6 missense genetic variants within the C-terminal domain have been associated with the occurrence of autoimmune disorders (Crohn disease and rheumatoid arthritis), and functional assays have demonstrated that these polymorphisms conferred decreased basal and/or ligand-induced CCR6 signaling.21 In several other experimental models, CCL20 and CCR6 interaction promoted intestinal carcinogenesis driven by macrophage recruitment into the intestine, while disruption of CCL20-CCR6 binding inhibited cutaneous T-cell lymphoma dissemination.19 These intriguing data provide the basis on which to define the functional role of CCR6 deregulation in MALT lymphoma. The C-terminal distribution of mutations in GPR34 and CCR6 is similar to that observed in two other oncogenic GPCRs: C-X-C chemokine receptor type 4 (CXCR4) and C-C chemokine receptor type 4 (CCR4).22,23 One-third of patients with Waldenstrรถm macroglobulinemia (WM), a rare lymphoplasmacytic lymphoma characterized by the constitutive MYD88 L265P activating mutation, exhibit CXCR4 mutations, which are also typical of the WHIM syndrome, an autosomal dominant immunodeficiency characterized by chronic neutropenia, hypogammaglobulinemia, recurrent infections, and myelokathexis.24 Functional characterization of WHIM-like mutations (i.e. S338X) in WM cells showed impaired CXCR4 receptor internalization following ligand binding, which led to enhanced AKT and ERK activation.25 Interestingly, such mutations promoted resistance to standard-of-care ibrutinib (a Bruton tyrosine kinase inhibitor) in WM cells, suggesting that additional therapies including proteasome inhibitors or CXCR4 targeting molecules could be of clinical benefit.22 While WHIM-like mutations have rarely been described in MZLs, increased expression of CXCR4 is frequently observed in splenic, nodal and MALT lymphomas, and a functional role of this Cxcr4 overexpression triggered by constitutive BCR signaling has been shown in experimental MZL models.15,26 On the other hand, CCR4 gain-of-function mutations located within the C-terminal domain are common in clinically aggressive adult T-cell leukemia/lymphoma, functionally leading to impaired receptor internalization and increased cell migration toward the CCL17 and CCL22 ligands.23 1254

Collectively, these data reveal the presence of functionally similar gain-of-function mutations in different GPCRs that are selectively observed in distinct lymphoid malignancies. However, a different type of GPCR mutations has been recently reported in nodal MZL, an entity closely related to MALT lymphoma. Spina et al. used whole-exome sequencing to identify novel mutations in G-protein coupled receptor 98 (GPR98) in 5 of 35 (14%) cases, all of which corresponded to missense changes in the large GPR98 extracellular N-terminus domain of 5800 amino acids (G1133E, K2010R, E2910Q, V2927G, K4232N).27 Interestingly, Usher syndrome type IIC, an autosomal recessive disorder characterized by congenital hearing loss and progressive retinitis pigmentosa (OMIM 605472), is caused by similar missense mutations and gene deletions within Calx-b extracellular GPR98 domains (i.e. Q2301X, S2764P, S2832X, I2906FS, M2931Fs)28 (Figure 1C). These data, together with the Usher syndrome phenotype developed by Gpr98 knock-out mice, suggest a loss-of-function role for extracellular GPR98 mutations that still has to be investigated. In line with these observations, two other GPCR members, the sphingosine-1-phosphate receptor 2 (S1PR2) and the P2Y receptor family member 8 (P2RY8), are recurrently mutated in germinal center (GC) mature B-cell lymphomas, preferentially by loss-of-function changes in the transmembrane domains.29 Further underscoring a deregulated role of GPCR signaling in GC B-cell-derived tumors, loss-of-signaling mutations disrupting the GNA13 gene (encoding the Ga13 coupled protein transmitting S1PR2/P2RY8 receptor signaling) and its effector ARHGEF1 are also frequently observed, together delineating a GPCR pathway that, when disrupted, promotes the growth and blocks the dissemination of GC B lymphocytes to induce the development of GC B-cell lymphoma.29 In summary, there is increasing evidence to support the implication of GPCR mutations in the pathogenesis of several lymphoid malignancies. Different clinical-pathological entities show functionally similar C-terminal domain mutations in specific GPCRs that seem to impair receptor internalization and induce constitutive receptor signaling, including CXCR4 in WM, CCR6 in T-cell leukemia/lymphoma, and GPR34 and CCR6 in MALT lymphoma. Conversely, mutations in transmembrane or extracellular domains of other GPCRs can be considered loss-of-function mutations that impair downstream receptor signaling, including GPR98 in nodal MZL, and S1PR2 and P2RY8 in GC B-cell lymphomas. Such unique genetic and territorial associations strongly suggest a role of tissue-specific extracellular cues that activate selective GPCR function and disturb cell dynamics to progressively cause genetic lesions. The functional consequences of most of these genetic changes are largely unknown, particularly in the context of the obvious synergistic co-operation of specific GPCR mutations with other selected genetic abnormalities on each particular tumor type. Among them, coexistence of CXCR4 and MYD88 mutations is currently a critical issue in the diagnosis and therapy of patients with WM, while other associations such as GPR34 and TBL1XR1 mutations in MALT lymphomas of the salivary glands remain to be characterized. Looking beyond this, the present study by Moody et al. provides a new insight haematologica | 2018; 103(8)

Editorials

into the pathogenesis of MALT lymphomas by linking, once again, immune receptor signaling activation and genetic abnormalities. Acknowledgments Supported by grants from ISCIII-FIS (PI16/00581), International Waldestrom Macroglobulinemia Foundation / Leukemia Lymphoma Society (MacroNext2017), and Worldwide Cancer Research (WCR15-1322).

References 1 Isaacson PG, Du MQ. MALT lymphoma: from morphology to molecules. Nat Rev Cancer. 2004;4(8):644-653. 2. Thieblemont C, Bertoni F, Copie-Bergman C, Ferreri AJ, Ponzoni M. Chronic inflammation and extra-nodal marginal-zone lymphomas of MALT-type. Semin Cancer Biol. 2014;24:33-42. 3. Zucca E, Bertoni F. The spectrum of MALT lymphoma at different sites: biological and therapeutic relevance. Blood. 2016;127(17):2082-2092. 4. Teixeira Mendes LS, Wotherspoon A. Marginal zone lymphoma: Associated autoimmunity and auto-immune disorders. Best Pract Res Clin Haematol. 2017;30(1-2):65-76. 5. Bertoni F, Rossi D, Zucca E. Recent advances in understanding the biology of marginal zone lymphoma. F1000Res. 2018;7:406. 6. Dierlamm J, Baens M, Wlodarska I, et al. The apoptosis inhibitor gene API2 and a novel 18q gene, MLT, are recurrently rearranged in the t(11;18)(q21;q21) associated with mucosa-associated lymphoid tissue lymphomas. Blood. 1999;93(11):3601-3609. 7. Sanchez-Izquierdo D, Buchonnet G, Siebert R, et al. MALT1 is deregulated by both chromosomal translocation and amplification in B-cell non-Hodgkin lymphoma. Blood. 2003;101(11):4539-4546. 8. Hailfinger S, Lenz G, Ngo V, et al. Essential role of MALT1 protease activity in activated B cell-like diffuse large B-cell lymphoma. Proc Natl Acad Sci USA. 2009;106(47):19946-19951. 9. Vicente-Dueñas C, Fontán L, Gonzalez-Herrero I, et al. Expression of MALT1 oncogene in hematopoietic stem/progenitor cells recapitulates the pathogenesis of human lymphoma in mice. Proc Natl Acad Sci USA. 2012;109(26):10534-10539. 10. Du MQ. MALT lymphoma: A paradigm of NF-κB dysregulation. Semin Cancer Biol. 2016;39:49-60. 11. Moody S, Thompson JS, Chuang SS, et al. Novel GPR34 and CCR6 mutation and distinct genetic profiles in MALT lymphomas of different sites. Haematologica. 2018;103(8):1329-1336. 12. Zhou XE, He Y, de Waal PW, et al. Identification of Phosphorylation Codes for Arrestin Recruitment by G Protein-Coupled Receptors. Cell. 2017;170(3):457-469.e13.

haematologica | 2018; 103(8)

13. Audet M, Bouvier M. Restructuring G-protein- coupled receptor activation. Cell. 2012;151(1):14-23. 14. Bar-Shavit R, Maoz M, Kancharla A, et al. G Protein-Coupled Receptors in Cancer. Int J Mol Sci. 2016;17(8). 15. Nugent A, Proia RL. The role of G protein-coupled receptors in lymphoid malignancies. Cell Signal. 2017;39:95-107. 16. Nieto Gutierrez A, McDonald PH. GPCRs: Emerging anti-cancer drug targets. Cell Signal. 2018;41:65-74. 17. Ansell SM, Akasaka T, McPhail E, et al. t(X;14)(p11;q32) in MALT lymphoma involving GPR34 reveals a role for GPR34 in tumor cell growth. Blood. 2012;120(19):3949-3957. 18. Schöneberg T, Meister J, Knierim AB, Schulz A. The G protein-coupled receptor GPR34 - The past 20 years of a grownup. Pharmacol Ther. 2018 Apr 22. [Epub ahead of print] 19. Williams IR. CCR6 and CCL20: partners in intestinal immunity and lymphorganogenesis. Ann N Y Acad Sci. 2006;1072:52-61. 20. Suan D, Kräutler NJ, Maag JLV, et al. CCR6 Defines Memory B Cell Precursors in Mouse and Human Germinal Centers, Revealing LightZone Location and Predominant Low Antigen Affinity. Immunity. 2017;47(6):1142-1153.e4. 21. Julian B, Gao K, Harwood BN, Beinborn M, Kopin AS. MutationInduced Functional Alterations of CCR6. J Pharmacol Exp Ther. 2017;360(1):106-116. 22. Roccaro AM, Sacco A, Jimenez C, et al. C1013G/CXCR4 acts as a driver mutation of tumor progression and modulator of drug resistance in lymphoplasmacytic lymphoma. Blood. 2014;123(26):41204131. 23. Nakagawa M, Schmitz R, Xiao W, et al. Gain-of-function CCR4 mutations in adult T cell leukemia/lymphoma. J Exp Med. 2014;211(13):2497-2505. 24. Balabanian K, Brotin E, Biajoux V, et al. Proper desensitization of CXCR4 is required for lymphocyte development and peripheral compartmentalization in mice. Blood. 2012;119(24):5722-5730. 25. Cao Y, Hunter ZR, Liu X, et al. The WHIM-like CXCR4(S338X) somatic mutation activates AKT and ERK, and promotes resistance to ibrutinib and other agents used in the treatment of Waldenstrom's Macroglobulinemia. Leukemia. 2015;29(1):169-176. 26. Robles EF, Mena-Varas M, Barrio L, et al. Homeobox NKX2-3 promotes marginal-zone lymphomagenesis by activating B-cell receptor signalling and shaping lymphocyte dynamics. Nature Commun. 2016;7:11889. 27. Spina V, Khiabanian H, Messina M, et al. The genetics of nodal marginal zone lymphoma. Blood. 2016;128(10):1362-1373. 28. Weston MD, Luijendijk MW, Humphrey KD, Möller C, Kimberling WJ. Mutations in the VLGR1 gene implicate G-protein signaling in the pathogenesis of Usher syndrome type II. Am J Hum Genet. 2004;74(2):357-366. 29. Muppidi JR, Schmitz R, Green JA, et al. Loss of signalling via Gα13 in germinal centre B-cell-derived lymphoma. Nature. 2014;516 (7530):254-258.

1255

REVIEW ARTICLE Ferrata Storti Foundation

Hide or defend, the two strategies of lymphoma immune evasion: potential implications for immunotherapy Marie de Charette1 and Roch Houot1,2

1 CHU Rennes, Service Hématologie Clinique, F-35033 and 2INSERM, U1236, F-35043, France

Haematologica 2018 Volume 103(8):1256-1268

ABSTRACT

vading immune eradication is a prerequisite for neoplastic progression and one of the hallmarks of cancer. Here, we review the different immune escape strategies of lymphoma and classify them into two main mechanisms. First, lymphoma cells may “hide” to become invisible to the immune system. This can be achieved by losing or downregulating MHC and/or molecules involved in antigen presentation (including antigen processing machinery and adhesion molecules), thereby preventing their recognition by the immune system. Second, lymphoma cells may “defend” themselves to become resistant to immune eradication. This can be achieved in several ways: by becoming resistant to apoptosis, by expressing inhibitory ligands that deactivate immune cells and/or by inducing an immunosuppressive (humoral and cellular) microenvironment. These immune escape mechanisms may have therapeutic implications. Their identification may be used to guide “personalized immunotherapy” for lymphoma.

Correspondence: roch.houot@chu-rennes.fr

Received: February 1, 2018. Accepted: April 24, 2018. Pre-published: July 13, 2018. doi:10.3324/haematol.2017.184192 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/01256 ©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.

1256

Introduction Since the hypothesis of “cancer immunosurveillance” proposed by Burnet and Thomas about 60 years ago,1 our knowledge about the interactions between cancer cells and the host immune system has dramatically increased. These interactions, referred to as “immunoediting”, are summarized in the three “Es” theory: Elimination, Equilibrium and Escape.2 Because of: i) genetic instability and tumor heterogeneity; and ii) immune selection pressure, tumor cells become progressively capable of avoiding immune destruction during carcinogenesis. This property of cancer cells is now recognized as a hallmark of cancer.3 The generation of an antitumor immune response requires several steps, elegantly summarized in the “cancer immunity cycle”.4 It consists of the release of tumor antigens (Ag), their capture by professional antigen-presenting cells (APC), and the priming of T cells. Then, effector T cells traffic to the tumor site, and recognize and kill cancer cells. To be effective, the priming of T cells needs two signals: i) the recognition of the MHC-Ag complex by the T-cell receptor (TCR) (signal 1); and ii) the co-stimulation by the CD80/CD86 molecules of CD28 (signal 2). Signal 1 without signal 2 leads to T-cell anergy.5 Only professional APC express both class I (MHC-I) and class II (MHC-II) major histocompatibility complex, and co-stimulatory molecules. All nucleated cells present endogenous Ag to CD8 T cells through MHC-I. Professional APC present exogenous Ag to CD4 T cells through MHC-II, but also exogenous Ag to CD8 T cells through MHC-I, a process called cross-presentation.6 B-cell lymphomas are unique among cancers because the tumor cells themselves are professional APC.7 With the advent of new immunotherapies including checkpoint inhibitors, bispecific antibodies and CAR T cells, understanding lymphoma immunity and immune evasion may be crucial to determine the optimal treatment and/or combinations for a given patient. Here, we review the different immune escape strategies of lymphoma and classify them into two main mechanisms. First, lymphoma cells may “hide” to become invisible to the immune system. Second, lymphoma cells may “defend” themselves to become resistant to immune eradication. Finally, we discuss how the understanding of haematologica | 2018; 103(8)

Immune escape mechanisms in lymphoma

these immune escape mechanisms may be used to determine the optimal immunotherapy for patients with lymphoma.

How lymphoma may hide from the immune system In order to evade immune eradication, tumor cells may first become “invisible”. This can be achieved by the loss or downregulation of molecules involved in antigen presentation (MHC), co-stimulation (CD80, CD86), and/or adhesion (CD54),8 thereby preventing their recognition by the immune system. Two types of mechanisms may be responsible for the loss of these molecules: i) “hard lesions” which consist of irreversible genetic alterations of the gene of interest or genes implicated in their transcriptional regulation; and ii) “soft lesions” which are reversible epigenetic changes that repress gene expression9 (Figure 1, "hide").

Prevention of antigen presentation MHC-I loss/downregulation Loss of MHC-I at the surface of lymphoma cells (total loss or miss-localization) occurs in 55-75% of diffuse large B-cell lymphoma (DLBCL)10,11 and 63% of Hodgkin lymphomas (HL).11 Most frequently, this results from mutations of the Beta2-microglobulin (b2M) gene which occurs in 29% of DLBCL,10 50% of primary mediastinal B-cell lymphoma (PMBCL),12 and at least 50% of classical HL (cHL).13 In immune-privileged lymphomas, MHC-I loss was found in 18% of primary central nervous system lymphomas (PCNSL) but not in primary testicular lymphomas (PTL).11 In HL, MHC loss is preferentially observed in EBV-negative rather than in EBV-positive HL (83% vs. 27%).11 Patients whose Reed Stenberg cells (RS) are negative for MHC-I or b2M have a shorter progression-free survival (PFS).14 Interestingly, 9p24.1 amplification (leading to PD-L1 overexpression, as discussed below) adversely impacts survival only in HL patients in whom RS have lost MHC-I.15 Loss of MHC-I is also observed in 30% of Burkitt lymphomas (BL) and 20% of follicular lymphoma (FL)16 with rare b2M mutations.17 In FL, the frequency of b2M mutations is higher after histological transformation18 and is associated with a lower infiltration of the tumor by CD8 T cells.19 Other irreversible mechanisms leading to MHC-I loss include alterations in MHC-I gene.16,20 Unlike non-hematologic cancers, epigenetic mechanisms do not seem to be frequently responsible for MHC-I loss/downregulation in lymphoma.7 Importantly, natural killer (NK) cells are activated in the absence of MHC-I and in the presence of CD58 (which stimulates NK cells through CD2). Interestingly, 67% of DLBCL lack CD58 surface expression, and 61% lack both CD58 and MHC-I expression, thereby preventing NK-cell activation.10 Of note, genetic alterations of CD58 are also found in transformed FL but not in FL.18 Genetic lesions disrupting the CD58 gene have been found only in 1021% of DLBCLs, suggesting alternative mechanisms.10,21,22

MHC- II loss/downregulation Transcriptional regulation Expression of MHC-II is regulated, through epigenetic mechanisms. CREBBP regulates CIITA by catalyzing haematologica | 2018; 103(8)

H3K27 acetylation at its promoter/enhancer in normal GC B cells and lymphoma cell lines.23-25 CREBBP may undergo loss-of-function mutation in the histone acetyl transferase domain. Thus, in FL and DLBCL, mutations of CREBBP prevent CIITA transcription, which in turn prevent MHCII transcription. HLA-DR expression is lost in 20% of DLBCL26 and is associated with a reduced T-cell infiltrate within the tumor27 and a poor outcome.27,28 Moreover, 19% of DLBCL have MHC-II intra-cytoplasmic aberrant localization which is associated with a worse outcome. This mislocalization is preferentially seen in BCL-2 and c-MYC double expresser lymphomas. Of note, c-MYC down-regulates enzymes implicated in the antigen presentation machinery (cf 2.1.3).29 The mechanisms of MHC-II downregulation remain incompletely understood but seem to occur at transcriptional level independently of genetic lesions on MHCII gene.30 Indeed, genes implicated in epigenetic regulation, including HMTs and HATs, are the most frequently altered genes in DLBCL (approx. 50% of GC-DLBCL and 30% of ABC-DLBCL).31 Moreover, DLBCL frequently harbor inactivating mutations of CREBP (19% of all DLBCL, 31% of GC-DLBCL and 6% of ABC-DLBCL)12 and CIITA (10% of DLBCL).12 CIITA is a target of somatic hypermutation (SHM) caused by AID.12 Finally, expression of CIITA and CREBP may be repressed through epigenetic silencing (i.e. independent of genetic alterations). Reduced expression of CIITA and CREBP is frequently found in DLBCL, leading to MHC-II downregulation and poor outcome.32-35 In some cases, MHC-II may be restored by lifting the repression of CIITA with HDAC inhibitors.33 MHC-II downregulation in DLBCL may also result from an overexpression of the transcription factor FOXP1 through a mechanism which, although not clearly elucidated, seems to be independent of CIITA.36 FOXP1 expression is associated with the nonGC phenotype (48% of GC-DLBCL vs. 71% of non-GCDLBCL)37 and a poor prognosis.38 The underlying mechanisms responsible for FOXP1 overexpression remain largely unknown. Genetic alterations on chromosome 3p leading to FOXP1 overexpression are found in a small subset of DLBCL.38 FOXP1 translocations are found in 5% of DLBCL and are associated with extra-nodal localizations and high proliferative index.39 Bea et al. also reported 15% of trisomy 3 and 31% of copy number gains of the chromosome 3p in ABC-DLBCL (versus 1% in GC-DLBCL), associated with MHC-II downregulation.40 In PMBCL, MHC-II downregulation also occurs at the transcriptional level and CIITA alterations is the most common mechanism:41 CIITA breaks are found in 38-56% of PMBCL and correlate with poor outcome;12,42 CREBP mutations are present in 11% of cases12 and abnormalities on chromosome 3 can be found, although rarely.40 However, loss of expression of MHC-II is found only in 12% of PMBCL.43 This is associated with poor survival.43 In FL, there is no evidence for mutation in MHC-II genes17 but CREBBP is mutated in 32-68% of cases17,34 and CIITA in 35%44 suggesting a downregulation at the transcriptional level. Furthermore, CREBBP mutation is an early event and a driver mutation in FL development.45 In HL, lack of MHC-II on RS occurs in 41% of cases and represents an independent prognosis factor.46 In 37.2% of cases, RS show aberrant localization in their cytoplasm.46 The mechanisms responsible for MHC-II loss in HL is not completely known but genomic CIITA break is found in 15% of HL42 and FOXP1 is not implicated.47 1257

M. de Charette et al.

Genetic alterations Direct, genetic alterations leading to MHC-II loss are mostly seen in DLBCL of immune-privileged sites. PTL and PCNSL have lost HLA-DR in 61% and 46% of cases, respectively.48 In contrast with other types of DLBCL, genetic lesions of MHC-II genes represent the main mechanism of HLA-DR loss:48,49 MHC-II is mutated in 78% of PTL and 50% of PCNSL.49 Transcription factors seem to be rarely implicated in HLA-II loss in PTL: CIITA and FOXP1 rearrangements are present in only 10% and 7% of cases, respectively.50 It is noteworthy that, when expressed, MHC-II may drive inhibitory signals. Indeed, lymphocyte-activation gene 3 (LAG-3), a member of immunoglobulin superfamily expressed on tumor infiltrating lymphocytes (TILs),51 binds to MHC-II with greater affinity than CD4, leading

to the inhibition of TCR signaling, proliferation and cytokine secretion by antigen-specific T cells. Exhausted LAG-3 positive TILs are present in the immune infiltrate of FL, DLBCL and HL (mostly in EBV positive cases, mixed cellularity and rich lymphocyte subtypes).52,53 Furthermore, circulating CD4 T cells from HL patients with active disease express LAG-3 at higher levels than healthy donors or patients in long-term remission.53 Antigen processing machinery alterations GILT and HLA-DM are enzymes of the antigen processing machinery (APM), located in the endocytic compartment of APC and B cells. Both are down-regulated by cMYC, leading to a defective antigen presentation that can be restored in vitro by cMYC inhibitors.54 GILT generates epitopes to be loaded on MHC-II. In

Figure 1. Lymphoma immune evasion mechanisms. (Top left panel) "Hide". Tumor cells may become “invisible” to the immune system by down-regulating MHC, costimulatory (CD80 and CD86) and/or adhesion (CD54) molecules. Downregulation of CD58 allows tumor cells to escape killing by natural killer (NK) cells, which are activated by self-missing signal (loss of MHC-I). (Right panel) "Defend". Tumor cells are seen by the immune system but avoid destruction through resistance to apoptosis signals and/or expression of inhibitory receptors. Tumor cells may resist apoptosis by different means: loss of FAS and/or TRAIL receptors (extrinsic pathway), hyperexpression of anti-apoptotic molecules such as BCL-2 (intrinsic pathway) or PI9 (Granzyme pathway). T cells can be inhibited by inhibitory ligands which are expressed by lymphoma cells or cells from their microenvironment such as PD-L1 or PD-L2/PD-1, LAG-3/MHC-II, CTLA-4/CD80 or CD86 and HLA-G/ILT. CD47 sends a “don’t eat me” signal to macrophages and DCs by interacting with its ligand SIRPa. Tumor cells may also express FAS-L to induce death of immune cells. Some molecules expressed by lymphoma cells may have dual roles: expression of MHC-II allows antigen presentation but also binds to the inhibitory receptor LAG-3; CD80 and CD86 stimulate T cells through CD28 but may also inhibit T cells through CTLA-4. (Bottom left panel) Immunosuppressive microenvironment. The tumor cells interact with their microenvironment to contribute to lymphoma immune evasion. IL-10 is a potent immunosuppressive cytokine that inhibits priming by dendritic cells (DC), promotes Th2 and Treg differentiation and M2 macrophages; TGF-b induces exhausted phenotype of CTL and Treg differentiation; IDO suppresses cytotoxic T lymphocyte (CTL) and NK immune response through degradation of tryptophan and production of kynurenine. Trp: tryptophan; Kyn: kynurenine; Gal: galectin; Ag: antigen.

1258

haematologica | 2018; 103(8)

Immune escape mechanisms in lymphoma

DLBCL patients treated with CHOP or rituximab-CHOP, Phipps-Yonas et al. identified lower GILT expression as an adverse prognostic factor for OS.55 Once formatted by GILT, peptides are loaded on MHC-II instead of CLIP fragment of invariant chain. This exchange is performed by HLA-DM. In absence of HLA-DM, antigens cannot be exposed and MHC-II present CLIP at the cell surface.11 HLA-DM is lost in 49% of cHL, 14% of DLBCL, and 2.9% of PTL and PCNSL.11

Prevention of co-stimulation: B7 molecule downregulation CD80 and CD86 are members of the B7 co-stimulatory family and are expressed on professional APC, including B cells. They have a dual specificity: they can bind to the stimulatory receptor CD28 promoting T-cell activation and to the inhibitory receptor CTLA-4 (with a much higher affinity than CD28) leading to T-cell inhibition.56 In B-cell lymphomas, CD80 and CD86 may be expressed on tumor cells and/or on cells from their microenvironment.57 CD80 is expressed in 97% of FL, 91% of marginal zone lymphomas (MZL), 90% of DLBCL, and 75% of mantle cell lymphomas (MCL).58 Interestingly, T and non-T cells present in the microenvironment of these tumors also express CD80.58 Loss of CD86 was found to be associated with decreased TIL infiltration in DLBCL.59 However, the prognostic value of CD80 and CD86 expression in lymphoma remains unclear, maybe because of their dual activity.

Prevention of adhesion Intercellular adhesion molecule 1 (ICAM-1 or CD54) plays a crucial role in cell-to-cell interaction, especially in the immune synapse and tumor cell adhesion and dissemination.8 Lower expression of CD54 compromises the interaction between tumor and immune cells. In DLBCL, lymphocyte infiltration is decreased in tumors which have lost CD54.59 However, in aggressive NHL, lower expression of CD54 correlates with more advanced stage of the disease, higher bone marrow infiltration and worse prognosis.60 Expression of CD54 is lost in 50%60 of non-Hodgkin lymphomas (NHL), but only 7% in DLBCL.59

How lymphoma may defend itself against the immune system Lymphoma cells may “defend” themselves to become resistant to immune eradication. This can be achieved in several ways: by becoming resistant to apoptosis and/or by expressing inhibitory ligands that deactivate immune cells (Figure 1, "defend").

Resistance to apoptosis Three apoptopic pathways may induce cell death: i) the perforin/granzyme pathway which results from the release of cytotoxic granules from NK cells or CTL activated through their TCR; ii) the extrinsic pathway, activated by T and NK cells through FAS or TRAIL death receptors; iii) the intrinsic pathway, involving BCL-2 family proteins and activated by intrinsic stress signals.61 By apoptopic gene profiling, Muris et al. identified two subsets of DLBCL with poor overall survival.62 The activated apoptosis cascade group (mostly ABC-DLBCL) was haematologica | 2018; 103(8)

characterized by high expression level of many pro- and anti-apoptotic genes of the intrinsic pathway, suggesting that these lymphoma cells are “primed for death” and their survival depends on the high expression level of antiapoptotic genes. The cellular cytotoxic response group was characterized by the expression of apoptosis-inducing effector molecules from CTL and NK cells (granzyme, TRAIL, FASL and other) and a high resistance to chemotherapy.63 The large immune cell infiltration in this subset suggests a selection of resistant lymphoma cells under the pressure of a strong cellular immune response.

Inhibition of granzyme The protease inhibitor 9 (PI9) was found to inhibit granzyme B and therefore to protect against apoptosis.64 PI9 is expressed in DLBCL, BL and HL (in RS), but is seems to be rarely found in low-grade lymphomas.57 Of note, few studies have analyzed PI9 expression in B-cell lymphomas and there is no evidence of relationship between PI9 expression and CTL infiltration or clinical outcome.65 To our knowledge, there is no mechanism of perforin inhibition in lymphoma.

Inactivation of death receptor extrinsic pathway: FAS/TRAIL-R FAS (CD95) belongs to the TNF receptor family and ligation of FASL (CD95L) induces apoptosis through its intracellular death domain and caspase activation. This mechanism plays a crucial role in affinity selection during the GC reaction.66 Immune cells also use this mechanism to kill cancer cells.67 In normal B cells, FAS is expressed on activated B cells from the GC and is absent in mantle zone or circulating B cells. CD95 is lost in 17% of FL68 and 27% of MALT lymphomas.69 In DLBCL, CD95 is lost in 51% of extra-nodal cases69 but rarely in cutaneous cases.70 CD95 expression on lymphoma cells is associated with improved survival and response to R-CHOP therapy in DLBCL.69-72 In HL, CD95 is rarely lost.73 Mutations in the CD95 gene are more commonly found in post-GC lymphomas, including 20% of DLBCL, and 44% of extra-nodal lymphomas (all types).74,75 Surprisingly, although derived from GC, no mutation of CD95 were found in BL.75 CD95 mutations are rare in FL (6%) and in pre-GC lymphomas (<2%) such as MCL.74,75 Only 5% of HL are associated with FAS mutation in RS.73 Müschen et al. hypothesized that FAS mutations are mostly found in post-GC lymphomas because CD95 mutations are target errors in the SHM process during the GC reaction.74 However, FAS mutations do not share features of AIDmediated activity and their underlying mechanism remains unclear. In some cases, lymphoma cells expressing CD95 are resistant to apoptosis, suggesting the existence of other mechanisms. For instance, HL resist to FASinduced apoptosis by expressing c-FLIP which is located at the cell membrane where it binds to the death domain of CD95.73 High levels of soluble CD95 are associated with poor outcome,76-78 supposedly because it binds to CD95L and prevents apoptosis. As discussed below, Galectin 3 also protects tumor cells from FAS-induced death. TRAIL is also a member of TNF receptor family, which triggers the extrinsic apoptotic pathway after ligation to death receptors (TRAIL receptors 1 and 2). The role of TRAIL in B-cell lymphomagenesis has been suggested by the association between TRAIL polymorphisms and higher risk of lymphoma79 and the rapid development of spon1259

M. de Charette et al.

taneous lymphoid malignancies in mice with TRAIL deficiency.80 Loss of TRAIL receptor was found in 6.8% of NHL.81 It is mainly caused by mutations of TRAIL death domain on chromosome 8p21.3 but may also occur at the transcriptional level by mutation of p53.82 Mutations of TRAIL receptor are found in 26% of MCL (55% of leukemic MCL vs. 19% of nodal MCL) and have a more aggressive phenotype.83

Inhibition of the stress-induced intrinsic pathway: BCL-2 overexpression BCL-2 family molecules are crucial regulators of the intrinsic pathway of mitochondrial apoptosis.84 BCL-2 itself is an anti-apoptotic protein but other members of the BCL-2 family are pro-apoptotic. BCL-2 is one of most commonly mutated genes in NHL, notably in DLBCL (37% of cases, particularly in GC subtype) and FL (54% of cases),85-87 whereas it is a rare event in peripheral T-cell lymphomas, MCL and PMBL.86 The t(14;18), present in almost all FL45 and 34% of GCDLBCL88 (vs. 17% of non-GC DLBCL), juxtaposes the BCL-2 gene and the enhancer of the heavy chain immunoglobulin. Thus, it induces a constitutive overexpression of BCL-2 and exposes BCL-2 oncogene to somatic hyper-mutations in the GC.84 Other mechanisms may explain genetic variations of the BCL-2 gene in t(14;18) negative DLBCL.84 In DLBCL, BCL-2 expression (but not mutation nor translocation) were historically associated with a worse prognosis but this negative impact seems to be overcome by the addition of rituximab to CHOP chemotherapy.86,89,90 Nevertheless, BCL-2 protein expression remains the strongest independent prognostic factor in primary cutaneous DLBCL.91 In FL, Correia et al. found that the presence of BCL-2 mutation at diagnosis was an independent risk factor of transformation and death, but patients were mostly treated without rituximab.92 This observation was not confirmed in another study in which FL patients were treated with a rituximab-containing regimen.87

Inhibition / killing of immune cells PD-L1/L2 expression PD-L1 and PD-L2 are members of the CD28 family and inhibit T cells through ligation to PD-1 receptor.56 Most FL contain a rich immune infiltrate of PD1+ cells, mostly in the inter-follicular areas, but tumor cells do not express PD-L1 (PD-L2 is weakly expressed in some rare tumor cells).52 In contrast, DLBCL often express PD-L1 and PD-L2 on tumor cells and in their microenvironment.52 PD-L1 and PD-L2 are more frequently expressed on tumor cells of ABC-DLBCL (36% and 60%, respectively) than GCDLBCL (4% and 26%, respectively).93 PD-L1 is also frequently expressed on tumor cells of PMBL (71% of cases)94 and HL (97% of cases).14 In immune-privileged lymphomas, level of PD-L1 protein expression is unknown in PTL and reported in a small study of PCNS lymphomas.95 The mechanisms responsible for PD-L1 and/or PD-L2 overexpression include: i) genetic alteration in 9p24; and ii) Epstein-Barr virus (EBV) infection. In the first case, the 9p24 amplicon contains the PD-L1 and PD-L2 genes that are directly amplified and over-expressed. It also contains the JAK2 gene that, indirectly, induces the transcription of 1260

the PD-L1 and PD-L2 genes. 9p24 alterations are found in all cases of HL,14 in most cases of PMBL (9p24 amplification in 63% of cases and translocation in 20% of cases),96,97 in 54% of PTL, and 52% of PCNSL (mainly due to copy number gain, whereas translocations are rare),98 and in 19% of DLBCL (mainly due to copy number gains) particularly in the non-GC subset.99 Structural variations disrupting the 3’ region of the PD-L1 gene are also implicated in 8% of DLBCL.100 Notably, immunoglobulin locus and CIITA are common partners of PD-L1 translocation.42,98,99 Finally, EBV infection (which is present in approx. 40% of HL tumors) also induces PD-L1 expression via the viral protein LMP1.101 PD-L1 expression in the tumor is an adverse prognostic factor for HL,14 PMBL,94 and DLBCL.93 Soluble PD-L1, although not correlated with PD-L1 expression by the tumor, is also associated with a poor prognosis in DLBCL.102,103 In these studies, high level of PD-L1 was associated with the clinical and histological aggressiveness of the disease.14,52,93,102

HLA-G expression HLA-G is a non-classical MHC-I molecule transcribed in membrane-bound or soluble (sHLA-G) isoforms. HLA-G binds to the inhibitory receptors ILT2 (on lymphoid cells, including B cells, and myeloid cells) and ILT4 (on myeloid cells). HLA-G also binds to CD8 co-receptor and induces FAS-mediated apoptosis of T and NK cells.104 HLA-G is expressed in 24% of DLBCL105 and 67% of cHL (on RS) at a higher level than healthy controls.73,106 In HL, HLA-G expression is associated with the loss of MHC-I on RS and the absence of EBV.107 sHLA-G is increased in lymphoproliferative disorders and contributes to immune escape.108,109 Indeed, sHLA-G purified from plasma of patients with lymphoproliferative disorders inhibits T-cell proliferation in vitro.108 However, there is no correlation between the level of sHLA-G and clinical or pathological characteristics of the disease108 or its prognosis.110 Thus, HLA-G may have ambivalent effects in lymphoma: on one hand, sHLA-G may inhibit the proliferation of tumor B cells through ILT2 receptor whereas, on the other hand, HLA-G expressed in the tumor may promote immune escape by inhibiting NK and CTL.104

CD47 expression CD47, the expression of which is ubiquitous, interacts with the inhibitory receptor SIRPa expressed by myeloid cells and macrophages. CD47-SIRPa interaction delivers a “don’t eat me” signal to the phagocytic cells which prevents phagocytosis.111 Thus, CD47 may lead to immune evasion in two ways: i) by inhibiting phagocytosis;112,113 and ii) by inhibiting cross-presentation by dendritic cells (DC).114 In NHL, CD47 is expressed at a higher level on tumor B cells compared to normal B cells.112 Additionally, CD47 expression is increased on lymphoma cells circulating in the blood compared to lymphoma cells in lymph nodes supporting the role of CD47 in lymphoma dissemination.113 Finally, high expression of CD47 is associated with poor prognosis in DLBCL and MCL.112

FASL expression Tumor cells may also “counter-attack” immune effector cells by expressing FASL in order to kill them.115 FASL was found to be strongly expressed in aggressive B-cell lymhaematologica | 2018; 103(8)

Immune escape mechanisms in lymphoma

phomas,116 secondary cutaneous DLBCL, primary cutaneous leg-type DLBCL,70 and HL,117 but seems to be weak in non-aggressive lymphomas (such as small lymphocytic lymphoma, lymphoplasmacytic lymphoma, and grade 1 FL) and MCL.116 In DLBCL, FASL expression is an adverse prognostic marker.69-72

Immunosuppressive microenvironment Lymphoma cells may evade immune eradication by inducing an immunosuppressive (humoral and cellular) microenvironment. Interactions between the lymphoma cells and their microenvironment have been reviewed in detail by Scott and Gascoyne.118 Here, we highlight the main immunosuppressive components present in the lymphoma microenvironment (Figure 1, "immunosuppressive microenvironment").

Cytokines IL-10 secretion IL-10 is an immunosuppressive cytokine which inhibits myeloid effector cells and priming functions of DC, promotes Th2 immune responses, induces Treg, and stimulates growth and differentiation of B cells.119 Thus, IL-10 may promote lymphoma in two ways: i) by stimulating the growth of tumor B cells; ii) by inducing an immunosuppressive environment. IL-10 serum level is higher in lymphoma patients than in healthy subjects and is associated with poor prognosis.120,121 Moreover, high levels of IL10 before treatment is associated with treatment failure and a worse outcome.120-122

TGF-b secretion TGF-b inhibits CTL function and promotes an immuno-

Table 1. Overview of lymphoma immune escape mechanisms. The respective contribution of each immune escape mechanism according to lymphoma subtype.

HL: Hodgkin lymphoma; BL: Burkitt lymphoma; DLBCL: diffuse large B-cell lymphoma; PMBL: primary mediastinal B-cell lymphoma; PTL: primary testicular lymphoma; PCNS: primary central nervous system lymphoma; MCL: mantle cell lymphoma; FL: follicular lymphoma; MZL: marginal zone lymphoma; MALT: mucosal associated lymphoid tissue; PTCL-NOS: primary T-cell lymphoma not otherwise specified; AITL: angio-immunoblastic T-cell lymphoma; ALCL: anaplastic large cell lymphoma; CTCL: cutaneous T-cell lymphoma; MF: mycosis fungoid; SS: Sezary syndrome; ATLL: acute T-cell lymphoma/leukemia; ENKTL: extranodal NK/T lymphoma; MDSC: myeloid-derived suppressor cell; TAM: tumor associated macrophage; APM: antigen processing machinery. °Refers to GILT. °°Refers to HLA-DM. *Refers to mycosis fungoid. #Refers to Sezary syndrome.

haematologica | 2018; 103(8)

1261

M. de Charette et al.

suppressive environment in several ways: i) it induces an exhausted phenotype in CTL (mostly on memory T cells) with a high PD-1 and TIM-3 expression;123 ii) it leads to FOXP3 expression, mostly in naĂŻve T CD4+ cells123 and induces the differentiation of Treg; and iii) represses the expression of CD95, perforin, granzyme and cytokines.124 Because TGF-b suppresses lymphoma growth by inhibiting proliferation and apoptosis, lymphoma cells may first acquire resistance or aberrant response to TGF-b.124 This may be achieved by several mechanisms including downregulation of TGF-b receptor on lymphoma cells125 through epigenetic mechanisms,126 abnormal signal transduction127 and expression of CD109, a negative regulator of TGF-b signaling.128 Thus, there is no clear prognostic impact of TGF-b in lymphoma.

IDO expression IDO is an enzyme, expressed by lymphoma cells and cells from the microenvironment, which suppresses CTL and NK immune responses and induces Treg through

degradation of tryptophan. The most important metabolite of tryptophan is kynurenine which inhibits antigen specific proliferation and induces T-cell death.129 IDO protein is expressed in stromal cells of HL130 and approximately 30% of NHL express IDO, and intratumoral levels are significantly higher than in reactive lymph nodes.131-133 In DLBCL131-133 and HL,130,134 IDO activity is associated with a more aggressive disease and a worse outcome. Upregulation of IDO is associated with Treg infiltration in both DLBCL and HL.130,133

Galectins expression Galectins (Gal) are key regulators of inflammation. These molecules act in the extra-cellular milieu by interacting with glycosylated receptors and, at the intra-cellular level, by modulating signalization and splicing.135 Among the 15 different galectins identified, types 1 and 3 have been implicated in lymphoma immune escape. Gal-1 is known to suppress Th1 responses and promote secretion of Th2 cytokines and expansion of Treg. Gal-1 is over-

Table 2. Strategies to reverse immune escape mechanisms in lymphoma.

Treg: regulatory T cells; mAb: monoclonal antibody; DC: dendritic cell. *No result available in lymphoma patients. #Pre-clinical data.

1262

haematologica | 2018; 103(8)

Immune escape mechanisms in lymphoma

expressed in EBV-associated lymphoma cells and is associated with an increased secretion of Th2 cytokines and infiltration by Tregs.135 Gal-3 can positively or negatively regulate T-cell survival, cytokine profiles and DC function. Gal-3 protects tumor cells from death induced by FAS,136 possibly through interaction with CD45.137 Gal-3 is over-expressed in 66% of DLBCL136 (but not in BL nor in FL).

imab era.146 This may be due to the antitumor activity of macrophages through phagocytosis of rituximab-coated tumor B cells.149 This observation was further supported by the GELA-GOELAMS study showing that macrophages were associated with adverse outcome only in patients treated without rituximab while there was no difference in survival in patients treated with rituximab.150 Finally, macrophages may also promote immune evasion by expression of PDL-1.146

Cells Regulatory T cells Tregs, which are characterized by the expression of CD4, FOXP3 and CTLA-4, are responsible for the prevention of autoimmunity.138 Tregs suppress immune cells through direct contact-dependent mechanisms, including induction of effector cell death, and indirect mechanisms by secreting inhibitory cytokines (IL-10, TGF-b) or interfering with effector T-cell metabolism.138 Tregs are more numerous in lymphoma tumors than in reactive lymph nodes139 and in the blood of lymphoma patients compared to healthy controls or cured patients.139,140 Tregs are recruited by CCR4 ligands (notably in cutaneous DLBCL, HL and EBV-associated lymphomas141) or converted from a conventional into a regulatory phenotype within the tumor microenvironment by modulation of tryptophan catabolism. Interestingly, Liu et al. demonstrated that Tregs found within the tumor microenvironment of FL are highly clonal.142 In this study, the diversity of Treg TCR repertoire inversely correlated with the TCR repertoire of CD8 T cells, suggesting an antigen-specific suppression of CTL by Tregs. High level of circulating Tregs at diagnosis is an adverse prognostic factor in DLBCL and correlates with elevated LDH, advanced stage of the disease,139 and poor survival.138,143

Myeloid-derived suppressor cells Myeloid-derived suppressor cells (MDSC) were recently described and remain poorly characterized. While their immunosuppressive properties are well established, only few mechanisms have been explored in lymphoma.144 Immunosuppressive functions of MDSC include: i) secretion of immunomodulatory factors and Treg expansion; ii) modulation of amino-acid metabolism and decrease of Tcell proliferation; iii) oxidative stress; iv) inhibition of T- or NK-cell viability and homing into the lymph nodes; and v) induction of T-cell apoptosis. In B-cell lymphoma, MDSC are involved in T-cell defect through PDL-1 expression, IL10 secretion, Treg expansion, and modulation of aminoacid metabolism.144 MDSC are increased in various B-cell lymphomas (including HL, DLBCL, FL) and correlate with poor prognosis.144,145

Immune escape mechanisms in T-cell lymphomas Mechanisms of immune evasion in T-cell lymphomas are less well characterized. Best described mechanisms result from resistance to apoptosis and from PD-L1 expression. PI9 granzyme inhibitor is expressed in 21% of anaplastic large cell lymphoma (ALCL), 27% of peripheral T-cell lymphoma not otherwise specified (PTCL-NOS), 80% of NK-/T-cell nasal type lymphoma (ENKTL), and 89% of enteropathy-type NHL.63 A defect in the extrinsic apoptosis (i.e. FAS) pathway is observed in many T-cell lymphomas which may be caused by three distinct mechanisms: i) FAS mutations, which are present in 50% of ENKTL151 and in some cases of MF (<20% of cases);152 ii) decreased expression of FAS through epigenetic mechanisms such as promoter methylation (45% of Sezary Syndrome) or splicing (43% of MF, 50% of CD30CTCL);152 iii) expression of c-FLIP inhibitory protein, which is seen in 90% of ALCL153 (although the underlying mechanism is not completely elucidated). Both PD1 and PD-L1 may be expressed in T-cell lymphomas, both on tumor cells and in their microenvironment. PD-L1 is expressed on tumor cells in less than 10% of ALCL and adult T-cell lymphoma / leukemia (ATLL), 27% of cutaneous T-cell lymphoma (CTCL), approximately 60% of PTCL-NOS, 56-80% of ENKTL and 7093% of angio-immunoblastic T-cell lymphoma (AITL).154 In both ALK negative and positive ALCL, and in CTCL, PD-L1 overexpression occurs through the STAT3 pathway.154 Like in B-cell lymphomas, structural variations disrupting the 3â&#x20AC;&#x2122; region of the PD-L1 gene (27% of ATLL) and EBV infection (particularly in ENKTL) are also responsible for PDL-1 expression. FAS-L is expressed in 12% of ALCL,153 81% of mycosis fungoid (MF),155 and a majority of CTCL156 which may lead to the elimination of CTL (through FAS-induced death) and to a worse outcome.155,156 Finally, IDO may also contribute to immune escape in ATLL and is associated with a worse outcome.157

Macrophages Macrophages are divided into M1 (pro-inflammatory, CD163-) and M2 (anti-inflammatory, CD163+) subsets. M2 macrophages are recruited into the tumor or differenced in situ (notably by IL-10) and promote tumor progression.146 In HL, a meta-analysis of 22 studies showed that a high density of CD68+/CD163+ macrophages was associated with poor survival.147 In DLBCL146 and MCL,148 CD163+ macrophages correlates with poor clinical outcome. In FL, a high density of CD68+ macrophages was associated with a poor prognosis in the pre-rituximab era while it was associated with a good prognosis in the post-rituxhaematologica | 2018; 103(8)

Implications for immunotherapy Restoring antigen recognition When tumor cells hide from the immune system by preventing Ag presentation, strategies to circumvent this escape mechanism depend on the type of lesions (Table 1). If antigen presentation deficiency results from genetic irreversible lesions, then immunotherapies that are MHCindependent may bypass the lack of antigen presentation. This can be achieved with bi-specific T-cell engager antibodies (BiTE) or CAR T cells which target surface antigens 1263

M. de Charette et al.

without the need for MHC.158,159 If antigen presentation deficiency results from epigenetic reversible lesions, then one may use therapies which can induce re-expression of MHC, co-stimulatory or adhesion molecules, such as epigenetic drugs, chemotherapy, radiotherapy or certain immunotherapies (e.g. CD40 agonists, CpG, IFN).7,160 Notably, the addition of histone deacetylase inhibitor (HDACI) to R-CHOP restored MHC-II expression161 and erased the negative prognostic value associated with MHC-II loss in DLBLC.162

depletion may be achieved with anti-CTLA4 mAbs (such as ipilimumab)166,167 or mAbs against CCR-4 (such as mogamulizumab) which is preferentially expressed by Th2 and Tregs.141,168 Treg infiltration may also be decreased by low doses of cyclophosphamide through downregulation of FOXP3.160 IDO enzyme may be down-regulated using IDO inhibitors or fludarabine.169,170

Restoring cell death

The recent success of ICP blocking antibodies in cancer patients confirmed the hypothesis of “cancer immunosurveillance” and demonstrated the potency of immunotherapy for the treatment of cancer. The goal of immunotherapy is to re-educate the immune system and to reverse the immune escape mechanisms to destroy the tumor cells. B-cell lymphoma is unique because tumor cells are professional APC and therefore can present their own antigens to the immune system. Immune escape in lymphoma may occur at the priming or at the effector phase. It may result from defects in antigen presentation (which may prevent the priming of T cells or the recognition of tumor cells at the effector phase), from resistance to immune killing, or from immunosuppressive mechanisms (either directly by the tumor cells or indirectly by their microenvironment). The advent of new classes of immunotherapies (including checkpoint inhibitors, bispecific antibodies and CAR T cells) offers novel opportunities to mobilize the immune system against lymphoma.159 However, we need to determine which of these immunotherapies will be optimal for a given patient. Furthermore, some immune escape mechanisms may dampen the efficacy of these immunotherapies and may require combination with other therapies to sensitize tumor cells to immune eradication. The characterization of immune escape mechanisms may be used to guide “personalized immunotherapy”, i.e. determine the optimal immunotherapy and/or combination in a given lymphoma patient.

BCL-2 inhibitors, such as venetoclax, may sensitize tumor cells to death induced through the intrinsic pathway. They have a strong efficacy in CLL and, to a lesser extent, in some NHL (MCL, FL, DLBCL).163 Surprisingly, despite the pathophysiological importance of BCL-2 translocation in FL, venetoclax demonstrated only poor efficacy in this disease. In pre-clinical models, Gal-3 inhibitor can disturb CD45/Gal-3 interaction and restore apoptosis.137

Blocking inhibitory signals Immune checkpoint (ICP) blockade releases inhibition of effector cells but requires an intact antigen presentation and a pre-existing anti-tumor immune response. Blockade of CTLA4, PD1 and PD-L1 have demonstrated efficacy in solid tumors and hematologic malignancies.158 Surprisingly, anti-PD1 mAbs were found to be particularly efficient in HL despite the fact that MHC expression was lost in most cases, suggesting an alternative mechanism of action. Phagocytosis may be blocked by CD47 signaling. Blocking antibodies against CD47 or SIRPa can disrupt CD47-SIRPa interaction and restore phagocytosis. Blocking CD47 signaling may also potentiate the efficacy of anti-CD20 mAb by increasing antibody-dependent cellular phagocytosis (ADCP).112-114

Modulating the tumor microenvironment Immunosuppressive macrophages may be depleted by chemotherapy164 or anti-CSF-1 receptor mAb.165 Treg

References 1. Burnet M. Cancer: a biological approach. III. Viruses associated with neoplastic conditions. IV. Practical applications. Br Med J. 1957;1(5023):841-847. 2. Dunn GP, Bruce AT, Ikeda H, Old LJ, Schreiber RD. Cancer immunoediting: from immunosurveillance to tumor escape. Nat Immunol. 2002;3(11):991-998. 3. Hanahan D, Weinberg RA. Hallmarks of Cancer: The Next Generation. Cell. 2011;144(5):646-674. 4. Chen DS, Mellman I. Oncology Meets Immunology: The Cancer-Immunity Cycle. Immunity. 2013;39(1):1-10. 5. Goodnow CC, Sprent J, de St Groth BF, Vinuesa CG. Cellular and genetic mechanisms of self tolerance and autoimmunity. Nature. 2005;435(7042):590-597. 6. Heath WR, Carbone FR. Cross-Presentation in Viral Immunity and Sefl-Tolerance. Nat Rev Immunol. 2001;1(2):126-134. 7. de Charette M, Marabelle A, Houot R.

1264

8. 9.

10.

11.

12.

Conclusion

Turning tumour cells into antigen presenting cells: The next step to improve cancer immunotherapy? Eur J Cancer. 2016;6813468147. Dustin ML. The Immunological Synapse. Cancer Immunol Res. 2014;2(11):1023-1033. Garrido F, Cabrera T, Aptsiauri N. “Hard” and “soft” lesions underlying the HLA class I alterations in cancer cells: Implications for immunotherapy. Int J Cancer. 2010;127249127256. Challa-Malladi M, Lieu YK, Califano O, et al. Combined Genetic Inactivation of 2Microglobulin and CD58 Reveals Frequent Escape from Immune Recognition in Diffuse Large B Cell Lymphoma. Cancer Cell. 2011;20(6):728-740. Nijland M, Veenstra RN, Visser L, et al. HLA dependent immune escape mechanisms in B-cell lymphomas: Implications for immune checkpoint inhibitor therapy? OncoImmunology. 2017;6(4):e1295202. Dubois S, Viailly P-J, Mareschal S, et al. Next-Generation Sequencing in Diffuse Large B-Cell Lymphoma Highlights

13.

14.

15.

16.

17.

Molecular Divergence and Therapeutic Opportunities: a LYSA Study. Clin Cancer Res. 2016;22(12):2919-2928. Reichel J, Chadburn A, Rubinstein PG, et al. Flow sorting and exome sequencing reveal the oncogenome of primary Hodgkin and Reed-Sternberg cells. Blood. 2015;125(7): 1061-1072. Roemer MGM, 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):2690-2697. Roemer MGM, Advani RH, Redd RA, et al. Classical Hodgkin Lymphoma with Reduced B2M/MHC Class I Expression Is Associated with Inferior Outcome Independent of 9p24.1 Status. Cancer Immunol Res. 2016;4(11):910-916. Fangazio M, Dominguez-Sola D, Tabbò F, et al. Genetic Mechanisms of Immune Escape in Diffuse Large B Cell Lymphoma. Blood. 2014;124(21):1692. Green MR, Kihira S, Liu CL, et al. Mutations in early follicular lymphoma progenitors are

haematologica | 2018; 103(8)

Immune escape mechanisms in lymphoma

18. 19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

associated with suppressed antigen presentation. Proc Natl Acad Sci USA. 2015;112 (10): E1116-E1125. Pasqualucci L, Khiabanian H, Fangazio M, et al. Genetics of Follicular Lymphoma Transformation. Cell Rep. 2014;6(1):130-140. Kridel R, Chan FC, Mottok A, et al. Histological Transformation and Progression in Follicular Lymphoma: A Clonal Evolution Study. PLOS Med. 2016;13(12):e1002197. Drenou B, Tilanus M, Semana G, et al. Loss of heterozygosity, a frequent but a nonexclusive mechanism responsible for HLA dysregulation in non-Hodgkin’s lymphomas. Br J Haematol. 2004;127(1):40-49. Lohr JG, Stojanov P, Lawrence MS, et al. Discovery and prioritization of somatic mutations in diffuse large B-cell lymphoma (DLBCL) by whole-exome sequencing. Proc Natl Acad Sci USA. 2012;109(10):3879-3884. Cao Y, Zhu T, Zhang P, et al. Mutations or copy number losses of CD58 and TP53 genes in diffuse large B cell lymphoma are independent unfavorable prognostic factors. Oncotarget. 2016;7(50):83294–83307. Jiang Y, Ortega-Molina A, Geng H, et al. CREBBP Inactivation Promotes the Development of HDAC3-Dependent Lymphomas. Cancer Discov. 2017;7(1):3853. Zhang J, Vlasevska S, Wells VA, et al. The CREBBP Acetyltransferase Is a Haploinsufficient Tumor Suppressor in Bcell Lymphoma. Cancer Discov. 2017;7(3):322-337. Hashwah H, Schmid CA, Kasser S, et al. Inactivation of CREBBP expands the germinal center B cell compartment, down-regulates MHCII expression and promotes DLBCL growth. Proc Natl Acad Sci USA. 2017;114(36):9701-9706. Tada K, Maeshima AM, Hiraoka N, et al. Prognostic significance of HLA class I and II expression in patients with diffuse large B cell lymphoma treated with standard chemoimmunotherapy. Cancer Immunol Immunother. 2016;65(10):1213-1222. Rimsza LM. Loss of MHC class II gene and protein expression in diffuse large B-cell lymphoma is related to decreased tumor immunosurveillance and poor patient survival regardless of other prognostic factors: a follow-up study from the Leukemia and Lymphoma Molecular Profiling Project. Blood. 2004;103(11):4251-4258. Rosenwald A, Wright G, Chan WC, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-Bcell lymphoma. N Engl J Med. 2002;346(25):1937-1947. Kendrick S, Rimsza LM, Scott DW, et al. Aberrant cytoplasmic expression of MHCII confers worse progression free survival in diffuse large B-cell lymphoma. Virchows Arch. 2017;470(1):113-117. Rimsza LM. Loss of major histocompatibility class II expression in non-immune-privileged site diffuse large B-cell lymphoma is highly coordinated and not due to chromosomal deletions. Blood. 2005;107(3):11011107. Pasqualucci L, Trifonov V, Fabbri G, et al. Analysis of the coding genome of diffuse large B-cell lymphoma. Nat Genet. 2011;43(9):830-837. Cycon KA, Rimsza LM, Murphy SP. Alterations in CIITA constitute a common mechanism accounting for downregulation of MHC class II expression in diffuse large Bcell lymphoma (DLBCL). Exp Hematol.

haematologica | 2018; 103(8)

2009;37(2):184-194.e2. 33. Cycon KA, Mulvaney K, Rimsza LM, Persky D, Murphy SP. Histone deacetylase inhibitors activate CIITA and MHC class II antigen expression in diffuse large B-cell lymphoma. Immunology. 2013;140(2):259272. 34. Pasqualucci L, Dominguez-Sola D, Chiarenza A, et al. Inactivating mutations of acetyltransferase genes in B-cell lymphoma. Nature. 2011;471(7337):189-195. 35. Autio M, Jäntti K, Cervera A, Hautaniemi S, Leppä S. Low Expression of the CIITA Gene Predicts Poor Outcome in Diffuse Large BCell Lymphoma. Blood 2016;128(22):2948. 36. Brown PJ, Wong KK, Felce SL, et al. FOXP1 suppresses immune response signatures and MHC class II expression in activated B-celllike diffuse large B-cell lymphomas. Leukemia. 2016;30(3):605-616. 37. Hans CP. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275282. 38. Koon HB, Ippolito GC, Banham AH, Tucker PW. FOXP1: a potential therapeutic target in cancer. Expert Opin Ther Targets. 2007;11(7):955-965. 39. Haralambieva E, Adam P, Ventura R, et al. Genetic rearrangement of FOXP1 is predominantly detected in a subset of diffuse large B-cell lymphomas with extranodal presentation. Leukemia. 2006;20(7):13001303. 40. Bea S. Diffuse large B-cell lymphoma subgroups have distinct genetic profiles that influence tumor biology and improve geneexpression-based survival prediction. Blood. 2005;106(9):3183-3190. 41. Mottok A, Woolcock B, Chan FC, et al. Genomic Alterations in CIITA Are Frequent in Primary Mediastinal Large B Cell Lymphoma and Are Associated with Diminished MHC Class II Expression. Cell Rep. 2015;13(7):1418-1431. 42. Steidl C, Shah SP, Woolcock BW, et al. MHC class II transactivator CIITA is a recurrent gene fusion partner in lymphoid cancers. Nature. 2011;471(7338):377-381. 43. Roberts RA. Loss of major histocompatibility class II gene and protein expression in primary mediastinal large B-cell lymphoma is highly coordinated and related to poor patient survival. Blood. 2006;108(1):311-318. 44. Loeffler M, Kreuz M, et al; on behalf of the HaematoSys-Project. Genomic and epigenomic co-evolution in follicular lymphomas. Leukemia. 2015;29(2):456-463. 45. Green MR, Gentles AJ, Nair RV, et al. Hierarchy in somatic mutations arising during genomic evolution and progression of follicular lymphoma. Blood. 2013;121(9): 1604-1611. 46. Diepstra A, van Imhoff GW, Karim-Kos HE, et al. HLA Class II Expression by Hodgkin Reed-Sternberg Cells Is an Independent Prognostic Factor in Classical Hodgkin’s Lymphoma. J Clin Oncol. 2007;25(21):31013108. 47. Brown P, Marafioti T, Kusec R, Banham AH. The FOXP1 Transcription Factor is Expressed in the Majority of Follicular Lymphomas but is Rarely Expressed in Classical and Lymphocyte Predominant Hodgkin’s Lymphoma. J Mol Histol. 2005;36(4):249-256. 48. Riemersma SA, Jordanova ES, Schop RF, et al. Extensive genetic alterations of the HLA region, including homozygous deletions of

49.

50.

51.

52.

53.

54.

55.

56. 57. 58.

59.

60.

61.

62.

63.

64.

HLA class II genes in B-cell lymphomas arising in immune-privileged sites. Blood. 2000;96(10):3569-3577. Mottok A, Steidl C. Genomic alterations underlying immune privilege in malignant lymphomas. Curr Opin Hematol. 2015;22(4):343-354. Twa DD, Mottok A, Chan FC, et al. Recurrent genomic rearrangements in primary testicular lymphoma: Genomic rearrangements in primary testicular lymphoma. J Pathol. 2015;236(2):136-141. He Y, Rivard CJ, Rozeboom L, et al. Lymphocyte-activation gene-3, an important immune checkpoint in cancer. Cancer Sci. 2016;107(9):1193-1197. Laurent C, Charmpi K, Gravelle P, et al. Several immune escape patterns in nonHodgkin’s lymphomas. OncoImmunology. 2015;4(8):e1026530. Gandhi MK. Expression of LAG-3 by tumorinfiltrating lymphocytes is coincident with the suppression of latent membrane antigen-specific CD8+ T-cell function in Hodgkin lymphoma patients. Blood. 2006;108(7):2280-2289. God JM, Cameron C, Figueroa J, et al. Elevation of c-MYC Disrupts HLA Class II– Mediated Immune Recognition of Human B Cell Tumors. J Immunol. 2015;194(4):14341445. Phipps-Yonas H, Cui H, Sebastiao N, et al. Low GILT Expression is Associated with Poor Patient Survival in Diffuse Large B-Cell Lymphoma. Front Immunol. 2013;4:425. Sharpe AH, Freeman GJ. THE B7–CD28 SUPERFAMILY. Nat Rev Immunol. 2002; 2(2):116-126. Greaves P, Gribben JG. The role of B7 family molecules in hematologic malignancy. Blood. 2013;121(5):734-744. Dakappagari N, Ho SN, Gascoyne RD, Ranuio J, Weng AP, Tangri S. CD80 (B7.1) is expressed on both malignant B cells and nonmalignant stromal cells in non-Hodgkin lymphoma. Cytometry B Clin Cytom. 2012;82B(2):112-119. Stopeck AT, Gessner A, Miller TP, et al. Loss of B7. 2 (CD86) and intracellular adhesion molecule 1 (CD54) expression is associated with decreased tumor-infiltrating T lymphocytes in diffuse B-cell large-cell lymphoma. Clin Cancer Res. 2000;6(10):3904-3909. Terol MJ, López-Guillermo A, Bosch F, et al. Expression of the adhesion molecule ICAM1 in non-Hodgkin’s lymphoma: relationship with tumor dissemination and prognostic importance. J Clin Oncol. 1998;16(1):35-40. Muris JJ, Meijer CJ, Ossenkoppele GJ, Vos W, Oudejans JJ. Apoptosis resistance and response to chemotherapy in primary nodal diffuse large B-cell lymphoma. Hematol Oncol. 2006;24(3):97-104. Muris JJF, Ylstra B, Cillessen SAGM, et al. Profiling of apoptosis genes allows for clinical stratification of primary nodal diffuse large B-cell lymphomas. Br J Haematol. 2007;136(1):38-47. Bladergroen BA, Meijer CJLM, ten Berge RL, et al. Expression of the granzyme B inhibitor, protease inhibitor 9, by tumor cells in patients with non-Hodgkin and Hodgkin lymphoma: a novel protective mechanism for tumor cells to circumvent the immune system? Blood. 2002;99(1):232-237. Bird CH, Sutton VR, Sun J, et al. Selective Regulation of Apoptosis: the Cytotoxic Lymphocyte Serpin Proteinase Inhibitor 9 Protects against Granzyme B-Mediated Apoptosis without Perturbing the Fas Cell

1265

M. de Charette et al.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

1266

Death Pathway. Mol Cell Biol. 1998;18(11): 6387-6398. Muris JJF, Meijer CJLM, Cillessen SAGM, et al. Prognostic significance of activated cytotoxic T-lymphocytes in primary nodal diffuse large B-cell lymphomas. Leukemia. 2004;18(3):589-596. van Eijk M, Defrance T, Hennino A, de Groot C. Death-receptor contribution to the germinal-center reaction. Trends Immunol. 2001;22(12):677-682. Afshar-Sterle S, Zotos D, Bernard NJ, et al. Fas ligand–mediated immune surveillance by T cells is essential for the control of spontaneous B cell lymphomas. Nat Med. 2014;20(3):283-290. Kondo E, Yoshino T, Yamadori I, et al. Expression of Bcl-2 protein and Fas antigen in non-Hodgkin’s lymphomas. Am J Pathol. 1994;145(2):330. Chatzitolios A, Venizelos I, Tripsiannis G, Anastassopoulos G, Papadopoulos N. Prognostic significance of CD95, P53, and BCL2 expression in extranodal nonHodgkin’s lymphoma. Ann Hematol. 2010;89(9):889-896. Zoi-Toli O, Meijer CJ, Oudejans JJ, de Vries E, van Beek P, Willemze R. Expression of Fas and Fas ligand in cutaneous B-cell lymphomas. J Pathol. 1999;189(4):533-538. Eser B, Sari I, Canoz O, et al. Prognostic significance of Fas (CD95/APO-1) positivity in patients with primary nodal diffuse large Bcell lymphoma. Am J Hematol. 2006;81(5): 307-314. Markovic O, Marisavljevic D, Cemerikic V, et al. Clinical and prognostic significance of apoptotic profile in patients with newly diagnosed nodal diffuse large B-cell lymphoma (DLBCL): Apoptosis in nodal diffuse large B-cell lymphoma. Eur J Haematol. 2011;86(3):246-255. Poppema S. Immunobiology and pathophysiology of Hodgkin lymphomas. Hematology Am Soc Hematol Educ Program. 2005;2005:231-238. Müschen M, Rajewsky K, Krönke M, Küppers R. The origin of CD95-gene mutations in B-cell lymphoma. Trends Immunol. 2002;23(2):75-80. Grønbaek K, Straten PT, Ralfkiaer E, et al. Somatic Fas mutations in non-Hodgkin’s lymphoma: association with extranodal disease and autoimmunity. Blood. 1998;92(9): 3018-3024. Niitsu N, Sasaki K, Umeda M. A high serum soluble Fas/APO-1 level is associated with a poor outcome of aggressive non-Hodgkin’s lymphoma. Leukemia. 1999;13(9):14341440. Hara T, Tsurumi H, Takemura M, et al. Serum-soluble fas level determines clinical symptoms and outcome of patients with aggressive non-Hodgkin’s lymphoma. Am J Hematol. 2000;64(4):257-261. Hara T, Tsurumi H, Goto N, et al. Serum soluble Fas level determines clinical outcome of patients with diffuse large B-cell lymphoma treated with CHOP and R-CHOP. J Cancer Res Clin Oncol. 2009;135(10):1421-1428. Heredia-Galvez B, Ruiz-Cosano J, TorresMoreno D, et al. Association of polymorphisms in TRAIL1 and TRAILR1 genes with susceptibility to lymphomas. Ann Hematol. 2014;93(2):243-247. Zerafa N, Westwood JA, Cretney E, et al. Cutting edge: TRAIL deficiency accelerates hematological malignancies. J Immunol. 2005;175(9):5586-5590. Lee SH, Shin MS, Kim HS, et al. Somatic mutations of TRAIL-receptor 1 and TRAIL-

82.

83.

84.

85.

86. 87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

receptor 2 genes in non-Hodgkin’s lymphoma. Oncogene. 2001;20(3):399. Young KH, Weisenburger DD, Dave BJ, et al. Mutations in the DNA-binding codons of TP53, which are associated with decreased expression of TRAILreceptor-2, predict for poor survival in diffuse large B-cell lymphoma. Blood. 2007;110(13):4396-4405. Rubio-Moscardo F, Climent J, Siebert R, et al. Mantle-cell lymphoma genotypes identified with CGH to BAC microarrays define a leukemic subgroup of disease and predict patient outcome. Blood. 2005;105(11):44454454. Singh K, Briggs JM. Functional Implications of the spectrum of BCL2 mutations in Lymphoma. Mutat Res Mutat Res. 2016;769:1-18. Morin RD, Mendez-Lago M, Mungall AJ, et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature. 2011;476(7360):298. Schuetz JM, Johnson NA, Morin RD, et al. BCL2 mutations in diffuse large B-cell lymphoma. Leukemia. 2012;26(6):1383. Huet S, Szafer-Glusman E, Tesson B, et al. BCL2 mutations do not confer adverse prognosis in follicular lymphoma patients treated with rituximab. Am J Hematol. 2017;92(6): 515-519. Iqbal J, Sanger WG, Horsman DE, et al. BCL2 translocation defines a unique tumor subset within the germinal center B-cell-like diffuse large B-cell lymphoma. Am J Pathol. 2004;165(1):159-166. 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. Akyurek N, Uner A, Benekli M, Barista I. Prognostic significance of MYC , BCL2 , and BCL6 rearrangements in patients with diffuse large B-cell lymphoma treated with cyclophosphamide, doxorubicin, vincristine, and prednisone plus rituximab: MYC, BCL2, BCL6 Rearrangements in DLBCL. Cancer. 2012;118(17):4173-4183. Grange F, Petrella T, Beylot-Barry M, et al. Bcl-2 protein expression is the strongest independent prognostic factor of survival in primary cutaneous large B-cell lymphomas. Blood. 2004;103(10):3662-3668. Correia C, Schneider PA, Dai H, et al. BCL2 mutations are associated with increased risk of transformation and shortened survival in follicular lymphoma. Blood. 2015;125(4): 658-667. Kiyasu J, Miyoshi H, Hirata A, et al. Expression of programmed cell death ligand 1 is associated with poor overall survival in patients with diffuse large B-cell lymphoma. Blood. 2015;126(19):2193-2201. Bledsoe JR, Redd RA, Hasserjian RP, et al. The immunophenotypic spectrum of primary mediastinal large B-cell lymphoma reveals prognostic biomarkers associated with outcome: Immunophenotypic Prognostic Markers in PMBL. Am J Hematol. 2016;91(10):E436-E441. Berghoff AS, Ricken G, Widhalm G, et al. PD1 (CD279) and PD-L1 (CD274, B7H1) expression in primary central nervous system lymphomas (PCNSL). Clin Neuropathol. 2014;33(1):42-49. 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. 97. Twa DDW, Chan FC, Ben-Neriah S, et al. Genomic rearrangements involving programmed death ligands are recurrent in primary mediastinal large B-cell lymphoma. Blood. 2014;123(13):2062-2065. 98. Chapuy B, Roemer MG, Stewart C, et al. Targetable genetic features of primary testicular and primary central nervous system lymphomas. Blood. 2016;127(7):869-881. 99. Georgiou K, Chen L, Berglund M, et al. Genetic basis of PD-L1 overexpression in diffuse large B-cell lymphomas. Blood. 2016;127(24):3026-3034. 100. Kataoka K, Shiraishi Y, Takeda Y, et al. Aberrant PD-L1 expression through 3 -UTR disruption in multiple cancers. Nature. 2016;534(7607):402-406. 101. Green MR, Rodig S, Juszczynski P, et al. Constitutive AP-1 activity and EBV infection induce PD-L1 in Hodgkin lymphomas and posttransplant lymphoproliferative disorders: implications for targeted therapy. Clin Cancer Res Off J Am Assoc Cancer Res. 2012;18(6):1611-1618. 102. Rossille D, Gressier M, Damotte D, et al. High level of soluble programmed cell death ligand 1 in blood impacts overall survival in aggressive diffuse large B-Cell lymphoma: results from a French multicenter clinical trial. Leukemia. 2014;28(12):2367-2375. 103. Rossille D, Azzaoui I, Feldman AL, et al. Soluble programmed death-ligand 1 as a prognostic biomarker for overall survival in patients with diffuse large B-cell lymphoma: a replication study and combined analysis of 508 patients. Leukemia. 2017;31(4):988. 104. Carosella ED, Rouas-Freiss N, Roux DT-L, Moreau P, LeMaoult J. HLA-G. In: Arun K Shukla, eds. Advances in Immunology. Elsevier; 2015; p.33-144. 105. Jesionek-Kupnicka D, Bojo M, ProchorecSobieszek M, et al. HLA-G and MHC Class II Protein Expression in Diffuse Large B-Cell Lymphoma. Arch Immunol Ther Exp (Warsz). 2016;64(3):225-240. 106. Caocci G, Greco M, Fanni D, et al. HLA-G expression and role in advanced-stage classical Hodgkin lymphoma. Eur J Histochem. 2016;60(2):2606. 107. Diepstra A, Poppema S, Boot M, et al. HLAG protein expression as a potential immune escape mechanism in classical Hodgkin’s lymphoma. Tissue Antigens. 2008;71(3): 219-226. 108. Sebti Y, Le Maux A, Gros F, et al. Expression of functional soluble human leucocyte antigen-G molecules in lymphoproliferative disorders. Br J Haematol. 2007;138(2):202–212. 109. Sebti Y, Le Friec G, Pangault C, et al. Soluble HLA-G molecules are increased in lymphoproliferative disorders. Hum Immunol. 2003;64(11):1093-1101. 110. Yong P, Kim SJ, Lee SJ, Kim BS. Serum level of soluble human leukocyte antigen-G molecules in non-Hodgkin lymphoma: Does it have a prognostic value? Leuk Lymphoma. 2008;49(8):1623-1626. 111. Barclay AN, van den Berg TK. The Interaction Between Signal Regulatory Protein Alpha (SIRP-a) and CD47: Structure, Function, and Therapeutic Target. Annu Rev Immunol. 2014;32(1):25-50. 112. Chao MP, Alizadeh AA, Tang C, et al. AntiCD47 Antibody Synergizes with Rituximab to Promote Phagocytosis and Eradicate NonHodgkin Lymphoma. Cell. 2010;142(5):699713. 113. Chao MP, Tang C, Pachynski RK, Chin R, Majeti R, Weissman IL. Extranodal dissemi-

haematologica | 2018; 103(8)

Immune escape mechanisms in lymphoma nation of non-Hodgkin lymphoma requires CD47 and is inhibited by anti-CD47 antibody therapy. Blood. 2011;118(18):48904901. 114. Liu X, Pu Y, Cron K, et al. CD47 blockade triggers T cell–mediated destruction of immunogenic tumors. Nat Med. 2015;21 (10):1209-1215. 115. Peter ME, Hadji A, Murmann AE, et al. The role of CD95 and CD95 ligand in cancer. Cell Death Differ. 2015;22(4):549-559. 116.Müllauer L, Mosberger I, Chott A. Fas ligand expression in nodal non-Hodgkin’s lymphoma. Mod Pathol. 1998;11(4):369-375. 117. Verbeke CS, Wenthe U, Grobholz R, Zentgraf H. Fas ligand expression in Hodgkin lymphoma. Am J Surg Pathol. 2001;25(3):388-394. 118. Scott DW, Gascoyne RD. The tumour microenvironment in B cell lymphomas. Nat Rev Cancer. 2014;14(8):517-534. 119. Moore KW, de Waal Malefyt R, Coffman RL, O’Garra A. Interleukin-10 and the interleukin-10 receptor. Annu Rev Immunol. 2001;19683-765. 120. Bien E, Balcerska A, AdamkiewiczDrozynska E, Rapala M, Krawczyk M, Stepinski J. Pre-treatment serum levels of interleukin-10, interleukin-12 and their ratio predict response to therapy and probability of event-free and overall survival in childhood soft tissue sarcomas, Hodgkin’s lymphomas and acute lymphoblastic leukemias. Clin Biochem. 2009;42(10–11):1144-1157. 121. Lech-Maranda E, Bienvenu J, BroussaisGuillaumot F, et al. Plasma TNF- and IL-10 Level-Based Prognostic Model Predicts Outcome of Patients with Diffuse Large BCell Lymphoma in Different Risk Groups Defined by the International Prognostic Index. Arch Immunol Ther Exp (Warsz). 2010;58(2):131-141. 122. Visco C, Vassilakopoulos TP, Kliche KO, et al. Elevated Serum Levels of IL-10 are Associated with Inferior Progression-Free Survival in Patients with Hodgkin’s Disease Treated with Radiotherapy. Leuk Lymphoma. 2004;45(10):2085-2092. 123. Yang Z-Z, Grote DM, Xiu B, et al. TGFupregulates CD70 expression and induces exhaustion of effector memory T cells in Bcell non-Hodgkin’s lymphoma. Leukemia. 2014;28(9):1872-1884. 124. Taylor JG, Gribben JG. Microenvironment abnormalities and lymphomagenesis: Immunological aspects. Semin Cancer Biol. 2015;3436-45. 125. Mao S, Yang W, Ai L, Li Z, Jin J. Transforming growth factor type II receptor as a marker in diffuse large B cell lymphoma. Tumor Biol. 2015;36(12):9903-9908. 126. Pan H, Jiang Y, Boi M, et al. Epigenomic evolution in diffuse large B-cell lymphomas. Nat Commun. 2015;66921. 127. Bakkebø M, Huse K, Hilden VI, Smeland EB, Oksvold MP. TGF- -induced growth inhibition in B-cell lymphoma correlates with Smad1/5 signalling and constitutively active p38 MAPK. BMC Immunol. 2010;11(1):57. 128. Yokoyama M, Ichinoe M, Okina S, et al. CD109, a negative regulator of TGF- signaling, is a putative risk marker in diffuse large B-cell lymphoma. Int J Hematol. 2017;105(5):614-622. 129. Frumento G, Rotondo R, Tonetti M, Damonte G, Benatti U, Ferrara GB. Tryptophan-derived catabolites are responsible for inhibition of T and natural killer cell proliferation induced by indoleamine 2,3dioxygenase. J Exp Med. 2002;196(4):459468.

haematologica | 2018; 103(8)

130. Choe J-Y, Yun JY, Jeon YK, et al. Indoleamine 2, 3-dioxygenase (IDO) is frequently expressed in stromal cells of Hodgkin lymphoma and is associated with adverse clinical features: a retrospective cohort study. BMC Cancer. 2014;14(1):1. 131. Ninomiya S, Hara T, Tsurumi H, et al. Indoleamine 2,3-dioxygenase in tumor tissue indicates prognosis in patients with diffuse large B-cell lymphoma treated with RCHOP. Ann Hematol. 2011;90(4):409-416. 132. Yoshikawa T, Hara T, Tsurumi H, et al. Serum concentration of L-kynurenine predicts the clinical outcome of patients with diffuse large B-cell lymphoma treated with RCHOP. Eur J Haematol. 2010;84(4):304-309. 133. Liu X-Q, Lu K, Feng L-L, et al. Up-regulated expression of indoleamine 2,3-dioxygenase 1 in non-Hodgkin lymphoma correlates with increased regulatory T-cell infiltration. Leuk Lymphoma. 2014;55(2):405-414. 134. Masaki A, Ishida T, Maeda Y, et al. Clinical significance of tryptophan catabolism in Hodgkin lymphoma. Cancer Sci. 2018;109(1):74-83. 135. Giordano M, Croci DO, Rabinovich GA. Galectins in hematological malignancies: Curr Opin Hematol. 2013;20(4):327–335. 136. Hoyer KK, Pang M, Gui D, et al. An antiapoptotic role for galectin-3 in diffuse large B-cell lymphomas. Am J Pathol. 2004;164 (3):893-902. 137. Clark MC, Pang M, Hsu DK, et al. Galectin3 binds to CD45 on diffuse large B-cell lymphoma cells to regulate susceptibility to cell death. Blood. 2012;120(23):4635-4644. 138. Lindqvist CA, Loskog ASI. T regulatory cells in B-cell malignancy - tumour support or kiss of death? Immunology. 2012;135(4): 255-260. 139. Mittal S, Marshall NA, Duncan L, Culligan DJ, Barker RN, Vickers MA. Local and systemic induction of CD4+ CD25+ regulatory T-cell population by non-Hodgkin lymphoma. Blood. 2008;111(11): 5359-5370. 140. Wu W, Wan J, Xia R, Huang Z, Ni J, Yang M. Functional role of regulatory T cells in B cell lymphoma and related mechanisms. Int J Clin Exp Pathol. 2015;8(8):9133. 141. Ishida T, Ueda R. CCR4 as a novel molecular target for immunotherapy of cancer. Cancer Sci. 2006;97(11):1139-1146. 142. Liu X, Venkataraman G, Lin J, et al. Highly clonal regulatory T-cell population in follicular lymphoma - inverse correlation with the diversity of CD8(+) T cells. Oncoimmunology. 2015;4(5):e1002728. 143. Chang C, Wu S-Y, Kang Y-W, et al. High Levels of Regulatory T Cells in Blood Are a Poor Prognostic Factor in Patients With Diffuse Large B-Cell Lymphoma. Am J Clin Pathol. 2015;144(6):935-944. 144. Roussel M, Irish JM, Menard C, Lhomme F, Tarte K, Fest T. Regulatory myeloid cells: an underexplored continent in B-cell lymphomas. Cancer Immunol Immunother. 2017;66(8):1103–1111. 145. Azzaoui I, Uhel F, Rossille D, et al. T-cell defect in diffuse large B-cell lymphomas involves expansion of myeloid-derived suppressor cells. Blood. 2016;128(8):1081-1092. 146. Komohara Y, Niino D, Ohnishi K, Ohshima K, Takeya M. Role of tumor-associated macrophages in hematological malignancies: TAMs in hematological malignancies. Pathol Int. 2015;65(4):170-176. 147. Guo B, Cen H, Tan X, Ke Q. Meta-analysis of the prognostic and clinical value of tumorassociated macrophages in adult classical Hodgkin lymphoma. BMC Med. 2016;14 (1):159.

148. Koh YW, Shin S-J, Park C, Yoon DH, Suh C, Huh J. Absolute monocyte count predicts overall survival in mantle cell lymphomas: correlation with tumour-associated macrophages. Hematol Oncol. 2014;32(4): 178-186. 149. Kessel A, Rosner I, Toubi E. Rituximab: Beyond Simple B Cell Depletion. Clin Rev Allergy Immunol. 2008;34(1):74-79. 150. Canioni D, Salles G, Mounier N, et al. High Numbers of Tumor-Associated Macrophages Have an Adverse Prognostic Value That Can Be Circumvented by Rituximab in Patients With Follicular Lymphoma Enrolled Onto the GELA-GOELAMS FL-2000 Trial. J Clin Oncol. 2008;26(3):440-446. 151. Takakuwa T, Dong Z, Nakatsuka S, et al. Frequent mutations of Fas gene in nasal NK/T cell lymphoma. Oncogene. 2002;21(30):4702. 152. Contassot E, French LE. Epigenetic Causes of Apoptosis Resistance in Cutaneous T-Cell Lymphomas. J Invest Dermatol. 2010;130(4): 922-924. 153. Oyarzo MP, Medeiros LJ, Atwell C, et al. cFLIP confers resistance to FAS-mediated apoptosis in anaplastic large-cell lymphoma. Blood. 2006;107(6):2544-2547. 154. Falchi L. Immune Dysfunction in NonHodgkin Lymphoma: Avenues for New Immunotherapy-Based Strategies. Curr Hematol Malig Rep. 2017;12(5):484-494. 155. Ni X, Hazarika P, Zhang C, Talpur R, Duvic M. Fas Ligand Expression by Neoplastic T Lymphocytes Mediates Elimination of CD8+ Cytotoxic T Lymphocytes in Mycosis Fungoides: A Potential Mechanism of Tumor Immune Escape? Clin Cancer Res. 2001;7(9):2682-2692. 156. Vermeer MH, van Doorn R, Dukers D, Bekkenk MW, Meijer CJLM, Willemze R. CD8+ T Cells in Cutaneous T-Cell Lymphoma: Expression of Cytotoxic Proteins, Fas Ligand, and Killing Inhibitory Receptors and Their Relationship With Clinical Behavior. J Clin Oncol. 2001;19(23):4322-4329. 157. Masaki A, Ishida T, Maeda Y, et al. Prognostic Significance of Tryptophan Catabolism in Adult T-cell Leukemia/Lymphoma. Clin Cancer Res. 2015;21(12):2830-2839. 158. Batlevi CL, Matsuki E, Brentjens RJ, Younes A. Novel immunotherapies in lymphoid malignancies. Nat Rev Clin Oncol. 2015;13(1):25-40. 159. Manson G, Houot R. Next generation immunotherapies for lymphoma: one foot in the future. Ann Oncol. 2018;29(3):588-601 160. Zitvogel L, Apetoh L, Ghiringhelli F, Kroemer G. Immunological aspects of cancer chemotherapy. Nat Rev Immunol. 2008;8(1):59-73. 161. Puvvada SD, Li H, Rimsza LM, et al. A phase II study of belinostat (PXD101) in relapsed and refractory aggressive B-cell lymphomas: SWOG S0520. Leuk Lymphoma. 2016;57 (10):2359-2369. 162. Persky DO, Li H, Rimsza LM, et al. A phase I/II trial of vorinostat (SAHA) in combination with rituximab-CHOP in patients with newly diagnosed advanced stage diffuse large B-cell lymphoma (DLBCL): SWOG S0806. Am J Hematol. 2018;93(4):486-493. 163 Davids MS. Targeting BCL-2 in B-cell lymphomas. Blood. 2017;130(9):1081-1088. 164. Galluzzi L, Buqué A, Kepp O, Zitvogel L, Kroemer G. Immunological Effects of Conventional Chemotherapy and Targeted

1267

M. de Charette et al. Anticancer Agents. Cancer Cell. 2015;28(6): 690-714. 165. Ries CH, Cannarile MA, Hoves S, et al. Targeting Tumor-Associated Macrophages with Anti-CSF-1R Antibody Reveals a Strategy for Cancer Therapy. Cancer Cell. 2014;25(6):846-859. 166. Romano E, Kusio-Kobialka M, Foukas PG, et al. Ipilimumab-dependent cell-mediated cytotoxicity of regulatory T cells ex vivo by nonclassical monocytes in melanoma patients. Proc Natl Acad Sci USA. 2015;112(19):6140-6145. 167. Arce Vargas F, Furness AJS, Litchfield K, et al. Fc Effector Function Contributes to the Activity of Human Anti-CTLA-4 Antibodies. Cancer Cell. 2018;33(4):649-663.e4. 168. Fuji S, Inoue Y, Utsunomiya A, et al. Pretransplantation Anti-CCR4 Antibody Mogamulizumab Against Adult T-Cell Leukemia/Lymphoma Is Associated With Significantly Increased Risks of Severe and Corticosteroid-Refractory Graft-VersusHost Disease, Nonrelapse Mortality, and Overall Mortality. J Clin Oncol. 2016;34(28):3426-3433.

1268

169. Ninomiya S, Narala N, Huye L, et al. Tumor indoleamine 2,3-dioxygenase (IDO) inhibits CD19-CAR T cells and is downregulated by lymphodepleting drugs. Blood. 2015;125 (25):3905-3916. 170. Hanafi L-A, Gauchat D, Godin-Ethier J, et al. Fludarabine Downregulates Indoleamine 2,3-Dioxygenase in Tumors via a Proteasome-Mediated Degradation Mechanism. PLoS One. 2014;9(6):e99211. 171. Boice M, Salloum D, Mourcin F, et al. Loss of the HVEM Tumor Suppressor in Lymphoma and Restoration by Modified CAR-T Cells. Cell. 2016;167(2):405-418.e13. 172. Launay E, Pangault C, Bertrand P, et al. High rate of TNFRSF14 gene alterations related to 1p36 region in de novo follicular lymphoma and impact on prognosis. Leukemia. 2012;26(3):559. 173. Cheung K-JJ, Johnson NA, Affleck JG, et al. Acquired TNFRSF14 Mutations in Follicular Lymphoma Are Associated with Worse Prognosis. Cancer Res. 2010;70(22):91669174. 174. Wu J, Wood GS. Reduction of Fas/CD95 Promoter Methylation, Upregulation of Fas

Protein, and Enhancement of Sensitivity to Apoptosis in Cutaneous T-Cell Lymphoma. Arch Dermatol. 2011;147(4):443-449. 175. Dong R, Dong R, Zhang M, et al. Galectin-3 as a novel biomarker for disease diagnosis and a target for therapy (Review). Int J Mol Med. 2018;41(2):599-614. 176. Jelinek T, Mihalyova J, Kascak M, Duras J, Hajek R. PD-1/PD-L1 inhibitors in haematological malignancies: update 2017. Immunology. 2017;152(3):357-371. 177. Llorente L, Richaud-Patin Y, García-Padilla C, et al. Clinical and biologic effects of anti– interleukin-10 monoclonal antibody administration in systemic lupus erythematosus. Arthritis Rheum. 2000;43(8):1790-1800. 178. Vicari AP, Chiodoni C, Vaure C, et al. Reversal of Tumor-induced Dendritic Cell Paralysis by CpG Immunostimulatory Oligonucleotide and Anti–Interleukin 10 Receptor Antibody. J Exp Med. 2002;196(4): 541-549. 179. Ni X, Langridge T, Duvic M. Depletion of regulatory T cells by targeting CC chemokine receptor type 4 with mogamulizumab. Oncoimmunology. 2015;4(7): e1011524.

haematologica | 2018; 103(8)

ARTICLE

Hematopoiesis

CARD10, a CEBPE target involved in granulocytic differentiation

Ferrata Storti Foundation

Pavithra Shyamsunder,1* Haresh Sankar,1 Anand Mayakonda,1 Lin Han,1,2 Hazimah Binte Mohd Nordin,1 Teoh Weoi Woon,1 Mahalakshmi Shanmugasundaram,1 Pushkar Dakle,1 Vikas Madan1*# and H. Phillip Koeffler1,3,4#

Cancer Science Institute of Singapore, National University of Singapore, Singapore; Department of Medicine, Yong Loo Lin School of Medicine, National University of Singapore; 3Cedars-Sinai Medical Center, Division of Hematology/Oncology, UCLA School of Medicine, Los Angeles, CA, USA and 4Department of Hematology-Oncology, National University Cancer Institute of Singapore (NCIS), National University Hospital, Singapore 1 2

Haematologica 2018 Volume 103(8):1269-1277

VM and HPK share senior authorship

ABSTRACT

aturation of granulocytes is dependent on controlled gene expression by myeloid lineage restricted transcription factors. CEBPE is one of the essential transcription factors required for granulocytic differentiation. Identification of downstream targets of CEBPE is vital to understand better its role in terminal granulopoiesis. In this study, we have identified Card10 as a novel target of CEBPE. We show that CEBPE binds to regulatory elements upstream of the murine Card10 locus, and expression of CARD10 is significantly reduced in Cebpe knock-out mice. Silencing Card10 in a human cell line and in murine primary cells impaired granulopoiesis, affecting expression of genes involved in myeloid cell development and function. Taken together, our data demonstrate for the first time that Card10 is expressed in granulocytes and is a direct target of CEBPE with functions extending to myeloid differentiation.

Correspondence: csips@nus.edu.sg or csivm@nus.edu.sg

Introduction Precise levels of progenitor cell proliferation versus lineage-committed differentiation is central to the balanced functioning of the hematopoietic system.1,2 Transcription factors are fundamental elements directing differentiation during hematopoietic development.3 The lineage and stage-restricted expression pattern of these factors underlines the need for precise regulation of their function. Major transcription factors regulating myeloid development include PU.1, GFI1, IRF8, RUNX1, SCL, TAL1 and the members of C/EBP family. Each of these lineage restricted factors are known to drive the expression of a panel of target genes.4-10 CEBPE is a member of the CCAAT/enhancer binding protein (C/EBP) family of transcription factors involved in hematopoietic cell development and induction of several inflammatory mediators.11,12 CEBPE is expressed in a stage-specific manner during myeloid differentiation and regulates transition from the promyelocyte to the myelocyte.13 This transcription factor is essential for secondary and tertiary granule formation in granulocytes.14 Germline mutations of the CEBPE gene have been detected in patients with neutrophil-specific granule deficiency. Their neutrophils display atypical bilobed nuclei, lack expression of granule proteins and these patients often have frequent bacterial infections.15,16 Cebpe knock-out mice resemble this clinical phenotype displaying a block in terminal differentiation and absence of secondary granule proteins. Cebpe KO mice develop normally, except that they fail to produce functional neutrophils and eosinophils. Neutrophils from these mice have impaired chemotaxis, bactericidal activity and mice typically die of infections between 3 and 5 months of age.9 The lack of secondary granule proteins in granulocytes from these mice impairs their responses to inflammatory signals characterized by an increase in circulating immature neutrophils and recurrent pyogenic infections.17 haematologica | 2018; 103(8)

Received: February 4, 2018. Accepted: May 14, 2018. Pre-published: May 17, 2018. doi:10.3324/haematol.2018.190280 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1269 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.

1269

P. Shyamsunder et al.

Identification of novel targets of CEBPE is imperative for a better understanding of granulocytic differentiation. In this study, we utilized ChIP-seq and RNA-seq data, the former to identify global DNA binding pattern of CEBPE and the latter to note expression changes of CEBPE deficient cells. Together, these techniques allowed identification of novel targets of CEBPE associated with myelopoesis. One of the interesting targets of CEBPE that we identified in this analysis is Card10 (also called Carma3). The CARMA family has three members, CARMA1, CARMA2, and CARMA3. They contain a N-terminal CARD domain, followed by a coiled-coil domain (C-C), a PDZ domain, a SH3 domain, and a Guanylate Kinase-like (GUK) domain in the C-terminus. Although the CARMA proteins share a high degree of sequence similarity, they display a distinct tissue expression pattern.18 A study by Xin Lin et al., analysed microarray data of 353 human tissue samples and found that CARMA1 is primarily expressed in hematopoietic tissues such as spleen, thymus, and peripheral blood leukocytes. Expression of CARMA2 is specific to placenta, and CARMA3 (CARD10) is expressed in a wide range of tissues with modest expression detected in hematopoietic cells.19 CARD10 has been described as a molecular link between G proteincoupled receptors and NF-κB. GPCR induced ubiquitination of IKK-γ (NEMO) with concomitant activation of the IKK complex is completely defective in CARD10 deficient cells.18 Studies have demonstrated that CARD10 deficient murine embryonic fibroblasts have diminished lysophosphatidic acid (LPA) induced NF-κB activation and lowered IL-8 production.18 Similarly, inhibition of CARD10 in airway epithelial cells reduces LPA-mediated NF-κB activity and the secretion of NF-κB dependent cytokines, TSLP and CCL20, thus pointing to a role for CARD10 in initiating allergic inflammation.20 In the present study, we show that Card10 is a direct target of CEBPE. We verify that CEBPE binds to a region upstream of Card10 gene and its expression is upregulated in the granulocytic population. We demonstrate that knock-down of Card10 in a human cell line and in murine progenitor cells causes a defect in granulocytic differentiation. Expression analysis of Card10 depleted murine progenitor cells revealed that knock-down of Card10 affected expression of key genes involved in granulopoiesis. Taken together, we identify Card10 as a novel target of CEBPE with a role in myeloid cell differentiation and function.

Methods Mice CEBPE knock out (KO) mice have been described previously.9 Mice were maintained on a C57BL/6J (B6) genetic background at the animal facility of Comparative Medicine Centre, National University of Singapore (NUS). CEBPE KO allele was genotyped using primers: GCTACAATCCCCTGCAGTCC, TGCCTTCTGCCCTTGTG and ATCGCCTTCTATCGCCTTCTTGACGAG. All mice experiments were approved by Institutional Animal Care and Use Committee, NUS, Singapore.

Flow cytometry Single cell suspensions were incubated with fluorochrome-conjugated antibodies for 30 min on ice. Cells were washed with 2% FBS/PBS and resuspended in SYTOX Blue Dead Cell Stain 1270

(ThermoFisher Scientific). Flow cytometric analysis was performed on FACS LSR II flow cytometer (BD Biosciences), and sorting of cells was performed on FACS Aria cell sorter (BD Biosciences). Data were analyzed using FACSDIVA software (BD Biosciences).

ChIP-PCR DNA-protein complexes were cross-linked with 1% formaldehyde at room temperature for 10 min, followed by quenching with 0.2 M glycine for 5 min. Cells were lysed and chromatin was sonicated in Lysis buffer (1% SDS, 50 mM Tris-HCl, 5mM EDTA) at 4°C using Diagenode Bioruptor. Sheared-sonicated chromatin was incubated overnight at 4°C with antibodies against Cebpe and a mixture of Dynabeads Protein A and Protein G (1:1). Bead-chromatin complexes were washed, and the chromatin was eluted in 1% SDS, 0.1M sodium bicarbonate and reverse-crosslinked at 65°C for 16 hours. Immunoprecipitated DNA was extracted using QIAquick PCR Purification Kit (Qiagen) and quantified using Qubit Fluorometer (Life Technologies). Input and immunoprecipitated DNA were amplified using three different primer pairs. Primer sequences used for ChIP-PCR can be found in Online Supplementary Table S4.

Electrophoretic mobility shift assay

293T cells were transfected with either 1 μg pcDNA3.1(-) empty vector or Cebpe expressing vector in 100 mm dishes using Jetprime transfection reagent (Polypus) according to manufacturer’s instructions. After transfection, cells were cultured for 48 hours, and nuclear extracts were prepared using NE-PER reagent (Thermo Scientific). Double-stranded oligonucleotide probes were labelled using 3’ Biotin End labelling kit (Thermo Scientific) following the manufacturer’s instructions. Following biotinylated probes were used: CEBP consensus (5’-GATCCATATCCCTGATTGCGCAATAGGCTCAAAA); Card10 (5'- GAATGAGCCGATTGCTGCAACCTGGAAGG); Mutant Card10 (5'- GAATGAGCCGGCCTTGGGGCCCTGGAAGG) along with 100 fold molar excess of corresponding cold competitor. EMSA was carried out using LightShift Chemiluminescent EMSA Kit (Thermo Scientific). DNA-protein complexes were resolved on native 10% polyacrylamide–TBE gels.

Luciferase reporter assay A 500 bp fragment upstream of Card10 gene (encompassing the CEBPE binding site identified in ChIP-seq) was amplified from genomic DNA extracted from murine bone marrow cells and subcloned into pGL4-Basic vector (Promega, Madison, WI). NIH/3T3 cells were transfected with pCDNA-Cebpe along with either pGL4 basic vector or pGL4-Card10 vector (-7kb peak) using Lipofectamine Plus (Life Technologies). Renilla basic vector was co-transfected as a control for normalization of luciferase activity; luciferase was measured 24 hours after transfection using Promega Dual-Glo assay kit, as per the manufacturer’s instructions.

Isolation, culture and differentiation of Lin–Kit+ bone marrow cells Murine Lin–Kit+ cells were isolated from the bone marrow of C57BL/6 mice. Briefly, bone marrow cells were flushed from femurs and tibias using PBS containing 5%FBS; red cells were lysed and anti-rat IgG Dynabeads (Invitrogen) were used to deplete lineage +ve cells. Lin–Kit+ cells were sorted on FACS Aria (BD Biosciences). Sorted cells were cultured in the presence of IL3, IL6 and SCF. To induce granulocytic differentiation, 10 ng/mL GM-CSF was added to the culture medium. Granulocytic differentiation was monitored using flow cytometry at days 3, 5 and 7 of GM-CSF treatment. haematologica | 2018; 103(8)

CARD10 regulates granulocytic differentiation

Short hairpin RNA (shRNA) interference

RNA sequencing and expression analysis

To obtain Card10 knock-down in NB4 cells and murine progenitor cells, human and murine Card10 shRNA were cloned in the pLKO.1 lentiviral vector. Briefly, for virus production, 293T cells were seeded in a 100 mm dish, one day before transfection in DMEM medium supplemented with 10% FBS. Plasmid DNA (either non-target shRNA or Card10 shRNA) was transfected along with pCMV-dr8.2 and pMD2.G using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA, USA). After 4 hours, the transfection medium was replaced with DMEM medium supplemented with 10% FBS. Cell culture supernatants containing the lentivirus were collected at 48 h and 72 h post-transfection. Cells were transduced with lentiviral particles in the presence of 8 μg/ml polybrene (Sigma-Aldrich) for 24 h. Transduced cells were selected with puromycin (1 μg/ml for NB4 and 2 μg/ml for mouse Lin–Kit+ BM cells) for a week.

cDNA libraries were prepared from poly-A selected RNA using Truseq RNA sample kit (Illumina). Libraries were sequenced on HiSeq 4000 and 100 bp paired-end reads were aligned to murine reference transcriptome (GRCm38/mm10; Ensemble version 84) using Kallisto (version 0.43.0). Transcript level fragment counts were summarized at the gene level using TxImport Bioconductor package, and differential analysis was performed using DESeq2. Gene expression was quantified in FPKM units using DESeq2 FPKM command and was used for all downstream analysis and plotting. All other test-statistics and plotting were performed using R 3.4.0. Gene Ontology (GO) was performed on differentially expressed genes using goseq Bioconductor package (version 1.20.0), which accounts for bias due to gene length. Resulting P-values were adjusted for False Discovery Rate (FDR). For GSEA analysis, we used all “active transcripts” with mean expression of 0.5 FPKM to identify significantly enriched gene sets among MSigDB C2 gene sets.

RNA isolation, cDNA conversion and QPCR RNA from sorted murine granulocytes, progenitor populations or NB4 cells was isolated using either RNeasy Micro or Mini Kit (Qiagen) depending on the number of cells. cDNA was prepared using MuLV Reverse Transcriptase (Thermo Fisher Scientific). Primer sequences used for quantitative RT-PCR are listed in Online Supplementary Table S1.

Statistical analysis Two-sided, unpaired Student t-test was used to determine the statistical significance of experimental results. Data represented as mean ± SD. P values <0.05 are considered statistically significant.

Figure 1. Characterization of CEBPE ChIP-Seq peaks. (A) Genomic annotation of the 40,517 CEBPE binding sites to the murine genome according to known RefSeq genes. (B) Box plots depict ChIP-seq signal intensities for histone marks in multiple hematopoietic cells22 for the 312 genes downregulated in Cebpe KO immature granulocytes. Histone marks were measured +/- 1kb around CEBPE peaks. Peaks were classified based on the location of CEBPE binding. (i) Promoter peak (+1kb to -100bp) (ii) Intronic peak (iii) Intergenic peak within 10kb of the TSS (iv) Intergenic peak beyond 10kb of the TSS.

haematologica | 2018; 103(8)

1271

P. Shyamsunder et al.

Results CEBPE predominantly binds at intergenic and intronic regions and regulates gene expression We previously identified genome-wide CEBPE binding sites in unfractionated mouse bone marrow cells using

chromatin immunoprecipitation followed by sequencing (ChIP-seq).21 Total bone marrow cells are composed of approximately 40-50% of mature granulocytes that express high levels of CEBPE. Further analysis of ChIP- seq data revealed that the overall binding pattern of CEBPE was similar to other CEBP factors, with increased binding

Figure 2. Genes bound by CEBPE have an activation signature in granulocytes. Heat maps of ChIP-Seq signal intensities of histone marks in long term hematopoetic stem cells (LT-HSC), common myeloid progenitors (CMP), granulocyte-monocyte progenitor (GMP) and granulocytes (Gr) populations for genes whose expression is downregulated in CEBPE KO cells and categorised according to location of binding of CEBPE to the promoter, intron, intergenic region <10kb and > 10kb of the nearest gene TSS.22 *in the figure represents Card10.

1272

haematologica | 2018; 103(8)

CARD10 regulates granulocytic differentiation

at intergenic (37%) and intronic regions (46%) and only 10% of the sites located in core promoter regions (Figure 1A). We integrated the gene expression profile of granulocytes from Cebpe WT and KO mice (RNA-seq) and CEBPE ChIP-seq analysis. A total of 312 genes were downregulated in the absence of CEBPE. Of these, 93 genes had a CEBPE promoter peak, 103 genes had an intronic peak and 116 genes had an intergenic peak. Of the 116 peaks, 18 genes had a peak within 10kb from the transcriptional start site (TSS), while 98 genes had a peak beyond 10kb from the TSS (Online Supplementary Table S1). As genome-wide distribution of enhancer and promoter associated histone modifications provides a reliable signature of cell identity, we further analysed the 312 genes for histone marks in the hematopoietic lineage using a previously published data set.22 We used ChIP-seq data for three histone marks; H3K4me3 (mark of active promoters), H3K4me1 (mark of poised enhancers) and H3K27ac (mark of poised and active enhancers) and analysed the distribution of these marks for the 312 genes in long term hematopoetic stem cells (LT-HSC), common myeloid progenitors (CMP), granulocyte-monocyte progenitor (GMP) and granulocytes (Gr). Expectedly, comparison of

H3K27ac, H3K4me1 and H3K4me3 profiles showed that the majority of genes with promoter binding of CEBPE had a transcriptional activation signature in the granulocytes but not in (LT-HSC), myeloid precursors (CMP and GMP populations) (Figure 1B). Cumulative assessment of Z scores revealed that in granulocytes, an overall increased signal intensity for all three active histone marks was detected in genes with either intronic or intergenic occupancy within 10kb of the TSS, as compared with genes with peaks beyond 10kb of the TSS (Online Supplementary Figure S1). This analysis further emphasizes the role of CEBPE as a transcription factor essential for granulopoiesis and identifies a set of genes that might be regulated by CEBPE, with ChIP-seq peaks at non-promoter regions.

Card10/Carma3 is regulated by CEBPE in the granulocytic population Among the 18 genes that had an intergenic binding of CEBPE within 10kb of the TSS, Card10 was an interesting target which harbors active histone marks and is exclusively expressed in granulocytes and monocytes (Online Supplementary Figure S2A).

Figure 3. Card10 is a direct target of CEBPE. (A) Illustration of CEBPE binding peaks at the mouse Card10 locus. Binding sites located around -1.5kb and -7kb of the Card10 transcription start site are depicted in red. (B) FPKM values (RNA-seq) of Cebpe and Card10 in LT-HSC, CMP, GMP and Granulocytes (Gr) (https://www.ebi.ac.uk/gxa/experiments/E-MTAB-3079). (C) RT-PCR validation of transcript levels of Card10 in sorted immature granulocytes from wildtype and Cebpe KO mice. (D) ChIP-PCR validates binding of CEBPE to the -7kb region of Card10 gene using three primer pairs. Data are presented as percentage of input. (E) pGL4 basic luciferase vector containing a 500bp fragment harboring the -7kb CEBPE binding region was co-transfected with different amounts (0.2, 0.5, 0.8, and 1 Îźg) of pcDNA-Cebpe into NIH/3T3 cells. Luciferase activities were assayed 24 hours after transfection. Results represent fold induction of relative luciferase activity after normalization to Renilla control of two independent experiments, each done in triplicate. (F) Left panel; EMSA was performed with wild-type CEBPE binding sequence (CARD10 oligo), mutant CEBPE binding sequences (mutant oligo) and consensus CEBP binding sequences (CEBP oligo). Biotin-labelled probe was mixed with protein extracts from 293T cells transfected with either an empty vector or an expression vector for CEBPE, and the reaction mixtures were resolved on native 10% polyacrylamide-TBE gel. Cold competition was carried out with 100- fold molar excess of unlabelled oligo. Right panel; Western blot of 293T cell lysate +/- transfection with CEBPE expression vector used for EMSA assay. Alpha tubulin (TUBA1A) was used as loading control. ****P<0.0001, *P<0.05

haematologica | 2018; 103(8)

1273

P. Shyamsunder et al.

Figure 4. Card10 locus exhibits transcriptional activation signature in granulocytes. H3K27ac, H3K4me1 and H3K4me3 ChIP-seq signals in LT-HSC, CMP, GMP and granulocytes (Gr) and IGV track of ATAC-seq profile of sorted granulocytes (bottom track). Tracks are represented in frame with the CEBPE ChIP-seq peak of Card10 gene.22

We identified that CEBPE binds upstream of Card10 gene (-1.5 kb and -7kb) (Figure 3A). Interestingly, expression profile of Card10 and Cebpe followed a similar pattern, with the highest level of expression observed in granulocytes (https://www.ebi.ac.uk/gxa/experiments/E-MTAB3079) (Figure 3B). Expression of Card10 was significantly lower in immature granulocytes from Cebpe KO mice compared with the same stage of differentiation of the wild type mice (Figure 3C, Online Supplementary Figure S2B). On scanning the ChIP-seq data, peaks occurred at the 7 kb and -1.5 kb regions upstream of Card10 gene. We also verified the binding of Cebpe to know targets such as Lactoferrin (Ltf) and Neutrophilic granule precursor protein (Ngp) (Online Supplementary Figure S3A). Examination of the nucleotide sequence using ConSite (an online software tool to predict transcription factor motifs) revealed two CEBPE motifs within the -7kb peak and one CEBPE motif was detected in the -1.5kb peak (Online Supplementary Figure S3B). The -7kb region had stronger binding of CEBPE compared with the -1.5 Kb region; and therefore, the region was further studied. CEBPE binding to the -7kb peak was validated using ChIP-PCR in bone marrow cells from Cebpe WT mice; and this binding was significantly reduced in the KO cells (Figure 3D and Online Supplementary Figure S4). To assess whether the interactions of CEBPE is functionally relevant, luciferase reporter assay and electrophoretic mobility shift assay (EMSA) were performed. Luciferase reporter assay with a 500 bp fragment containing the CEBPE motif sequence transfected into NIH/3T3 cells revealed that CEBPE was able to transactivate the luciferase reporter in a dose-dependent manner (Figure 3E). Next, biotinylated oligos encompassing the putative CEBP motif were designed and EMSA assays were performed. Incubation of biotinylated oligos with nuclear extract from cells ectopically expressing CEBPE caused a shift in migration of the biotinylated oligos. This shift dis1274

appeared when the nuclear extract complex was incubated with either 100-fold molar excess of unlabelled competitor oligos (cold competition) or with biotinylated oligos harboring a mutation in CEBP motif (Figure 3F). Taken together, these results indicate that Card10 is a direct target of CEBPE.

Epigenetic landscape and expression of Card10 reveals an exclusive signature in granulocytes Comparison of H3K27ac, H3K4me1 and H3K4me3 occupancy at Card10 locus (at the core promoter and -7kb CEBPE binding region) in LT-HSC, CMP, GMP and granulocytes, revealed transcriptional activation signatures exclusively in granulocytes (Figure 4). ATAC-Seq of sorted granulocytes22 also indicated that the region bound by CEBPE is located in open chromatin (Figure 4). These findings along with exclusive expression of Card10 in granulocytes suggests a role in myeloid differentiation.

CARD10 regulates neutrophil differentiation in vitro Since Card10 expression is controlled by CEBPE, we hypothesized that CARD10 may have a role in granulopoesis. To test this hypothesis, CARD10 expression was stably silenced using short hairpin RNA (shRNA) in NB4 cells, an acute promyelocytic leukemia cell line that can be differentiated into granulocytes in the presence of all-trans retinoic acid (ATRA) (Figure 5A). We observed that knockdown of CARD10 resulted in a reduced proportion of CD11b+ cells upon induction with ATRA (control shRNA: 93% and CARD10 KD: 83% (sh4) and 63% (sh5) 4 days incubation). Although the data are not statistically significant, we did observe a trend towards impaired differentiation following Card10 knock-down (Fig. 5B). In parallel as a control, we also verified granulocytic differentiation after ATRA induction of CEBPE knock-down cells (Online Supplementary Figure S5A and S5B). Viability assay with control and Card10 KD cells revevaled no major effect on cell proliferation (Online Supplementary Figure S5C). haematologica | 2018; 103(8)

CARD10 regulates granulocytic differentiation

Figure 5. Knock down of CARD10 impairs myeloid differentiation. (A) RT-PCR analysis of CARD10 expression in human NB4 cells transduced with either non-target shRNA (NT) or CARD10 specific shRNAs (sh4 and sh5). Y-axis represents relative expression of CARD10 normalized to transcript levels of GAPDH (B) Proportion of CD11b+ NB4 cells stably transduced with either control (NT) or CARD10 shRNA and cultured with 1 μM ATRA for different duration. (C) RT-PCR validation of Card10 knock-down in murine Lin–Kit+ bone marrow (BM ) cells. Y-axis represents relative expression of Card10 normalized to Gapdh. (D) Granulocytic differentiation of Lin–Kit+ BM cells stably expressing either NT or Card10 shRNAs (sh2 or sh5) in response to 10ng/mL of GM-CSF. Differentiation was monitored by flow cytometric quantification of CD11b+Gr1+ cells. Results are average of three independent experiments. *P<0.05, **P<0.01.

Knock-down of CARD10 impairs granulocytic differentiation of murine Lin–Kit+ BM cells We evaluated further the role of CARD10 in granulocytic differentiation. shRNAs targeting murine Card10 were initially screened in NIH/3T3 cells (Online Supplementary Figure S6A) and two shRNAs that silenced Card10 were used to transduce murine Lin–Kit+ bone marrow (BM) cells. These shRNA sequences also robustly reduced the expression of Card10 in murine Lin–Kit+ BM cells (Figure 5C). Lin–Kit+ myeloid progenitors were transduced with either control or mouse Card10 shRNAs and grown in the presence of IL3, IL6, SCF and GM-CSF. The proportion of CD11b+Gr1+ cells were analysed at different time points using flow cytometry. Our analysis revealed that Card10 knock-down resulted in significantly lower proportion of granulocytes at all the time points tested (control shRNA: 50%, Card10 KD: 19% (sh2) and 11% (sh5) at 7 days) (Figure 5D and Online Supplementary Figure S6B). This further illustrated a role for CARD10 in myeloid cell differentiation.

Knock-down of Card10 affects expression of myeloid-specific genes To understand the changes in gene expression following Card10 knock-down, RNA-Seq of bulk control and Card10 knock-down Lin–Kit+ BM cells was performed. Ninetyfour genes were downregulated and 68 genes were upregulated following Card10 knock-down (FDR<0.1) (Figure 6A). Gene Ontology analysis of differentially expressed genes revealed enrichment of genes involved in immune response, inflammatory response, leukocyte migration and response to external stimuli (Figure 6B and Online haematologica | 2018; 103(8)

Supplementary Table S2). Furthermore, GSEA analysis of downregulated genes revealed a strong enrichment for genes involved in myeloid development (Figure 6C). Expression signature of Card10 knock-down Lin–Kit+ BM cells (carried out with two independent shRNAs) was compared to that of Cebpe KO immature granulocytes. Twenty-nine of the 94 genes downregulated in Card10 knock-down cells were also downregulated in Cebpe KO cells (Figure 6D). We validated the lower expression of prominent myeloid specific genes using RT-PCR in Card10 KD Lin–Kit+ BM cells (Figure 6E). These results show that loss of CARD10 affects expression of genes implicated in myelopoesis.

Discussion CEBPE is a transcription factor essential for functional maturation of granulocytes. Patients with neutrophil-specific granule deficiency have mutations of the CEBPE gene.15,16 Cebpe knockout mice recapitulate the disease and fail to produce terminally differentiated granulocytes. Previous studies from our group and others have identified several key targets of CEBPE,23-26 including a comprehensive approach that curated a list of CEBPE targets by comparing CEBPE binding sites to the gene expression changes in sorted granulocytes from Cebpe WT and KO cells.21 The present study extends the analysis to identify additional novel targets of CEBPE including genes with intronic or intergenic binding. Analysis of histone modifications at loci of genes with intronic/intergenic peaks showed that a majority of them had an active epigenetic signatures in the granulocytic population, suggesting these 1275

P. Shyamsunder et al.

Figure 6. Gene expression changes in Card10 deficient murine Lin–Kit+ bone marrow cells. (A) Volcano plot depicts genes differentially expressed in Card10 knockdown Lin–Kit+ BM cells compared to the WT cells (RNA-Seq). (B and C) Gene Ontology analysis (B) and GSEA plot (C) reveals enrichment of genes involved in myeloid development. (D) Venn diagram shows overlap of downregulated genes in WT vs. Cebpe KO (sorted immature granulocytes) compared to Card10 KD Lin–Kit+ BM cells. (E) RT-PCR measuring the transcript levels of myeloid specific genes in NT and Card10 knock-down Lin–Kit+ BM cells. Y-axis represents relative expression of genes normalized to Gapdh. *P<0.05, **P<0.01, ***P<0.001.

genes are also positively regulated by CEBPE during differentiation of granulocytes. Amongst these, Card10 was investigated as a putative novel target of CEBPE. We identified binding of CEBPE at two sites upstream of the Card10 gene and hypothesized that CEBPE might regulate transcription of Card10. We verified occupancy of CEBPE to the -7kb region upstream of Card10 gene using ChIP-PCR, luciferase reporter assay and EMSA. We observed impaired granulocytic diffferentiation following Card10 knock-down in both human (NB4) and murine (Lin–Kit+) BM cells. These findings suggest that CARD10 may be an important mediator of granulocytic differentiation. To understand how deficiency of CARD10 causes perturbation in myeloid differentiation, we performed expression analysis of Card10 knock-down and murine Lin–Kit+ bone marrow cells. RNA-seq data showed that knock-down of Card10 affected expression of genes involved in myeloid development and function. Among the genes downregulated in Card10 knock-down cells, 29 were also downregulated in Cebpe KO granulocytes and analysis of their gene expression pattern revealed that these genes were exclusively expressed in the granulocytic population (Online Supplementary Table S3). Migration of neutrophils to an inflammatory site is modulated by the activation of NFκB signalling.27,28 Published literature has documented a role for CARD10 as a scaffold protein for the NFκB signalling pathway, with no reference of its expression or role in granulopoiesis.19 The present study highlights that Card10 expression is 1276

regulated by CEBPE and that Card10 deficient cells have impaired granulopoiesis accompanied by reduced transcript levels of genes important in myeloid cell function. The impaired migration and phagocytosis observed in Cebpe KO mice may be contributed by lowered levels of Card10. We have previously shown that stimulating activity of CEBPE in vivo can enhance the antimicrobial function of the body.29 As microbes evolve to discover new ways to become resistant to antibiotics, enhancing innate immunity becomes more important. Whether the transcriptional changes observed upon Card10 knock-down is an effect of loss of mature granulocytes or a direct effect of Card10 function, warrants further analysis. In summary, this study demonstrates that Card10 is a novel CEBPE target gene and is implicated in granulocytic differentiation. Acknowledgments We would like to thank the staff of Comparative Medicine, NUS for their support in maintenance of mouse colonies and experiments involving mice. We would also like to acknowledge the expert help and support from the FACS facility at CSI, Singapore. We thank the Melamed Family and Reuben Yeroushalmi for their generous support. Funding This work was also funded by the Leukemia Lymphoma Society of America, (the Singapore Ministry of Health’s National Medical Research Council (NMRC) under its Singapore Translational Research (STaR) Investigator Award haematologica | 2018; 103(8)

CARD10 regulates granulocytic differentiation

to H. Phillip Koeffler (NMRC/STaR/0021/2014), Singapore Ministry of Education Academic Research Fund Tier 2 (MOE2013-T2-2-150), the NMRC Centre Grant awarded to National University Cancer Institute of Singapore (NMRC/CG/012/2013) and the National Research Foundation Singapore and the Singapore Ministry of Education

References 1. Orkin SH, Zon LI. Hematopoiesis: an evolving paradigm for stem cell biology. Cell. 2008;132(4):631-644. 2. Novershtern N, Subramanian A, Lawton LN, et al. Densely interconnected transcriptional circuits control cell states in human hematopoiesis. Cell. 2011;144(2):296-309. 3. Tenen DG, Hromas R, Licht JD, Zhang DE. Transcription factors, normal myeloid development, and leukemia. Blood. 1997;90(2):489-519. 4. Iwasaki H, Somoza C, Shigematsu H, et al. Distinctive and indispensable roles of PU.1 in maintenance of hematopoietic stem cells and their differentiation. Blood. 2005; 106(5):1590-1600. 5. Hock H, Hamblen MJ, Rooke HM, et al. Intrinsic requirement for zinc finger transcription factor Gfi-1 in neutrophil differentiation. Immunity. 2003;18(1):109-120. 6. Holtschke T, Lohler J, Kanno Y, et al. Immunodeficiency and chronic myelogenous leukemia-like syndrome in mice with a targeted mutation of the ICSBP gene. Cell. 1996;87(2):307-317. 7. Growney JD, Shigematsu H, Li Z, et al. Loss of Runx1 perturbs adult hematopoiesis and is associated with a myeloproliferative phenotype. Blood. 2005;106(2):494-504. 8. Shivdasani RA, Mayer EL, Orkin SH. Absence of blood formation in mice lacking the T-cell leukaemia oncoprotein tal-1/SCL. Nature. 1995;373(6513):432-434. 9. Yamanaka R, Barlow C, Lekstrom-Himes J, et al. Impaired granulopoiesis, myelodysplasia, and early lethality in CCAAT/enhancer binding protein epsilondeficient mice. Proc Natl Acad Sci USA. 1997;94(24):13187-13192. 10. Zhang DE, Zhang P, Wang ND, Hetherington CJ, Darlington GJ, Tenen DG. Absence of granulocyte colony-stimulating factor signaling and neutrophil development in CCAAT enhancer binding protein alpha-deficient mice. Proc Natl Acad Sci USA. 1997;94(2):569-574.

haematologica | 2018; 103(8)

under its Research Centres of Excellence initiatives. This research is also supported by the RNA Biology Centre at the Cancer Science Institute of Singapore, NUS, as part of funding under the Singapore Ministry of Educationâ&#x20AC;&#x2122;s Tier 3 grants, grant number MOE2014-T3-1-006. This paper is dedicated to the memory of Parker Hughes.

11. Akagi T, Thoennissen NH, George A, et al. In vivo deficiency of both C/EBPbeta and C/EBPepsilon results in highly defective myeloid differentiation and lack of cytokine response. PLoS One. 2010; 5(11):e15419. 12. Chumakov AM, Grillier I, Chumakova E, Chih D, Slater J, Koeffler HP. Cloning of the novel human myeloid-cell-specific C/EBPepsilon transcription factor. Mol Cell Biol. 1997;17(3):1375-1386. 13. Lekstrom-Himes J, Xanthopoulos KG. Biological role of the CCAAT/enhancerbinding protein family of transcription factors. J Biol Chem. 1998;273(44):2854528548. 14. Morosetti R, Park DJ, Chumakov AM, et al. A novel, myeloid transcription factor, C/EBP epsilon, is upregulated during granulocytic, but not monocytic, differentiation. Blood. 1997;90(7):2591-2600. 15. Gombart AF, Shiohara M, Kwok SH, Agematsu K, Komiyama A, Koeffler HP. Neutrophil-specific granule deficiency: homozygous recessive inheritance of a frameshift mutation in the gene encoding transcription factor CCAAT/enhancer binding protein--epsilon. Blood. 2001; 97(9):2561-2567. 16. Wada T, Akagi T, Muraoka M, et al. A novel in-frame deletion in the leucine zipper domain of C/EBPepsilon leads to neutrophil-specific granule deficiency. J Immunol. 2015;195(1):80-86. 17. Gombart AF, Krug U, O'Kelly J, An E, Vegesna V, Koeffler HP. Aberrant expression of neutrophil and macrophage-related genes in a murine model for human neutrophil-specific granule deficiency. J Leukoc Biol. 2005;78(5):1153-1165. 18. Grabiner BC, Blonska M, Lin PC, et al. CARMA3 deficiency abrogates G proteincoupled receptor-induced NF-{kappa}B activation. Genes Dev. 2007;21(8):984-996. 19. Blonska M, Lin X. NF-kappaB signaling pathways regulated by CARMA family of scaffold proteins. Cell Res. 2011;21(1):5570. 20. Medoff BD, Landry AL, Wittbold KA, et al.

21.

22.

23.

24.

25.

26.

27.

28. 29.

CARMA3 mediates lysophosphatidic acidstimulated cytokine secretion by bronchial epithelial cells. Am J Respir Cell Mol Biol. 2009;40(3):286-294. Suh HC, Benoukraf T, Shyamsunder P, et al. LPS independent activation of the proinflammatory receptor Trem1 by C/EBPepsilon in granulocytes. Sci Rep. 2017;7:46440. Lara-Astiaso D, Weiner A, Lorenzo-Vivas E, et al. Immunogenetics. Chromatin state dynamics during blood formation. Science. 2014;345(6199):943-949. Tanaka M, Gombart AF, Koeffler HP, Shiohara M. Expression of bactericidal/permeability-increasing protein requires C/EBP epsilon. Int J Hematol. 2007;85(4):304-311. Gombart AF, Kwok SH, Anderson KL, Yamaguchi Y, Torbett BE, Koeffler HP. Regulation of neutrophil and eosinophil secondary granule gene expression by transcription factors C/EBP epsilon and PU.1. Blood. 2003;101(8):3265-3273. Chumakov AM, Kubota T, Walter S, Koeffler HP. Identification of murine and human XCP1 genes as C/EBP-epsilondependent members of FIZZ/Resistin gene family. Oncogene. 2004;23(19):3414-3425. Khanna-Gupta A, Zibello T, Sun H, Gaines P, Berliner N. Chromatin immunoprecipitation (ChIP) studies indicate a role for CCAAT enhancer binding proteins alpha and epsilon (C/EBP alpha and C/EBP epsilon ) and CDP/cut in myeloid maturation-induced lactoferrin gene expression. Blood. 2003;101(9):3460-3468. McDonald PP, Bald A, Cassatella MA. Activation of the NF-kappaB pathway by inflammatory stimuli in human neutrophils. Blood. 1997;89(9):3421-3433. Cassatella MA. The production of cytokines by polymorphonuclear neutrophils. Immunol Today. 1995;16(1):21-26. Kyme P, Thoennissen NH, Tseng CW, et al. C/EBPepsilon mediates nicotinamideenhanced clearance of Staphylococcus aureus in mice. J Clin Invest. 2012; 122(9):3316-3329.

1277

ARTICLE

Bone Marrow Failure

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1278-1287

Natural history of GATA2 deficiency in a survey of 79 French and Belgian patients Jean Donadieu,1 Marie Lamant,2 Claire Fieschi,3,4 Flore Sicre de Fontbrune,5 Aurélie Caye,6 Marie Ouachee,7 Blandine Beaupain,8 Jacinta Bustamante,9,10,11,12 Hélène A. Poirel,13 Bertrand Isidor,14 Eric Van Den Neste,15 Antoine Neel,16 Stanislas Nimubona,17 Fabienne Toutain,18 Vincent Barlogis,19 Nicolas Schleinitz,20 Thierry Leblanc,7 Pierre Rohrlich,21 Felipe Suarez,22 Dana Ranta,23 Wadih Abou Chahla,24 Bénédicte Bruno,24 Louis Terriou,25 Sylvie Francois,26 Bruno Lioure,27 Guido Ahle,28 Françoise Bachelerie,29 Claude Preudhomme,30 Eric Delabesse,31,32 Hélène Cave,6 Christine BellannéChantelot,33 Marlène Pasquet2,32 and the French GATA2 study group.

Department of Paediatric Haematology and Oncology, Registre National des Neutropénies Chroniques, AP-HP Trousseau Hospital, Paris, France; 2Department of Paediatric Haematology and Immunology, CHU Toulouse, France; 3Department of Clinical Immunology Assistance Publique – Hôpitaux de Paris (AP-HP) Saint-Louis Hospital, France; 4INSERM UMR1126, Centre Hayem, Université Paris Denis Diderot, Sorbonne Paris Cité, France; 5 Department of Haematology and Bone Marrow Transplantation, AP-HP Saint-Louis Hospital, Paris, France; 6Genetic Laboratory, AP-HP Robert Debré Hospital, Paris, France; 7 Department of Haematology, AP-HP Robert Debré Hospital, Paris, France; 8French Neutropenia Registry, AP-HP Trousseau Hospital, Paris, France; 9Laboratory of Human Genetics of Infectious Diseases, Necker Branch, INSERM UMR 1163, Necker-Enfants Malades Hospital, Paris, France; 10Centre for the Study of Primary Immunodeficiencies, Necker-Enfants Malades Hospital, AP-HP, Paris, France; 11St Giles Laboratory of Human Genetics of Infectious Diseases, Rockefeller Branch, New York, NY, USA; 12Paris Descartes University, Imagine Institute, Paris, France; 13Centre for Human Genetics, Cliniques Universitaires Saint-Luc & Human Molecular Genetics (GEHU), de Duve Institute Université Catholique de Louvain, Brussels, Belgium; 14Department of Genetics, CHU Nantes, France; 15Department of Haematology, St Luc Hospital, Brussels, Belgium; 16 Department of Internal Medicine, CHU Nantes, France; 17Department of Haematology, CHU de Rennes, France; 18Department of Paediatric Haematology and Oncology, CHU de Rennes, France; 19Department of Paediatric Haematology, CHU de Marseille, Hopital La Timone, Université Aix-Marseille, France; 20Internal Medicine, CHU de Marseille, Hopital La Timone, Université Aix-Marseille, France; 21Department of Haematology, CHU de Besançon, France; 22Department of Haematology, AP-HP Necker-Enfants Malades, INSERM UMR 1163 and CNRS ERL 8254 Institut Imagine, Sorbonne Paris Cité, Université Paris Descartes, France; 23Department of Haematology, CHU de Nancy, France; 24Department of Paediatric Haematology, CHU de Lille, France; 25Department of Internal Medicine and Immunology, CHU Lille, France; 26Department of Haematology, CHU d'Angers, France; 27Department of Haematology, CHU de Strasbourg, France; 28Department of Neurology, Hôpitaux Civils de Colmar, France; 29Inflammation Chimiokines et Immunopathologie, INSERM, Faculté de Médecine, Université Paris-Sud, Université Paris-Saclay, Clamart, France; 30Laboratory of Haematology, CHU de Lille, France; 31Laboratory of Haematology, IUCT-Oncopole, Toulouse, France; 32Centre of Research in Oncology, INSERM U1037, Team 16, IUCT-Oncopole, Toulouse, France and 33Department of Genetics, AP-HP Pitié Salpêtrière Hospital, Faculté de Médecine Sorbonne Université, Paris, France 1

Correspondence: pasquet.m@chu-toulouse.fr

Received: October 28, 2017. Accepted: April 27, 2018. Pre-published: May 3, 2018.

doi:10.3324/haematol.2017.181909 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1278 ©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.

1278

*JD and ML, and CBC and MP contributed equally to this study

ABSTRACT

eterozygous germline GATA2 mutations strongly predispose to leukemia, immunodeficiency, and/or lymphoedema. We describe a series of 79 patients (53 families) diagnosed since 2011, made up of all patients in France and Belgium, with a follow up of 2249 patients/years. Median age at first clinical symptoms was 18.6 years (range, 0-61 years). Severe infectious diseases (mycobacteria, fungus, and human papilloma virus) and hematologic malignancies were the most common first manifestations. The probability of remaining symptomfree was 8% at 40 years old. Among the 53 probands, 24 had missense mutations including 4 recurrent alleles, 21 had nonsense or frameshift mutations, 4 had a whole-gene deletion, 2 had splice defects, and 2 patients had complex mutations. There were significantly more cases of leukemia in patients with missense mutations (n=14 of 34) than in haematologica | 2018; 103(8)

GATA2 deficiency French/Belgian cohort

patients with nonsense or frameshift mutations (n=2 of 28). We also identify new features of the disease: acute lymphoblastic leukemia, juvenile myelomonocytic leukemia, fatal progressive multifocal leukoencephalopathy related to the JC virus, and immune/inflammatory diseases. A revised International Prognostic Scoring System (IPSS) score allowed a distinction to be made between a stable disease and hematologic transformation. Chemotherapy is of limited efficacy, and has a high toxicity with severe infectious complications. As the mortality rate is high in our cohort (up to 35% at the age of 40), hematopoietic stem cell transplantation (HSCT) remains the best choice of treatment to avoid severe infectious and/or hematologic complications. The timing of HSCT remains difficult to determine, but the earlier it is performed, the better the outcome.

Introduction GATA2 gene encodes a transcription factor critical to hematopoiesis characterized by the presence of two zinc finger domains. Since 2011, heterozygous germline mutations in GATA2 have been reported to cause a complex and heterogeneous syndrome consisting of myelodysplasia (MDS), acute myeloid leukemia (AML),1 monocytopenia mycobacterial infections/dendritic cell,2 monocyte, B and natural killer (NK) cell deficiency (MonoMAC3,4/DMLC),5 and lymphoedema (Emberger syndrome).6 The mutational spectrum of GATA2 is heterogeneous, consisting of missense mutations mostly located within the highly conserved C-terminal zinc finger domains, null mutations mostly located upstream the zinc finger domains, splice site defects, mutations in the enhancer located in the intron 4,7 and, more rarely, exonic and whole-gene deletions. Apart from hematologic and infectious phenotypes, additional clinical presentations have been described in the last six years, such as aplastic anemia,8 pulmonary alveolar proteinosis,9 dermatological,10 autoimmune or vascular features. So far, a total of 158 patients with a germline GATA2 mutation have been reported in 4 large surveys.11-14 Except for lymphoedema which is more frequent in patients with null or regulatory mutations, no correlation between the type or location of the GATA2 mutation and the clinical/biological phenotype have been established in previous reports. We now report a large multicenter survey which brings together all the patients that have been diagnosed in France and Belgium since 2011, i.e. 79 patients with a heterozygous germline GATA2 mutation from 53 pedigrees. These 79 patients include 14 previously identified patients through the French Chronic Neutropenia Registry;11 their follow up has been up-dated. This survey allows human GATA2 deficiency to be more accurately defined, taking advantage of a very long follow-up period (2249 patients/years). We describe the initial manifestations, their evolution, and their outcome regarding the onset of severe manifestations (leukemia, severe infections, vascular defects). This large cohort also allows new features of the disease to be described.

Methods Patients All patients with heterozygous germline GATA2 mutations diagnosed between 2011 and 2016 in France and Belgium were enrolled in this survey, secondary to their identification through haematologica | 2018; 103(8)

the laboratories which performed their genetic diagnosis in France and Belgium (JB, CBC, HC-AC, ED, CP). Fourteen patients with chronic neutropenia, and who were registered in the French Severe Chronic Neutropenia Registry, had been reported previously.11 This registry has been recognized as a national registry by the French health authorities since 2008, and has contributed to several studies.15,16 The database was approved by the French national data protection agency (CNIL, certificate n. 97.075). This registry was primary established to enrol all the patients with chronic neutropenia in France. By extension, all patients identified with a given genetic disease (e.g. GATA2 ) occasionally associated with a chronic neutropenia can be enrolled in the registry. With regards to GATA2 mutations, we systemically seek additional sources of enrollment, extending the borders of the initial network to internal medicine, infectious diseases and genetics, as well as from adult hematopoietic stem cell transplantation (HSCT) units.

Genetic analysis The patient or his/her parents gave their written informed consent to undergo genetic testing and participate in the study. Genomic DNA was extracted from a blood sample. Genetic analysis included the Sanger sequencing of exons 2 to 6, the intronic regulatory region (intron 4) of the GATA2 gene (NM_032631.4), and the search for exonic or large genomic deletions by quantitative PCR and/or MLPA. The germline status of the identified GATA2 mutation was confirmed by analyzing non-hematopoietic tissue (cultured skin fibroblasts, hair follicles or nails) in 30 probands. Null mutations (nonsense, frameshift, multi-exon deletion) were considered to be diseasecausing. The pathogenicity of missense mutations and splicesite variants that did not affect the canonical +1 and +2 splice site bases were based on the following criteria: frequency in the general-population database [Exome Aggregation Consortium (ExAC): http://exac.broadinstitute.org], literature that took into account mutations that were previously reported as a GATA2associated defect, functional studies supporting a damaging effect, a de novo occurrence, family segregation analysis, and finally predictive algorithms of pathogenicity for missense mutations [SIFT, Align GVGD, PolyPhen-2 and Combined Annotation-Dependent Depletion (CADD) score, and for splice-site defects (MaxEntScan and Human Splicing Finder)].17

Clinical investigation Demographics, immuno-hematologic parameters and infectious status were recorded. Septicemia, cellulitis, pneumonia, osteitis, and liver abscess were considered to be severe infections. Computed tomography (CT) scans, bronchioalveolar lavage and pulmonary function investigations were performed in patients with lung disease. Profuse skin or genito-anal warts were considered to be a specific event. Mycobacterial infections were considered if mycobacteria were identified in a pathological tissue (Ziehl 1279

J. Donadieu et al.

coloration and/or culture in 14 of 16 patients). Mycobacterial infection was suspected if the tissue sample demonstrated granuloma, and/or clinical symptoms were cured by antimycobacterial drugs (2 out of 16 patients). Immunoglobulin levels were analyzed according to the patient's age.18 Age at first symptom was defined by the age at the first clinical pathological manifestation among the following list: myelodysplastic syndromes (MDS) or acute leukemia (AL), any severe and potentially life-threatening infection, lymphoedema, pulmonary proteinosis, or profuse human papillomavirus (HPV) infection. A GATA2 mutation carrier was considered asymptomatic if no clinical and/or biological symptoms were described at the last follow-up visit. Siblings or parents of probands were considered as carriers of the familial GATA2 mutation if they presented with one of the typical manifestations of the GATA2 deficiency, even in the absence of genetic testing.

Hematologic features Blood counts were recorded at baseline if available, at any period following a hematologic complication, and after HSCT (if applicable). Bone marrow studies were performed in the event of blood count abnormality. Hematologic malignancies were classified according to the 2008 World Health Organization (WHO) classification.19,20 MDS was classified according to the revised version of the International Prognostic Scoring System (IPSS)21 and juvenile myelo-monocytic leukemia (JMML) was classified according to the 2016 WHO classification.22

Statistical analysis Stata® software (v.13) was used for all the statistical analyses. Lower and upper interquartile and median values express the distribution of quantitative variables. Differences between groups of patients were analyzed using Fisher’s exact test if the event was discrete and Wilcoxon’s test for quantitative variables. Survival was compared between the groups of subjects using the log-rank test, and Cox's model was used for the multivariate analysis. As we performed repeated tests, P<0.01 was considered significant, unless otherwise stated. For survival, the end points were death, MDS or AL; the time-period started at birth until an event or the day of last news. We also analyzed survival after onset of a clonal event. The time period started from the first clonal event (MDS or AL) until death or the day of last news. The Kaplan-Meier method was used to estimate survival rates. The cut-off date was 30th September 2016.

Results Early onset of severe infections and/or hematologic diseases in GATA2 deficiency Forty males and 39 females from 53 families with a heterozygous germline GATA2 mutation are herein reported (Table 1 and Online Supplementary Table S1), including 14 previously described patients,11 whose clinical and biological data have been up-dated. The patients were enrolled in France (n=72) and Belgium (n=7). Median age at the last follow up was 24.5 years old (range, 3.9-73). The probability of remaining symptom-free was 38% at the age of 20 (95%CI: 27-48.7%) and 8% at the age of 40 (95%CI: 3.315%) (Figure 1A). All patients except 5 were symptomatic at the time of the last follow up. These 5 individuals were first-degree relatives of symptomatic patients with a GATA2 mutation (Online Supplementary Table S2). Median age at onset of the first clinical symptom was 18.6 years (range, 0-61) (Figure 1A and Online Supplementary Table 1280

S1). Initial manifestations were a hematologic malignancy in 19 patients (26%), a severe bacterial infection in 17 (23%), profuse warts or HPV in 15 (20%), lymphoedema in 7 (9.4%), or a mycobacterial infectious disease in 6 (8.1%). Blood counts of patients with opportunistic infections (HPV, mycobacteria, mycosis, the JC virus) were systematically abnormal (monocytopenia, neutropenia, pancytopenia, severe anemia).

Additional clinical features in GATA2 deficiency Outside hematologic and infectious clinical presentations, erythema nodosum/panniculitis (4 patients), mental retardation (1 patient), transient ischemic cerebral palsy (1 patient), and progressive multifocal leukoencephalopathy linked to the JC virus (PML, 1 patient) were the initial symptoms in 7 patients. Over the course of the disease, 9 patients had systemic inflammatory manifestations with panniculitis, vasculitis, Sweet's syndrome, lupus-like disease or granulomatous disease mimicking sarcoidosis. Of note, auto-immune markers were present in 12 patients, which may be an underestimation because they were not sought for in all patients (Table 1). Chronic lymphoedema was noted in 12 patients (15%). Vascular and/or thrombotic complications were observed in 7 patients: 2 patients presented with transient cerebral palsy, one patient presented with splenic-vein thrombosis after a splenectomy and mycobacteriosis, one patient presented with 3 deep-vein thromboses in a context of AL,

Table 1. Clinical and biological presentation of GATA2 deficient patients.

Diagnostic features Hematologic features

Clinical and biological aspects

MDS AML ALL Aplastic anemia Juvenile myelomonocytic leukemia Recurrent infections Monocytopenia (viral, mycobacterial, fungal) B lymphopenia NK lymphopenia Warts HPV-related (genital and cutaneous) Oncogenesis Lymphoedema Pulmonary features Pulmonary alveolar proteinosis Recurrent bacterial infections Vascular features Thrombosis, myocardial infarction Deafness Autoimmune features Panniculitis, erythema nodosum, vasculatis, lupus-like and sarcoidosis-like syndrome, Sweet's syndrome Other features Urinary-system malformation Premature labor, miscarriage Hypothyroidism

Incidence in our survey 70% (55/79) 19% (15/79) 1.3% (1/79) 2.5% (2/79) 1.3% (1/79) 49% (24/49) 100% (38/38) 7.8% (3/38) 40% (32/79) 3.8% (3/79) 15% (12/79) 3.8% (3/79) 56% (44/79) 9% (7/79) 1.3% (1/79) 11% (9/79)

5% (4/79) 6.3% (5/79) 1.3% (1/79)

MDS: myelodysplastic syndromes; AML: acute myeloid leukemia; ALL: acute lymphoblastic leukemia; NK: natural killer cell; HPV: human papillomavirus.

haematologica | 2018; 103(8)

GATA2 deficiency French/Belgian cohort

one patient presented with deep-vein thrombosis and pulmonary embolism while receiving treatment for breast cancer and MDS, one patient presented with myocardial infarction at the age of 40, and one patient died from aortic dissection at the age of 33 years (Table 1). Only 3 patients had pulmonary alveolar proteinosis and one patient in the cohort was deaf. Four patients had urogenital abnormalities. Three patients had been born pre-

maturely, 2 women suffered from miscarriages, and one patient had hypothyroidism.

At diagnosis, the majority of GATA2 deficient patients had abnormal blood parameters Sixty of the 74 symptomatic patients were free of hematologic malignancy at diagnosis. A blood count was available for 49 patients before hematologic evolution. The

Figure 1. Onset of disease, hematologic and infectious complications. (A) Kaplan Meier curves are shown for the onset of disease in 74 symptomatic patients with a GATA2 mutation. (B) Occurrence of myelodysplastic leukemia (MDS) / acute leukemia (AL).(C-F) Rate in mycobacterial, human papillomavirus (HPV), bacterial and mycotic infections. Confidence intervals of 95% are shaded gray.

haematologica | 2018; 103(8)

1281

J. Donadieu et al.

blood count was abnormal in 36 patients (73%): 24 patients (49%) had monocytopenia lower than 0.1 G/L, 19 patients (39%) had neutropenia lower than 1.5 G/L, 9 patients (18%) had platelet levels lower than 100x109/L, 7 patients (14%) had macrocytosis, and 5 patients (10.2%) had anemia lower than 9 g/dL (Online Supplementary Figure S1). Consequently, 36 out of these 49 evaluable patients (73%) had an abnormal blood count. Immunological data were available in 38 patients: T-cell counts were slightly decreased (median 0.97 G/L T CD3 (range, 0.1-7.5), 0.37 G/L T CD4 (range, 0.05-5.6), and 0.49 G/L T CD8 (range, 0.02-2.3), NK cells (CD16+/CD56+) were preserved (median 0.12 G/L, range: 0-0.34); B-cell levels were consistently low (median 0.02 G/L, range, 01.51) although immunoglobulin levels were within normal ranges (median IgG=9.3 g/L, range, 4-40; IgA=0.9 g/L, range, 0.33-3.4; IgM=1 g/L, range, 0.05-2.40). Overall, GATA2 defects are mainly associated with a monocytopenia and a B-cell lymphopenia.

More than 80% of patients presented with a hematologic malignancy at the age of 40 years Among the 74 symptomatic patients, 64 developed a hematologic malignancy (MDS and/or AL). The risk of developing MDS/AL rapidly increased from 6% at the age of 10 years to 39% at the age of 20, and 81% at the age of 40 (Figure 1B). Among the 64 patients, the initial diagnosis was MDS in 55 patients (69%), AL in 7 patients (9%), and chronic leukemia in 2 patients (3%). Among the 55 patients with an MDS, a progression to AL was observed in 9 patients (16%) (Figure 2A). The AL were mainly myeloid (AML), but we observed a case of T-cell acute

lymphoblastic leukemia with a monosomy 7. In addition to these hematologic complications, a juvenile myelomonocytic leukemia (JMML) case occurred in a neonate. This patient received neither chemotherapy nor allograft. This patientâ&#x20AC;&#x2122;s blood count is normal four years after diagnosis without any treatment. Karyotypes were abnormal in 43 of 66 patients (65%), with a complete or a partial loss of chromosome 7 in 27 cases (35%), trisomy-8 in 16 cases (18%), 4 patients combining the two (Figure 2B). In order to better define the prognosis of MDS, the IPSSR21 was calculated for 47 of 55 patients upon diagnosis of MDS. The prognosis was mainly intermediate (1.5-4.5) in 24 patients (51%), high (>4.5) in 13 patients (27%), and low (<1.5) in 10 patients (21%). There was no significant difference in the age of the patients between these 3 groups.

Low frequency of solid neoplasia Solid tumors were identified in 6 patients only, mainly secondary to HPV (3 cases). In addition, one woman developed breast cancer, one patient developed a metastatic adenocarcinoma, and one patient developed an epidermoid carcinoma.

Severe infectious diseases explain the high mortality Severe bacterial infections were the most frequent feature, occurring at some time over the patientsâ&#x20AC;&#x2122; lives in 44 cases (56%). The 20-year cumulative rate of bacterial infection was 33%, rising to 64% at the age of 40 years (Figure 1E). Lung infections were the most frequent (two-thirds of all cases) and, although they evolved

Figure 2. Hematologic features of the 79 patients. (A) Hematologic malignancies among the 79 patients. Progression to acute myeloid leukemia (AML) from myelodysplastic syndromes (MDS) is indicated. (B) Karyotypes availables for 66 out of 79 GATA2-deficient patients. *4 patients with monosomy 7 also had trisomy 8. (C) Hematologic complications and the outcome of patients older than the age of 40. JMML: juvenile myelomonocytic leukemia; CMML: chronic myelomonocytic leukemia; LEMP: leukoencephalomyelopathy; HSCT: hematopoietic stem cell transplantation.

1282

haematologica | 2018; 103(8)

GATA2 deficiency French/Belgian cohort

favorably with antibiotics, recurrences were frequent. Nine patients had bacterial soft-tissue infections, and 5 had ENT infections. Twelve patients had a non-tuberculous mycobacterial infection (Figures 1C and 3) (Mycobacterium avium, kansasii, chelonae, genavense), and 4 patients developed tuberculosis. These mycobacterial infections were concomitant with MDS in 7 cases. The risk of acquiring a mycobacterial infection increased with age: 9% at the age of 20 years to 42% at the age of 40 years. Severe viral infections led to death in 4 patients: H1N1 influenza five years after AML treatment (Figure 3), Epstein-Barr virus (EBV) lymphoproliferative disease after HSCT, HPV-related metastatic carcinoma and a progressive multifocal leukoencephalopathy caused by the JC virus which was the first manifestation of the disease. Cutaneous or genital recurrent HPV-induced warts were often the first reported symptom (32 cases, 40%), with 20year and 40-year rates of 25% and 50%, respectively

(Figures 1 and 3). A high resistance to local treatment and frequent recurrences were common. Two patients developed a neoplasia. Eighteen fungal infections were observed in 16 patients (11 cases of aspergillosis, 5 of candidosis, and 2 of mucormycosis). Eight of these 18 infections were diagnosed during chemotherapy (n=5) or HSCT (n=3) (Figure 1F). Several infectious complications appeared post HSCT (3 fungal, 1 viral and 2 bacterial infections related to HSV, 2 patients with EBV prior to the HSCT had recurrence of this virus after HSCT, which evolved to lymphoproliferative disease in 1 patient) (Online Supplementary Table S1). The course of infection was complicated by hemophagocytic syndrome in 6 patients (2 mycobacterial, 1 fungal and 3 viral infections).

A poor survival rate was observed in GATA2-deficient patients despite aggressive treatments In our cohort, 27 patients (34%) died at a median age of

Figure 3. Clinical, radiographic, and cytological features of GATA2 syndrome. (A) Cutaneous warts on the hands of a woman with myelodysplastic syndromes (MDS). (B) Hand rheumatism (C) Bilateral lymphoedema post-hematopoietic stem cell transplantation. (D) Acute respiratory distress syndrome in a H1N1 infection. (E) Osteomyelitis at presentation. (F) Progressive multifocal leukoencephalopathy in a 43-year old man. (G) Pulmonary alveolar proteinosis in a woman with MDS and warts. (H) Disseminated mycobacteriosis. (I and J) Bone marrow smears of pedigree 46. (I) Dysgranulopoiesis and blasts in a woman with MDS evolving to acute myeloid leukemia.2 (J) Dyserythropoiesis and dysgranulopoiesis in her son with MDS. (K) Macrophage activation secondary to flu infection.

haematologica | 2018; 103(8)

1283

J. Donadieu et al.

29 years (range, 10.2-72.6). Survival analyses demonstrated a poor outcome: mortality was 6% at the age of 20, 42% at the age of 40, and 69% at the age of 60 (Figure 4A). Probability of survival after a clonal event (MDS and/or AL) was 60% by the age of 40 (Figure 4B). The 5-year survival rate in patients with MDS regarding the 3 IPSS-R groups was: 30% in the high-risk group, 80% in the intermediate-risk group, and 100% in the low-risk group (P<0.001) (Figure 4D). Of note, severe bacterial and/or viral complications were the main causes of death in patients over the age of 40 (Figure 2C). Myelodysplastic syndromes and AL were the main causes of death in 15 patients: 8 cases after chemotherapy, and 7 after HSCT. Ten patients had lethal infections: disseminated mycobacterial infections in 3, bacterial infections in 3, and severe viral infections in 4 (JC virus encephalitis, oncogenic HPV, H1N1 flu, and EBV lymphoproliferative disease

post HSCT). One patient died from an aortic dissection and another from metastatic carcinoma. Twenty-eight patients underwent HSCT for MDS or AL and/or immune deficiency. The overall survival rate of these patients was 73% after one year and 62% after five years, which then plateaued. Nine of the 28 patients died from severe infections or graft-versus-host disease. Survival after HSCT was dependent on the age at transplantation (Figure 4): the earlier the HSCT was performed, the better the outcome, even if the difference was not statistically significant. In cases of AL (n=18), an aggressive chemotherapy induction regimen was proposed for 16 patients, with primary failure in 12 and severe infectious toxicity in 9 cases (5 cases of fungal infection). A demethylating agent was given to 3 patients and has allowed long-term disease management for MDS (n=1) and AML (n=2).

Figure 4. Survival in 79 patients. (A) Kaplan-Meier curves showing overall survival of the whole cohort. (B) Survival after a clonal event. (C) Overall survival was studied after hematopoietic stem cell transplantation (HSCT) depending on the age (years) at transplantation (P>0.05). (D) Overall survival of patients with myelodysplastic syndromes (MDS) and/or acute leukemia (AL) was plotted according to the revised International Prognostic Scoring System (IPSS-R) score at the time of the diagnosis of the malignancy. Survival significantly depends on the IPSS score (P<0.001). Confidence intervals of 95% are shaded gray. y: years.

1284

haematologica | 2018; 103(8)

GATA2 deficiency French/Belgian cohort

Genotype/phenotype correlation: leukemia was more frequently observed in patients with missense mutations

Discussion

Among the 53 probands, 45 different mutated alleles were identified (Online Supplementary Table S1). Mutations were mainly located in exons 4 and 5 (Figure 5). Four patients (8% of the cohort) had a complete heterozygous GATA2 locus deletion. Five mutations were recurrent in unrelated families (R362X, R361H, A372T, R396W, R398Q). Some residues tended to be mutated: T354 (P or R), R361 (G or H), R362 (P or X), R396 (W or Q), R398 (W or Q) (Figure 5). We identified 19 different missense mutations in 24 probands and 14 relatives (46%), 7 nonsense mutations in 10 probands and 4 relatives (17%), and 11 small deletions or insertions leading to predicted stop codons (21%). There were 2 splice defects (4%), one in frame duplication (2%) and one intronic variant (2%) located in the regulatory element of intron 4, (Figure 5 and Online Supplementary Table S1). The germline status of the GATA2 mutations was confirmed in 30 probands. The other 23 mutations were highly suspected to be germline as the variant allele frequency was close to 50%. Parental segregation was analyzed in 27 pedigrees. In 6 probands, the GATA2 mutation occurred de novo (P1, P9, P33, P35, P47 and P52). There was no significant difference in median age at diagnosis between probands and relatives. If we consider 2 groups of mutations based on the type of mutation (missense vs. nonsense/frameshift), no genotype/phenotype correlation could be highlighted regarding infection, warts, MDS, neoplasia or inflammatory complications. By contrast, there was a significant risk of developing leukemia in the group of patients with the missense mutations (14 of 38) versus the group with nonsense or frameshift mutations (2 of 28; P=0.007, Fisherâ&#x20AC;&#x2122;s exact test) (Table 2).

This large cohort with germline GATA2 mutation has the longest follow up (2249 patients/year) of any study. The homogeneous and exhaustive available clinical and biological data allow key clinical points regarding the disease to be described: the majority of patients (>90%) will present with a life-threatening hematologic and/or infectious manifestation by the age of 40. Within the first decade, disease presentation is limited to common bacterial infections or lymphoedema. During the second decade, patients may present with infections and/or inflammatory disease and/or hematologic transformation. Monocytopenia was frequent even without any other detectable hematologic disease. Our study also confirmed that hematologic complications are the major issue of the GATA2 deficiency: the probability of developing MDS and/or AML rapidly increased from 39% at the age of 20 to 80% at the age of

Table 2. Genotype/phenotype associations. The missense mutation group was associated with a significant risk of leukemia (*P=0.007, Fisher's exact test). MDS: myelodysplasia.

Patient pedigree Missense mutations (n=38) Frameshift + nonsense mutations (n=28) P

Leukemia

MDS

Warts

Fungus Mycobacteria

0.007*

0.793

0.081

0.331

0.237

Figure 5. Schematic organization of the GATA2 locus and protein. The protein is composed of 480 amino acids (top) encoded by 1443 nucleotides (bottom). The 5 coding exons (E2 to E6) are separated by blue lines. Fortyone mutations are shown: whole locus deletion including the del3q21 are in orange, nonsense mutation in the top of the schematic protein is in red, missense mutations are in green, small deletion in the bottom of the scheme is in brown, intronic mutation and splice defects are in blue, at the bottom, and mutation of the enhancer is in dark blue. Recurrence of mutations is specified in brackets.

haematologica | 2018; 103(8)

1285

J. Donadieu et al.

40. Eighteen patients developed acute or chronic leukemia. In addition, we identified a second patient with acute lymphoblastic leukemia (ALL).23 The GATA2 transcription factor is crucial for hematopoietic stem cell selfrenewal and differentiation,24,25 but also for B- and T-cell development in vitro, as shown in a recent murine lowlevel GATA2 overexpression model.26 Only 4 of 18 patients survived (2 after HSCT, one after azacitidine treatment, one JMML). Most of the other 14 died from infections and/or progressive hematologic disease. Longterm survival of our cohort is poor, with a high rate of mortality (probability of 42% at the age of 40, 69% at the age of 60). Classic chemotherapy strategies were revealed to be toxic and poorly efficient, and HSCT is hampered by the very high rates of toxicity in these patients. Early deaths were caused by the association of hematologic malignancies with severe infections. We propose to identify the patients at risk of evolution towards leukemia using the IPSS-R score.21 Moreover, secondary somatic mutations occur, which leads to leukemic transformation in patients with a GATA2 mutation. ASXL1 mutations were implicated in the first reports,27,28 then mutations in the RAS pathway and in the AML/MDS mutated genes.29 Some patients have a long history of low-risk MDS before it evolves into an aggressive disease, thus underlying the importance of identifying the markers that precede this hematologic evolution to help clinicians. Importantly, our study showed that the earlier HSCT is performed, the better the outcome. The question of timing of pre-emptive transplantation is still a subject of debate, but the improved overall survival of patients with refractory cytopenias suggests that early HSCT is a reasonable approach. Identification of additional somatic mutation in patients with MDS may prompt clinicians to perform HSCT. Our series confirmed the heterogeneity of the GATA2 mutational spectrum, with 45 different alleles including 26 new mutations. The previously published series have not reported correlations between genotype and phenotype, with the exception of null mutants which seems to be associated with an increased risk of lymphoedema in the US cohort.13 In our cohort, patients with missense mutations had a higher risk of developing leukemia than patients with frameshift or nonsense mutations. These data may suggest that the translated mutated GATA2 protein resulting from missense mutations is dominant negative and/or promotes leukemogenesis in contrast to frameshift or nonsense mutations, which may lead to haploinsufficiency. Recently, Chong et al. reported that the most prevalent GATA2 missense mutations (gT354M, gR396Q and gR398W) exhibit differences in the age of leukemia onset, supporting the concept of different functional consequences of GATA2 mutants.30 Our observation is reported for the first time and may also help clinicians to choose the best therapeutic option, especially an aggressive treatment for the disease. Further functional studies are needed to demonstrate this hypothesis. Severe and recurrent bacterial infections are frequent at diagnosis, and persist throughout the patient’s life. A mild defect of immunoglobulin production or a weak vaccinal response had also been reported in patients with a GATA2 deficiency.31 There is a lower incidence of mycobacterial diseases in our cohort (40% of the patients at the age of 40) than previously reported,13,14 occurring after the age of 20 in the majority of patients. 1286

All patients with mycobacterial disease have abnormal blood counts, monocytopenia (10 of 16), MDS (9 of 15) or both (7 of 15). The relatively low frequency of mycobacterial infection may be explained by the severity of disseminated infection leading to death or drastic treatment (such as HSCT) to avoid recurrence. Some patients experienced successive diseases with different species of environmental mycobacteria, suggesting that immunological memory is not efficient in patients with GATA2 mutations. Fungal infections occurred in 18 patients. Aspergillosis was always associated with neutropenia, as a consequence of GATA2 deficiency or secondary to the chemotherapy. Multiple cutaneous and genital warts at presentation are frequent (32 patients). Recurrent and life-threatening oncogenic HPV lesions led us to recommend early HPV vaccination, as proposed in WILD syndrome,32,33 which is maybe a clinical variant of GATA2 deficiency.34 Interestingly, one patient developed new HPV lesions after HSCT, raising the question of HPV genome persistence in epithelial cells, or a specific role for GATA2 in keratinocytes in the host control of HPV. It suggests that early HPV vaccination should be proposed in mutated patients. Susceptibility to severe viral infections led to 4 deaths in our cohort. One patient died from PML caused by the JC virus as the first manifestation of GATA2 deficiency; NK cell deficiency,35 monocytopenia and dendritic cell deficiency5 probably contribute to this immunodeficiency. New clinical presentations were identified in our survey. Auto-immune or chronic inflammatory disorders, such as lupus, sarcoidosis-like disease, Sweet's syndrome, panniculitis are recurrent. Lupus-like symptoms and autoimmune hepatitis have also been described in GATA2 deficiency.14,36 Given the occurrence of mycobacterial disease, infection should be investigated in patients with proven granuloma. Beyond the marked clinical heterogeneity of GATA2 deficiency, we also described 5 asymptomatic cases, including that of a 60-year old patient, raising the possibility that clinical penetrance is not complete. To evaluate clinical penetrance, genotypes of all first degree relatives of patients must be available. Moreover, these observations should lead to systematically testing a potential relative considered for donation when an HSCT with a sibling donor is feasible. This multicenter study was a unique opportunity to provide an extended and detailed clinical picture of GATA2 deficiency, which is a severe disorder that combines immunodeficiency, hematologic malignancy, pulmonary, dermatological and vascular diseases. It highlighted the fact that patients with GATA2 missense mutations have a high risk of developing leukemia and that this may be prevented by early HSCT with the help of new markers (identification of additional somatic mutations). Acknowledgments The Authors thank the patients and families for their participation in this study. The French registry is supported by grants from Amgen SAS, Chugai SA, Novartis, and by a grant from the Inserm. This project is supported by grants from Associations Laurette Fugain, 111 les Arts, Société Française des Cancers de l’Enfant, Enfanfare, Association Sportive de Saint Quentin Fallavier, and Barth France. The Authors thank the association IRIS for its support. haematologica | 2018; 103(8)

GATA2 deficiency French/Belgian cohort

Reference 1. Hahn CN, Chong C-E, Carmichael CL, et al. Heritable GATA2 mutations associated with familial myelodysplastic syndrome and acute myeloid leukemia. Nat Genet. 2011;43(10):1012-1017. 2. Vinh DC, Patel SY, Uzel G, et al. Autosomal dominant and sporadic monocytopenia with susceptibility to mycobacteria, fungi, papillomaviruses, and myelodysplasia. Blood. 2010;115(8):1519-1529. 3. Calvo KR, Vinh DC, Maric I, et al. Myelodysplasia in autosomal dominant and sporadic monocytopenia immunodeficiency syndrome: diagnostic features and clinical implications. Haematologica. 2011;96(8):1221-1225. 4. Hsu AP, Sampaio EP, Khan J, et al. Mutations in GATA2 are associated with the autosomal dominant and sporadic monocytopenia and mycobacterial infection (MonoMAC) syndrome. Blood. 2011;118(10):2653-2655. 5. Bigley V, Haniffa M, Doulatov S, et al. The human syndrome of dendritic cell, monocyte, B and NK lymphoid deficiency. J Exp Med. 2011;208(2):227-234. 6. Ostergaard P, Simpson MA, Connell FC, et al. Mutations in GATA2 cause primary lymphedema associated with a predisposition to acute myeloid leukemia (Emberger syndrome). Nat Genet. 2011;43(10):929-931. 7. Hsu AP, Johnson KD, Falcone EL, et al. GATA2 haploinsufficiency caused by mutations in a conserved intronic element leads to MonoMAC syndrome. Blood. 2013;121(19):3830-3837, S1-7. 8. Ganapathi KA, Townsley DM, Hsu AP, et al. GATA2 deficiency-associated bone marrow disorder differs from idiopathic aplastic anemia. Blood. 2015;125(1):56-70. 9. Griese M, Zarbock R, Costabel U, et al. GATA2 deficiency in children and adults with severe pulmonary alveolar proteinosis and hematologic disorders. BMC Pulm Med. 2015;1587. 10. Polat A, Dinulescu M, Fraitag S, et al. Skin manifestations among GATA2-deficient patients. Br J Dermatol. 2018;178(3):781785. 11. Pasquet M, Bellanné-Chantelot C, Tavitian S, et al. High frequency of GATA2 mutations in patients with mild chronic neutropenia evolving to MonoMac syndrome, myelodysplasia, and acute myeloid leukemia. Blood. 2013;121(5):822-829. 12. Wlodarski MW, Hirabayashi S, Pastor V, et al. Prevalence, clinical characteristics, and prognosis of GATA2-related myelodysplastic syndromes in children and adolescents. Blood. 2016;127(11):1387-1397; quiz 1518. 13. Spinner MA, Sanchez LA, Hsu AP, et al.

haematologica | 2018; 103(8)

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

GATA2 deficiency: a protean disorder of hematopoiesis, lymphatics, and immunity. Blood. 2014;123(6):809-821. Collin M, Dickinson R, Bigley V. Haematopoietic and immune defects associated with GATA2 mutation. Br J Haematol. 2015;169(2):173-187. Donadieu J, Beaupain B, Mahlaoui N, Bellanné-Chantelot C. Epidemiology of congenital neutropenia. Hematol Oncol Clin North Am. 2013;27(1):1-17, vii. Donadieu J, Leblanc T, Bader Meunier B, et al. Analysis of risk factors for myelodysplasias, leukemias and death from infection among patients with congenital neutropenia. Experience of the French Severe Chronic Neutropenia Study Group. Haematologica. 2005;90(1):45-53. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med Off J Am Coll Med Genet. 2015;17 (5):405-424. Plebani A, Ugazio AG, Avanzini MA, et al. Serum IgG subclass concentrations in healthy subjects at different age: age normal percentile charts. Eur J Pediatr. 1989;149(3): 164-167. Niemeyer CM, Baumann I. Classification of childhood aplastic anemia and myelodysplastic syndrome. Hematol Am Soc Hematol Educ Program. 2011;2011:84-89. Vardiman JW, Thiele J, Arber DA, et al. The 2008 revision of the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009;114(5):937-951. Greenberg PL, Tuechler H, Schanz J, et al. Revised international prognostic scoring system for myelodysplastic syndromes. Blood. 2012;120(12):2454-2465. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. Koegel AK, Hofmann I, Moffitt K, Degar B, Duncan C, Tubman VN. Acute lymphoblastic leukemia in a patient with MonoMAC syndrome/GATA2 haploinsufficiency. Pediatr Blood Cancer. 2016;63(10):18441847. Tsai FY, Keller G, Kuo FC, et al. An early haematopoietic defect in mice lacking the transcription factor GATA-2. Nature. 1994;371(6494):221-226. Rodrigues NP, Tipping AJ, Wang Z, Enver T. GATA-2 mediated regulation of normal hematopoietic stem/progenitor cell func-

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

tion, myelodysplasia and myeloid leukemia. Int J Biochem Cell Biol. 2012;44(3):457-460. Nandakumar SK, Johnson K, Throm SL, Pestina TI, Neale G, Persons DA. Low-level GATA2 overexpression promotes myeloid progenitor self-renewal and blocks lymphoid differentiation in mice. Exp Hematol. 2015;43(7):565-577.e1-10. Bödör C, Renneville A, Smith M, et al. Germ-line GATA2 p.THR354MET mutation in familial myelodysplastic syndrome with acquired monosomy 7 and ASXL1 mutation demonstrating rapid onset and poor survival. Haematologica. 2012;97(6):890-894. West RR, Hsu AP, Holland SM, CuellarRodriguez J, Hickstein DD. Acquired ASXL1 mutations are common in patients with inherited GATA2 mutations and correlate with myeloid transformation. Haematologica. 2014;99(2):276-281. Wang X, Muramatsu H, Okuno Y, et al. GATA2 and secondary mutations in familial myelodysplastic syndromes and pediatric myeloid malignancies. Haematologica. 2015;100(10):e398-401. Chong C-E, Venugopal P, Stokes PH, et al. Differential effects on gene transcription and hematopoietic differentiation correlate with GATA2 mutant disease phenotypes. Leukemia. 2018;32(1):194-202. Chou J, Lutskiy M, Tsitsikov E, Notarangelo LD, Geha RS, Dioun A. Presence of hypogammaglobulinemia and abnormal antibody responses in GATA2 deficiency. J Allergy Clin Immunol. 2014;134(1):223-226. Kreuter A, Hochdorfer B, Brockmeyer NH, et al. A human papillomavirus-associated disease with disseminated warts, depressed cellmediated immunity, primary lymphedema, and anogenital dysplasia: WILD syndrome. Arch Dermatol. 2008;144(3):366-372. Ostrow RS, Manias D, Mitchell AJ, Stawowy L, Faras AJ. Epidermodysplasia verruciformis. A case associated with primary lymphatic dysplasia, depressed cellmediated immunity, and Bowen’s disease containing human papillomavirus 16 DNA. Arch Dermatol. 1987;123(11):1511-1516. Dorn JM, Patnaik MS, Van Hee M, et al. WILD syndrome is GATA2 deficiency: A novel deletion in the GATA2 gene. J Allergy Clin Immunol Pract. 2017;5(4):11491152.e1. Mace EM, Hsu AP, Monaco-Shawver L, et al. Mutations in GATA2 cause human NK cell deficiency with specific loss of the CD56(bright) subset. Blood. 2013;121(14): 2669-2677. Webb G, Chen Y-Y, Li K-K, et al. Single-gene association between GATA-2 and autoimmune hepatitis: A novel genetic insight highlighting immunologic pathways to disease. J Hepatol. 2016;64(5):1190-1193.

1287

ARTICLE

Myelodysplastic Syndrome

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1288-1297

The interleukin-3 receptor CD123 targeted SL-401 mediates potent cytotoxic activity against CD34+CD123+ cells from acute myeloid leukemia/myelodysplastic syndrome patients and healthy donors Rajeswaran Mani,1 Swagata Goswami,1 Bhavani Gopalakrishnan,1 Rahul Ramaswamy,1 Ronni Wasmuth,1 Minh Tran,1 Xiaokui Mo,2 Amber Gordon,1 Donna Bucci,1 David M. Lucas,1,3 Alice Mims,3 Christopher Brooks,4 Adrienne Dorrance,1,3 Alison Walker,1,3 William Blum,1,3 John C. Byrd,1,3 Gerard Lozanski,1,5 Sumithira Vasu1,3,* and Natarajan Muthusamy1,3,*

Comprehensive Cancer Center, The Ohio State University, Columbus, OH; 2Center for Biostatistics, The Ohio State University, Columbus, OH; 3Division of Hematology, Department of Internal Medicine, College of Medicine, The Ohio State University, Columbus, OH; 4Stemline Therapeutics, Inc., New York, NY and 5Department of Pathology, College of Medicine, The Ohio State University, Columbus, OH, USA 1

*SV and NM contributed equally as senior authors

ABSTRACT

Correspondence: raj.muthusamy@osumc.edu

Received: January 10, 2018. Accepted: May 15, 2018. Pre-published: May 17, 2018. doi:10.3324/haematol.2018.188193 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1288

iseases with clonal hematopoiesis such as myelodysplastic syndrome and acute myeloid leukemia have high rates of relapse. Only a small subset of acute myeloid leukemia patients are cured with chemotherapy alone. Relapse in these diseases occurs at least in part due to the failure to eradicate leukemic stem cells or hematopoietic stem cells in myelodysplastic syndrome. CD123, the alpha chain of the interleukin-3 receptor heterodimer, is expressed on the majority of leukemic stem cells and myelodysplastic syndrome hematopoietic stem cells and in 80% of acute myeloid leukemia. Here, we report indiscriminate killing of CD123+ normal and acute myeloid leukemia / myelodysplastic syndrome cells by SL-401, a diphtheria toxin interleukin-3 fusion protein. SL-401 induced cytotoxicity of CD123+ primary cells/blasts from acute myeloid leukemia and myelodysplastic syndrome patients but not CD123– lymphoid cells. Importantly, SL-401 was highly active even in cells expressing low levels of CD123, with minimal effect on modulation of the CD123 target in acute myeloid leukemia. SL-401 significantly prolonged survival of leukemic mice in acute myeloid leukemia patient-derived xenograft mouse models. In addition to primary samples, studies on normal cord blood and healthy marrow show that SL-401 has activity against normal hematopoietic progenitors. These findings indicate potential use of SL-401 as a “bridge-to-transplant” before allogeneic hematopoietic cell transplantation in acute myeloid leukemia / myelodysplastic syndrome patients.

©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.

1288

Introduction Acute myeloid leukemia (AML) incidence increases with age, and about 21,000 new cases are expected in 2017.1,2 Significant heterogeneity exists in AML as shown by diversity of karyotype, genetic mutations and epigenetic aberrations. Standard chemotherapies and immunotherapies have only limited efficacy, and most AML patients relapse partly due to failure to eradicate AML leukemic stem cells (LSC) which undergo clonal evolution and serve as a reservoir for relapse.3 Up to 47% of patients older than 60 years who undergo allogeneic transplantation for AML will relapse.4 Myelodysplastic syndrome (MDS) incidence also increases with age with an expected incidence of 15,000 cases annually.5 Upon transformation to AML, MDS patients have a poor prognosis as compared to AML cases that occur de novo.6 haematologica | 2018; 103(8)

Activity of SL-401 in AML and MDS

Patients with MDS who are refractory to hypomethylating agents also have very limited therapeutic options.7 CD33 is a widely expressed myeloid marker present on the majority of AML cells, and CD33-targeted immunotherapies have shown promising results.8,9 However, LSC of immature AML lack CD33 expression and are not eliminated by CD33 targeted agents.8 CD123 is the a chain of the interleukin-3-receptor (IL-3R) heterodimer that has affinity and specificity for interleukin-3 (IL-3).10 LSC are known to express CD12311,12 and several efforts to target CD123 to eradicate these cells have emerged.13-18 To date, however, each of these agents exhibited shortcomings that limited their development. While a prototype monoclonal antibody against CD123 showed no promising response in AML, new clones of antibodies/single chain fragment variable (ScFv) are currently under investigation for antibody-drug conjugates (ADC) and antibody-dependent cellular cytotoxicity (ADCC) functions.13-15 CD123 targeted chimeric antigen receptor T cells (CART cells) are now being evaluated in early phase clinical trials and fatalities from cytokine release syndrome in patients receiving CD123-CART therapy were reported as of this writing. Recently, preclinical effects of a dual affinity retargeting (DART) molecule, generated from CD3 and CD123 antibodies, were described in AML.18 SL-401 is a recombinant fusion protein consisting of human IL-3 and truncated diphtheria toxin, previously reported to be active in CD123+ neoplasms.19,20 The IL-3 domain dictates the specificity for CD123 expressing cells, and the catalytic unit of diphtheria toxin upon internalization inhibits the translational machinery to initiate cell death. Although, most chemotherapy regimens are effective in eradicating tumor bulk in the periphery, progenitor tumor cells in the marrow are less exposed to treatments and receive protective support from the niche. Bone marrow stromal cells are a part of the bone marrow microenvironment and play an important role in the proliferation of AML by offering various forms of protection mediated through soluble factors and contact-dependent survival signals to leukemic cells.21-24 Co-culturing of AML cells on human bone marrow stromal cell line HS-5 monolayers increases proliferation, viability, and colony formation of the AML cells, while diminishing chemotherapy-induced apoptosis.25 Furthermore, bone marrow-derived mesenchymal stromal cells (MSC) from AML patients exhibit overlapping as well as distinct cytogenetic abnormalities versus AML blasts and provide a conducive cytokine and growth milieu for leukemia cells.24,26 Therefore, it is critical to assess potential AML therapies in the context of the bone marrow microenvironment with autologous MSCs, which might be unique for each patient. In the present study, we report the potent activity of SL-401 in CD123 expressing primary AML blasts and MDS samples and its relation to CD123 levels. We also show in vitro studies of SL-401 when AML cells are co-cultured with MSCs and in vivo studies using patient derived xenograft (PDX) mouse models. Whether CD123 is sufficiently specific for leukemic stem cells is controversial. We show here definitively that CD123 targeted SL-401 is cytotoxic to both normal cord blood-derived hematopoietic stem cells and CD123+ blasts in AML and MDS. These findings suggest that CD123 targeting may cause pancytopenia as a consequence of ontarget off-tumor effects and have translational relevance haematologica | 2018; 103(8)

for use of CD123 targeting as a “bridge to transplant” in AML and MDS. Whether MDS may be less likely to develop on-target and off-tumor side effects is being explored in combination studies of SL-401 and hypomethylating agents in early phase clinical trials (clinicaltrials.gov identifier 03113643).

Methods Cells AML cell lines, MV4-11 and MOLM-13 were obtained from DSMZ in 2015, confirmed to have FLT3-ITD mutation by DNA sequencing, and routinely tested for mycoplasma using MycoAlert™ Mycoplasma Detection Kit (Lonza). Primary AML and MDS cells were obtained from patient apheresis products or bone marrow aspirates following written informed consent under an institutional review board (IRB) approved protocol, according to the Declaration of Helsinki. Umbilical cord blood samples were received through the Leukemia tissue bank and collected under an IRB-approved protocol. AML primary cells were cultured in RPMI 1640 (Life Technologies, Grand Island, NY) with 10-20% fetal bovine serum (FBS) (Sigma-Aldrich, St Louis, MO); 2mM L-glutamine; penicillin (100 U/ml); streptomycin (100 μg/ml) (Life Technologies) and supplemented with 10ng/ml GM-CSF and SCF (R&D Systems) overnight before treatment with SL-401. MDS samples were cultured similarly, with additional IL-3 (10ng/ml; R&D Systems). AML subtype and mutational characteristics of AML patient samples used in this study are listed in Online Supplementary Table S1.

Bone marrow-derived mesenchymal stromal cell expansion For AML-stromal cell co-culture experiments, AML cells were cultured in RPMI 1640 base medium with 10% FBS without any growth factors on the monolayer of the HS-5 human fibroblast cell line expressing green fluorescent protein (GFP) or with autologous mesenchymal stromal cells (MSC) derived from bone marrow of AML patients. To derive MSC for each AML, bone marrow cells were incubated for 24-36 hours in DMEM (Life Technologies) with 10% FBS and antibiotics. The plastic-adherent, elongated spindle- to stellate- shaped cells were expanded following aspiration of non-adherent cells and weekly exchange of media and passaging when confluent. MSC in passage 3 or 4 were trypsinized and stained for CD45, CD73, CD90 and CD105 with a viability stain and assessed via flow cytometry for purity.

Chemicals and reagents SL-401 was provided by Stemline Therapeutics Inc. and is now in clinical trials20 (clinicaltrials.gov identifier 02270463). Additional information in Online Supplementary Methods.

Cell viability and staining Apoptosis was measured by annexin-V-FITC and propidium iodide as described previously.27 For co-culture experiments, stromal cells were gated out based on GFP+ for HS-5 and CFSE+ for MSC, and the viability of leukemic cells was determined by annexin-V PE and 7-AAD staining. Cell counts were measured using a Tali cytometer (Invitrogen) according to the manufacturer’s instruction. LIVE/DEAD stain (Invitrogen) was used in combination with surface markers for gating live cells in flow cytometric experiments. CD123 Molecules of Equivalent Soluble Fluorochrome (MESF) on AML cells was calculated using Quantum™ MESF microsphere kit (Bangs Laboratories, Fishers, IN) and CD123 antibody (Clone: SSDCLY107D2; Beckman 1289

R. Mani et al.

Coulter) after establishing calibration curve for each experiment per the manufacturerâ&#x20AC;&#x2122;s instructions. These CD123-MESF experiments were performed at the OSU Wexner Medical Center Clinical Cytometry Facility. Flow cytometric data were analyzed using Kaluza software (Beckman-Coulter).

AML Patient-Derived Xenograft (PDX) models Animal experiments were performed under a protocol approved by the OSU Institutional Animal Care and Use Committee (IACUC) in NRG-SGM3 (NRGS) mice. Detailed in Online Supplementary Methods.

Statistics Detailed in Online Supplementary Methods.

Results AML cells express varying levels of CD123 and are sensitive to SL-401 While more than 80% of AML cases show CD123 expression on blasts,28 the levels of expression are variable. Therefore, we tested the cell surface expression of CD123 on primary AML blasts. As expected, CD123 was found on the majority of AML samples tested (Table 1, and Online Supplementary Figure S1). Rarely, CD33-/CD123+ AML cell populations were also observed (Online Supplementary Figure S2A). SL-401 induced potent cytotoxicity on AML primary cells as seen by dose-dependent reduction in viability of AML cells (Figure 1A) and absolute cell numbers (Online Supplementary Figure S2B). The clonogenicity of AML samples was also significantly reduced by SL-401 as seen in colony forming assays (Figure 1B). We next determined if the activity of SL-401 was diminished in the presence of IL-3. In AML cell cultures containing 10ng/ml IL-3, a concentration > 100 fold of physiological levels, SL-401 was effective in inducing cytotoxicity and reducing the cell counts (Online Supplementary Figure S2 B-C). However, SL-401 (100ng/ml) cytotoxicity was inhibited by IL-3 in dose dependent manner, especially at doses >1000 fold of physiological levels, proving target specificity (Online Supplementary Figure S2D). Blasts derived from high-risk FLT3-ITD mutated AML often express high levels of CD123,28 so we next tested the activity of SL-401 on AML primary cells and cell lines with this mutation. Importantly, high risk FLT3-ITD+ AML tend to express similar or higher than median CD123-MESF (22825) and responded to SL-401 (Figure 1C). The FLT3-ITD+AML cell lines MV4-11 and MOLM-13 express CD123 with other myeloid markers CD45, CD33 (Figure 1D) and SL-401 strongly inhibited the growth of these cell lines (P<0.0001 for dose trends of MOLM-13 and MV4-11) (Figure 1E). However, SL-401 did not induce cytotoxicity in CD123 cell line K562 proving target specificity (Online Supplementary Figure S3A). Interestingly, cell density did not affect cytotoxicity in CD123+ MV4-11 cells (Online Supplementary Figure S3B).

SL-401 overcomes autologous stromal cell protection in co-cultures As AML cells and their progenitors derive growth and survival support from the bone marrow stromal cells and microenvironment,22,23 we evaluated the effect of stromal protection on SL-401-induced cytotoxicity. Although 1290

Table 1. Acute myeloid leukemia blasts express varying levels of cell surface CD123.

AML

AML 1 AML 2 AML 3 AML 4 AML 5 AML 6 AML 7 AML 8 AML 9 AML 10 AML 11 AML 12 AML 13 AML 14 AML 15 AML 16

CD123-MFI

CD123-MESF on blasts

Blasts (B)

Lymphs (L)

B/L Ratio

2.03 1.73 2.45 2.35 1.48 1.29 1.77 1.59 1.13 1.02 0.83 1.44 1.60 1.29 0.46 0.41

0.28 0.32 0.58 0.61 0.58 0.28 0.35 0.40 0.37 0.36 0.40 0.37 0.37 0.39 0.35 0.37

7.38 5.49 4.21 3.86 2.53 4.56 5.00 3.94 3.10 2.81 2.06 3.86 4.32 3.33 1.30 1.11

ND ND 13672 10692 4396 ND 47791 61863 66972 33137 14117 22825 19297 63204 35980 18636

CD123-MFI and CD123-MESF on blasts and lymphocytes (N=16) are shown. MFI: Mean Fluorescence Intensity; MESF: Molecules of Equivalent Soluble Fluorochrome; ND: Not determined.

MV4-11 cells did not receive additional protection, the protective effect of stromal HS-5 on MOLM-13 and AML cells was reduced by SL-401 treatment (Figure 2 A-B). We next tested the effect of SL-401 on AML cells cultured on primary autologous bone marrow stromal cells from AML patients. For this purpose, we derived MSC from individual AML patient marrow samples. MSCs grown in flasks exhibited stellate- to spindle-shaped large cells and were adherent. Purity and phenotype were confirmed by multicolor flow cytometry (Online Supplementary Figure S4A). Importantly, SL-401 was able to reduce the growth of AML regardless of culture with autologous MSC (Figure 2C and Online Supplementary Figure S4B).

Low levels of CD123 are sufficient for SL-401 mediated cytotoxicity Although earlier studies utilizing variant/wild type IL-3diphtheria toxin have reported direct correlation of cytotoxicity to the levels of IL-3R subunits, the clinical trials have shown intriguing results with lack of correlation between pharmacokinetics and pharmacodynamics.19, 20,2934 To study this relationship between SL-401 and its target, we compared the sensitivity of AML cells to SL-401 vs. their expression of CD123. As shown in Figure 3A, there was no significant correlation between the relative viability of AML samples treated with SL-401 for 48 hours with their CD123 expression (CD123-MESF on AML cells). Interestingly, AML cells expressing as few as 10,000 CD123 molecules were still sensitive to SL-401-mediated cytotoxicity. As leukemic cells can acquire resistance through target down regulation during the course of treatment, we sought to determine if SL-401 modulated cell surface CD123 levels after exposure. Although SL-401 treated samples had slightly reduced CD123 (CD123 haematologica | 2018; 103(8)

Activity of SL-401 in AML and MDS

Figure 1. AML cells express CD123 and can be targeted with SL-401. (A) SL-401 induces cytotoxicity in patient-derived AML blasts. AML blasts were cultured with varying doses of SL-401 and viability was measured at 24 and 48 hr (N=16; Trend: 24hr difference = -42.13, P<0.0001 and 48hr difference =-91.59, P<0.0001). Only AML samples >50% viable by PI staining were used for the analysis. (B) SL-401 inhibits clonogenicity of AML cells. Representative images of AML colony from vehicle treated plates are shown (PH 4X EVOS® XL Core imaging system). Leukemic colonies were counted 10-14 days after plating AML cells with continuous presence of vehicle or SL-401(1 μg/ml) in duplicates. Only AML samples that formed at least 15 colonies (more than 20 cells per colony) or clusters (5-20 cells per cluster) in vehicle treated plates were used for the analysis. Each dot represents average of duplicate plates for an AML sample (N=5 AML). (C) FLT3-ITD+ AML are sensitive to SL-401(same samples in Figure 1A). (D) Expression of CD123 in AML cell lines MV4-11 and MOLM-13 as determined by flow cytometry. (E) Growth inhibition by SL-401 in MV4-11 and MOLM-13 cells. Cells (0.5x106/ml) were treated with vehicle or SL-401 (100ng/ml and 1 μg/ml) and cell viability was measured after 72 hrs.

haematologica | 2018; 103(8)

1291

R. Mani et al. A

Figure 2. SL-401 is active despite the presence of stroma in co-cultures. (A-B) Anti-leukemic activity of SL-401 in AML-HS5 stromal co-cultures. Cell lines (0.25x106/ml) (A) or patient-derived AML blasts (1x106/ml) (N=7) (B) were seeded onto pre-cultured GFP+ HS5 stromal cells and treated with vehicle or SL-401 (100 ng/ml; 1 μg/ml). Viable AML cell counts were measured at 48 hr by annexin-V-PE and 7-AAD staining after exclusion of GFP+ stromal cells. (C) SL-401 abrogates AML growth in autologous bone marrow-derived MSC co-cultures. AML cells were seeded onto CFSE-labeled marrow-derived autologous MSC and cultured with varying concentration of SL-401 for 120 hrs. Viability and cell counts were measured using a flow cytometric panel after staining with annexin-V-PE, 7-AAD and adding Countbrite beads (N=6; protective effect of MSC P=0.0107). Presence of MSC had no significant inhibitory effect on drug (P=0.7297, SL-401 100 ng/ml ± MSC).

ΔMFI= 0.689 Vehicle vs. SL-401 at 24 hr; N=16 AML; P= 0.001), it should be noted that a proportion of the leukemic cells were killed at this time point and CD123 levels were calculated from the viable cells. Moreover, the leukemic cells still maintained CD123 expression >10,000 MESF (Figure 3B) and showed further decrease in viability at 48 hr (same samples as in Figure 1A). Together, these data suggest a lack of an SL-401 escape mechanism for malignant cells via target down-modulation.

Effects of SL-401 on normal CD123+ cells derived from cord blood To further assess the effect of SL-401 on normal CD123+ cells, we selected CD34+ cells from umbilical cord blood samples. Cord blood derived CD34+ cells expressed dim to high expression of CD123 (Figure 3C). In long-term cultures using semisolid methylcellulose media and myeloid growth factors, SL-401 completely abrogated myeloid colonies suggesting that SL-401 can compromise hematopoiesis of normal CD34+ cells (Figure 3C and Online Supplementary Figure S5A). Moreover, in short-term liquid culture of cord blood samples, SL-401 mediated moderate toxicity of CD34+ cells (Figure 3D). This was 1292

also confirmed using CD34+CD38–Lineage– sorted cells from normal donor bone marrow where SL-401 compromised the clonogenicity of hematopoietic stem cells (Online Supplementary Figure S5B).

Blasts from high risk MDS express CD123 and are sensitive to SL-401 As SL-401 exhibited cytotoxic effects in AML blasts expressing low levels of CD123, we evaluated if SL-401 is active in high-risk MDS, where blasts % is < 20%. We first tested CD123 expression in MDS with refractory anemia and excess blasts (RAEB)/ (MDS-EB) samples by flow cytometry. The majority of these samples (6/7) tested positive for CD123 expression, as determined by MFI. Importantly, CD123+ MDS blasts were sensitive to SL-401 mediated cytotoxicity, while CD123– lymphoid cells in the same cultures were spared (Figure 4 A-B and Online Supplementary Figure S6).

SL-401 prolongs survival in AML PDX model To demonstrate the therapeutic effect of SL-401 in vivo, we used AML patient-derived xenograft (PDX) models. As described in the methods, cells from CD123+ AML haematologica | 2018; 103(8)

Activity of SL-401 in AML and MDS

patients (n=3) pre-tested for engraftment potential were used to engraft NRGS mice (one AML donor per mouse) and randomized to treatment cohorts. Despite the selection of patient samples containing a high percentage of CD123+AML cells (Online Supplementary Figure S7), we initially encountered T-cell expansion in vivo due to contaminating T cells in our preliminary studies (data not shown). We therefore tested the effect of anti-CD3 antibody (OKT3) on AML cultures in vitro in ablating T cells, and confirmed that OKT3 reduced both absolute T-cell numbers and CD3 expression (data not shown). Engraftment of haematologica | 2018; 103(8)

Figure 3. Low levels of CD123 are sufficient for SL-401-mediated cytotoxicity. (A) Correlation between CD123 levels and SL-401 induced cytotoxicity in patient AML cells. CD123-MESF on vehicle-treated AML samples (N=13) and viability after 48 hr of SL-401 (1 μg/ml) treatment were used for correlation analysis. (B) Changes in CD123 molecules in AML blasts after SL-401 treatment. CD123-MESF was determined on viable AML blasts after 24 hours of treatment with vehicle or SL-401 (N=13). Mean difference = 11086 (95% CI for mean difference: 2006, 20166), P=0.021. (C-D) CD123 expression and effect of SL-401 on umbilical cord blood derived CD34+ cells. (C) CD34 positive selected cells were used for multicolor flow cytometry analysis and colony formation assays. Colonies were counted 10-14 days after plating CD34+ cells with continuous presence of vehicle or SL401(1 μg/ml) P=0.019. (D) Effect of SL-401 on umbilical cord blood liquid cultures. Non-enriched Cord blood samples (N=4 CB) were ficoll processed to obtain mononuclear cells and cultured in RPMI media with 20% FBS and GM-CSF, SCF and IL-3 (10 ng/ml) in presence of SL-401 and the cells were counted and immunophenotyped after 48 hours. Live CD34+ cell counts per ml of culture are shown in the graph. Vehicle vs. SL-401 10ng/ml not significant; Vehicle vs. SL-401 100ng/ml; P=0.0356 and Vehicle vs. SL-401 1 μg/ml; P=0.0089.

AML was confirmed in the peripheral blood of mice by week 4. Treatment of (AML 28/AML29) xenografted mice with SL-401 significantly increased survival compared to the vehicle control (Figure 5A). In the absence of busulfan preconditioning (AML 30), the development of AML was delayed and the mouse reached removal criteria at 102 days after engraftment (versus with busulfan 48days; data not shown). In this group, SL-401 treatment increased the survival time in the treated mouse (survival: vehicle, 102 days; SL401, 154 days; not shown). Further, evaluation of bone marrow after treatment with SL-401 in AML PDX 1293

R. Mani et al.

mice revealed reduced leukemic burden compared to vehicle treated control mice (Figure 5B). The presence of AML in the animals was confirmed in spleen, bone marrow and blood using a multi-color flow cytometric panel (CD45 /CD33/CD34/CD123/CD19/CD3/HLA-DR/murineCD45 /viability stain) and CD123 expression was found to be maintained in vivo in engrafted mice (Figure 5C and Online Supplementary Figure S8-S9).

Discussion Relapse-free survival is low in AML and MDS in the absence of allogeneic hematopoietic stem cell transplantation. Current therapies, even if targeted for specific subtypes and mutational groups, provide only transient remissions, likely due in part to persistent stem cells.3 The expression of CD123, the alpha chain of IL-3 receptor, is

A P

P P

Figure 4. Blasts from high-risk MDS express CD123 and are sensitive to SL401. (A-B) SL-401 depletes CD123+ MDS blasts. MDS samples cultured with vehicle or SL-401 (1 μg/ml, 2 μg/ml) for 120 hours were stained for various myeloid markers (CD45, CD33, CD34, CD123, and viability stain) and live cells were counted. (A) Representative MDS-RAEB cells cultured with varying doses of SL-401 for 120 hours. Density plots gated on live CD45+ population are shown. (B) Live CD34+ blast concentration (bars), live CD34+ blast % (red lines) and live lymphoid % (green lines) are shown for each sample (N=6 CD123+ MDS) P=0.0183 for live CD34+ blast concentration between SL401 1 μg/ml vs. vehicle, and P=0.0063 for live CD34+ blast concentration between SL401 2 μg/ml vs. vehicle. CD123– MDS sample (MDS 1) was included as a control.

1294

haematologica | 2018; 103(8)

Activity of SL-401 in AML and MDS

limited to a sub-population of normal stem cells, few lymphoid progenitors, basophils and AML LSCs.35-37 Moreover, blasts in about 10% of AML cases express CD123 without concurrent CD33 expression, and overall about 80% of AML patients express CD123.28 Furthermore, high numbers of CD34+CD38low/-CD123+ blasts predict poor outcome in AML patients and CD34+CD38â&#x20AC;&#x201C;CD123+ population is increased in AML patients at relapse.37 While favorable and intermediate risk AML subtypes have comparable levels of CD123 expression, high-risk AML sub-groups harboring a FLT3-ITD mutation tend to have higher CD123 expression.38 It was recently shown that CD123 targeted chimeric antigen receptor (CAR) T-cell therapy leads to myeloablation in primary AML xenografts.16 Therefore, CD123 serves as an important target for treating AML. Here, we have shown that AML cells express varying levels of cell surface CD123. Importantly, we show that SL-401 is potent in killing AML cells that express even low levels of CD123 and activity of SL-401 is not dependent on surface density of CD123. Earlier reports have used IL3 or variant IL-3-fused toxins to investigate the CD123 targeted killing of leukemic cells.29-34 These studies did not include a detailed analysis on AML cell survival or activity

using in vivo PDX survival models. Previous studies involving IL-3-protein-toxin have used fluorescence intensities or transcript levels to compare the CD123 expression on different cell types and colony forming ability of AML as a prime read out. To control for variability between experiments, we used microspheres with a standardized fluorochrome to derive the CD123-MESF, thus minimizing variations due to instrument or time points. In our study, we saw no correlation between CD123-MESF and sensitivity to SL-401 cytotoxicity. It is important to note that the previous studies utilized a different clone of antibody, assay for receptor subunits, AML culture methods and cytotoxicity assays and end points. The use of high serum containing medium to culture AML in our studies may have affected CD123 expression less likely (Online Supplementary Figure S10) and possible induction of LSC differentiation. The fact that AMLs expressing less than 10,000 CD123MESF are sensitive to SL-401 underscores the potency of this agent. This is corroborated by a clinical trial with SL401 that showed no correlation between robust responses and SL-401 pharmacokinetics.19,20 Interestingly, SL-401 mediated cytotoxicity in the presence of exogenous IL-3 (10ng/ml) in in vitro AML/MDS cultures. We think this

P=0.0069

Figure 5. In vivo activity of SL-401 in AML PDX models. (A) Survival curves of treatment groups from busulfan preconditioned NRGS mice engrafted with primary AML (AML 28 and AML 29). AML 28 was used to engraft three animal in each group and AML 29 was used to engraft one animal in each group. Total mice used are four per group. Ten days after engraftment, mice were randomized and treated with vehicle or SL-401 (50 Îźg/kg administered intraperitoneally, 3 doses: M/W/F per week for 5 weeks). (B) AML burden in bone marrow of NRGS mice engrafted with AML and treated with vehicle or SL-401 during week 3-6. Mice were treated blindly and sacrificed on week 7. Bone marrow was harvested, counted and immunophenotyped by multicolor flow cytometry. P=0.004 for Vehicle vs. SL-401 treatment. (C) Representative flow cytometric dot plots of bone marrow showing tumor burden of Vehicle and SL-401 treated PDX mice, engrafted with AML 28, at week 7.

haematologica | 2018; 103(8)

1295

R. Mani et al.

may be due to high potency of diphtheria toxin, as few molecules are sufficient for cytotoxicity and/or independent binding sites of IL-3 and SL-401. However, the results we saw with decreased myeloid colony formation ability of cord blood derived CD34+ cells and normal bone marrow derived CD34+CD38– cells with SL-401 suggests that SL-401 can impact normal stem cells and myeloid progenitors. These findings suggest that SL-401 could be used to target leukemic progenitors and decrease leukemic burden with potential use as a “bridge-to-transplant” before allogeneic hematopoietic cell transplantation.39 These results are consistent with previous studies showing reduction in normal hematopoiesis when CD123-targeted CAR-T cells were used in mouse models.16 Bone marrow MSC play a critical role in leukemic cell survival and drug resistance. Recently, bone marrowderived MSC from primary AML samples were used to establish faithful PDX models by coating ossicle/bioscaffold with MSC before inoculation into the mice by investigators.40-42 We therefore evaluated SL-401 in AML co-cultures using HS5 stromal cell lines and autologous MSC derived from bone marrow aspirates. Despite variabilities and inconsistencies in culturing patient marrow-derived MSCs, we were able to successfully propagate six AML MSC and co-cultured AML blasts on these monolayers. This method accounts for the unique MSC cytogenetic/mutational aberration-dependent survival of AML blasts. Despite this MSC protection, SL-401 retained its activity, supporting its therapeutic potential in patients. MDS are a group of myeloid disorders characterized by dysplastic changes in the bone marrow and peripheral cytopenias, which are manifestations of ineffective hematopoiesis.6 The MDS-RAEB / MDS-EB groups represent high risk MDS and often show high levels of CD34+ cells. Thus, we interrogated the expression of CD123 in these groups. Almost all MDS samples tested (6/7) were positive for CD123 expression. Due to limited number of cells, we did not perform CD123 MESF analysis on MDS samples. However, MDS blasts responded to SL-401 as shown by decreases in both the blast percentage and the absolute count. Moreover, we showed that SL-401 reduced the numbers of CD123+ myeloid cells but not CD123– lymphoid cells in long-term cultures, supporting the selectivity of SL-401. Our evaluation of SL-401 in mice engrafted with primary AML cells is particularly relevant to the therapeutic potential of this agent in AML. Although difficult to establish, AML PDX models capture the patient heterogeneity and genomic landscape of patients and form an important tool for evaluation of therapies.43 Consistent with previously described variability in engraftment capabilities of

References 1. Siegel RL, Miller KD, Jemal A. Cancer Statistics, 2017. CA Cancer J Clin. 2017;67(1):7-30. 2. Estey EH. Acute myeloid leukemia: 2014 update on risk-stratification and management. Am J Hematol. 2014;89(11):1063-1081. 3. Shlush LI, Mitchell A, Heisler L, et al. Tracing the origins of relapse in acute

1296

primary AML samples, only 3 of 7 AML samples engrafted successfully in NRGS mice, although the inclusion of busulfan as conditioning regimen allowed accelerated leukemic engraftment. Recently it was shown that inclusion of OKT3 either in in vitro cultures or in mice prevented T-cell mediated GvHD and improved human hematopoietic cell engraftment. Thus, for our in vivo studies, we cultured AML with growth factors and OKT3 prior to engraftment to eliminate CD3+ cells. SL-401 treated mice had significantly longer mean survival compared to the vehicle-treated controls. However, there was no difference in the proportion of human leukemic cells in the spleen, bone marrow or peripheral blood of mice treated with vehicle and SL-401 at end points in the survival study, presumably due to discontinuation of treatment in SL-401 group and repopulation of leukemic cells in subsequent accruing days and late time points the SL-401 group mice died. To assess the effect of SL-401 on tumor burden, we designed another study using same AML donors, but sacrificed both the groups on week 7 where we found that SL-401 treated mice had reduced human AML cells in marrow. A first-in-human clinical trial with SL-401 in patients with blastic plasmacytoid dendritic cell neoplasm (BPDCN) noted that patients’ humoral immune responses against SL-401 limited the treatment cycles.20 However, in myeloid neoplasms where memory immune responses are poor, this is less likely to be a limitation. Furthermore, SL-401 is not a substrate of any known drug efflux pumps, making resistance due to this mechanism unlikely.44 Capillary leak syndrome is a known, severe side effect of SL-401 and needs to be managed for successful utilization of this agent in MDS and AML. Together, these data indicate that SL-401 is active against CD123+ AML/MDS and normal hematopoietic progenitors. These findings have translational relevance in regards to management of potential side effects such as marrow aplasia. This agent is currently being evaluated in a Phase 2 trial of AML patients in remission with minimal residual disease, a setting associated with the persistence of CD123+ LSCs (clinicaltrials.gov identifier 02270463) and a phase I trial of azacitidine and SL-401 in MDS (clinicaltrials.gov identifier 03113643). Acknwoledgments The authors are grateful to the AML and MDS patients who contributed to these studies, the OSU Comprehensive Cancer Center Leukemia Tissue Bank Shared Resource (P30 CA016058), the Clinical Flow Cytometry facility of the OSU Wexner Medical Center, R01 CA197844, R35 CA197734-01, D Warren Brown Foundation and the Lauber AML fund.

myeloid leukaemia to stem cells. Nature. 2017;547(7661):104-108. 4. Devine SM, Owzar K, Blum W, et al. Phase II Study of allogeneic transplantation for older Patients with acute myeloid leukemia in first complete remission using a reducedintensity conditioning regimen: results from Cancer and Leukemia Group B 100103 (Alliance for Clinical Trials in Oncology)/Blood and Marrow Transplant Clinical Trial Network 0502. J Clin Oncol.

2015;33(35):4167-4175. 5. Cogle CR. Incidence and burden of the myelodysplastic syndromes. Curr Hematol Malig Rep. 2015;10(3):272-281. 6. Ades L, Itzykson R, Fenaux P. Myelodysplastic syndromes. Lancet. 2014;383(9936):2239-2252. 7. Shastri A, Will B, Steidl U, Verma A. Stem and progenitor cell alterations in myelodysplastic syndromes. Blood. 2017; 129(12): 1586-1594.

haematologica | 2018; 103(8)

Activity of SL-401 in AML and MDS

8. Walter RB, Appelbaum FR, Estey EH, Bernstein ID. Acute myeloid leukemia stem cells and CD33-targeted immunotherapy. Blood. 2012;119(26):6198-6208. 9. Vasu S, He S, Cheney C, et al. Decitabine enhances anti-CD33 monoclonal antibody BI 836858-mediated natural killer ADCC against AML blasts. Blood. 2016;127(23): 2879-2889. 10. Sato N, Caux C, Kitamura T, et al. Expression and factor-dependent modulation of the interleukin-3 receptor subunits on human hematopoietic cells. Blood. 1993;82(3):752-761. 11. Jordan CT, Upchurch D, Szilvassy SJ, et al. The interleukin-3 receptor alpha chain is a unique marker for human acute myelogenous leukemia stem cells. Leukemia. 2000; 14(10):1777-1784. 12. Munoz L, Nomdedeu JF, Lopez O, et al. Interleukin-3 receptor alpha chain (CD123) is widely expressed in hematologic malignancies. Haematologica. 2001;86(12):1261-1269. 13. Busfield SJ, Biondo M, Wong M, et al. Targeting of acute myeloid leukemia in vitro and in vivo with an anti-CD123 mAb engineered for optimal ADCC. Leukemia. 2014;28(11):2213-2221. 14. Du X, Ho M, Pastan I. New immunotoxins targeting CD123, a stem cell antigen on acute myeloid leukemia cells. J Immunother. 2007;30(6):607-613. 15. Stein C, Kellner C, Kugler M, et al. Novel conjugates of single-chain Fv antibody fragments specific for stem cell antigen CD123 mediate potent death of acute myeloid leukaemia cells. Br J Haematol. 2010; 148(6):879-889. 16. Gill S, Tasian SK, Ruella M, et al. Preclinical targeting of human acute myeloid leukemia and myeloablation using chimeric antigen receptor-modified T cells. Blood. 2014; 123(15):2343-2354. 17. Mardiros A, Dos Santos C, McDonald T, et al. T cells expressing CD123-specific chimeric antigen receptors exhibit specific cytolytic effector functions and antitumor effects against human acute myeloid leukemia. Blood. 2013;122(18):3138-3148. 18. Al-Hussaini M, Rettig MP, Ritchey JK, et al. Targeting CD123 in acute myeloid leukemia using a T-cell-directed dual-affinity retargeting platform. Blood. 2016;127(1):122-131. 19. Frankel A, Liu JS, Rizzieri D, Hogge D. Phase I clinical study of diphtheria toxininterleukin 3 fusion protein in patients with acute myeloid leukemia and myelodysplasia. Leuk Lymphoma. 2008;49(3):543-553. 20. Frankel AE, Woo JH, Ahn C, et al. Activity of SL-401, a targeted therapy directed to interleukin-3 receptor, in blastic plasmacytoid dendritic cell neoplasm patients. Blood. 2014;124(3):385-392. 21. Macanas-Pirard P, Leisewitz A, Broekhuizen R, et al. Bone marrow stromal cells modulate mouse ENT1 activity and

haematologica | 2018; 103(8)

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

protect leukemia cells from cytarabine induced apoptosis. PLoS One. 2012;7(5): e37203. Konopleva M, Konoplev S, Hu W, Zaritskey AY, Afanasiev BV, Andreeff M. Stromal cells prevent apoptosis of AML cells by up-regulation of anti-apoptotic proteins. Leukemia. 2002;16(9):1713-1724. Binato R, de Almeida Oliveira NC, Du Rocher B, Abdelhay E. The molecular signature of AML mesenchymal stromal cells reveals candidate genes related to the leukemogenic process. Cancer Lett. 2015; 369(1):134-143. Huang JC, Basu SK, Zhao X, et al. Mesenchymal stromal cells derived from acute myeloid leukemia bone marrow exhibit aberrant cytogenetics and cytokine elaboration. Blood Cancer J. 2015;5:e302. Garrido SM, Appelbaum FR, Willman CL, Banker DE. Acute myeloid leukemia cells are protected from spontaneous and druginduced apoptosis by direct contact with a human bone marrow stromal cell line (HS5). Exp Hematol. 2001;29(4):448-457. Blau O, Baldus CD, Hofmann WK, et al. Mesenchymal stromal cells of myelodysplastic syndrome and acute myeloid leukemia patients have distinct genetic abnormalities compared with leukemic blasts. Blood. 2011;118(20):5583-5592. Mani R, Mao Y, Frissora FW, et al. Tumor antigen ROR1 targeted drug delivery mediated selective leukemic but not normal Bcell cytotoxicity in chronic lymphocytic leukemia. Leukemia. 2015;29(2):346-355. Ehninger A, Kramer M, Rollig C, et al. Distribution and levels of cell surface expression of CD33 and CD123 in acute myeloid leukemia. Blood Cancer J. 2014;4:e218. Alexander RL, Kucera GL, Klein B, Frankel AE. In vitro interleukin-3 binding to leukemia cells predicts cytotoxicity of a diphtheria toxin/IL-3 fusion protein. Bioconjug Chem. 2000;11(4):564-568. Feuring-Buske M, Frankel AE, Alexander RL, Gerhard B, Hogge DE. A diphtheria toxin-interleukin 3 fusion protein is cytotoxic to primitive acute myeloid leukemia progenitors but spares normal progenitors. Cancer Res. 2002;62(6):1730-1736. Black JH, McCubrey JA, Willingham MC, Ramage J, Hogge DE, Frankel AE. Diphtheria toxin-interleukin-3 fusion protein (DT(388)IL3) prolongs disease-free survival of leukemic immunocompromised mice. Leukemia. 2003;17(1):155-159. Testa U, Riccioni R, Biffoni M, et al. Diphtheria toxin fused to variant human interleukin-3 induces cytotoxicity of blasts from patients with acute myeloid leukemia according to the level of interleukin-3 receptor expression. Blood. 2005;106(7): 25272529. Hogge DE, Yalcintepe L, Wong SH, Gerhard B, Frankel AE. Variant diphtheria

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

toxin-interleukin-3 fusion proteins with increased receptor affinity have enhanced cytotoxicity against acute myeloid leukemia progenitors. Clinical Cancer Res. 2006;12(4):1284-1291. Yalcintepe L, Frankel AE, Hogge DE. Expression of interleukin-3 receptor subunits on defined subpopulations of acute myeloid leukemia blasts predicts the cytotoxicity of diphtheria toxin interleukin-3 fusion protein against malignant progenitors that engraft in immunodeficient mice. Blood. 2006;108(10):3530-3537. Hassanein NM, Alcancia F, Perkinson KR, Buckley PJ, Lagoo AS. Distinct expression patterns of CD123 and CD34 on normal bone marrow B-cell precursors ("hematogones") and B lymphoblastic leukemia blasts. Am J Clin Pathol. 2009;132(4):573-580. Hwang K, Park CJ, Jang S, et al. Flow cytometric quantification and immunophenotyping of leukemic stem cells in acute myeloid leukemia. Ann Hematol. 2012;91(10):1541-1546. Ho TC, LaMere M, Stevens BM, et al. Evolution of acute myelogenous leukemia stem cell properties after treatment and progression. Blood. 2016;128(13):1671-1678. Riccioni R, Pelosi E, Riti V, Castelli G, LoCoco F, Testa U. Immunophenotypic features of acute myeloid leukaemia patients exhibiting high FLT3 expression not associated with mutations. Br J Haematol. 2011; 153(1):33-42. Warlick ED, Cioc A, Defor T, Dolan M, Weisdorf D. Allogeneic stem cell transplantation for adults with myelodysplastic syndromes: importance of pretransplant disease burden. Biol Blood Marrow Transplant. 2009;15(1):30-38. Antonelli A, Noort WA, Jaques J, et al. Establishing human leukemia xenograft mouse models by implanting human bone marrow-like scaffold-based niches. Blood. 2016;128(25):2949-2959. Reinisch A, Thomas D, Corces MR, et al. A humanized bone marrow ossicle xenotransplantation model enables improved engraftment of healthy and leukemic human hematopoietic cells. Nat Med. 2016; 22(7):812-821. Abarrategi A, Foster K, Hamilton A, et al. Versatile humanized niche model enables study of normal and malignant human hematopoiesis. The Journal of clinical investigation. 2017;127(2):543-548. Wang K, Sanchez-Martin M, Wang X, et al. Patient-derived xenotransplants can recapitulate the genetic driver landscape of acute leukemias. Leukemia. 2017; 31(1):151-158. Frankel AE, Hall PD, McLain C, Safa AR, Tagge EP, Kreitman RJ. Cell-specific modulation of drug resistance in acute myeloid leukemic blasts by diphtheria fusion toxin, DT388-GMCSF. Bioconjugate chemistry. 1998;9(4):490-496.

1297

ARTICLE

Chronic Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1298-1307

Safety and efficacy of second-line bosutinib for chronic phase chronic myeloid leukemia over a five-year period: final results of a phase I/II study Carlo Gambacorti-Passerini,1 Jorge E. Cortes,2 Jeff H. Lipton,3 Hagop M. Kantarjian,2 Dong-Wook Kim,4 Philippe Schafhausen,5 Rocco Crescenzo,6 Nathalie Bardy-Bouxin,7 Mark Shapiro,8 Kay Noonan,9 Eric Leip,8 Liza DeAnnuntis,6 Tim H. Brümmendorf,5,10 and H. Jean Khoury11*

University of Milano-Bicocca, Monza, Italy; 2University of Texas MD Anderson Cancer Center, Houston, TX, USA; 3Princess Margaret Cancer Centre, Toronto, ON, Canada; 4Seoul St. Mary’s Hospital, South Korea; 5Department of Internal Medicine II, Hubertus Wald Tumor Center - University Cancer Center Hamburg, Germany; 6Pfizer Inc., Collegeville, PA, USA; 7Pfizer Global Research and Development, Paris, France; 8Pfizer Inc., Cambridge, MA, USA; 9Pfizer Inc., Groton, CT, USA; 10Universitätsklinikum RWTH Aachen, Germany and 11 Winship Cancer Institute of Emory University, Atlanta, GA, USA. 1

*Deceased

ABSTRACT

B Correspondence: carlo.gambacorti@unimib.it

Received: April 21, 2017. Accepted: May 10, 2018. Pre-published: May 17, 2018.

doi:10.3324/haematol.2017.171249 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1298 ©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.

1298

osutinib is a Src/Abl tyrosine kinase inhibitor indicated for adults with newly-diagnosed Philadelphia positive chronic myeloid leukemia or with resistant/intolerant disease. We report the final results of a phase I/II study of second-line bosutinib in chronic phase chronic myeloid leukemia patients after imatinib failure (n=284). Median follow up and treatment durations were 54.8 (range 0.6-96.3) and 25.6 (0.2-96.3) months, respectively. At years 2 and 5, 54% and 40% of patients, respectively, remained on bosutinib. Cumulative major cytogenetic response and complete cytogenetic response rates (newly-attained or maintained from baseline) were 58% and 46%, respectively, by year 2 and 60% and 50% by year 5. Kaplan-Meier probability of maintaining major and complete cytogenetic response was 76% and 78%, respectively, at year 2 and 71% and 69% at year 5. Cumulative incidence of ontreatment disease progression/death was similar at years 5 (19%) and 2 (15%); Kaplan-Meier overall survival was 91% at year 2 and 84% at year 5. Of 169 patients who had discontinued bosutinib by year 5, 38 did so after year 2, most commonly for disease progression (n=11). Most adverse events initially occurred within two years. Overall, gastrointestinal events were the most common (diarrhea 86%, nausea 46%, vomiting 37%); the most common grade 3/4 toxicity was thrombocytopenia (25%). None of the 4 on-treatment deaths in years 3-5 were related to bosutinib. Bosutinib demonstrated durable efficacy and manageable toxicity through year 5 confirming its importance in the treatment of chronic phase chronic myeloid leukemia patients resistant/intolerant to prior imatinib. This trial was registered at clinicaltrials.gov identifier: 00261846.

Introduction With the success of BCR-ABL1 tyrosine kinase inhibitors (TKIs), patients with Philadelphia chromosome positive (Ph+) chronic myeloid leukemia (CML) can potentially enjoy a normal life expectancy.1,2 Therefore, information regarding longterm efficacy and tolerability of TKIs is important for informing treatment selection. haematologica | 2018; 103(8)

Long-term second-line bosutinib for CP CML

Several TKIs are approved for treatment of newly diagnosed chronic phase (CP) CML, each associated with a distinct safety profile. First-line TKI therapies include imatinib3 and the second-generation TKIs, nilotinib,4 dasatinib,5 and, most recently, bosutinib.6 Although response rates are high with TKI therapy, patients develop treatment resistance or experience intolerable toxicities.7,8 Determining the most appropriate therapy following treatment failure is critical to achieve optimal outcomes and prevent disease progression. Second- and third-line treatment decisions are based on the prior therapy, the reason for failure (primary or secondary resistance, intolerance), BCR-ABL1 mutation status, comorbidities, and prior toxicities.8,9 Bosutinib was initially approved in 2012 for the treatment of patients with Ph+ CP, accelerated-phase (AP), or blast-phase (BP) CML resistant or intolerant to prior TKI therapy.6 This approval was based primarily on the results of a pivotal phase I/II trial of bosutinib in CP CML patients following failure of imatinib.10 Results of a preliminary analysis approximately 15 months after the last enrolled patient demonstrated potent activity with second-line bosutinib across a spectrum of BCR-ABL1 mutations and a toxicity profile distinct from those of other TKIs.10 The durability of response was confirmed with subsequent analyses 24 and 48 months after the last enrolled patient.11,12 The final results presented here from the phase I/II trial of bosutinib for imatinib-resistant or imatinibintolerant CP CML are assessed after at least five years from the time the last patient was enrolled.

molecular response due to logistical constraints. For the purpose of this study, responders for major molecular response (MMR) had a ≥3-log reduction from standardized baseline, a detectable BCR-ABL1 transcript at baseline or postbaseline, and must have maintained or attained a CCyR. Duration of response (date of first response until confirmed response loss or PD/death) was evaluated through 30 days after last dose using the KaplanMeier method; patients without events were censored at their last assessment visit. See the Online Supplementary Methods for further details of the statistical methods used for this report. Disease progression was assessed as described previously.10 Time to PD or death, and time to on-treatment transformation to AP/BP CML, were evaluated through 30 days after the last bosutinib dose using cumulative incidence adjusting for the competing risk of treatment discontinuation without an event; patients without events were censored at their last assessment visit. Progression-free survival (PFS; time to PD or death within 30 days of last bosutinib dose) was analyzed for prognostic factors. Overall survival (OS) was evaluated using the Kaplan-Meier method; patients still alive were censored at the last known date on which they were alive. Per protocol, patients were followed for OS for up to two years after discontinuation of study treatment; analysis of follow up and OS includes data from patients enrolled in an ongoing extension study (clinicaltrials.gov identifier: 01903733). Safety was assessed in patients who received at least one bosutinib dose; AEs were reported at each visit up to 30 days after last dose and graded according to National Cancer Institute Common Terminology Criteria for Adverse Events v.3.0. Treatment-emergent AEs (TEAEs) were assessed overall and by year of first occurrence.

Methods Results Study design and patients This phase I/II, open-label, multicenter study was initiated in January 2006; the design has been described previously.10,12 Part 1 (dose-escalation phase) determined the recommended Part 2 starting dose of bosutinib 500 mg/day; in Part 2, the efficacy, safety, and tolerability of bosutinib were evaluated. Patients without a complete hematologic response (CHR) by week 8 or complete cytogenetic response (CCyR) by week 12 were allowed to receive bosutinib 600 mg/day unless treatment-related grade ≥3 adverse events (AEs) occurred. Patients continued bosutinib treatment until disease progression (PD), death, unacceptable toxicity, or withdrawal of consent. The protocol was approved by central or institutional review boards for each site and was conducted in accordance with Good Clinical Practice principles and the Declaration of Helsinki. Eligible patients were aged 18 years or over with a confirmed diagnosis of Ph+ CP CML resistant to full-dose imatinib (IM-R; ≥600 mg/day) or intolerant to any dose of imatinib (IM-I). Additional eligibility criteria are provided in the Online Supplementary Methods.

Safety and efficacy analyses Cytogenetic response was assessed as previously described10 and defined as newly-achieved on study or maintained from baseline for four weeks or more (earliest time point for assessment). Evaluable patients received at least one dose and had a valid baseline assessment for the respective end point. Molecular response data were assessed at a central laboratory; however, the International Scale (IS) was not used. Patients from sites in China, India, Russia, and South Africa were not assessed for haematologica | 2018; 103(8)

Patients and treatment A total of 284 patients with CP CML (IM-R, n=195; IMI, n=89) were enrolled and treated with second-line bosutinib (Online Supplementary Table S1). The study was closed as of August, 2015; patients still on study were offered enrollment on an extension study. As for the final database lock for this study (2nd October 2015), the time from the last second-line patient’s first dose was 60 months or more. The median (range) duration of OS follow up was 54.8 (0.6-96.3) months, and the median treatment duration was 25.6 (0.2-96.3) months (Table 1). At year 2, 153 (54%) patients were receiving bosutinib and at year 5, 115 (40%; IM-R, n=81 and IM-I, n=34) patients still remained on bosutinib treatment (1 year=48 weeks). Discontinuation from treatment was most common within the first two years, with 131 (46%) patients discontinuing by the end of year 2, and 38 (13%) patients discontinuing treatment in years 3 through 5 (Online Supplementary Table S2). The most common primary reasons for discontinuation from treatment through year 5 were lack of efficacy [categorized by the investigator separately as PD and unsatisfactory response; n=47 (17%) and n=21 (7%), respectively], AE [n=64 (23%)], and patient request [n=19 (7%)]. The most common reasons for discontinuation from treatment in years 3 through 5 were PD and unsatisfactory response in year 3, AE and PD in year 4, and unsatisfactory response and death in year 5. Overall, younger patients (aged <65 years vs. ≥65 years) were less likely to permanently discontinue treatment because of AE (21% vs. 32%), patient request (7% vs. 14%), or death (1% vs. 1299

C. Gambacorti-Passerini et al. Table 1. Treatment summary.*

Parameter Median (range) duration of follow up,† mo Median (range) duration of treatment,† mo Patients with ≥1 dose interruption due to AEs, n (%) Median (range) duration of events of dose interruptions, d Median (range) cumulative duration of dose interruptions, d Patients with ≥1 dose reduction due to AEs, n (%) Median (range) time to first dose reduction due to AEs, d Any dose reduction Reduction to 400 mg/d Reduction to 300 mg/d Median (range) cumulative duration of dose reduction due to AEs, d Any dose reduction Reduction to 400 mg/d Reduction to 300 mg/d Discontinued treatment, n (%) Enrolled in extension study AE PD Patient request Unsatisfactory response (efficacy)‡ Death Investigator request Lost to follow up Symptomatic deterioration Discontinuation of study by sponsor Other

IM-R (n=195)

IM-I (n=89)

Total (n=284)

46.8 (0.6–96.3) 27.6 (0.2–96.3) 133 (68) 8 (1–981) 24 (1–983) 89 (46)

58.8 (0.6–93.2) 24.2 (0.3–84.3) 76 (85) 14 (1–280) 24 (1–429) 52 (58)

54.8 (0.6–96.3) 25.6 (0.2–96.3) 209 (74) 10 (1–981) 24 (1–983) 141 (50)

48 (8–2166) 49 (8–1800) 194 (58–2166)

52 (7–1875) 55 (11–1875) 107 (29–1296)

49 (7–2166) 52 (8–1875) 162 (29–2166)

349 (3–2881) 238 (3–2667) 114 (3–1641) 195 (100) 61 (31) 32 (16) 43 (22) 12 (6) 19 (10) 5 (3) 7 (4) 4 (2) 2 (1) 1 (1) 9 (5)

285 (2–2368) 84 (1–2368) 168 (5–1473) 89 (100) 22 (25) 35 (39) 8 (9) 12 (13) 4 (4) 3 (3) 1 (1) 0 0 0 4 (4)

346 (2–2881) 198 (1–2667) 116 (3–1641) 284 (100) 83 (29) 67 (24) 51 (18) 24 (8) 23 (8) 8 (3) 8 (3) 4 (1) 2 (1) 1 (<1) 13 (5)

AE: adverse event; d: days; IM-I: imatinib-intolerant; IM-R: imatinib-resistant; n: number; mo: months; PD: progressive disease. *In both Parts 1 and 2, patients received treatment until PD, death, unacceptable toxicity, or withdrawal of consent. †One month is defined as 30.4 days. ‡Defined as failure to achieve an optimal response as determined by the investigator.

8%), and more likely to enroll in the extension study (32% vs. 19%) (Online Supplementary Table S3). Ninety-nine (35%) patients completed the 2-year follow up after treatment discontinuation (IM-R, n=60; IM-I, n=39). Thirtytwo (11%) patients discontinued treatment after year 5, and 83 (29%) patients continued treatment in the extension study.

Efficacy Most cytogenetic responses to bosutinib occurred within two years of initiating treatment (Table 2). By week 12, the cumulative major cytogenetic response (MCyR) rate was 35% (n=93 of 262 evaluable patients), including 22% (n=57) who attained/maintained a CCyR. Of 246 evaluable patients without a CCyR at baseline, 76 (31%) attained an MCyR and 45 (18%) attained a CCyR. The cumulative MCyR rate observed by year 2 was 58%, including 46% with a CCyR. Patients continued to attain a CCyR after two years, with 10 patients having initial ontreatment CCyR during years 3-5. By year 5, the cumulative MCyR CCyR rates were 60% (n=156 of 262 evaluable patients) and 50% (n=130), respectively; 57% (n=141 of 246 evaluable patients) newly-attained an MCyR and 47% (n=116) newly-attained a CCyR. The cumulative MMR rate was 42% (n=82 of 197 evaluable patients). Cytogenetic responses by both two and five years were similar in the IM-R and IM-I subgroups, whereas MMR rates were higher among IM-R patients at both time points (Table 2). Younger patients were more likely to 1300

have at least an MCyR (61% vs. 54%); however, rates of CCyR were similar among patients aged under 65 years and 65 years or over (Online Supplementary Table S3). Among responders, the Kaplan-Meier estimated probability of maintaining a response was similar at years 5 and 2: 71% and 76%, respectively, for an MCyR; 69% and 78% for a CCyR; and 68% and 70% for MMR (Table 2 and Figure 1). Overall, 41 of 156 (26%) responders lost MCyR, 37 of 130 (28%) lost CCyR, and 25 of 82 (30%) lost MMR. Few patients lost response after year 2 (7 lost MCyR, 10 lost CCyR, and 2 lost MMR). At the last assessment prior to discontinuation, 81% (n=127) of all 156 responders had an MCyR and 68% (n=106) had a CCyR; of 141 responders without a baseline CCyR, 111 attained an MCyR and 93 attained a CCyR. Among 132 patients (IM-R, n=81; IM-I, n=51) who required a dose reduction to 400 mg/day due to an AE, 81 (61%; IM-R, 63%; IM-I, 59%) had an MCyR, including 67 of 110 (61%) who did not have a CCyR at baseline (Online Supplementary Table S4). Fifty-seven (43%) patients first achieved an MCyR after dose reduction, 19 (14%) achieved an MCyR before and maintained it after dose reduction, and 5 (4%) lost their MCyR after dose reduction. Among 50 patients (IM-R, n=32; IM-I, n=18) who had a dose reduction to 300 mg/day due to an AE, 29 (58%; IM-R, 69%; IM-I, 39%) had an MCyR, including 25 of 43 (58%) without a CCyR at baseline. Twenty (40%) patients achieved an MCyR before and maintained it after dose reduction, 8 (16%) first achieved MCyR after a dose haematologica | 2018; 103(8)

Long-term second-line bosutinib for CP CML

Table 2. Efficacy outcomes at two and five years.

Parameter

IM-R (n=195) Year 2

IM-I (n=89) Year 5

Year 2

Year 5

107/182 (59) [51.3‒66.0] 88/182 (48) [40.9‒55.9] 57/127 (45) [36.1‒54.0] 67.2 (56.8−75.6) 69.6 (57.9−78.6) 62.9 (48.5−74.2) 23.1 (17.9–29.8) 6.2 (3.6–10.6) 80.8 (73.8–86.0)

49/80 (61) [49.7‒71.9] 41/80 (51) [39.8‒62.6] 20/70 (29) [18.4‒40.6] 86.3 (72.0−93.6) 78.9 (62.2−88.9) 87.2 (65.3−95.7) 6.7 (3.1–14.6) 2.2 (0.6–8.8) 97.7 (91.0–99.4)

49/80 (61) [49.7‒71.9] 42/80 (53) [41.0‒63.8] 25/70 (36) [24.6‒48.1] 79.8 (63.1−89.5) 68.2 (49.5−81.3) 81.4 (57.1−92.7) 10.1 (5.4–18.8) 2.2 (0.6–8.8) 89.6 (79.1–95.0)

Total (n=284) Year 2 Year 5

*†

Response, n. of evaluable patients (%) [95% CIs] Cumulative MCyR

102/182 (56) [48.5‒63.4] Cumulative CCyR 79/182 (43) [36.1‒50.9] Cumulative MMR 42/127 (33) [25.0‒42.0] Probability of maintaining 72.0 MCyR, % (95% CI)‡ (62.1−79.8) Probability of maintaining 77.3 CCyR, % (95% CI)‡ (66.8−84.9) Probability of maintaining 62.9 MMR, % (95% CI)‡ (48.5−74.2) Cumulative incidence of progression or death,§ 19.0 % (95% CI) (14.2–25.4) Cumulative incidence of transformation to AP/BP CML,ǁ 5.6 % (95% CI) (3.2–10.0) OS,‡,# % (95% CI) 88.2 (82.7–92.1)

151/262 (58) [51.4‒63.7] 120/262 (46) [39.7‒52.0] 62/197 (31) [25.1‒38.5] 76.4 (68.5−82.5) 77.8 (69.2−84.2) 70.0 (58.4−79.0) 15.1 (11.5–19.9) 4.6 (2.7–7.8) 91.2 (87.1–94.0)

156/262 (60) [53.3‒65.5] 130/262 (50) [43.4‒55.8] 82/197 (42) [34.7‒48.8] 71.1 (62.6−78.0) 69.3 (59.7−77.0) 68.3 (56.4−77.5) 19.0 (15.0–24.2) 4.9 (3.0–8.2) 83.5 (78.1–87.7)

n: number; MCyR: major cytogenetic response (complete + partial); CCyR: complete cytogenetic response; MMR: major molecular response; CI: Confidence Interval; AP: accelerated phase; BP: blast phase; CML: chronic myeloid leukemia; IM-I: imatinib intolerant; IM-R: imatinib-resistant; OS: overall survival. *To be considered a responder for MCyR or CCyR, the patient must have improved from their baseline assessment or maintained their baseline response. To be considered a responder for MMR, a patient must have had a ≥3-log reduction from standardized baseline, a detectable BCR-ABL1 transcript at baseline or postbaseline, and must have maintained or attained a CCyR. Patients from sites in China, India, Russia and South Africa were not assessed for molecular response. Three patients achieved their initial MMR after year 5. †Evaluable patients had received ≥1 bosutinib dose and had a valid baseline cytogenetic assessment. ‡Based on Kaplan-Meier estimates. Results for MCyR and CCyR are based on both maintained and achieved responses. §Based on cumulative incidence adjusting for competing risk of treatment discontinuation without progressive disease or death; progressive disease was defined as transformation to accelerated or blast phase CML, increasing white blood cell count (doubling over ≥1 month with second count >20×109/L and confirmed ≥1 week later), or loss of confirmed complete hematologic response or unconfirmed MCyR. ǁBased on cumulative incidence adjusting for competing risk of treatment discontinuation without transformation. #Per protocol, patients were followed for OS for two years after treatment discontinuation. Analysis includes data from long-term extension study.

reduction to 300 mg/day, and one (2%) patient lost MCyR. Among patients who achieved an MCyR after a dose reduction, median duration of response (non-KaplanMeier) was longer for patients receiving 400 mg/day versus 300 mg/day (167 days vs. 17 days); median MCyR durations (non-Kaplan-Meier) were similar in patients who had a response before and after dose reduction (283 days vs. 260 days) (Online Supplementary Table S4). Of the 224 (79%) patients with a known baseline mutation status, 79 (35%) had at least one mutation in the BCR-ABL1 kinase domain, most of whom were in the IMR cohort [IM-R, n=73 of 156 (47%); IM-I, n=6 of 68 (9%)] (Online Supplementary Table S5). Thirteen patients had multiple mutations, all of whom were IM-R. A total of 43 unique BCR-ABL1 mutations were evident, most commonly F359V (n=9), M351T (n=8), M244V (n=6), G250E (n=6), and T315I (n=9). Among evaluable patients with a mutation other than T315I, most (44 of 69, 64%) attained/maintained an MCyR; response rates were similar among patients without a mutation (75 of 130, 58%) and appeared lower among those with multiple mutations (6 of 12; 50%). Among evaluable patients with mutations that are sensitive (n=30), moderately resistant (n=12), and highly resistant (n=12) to bosutinib (see Figure 2 legend for definitions), the cumulative MCyR rates were 67%, 58%, and 33%, respectively, with corresponding CCyR rates of 57%, 42%, and 33%. Among evaluable patients with mutations of unknown sensitivity (n=24) and patients for whom mutation status was unknown (n=54), the MCyR haematologica | 2018; 103(8)

rates were 63% and 65%, respectively, with corresponding CCyR rates of 50% and 57%. Of 104 (37%) patients evaluated for BCR-ABL1 kinase domain mutations before and during bosutinib therapy, 26 had at least one newly-emerging mutation; this was most commonly T315I (n=9), V299L (n=5), and M244V (n=2). Fourteen of the 26 patients with newly-emerging mutations also had at least one BCR-ABL1 mutation present at baseline. Four patients were enrolled in the extension study; the remaining 22 discontinued because of PD (n=12), unsatisfactory response (n=6), AE (grade 2 liver toxicity; n=1), death (sepsis unrelated to bosutinib; n=1), and other (n=2). Four patients achieved a best response of partial cytogenetic response (PCyR) and 8 achieved a CCyR; 12 patients achieved only a CHR as a best response, and 2 patients had no response.

Transformations and survival outcomes On-treatment transformation to AP/BP CML occurred in 15 (IM-R, n=13; IM-I, n=2) patients overall, with 9 patients transforming to AP (IM-R, n=7; IM-I, n=2), and 6 patients transforming to BP (all IM-R). The cumulative incidence at year 5 was 5% overall (IM-R, 6%; IM-I, 2%); 95% of patients discontinued treatment without transformation. Among the baseline and on-treatment factors examined, only higher baseline peripheral blood blasts was predictive of on-treatment transformation (P=0.0011). Despite being initially classified as having progressed to AP, 2 patients did well on bosutinib therapy for two years 1301

C. Gambacorti-Passerini et al.

or more after transformation: one patient who transformed on day 14 continued bosutinib treatment for another six years and subsequently continued treatment in the extension study; another patient who transformed on day 246 discontinued treatment two years later for PD. Among 153 patients remaining on treatment after year 2, only 2 (both IM-R) had on-treatment transformation to AP

after this time (on days 734 and 2165). Eleven of the 15 patients with on-treatment transformation had responses to bosutinib, including 4 with best responses of MCyR, 3 with CCyR, and 4 with CHR. The cumulative incidence of on-treatment PD/death was higher by 4% at year 5 [19% (IM-R, 23%; IM-I, 10%)] versus year 2 [15% (IM-R, 19%; IM-I, 7%)]; 42% of patients discontinued treatment

Figure 1. Duration of response. Duration of major cytogenetic response (MCyR) (A) and complete cytogenetic response (CCyR) (B) among responders. Open circles indicate censored observations. IM-I: imatinibintolerant; IM-R: imatinib-resistant; n: number; d: day.

1302

haematologica | 2018; 103(8)

Long-term second-line bosutinib for CP CML

without on-treatment PD/death before year 5. Long-term outcomes are reported according to age in Online Supplementary Table S3. Kaplan-Meier probability of OS at year 5 was 84% (IMR, 81%; IM-I, 90%) versus 91% (IM-R, 88%; IM-I, 98%) at year 2; 40% of patients were censored prior to year 5, and

31% enrolled in the extension study for continued treatment or follow up for longer-term survival (Figure 3). A total of 45 (16%) deaths occurred on study, 24 through year 2, 5 after year 5, and 10 within 30 days of the last bosutinib dose. Patients aged under 65 years had a higher OS rate compared with those aged 65 years or over (85%

Figure 2. Predictors of response loss, disease progression, and death. Closed circles represent major cytogenetic response (MCyR) duration and open circles represent complete cytogenetic response. Based on multivariate Cox regression models. Parameters failing to meet elimination criteria (0.20) not shown. Hazard ratios >1 indicate worse outcome. P-values were not adjusted for multiple comparisons; significant P-values are in bold. Definitions of covariates can be found in the Online Supplementary Methods. On-treatment characteristics are Cox time-dependent covariates. *Baseline factor for durable response model. BOS: bosutinib; IM: imatinib; LFT: liver function test; Ph+: Philadelphia chromosome positive; y: years; CI: confidence interval.

Figure 3. Kaplan-Meier estimated overall survival. Open circles indicate censored observations. Overall survival was calculated as the first date of study dosing until the date of death; patients without events were censored at the last contact. Per protocol, patients were followed for overall survival for two years after treatment discontinuation. Analysis includes data from a long-term extension study. IM-I: imatinibintolerant; IM-R: imatinib-resistant; n: number.

haematologica | 2018; 103(8)

1303

C. Gambacorti-Passerini et al.

vs. 77%) (Online Supplementary Table S3). Causes of death were PD [n=26 (58%); IM-R: n=23; IM-I: n=3], AE unrelated to bosutinib [n=16 (36%); IM-R: n=14; IM-I: n=2], and unknown cause [n=3 (7%); all IM-I]. None of the 45 deaths were assessed as treatment-related. Four deaths occurred within 30 days of the last bosutinib dose through year 2 (all IM-R) and 4 occurred during years 3-5 (2 IM-R and 2 IM-I patients).

factors examined, experiencing an abnormal liver function test (LFT) was predictive of increased OS. Notably, prior response or resistance to imatinib did not predict duration of cytogenetic response or long-term survival outcomes. Factors predictive of decreased PFS were Ph+ ratio ≥95% versus ≤35%, lack of MCyR by week 12, and higher baseline peripheral blood blasts (Figure 2). The on-treatment factor of receiving a bosutinib dose reduction to 400 mg due to AEs was predictive of increased PFS.

Predictors of response duration, PFS, and OS Significant (P<0.05) baseline factors predictive of MCyR or CCyR loss were Ph+ ratio ≥95% versus ≤35% and late versus early disease stage (Figure 2). No on-treatment factors were predictive of MCyR duration; however, experiencing treatment-emergent thrombocytopenia was predictive of loss of CCyR (P=0.0471). Several baseline factors predictive of decreased OS were identified: age ≥65 years versus <65 years, Ph+ ratio ≥95% versus ≤35%, lack of an MCyR by week 12, higher baseline peripheral blood blasts, and having a BCR-ABL1 mutation at baseline that is either sensitive or highly resistant to bosutinib. Among on-treatment

Safety and tolerability The most common any grade hematologic treatmentemergent AEs (TEAEs) were thrombocytopenia [42% (Grade 3/4, 25%)], anemia [29% (Grade 3/4, 13%)], and neutropenia [16% (Grade 3/4, 10%)] (Online Supplementary Table S6). The most common non-hematologic TEAEs were diarrhea [86% (Grade 3/4, 10%)], nausea [46% (Grade 3/4, 2%)], vomiting [37% (Grade 3/4, 4%)], and rash [36% (Grade 3/4, 9%)]. Most newly-occurring AEs (MedDRA preferred terms not reported for the same patient previously for those on treatment during a given

Figure 4. Incidence of newly-occurring adverse events (AEs) over time. Denominators are the number of patients on treatment during the indicated years (NB: incidences of certain AEs appear higher in later years compared with previous years due to a lower number of patients on treatment). Newly-occurring AEs were those not experienced by the same patient previously among patients on treatment during a given year (1 year = 365.25 days). *Includes the high-level group terms (HLGTs) cardiac arrhythmias, pericardial disorders, and heart failures under the cardiac disorders system organ class (SOC); relevant preferred terms (PTs) (cardiac death, sudden cardiac death, sudden death) under the general disorders and administration site SOC conditions; relevant PTs (decreased ejection fraction, abnormal electrocardiogram QT interval, prolonged electrocardiogram QT, long QT syndrome, congenital long QT syndrome, Torsade de pointes, ventricular tachycardia) under the SOC investigations. † HLGTs included: coronary artery disorders, atherosclerosis, stenosis, vascular insufficiency and necrosis, embolism and thrombosis; high-level terms (HLTs) included arterial therapeutic procedures (excluding aortic), central nervous system hemorrhages and cerebrovascular accidents, central nervous system vascular disorders not elsewhere classified (NEC), non-site specific vascular disorders NEC, peripheral vascular disorders NEC (excluding the PTs flushing and hot flash), transient cerebrovascular events, vascular imaging procedures NEC, and vascular therapeutic procedures NEC and all subordinate terms. ‡HLGTs included: vascular hypertensive disorders and cardiac and vascular investigations (excluding enzyme tests), the HLT vascular tests NEC (including blood pressure); PTs included: abnormal blood pressure, abnormal ambulatory blood pressure, increased ambulatory blood pressure, abnormal diastolic blood pressure, increased diastolic blood pressure, increased blood pressure, abnormal systolic blood pressure, and increased systolic blood pressure. §HLT included: renal failure and impairment; PTs included: blood creatinine abnormal, blood creatinine increased, creatinine renal clearance abnormal, creatinine renal clearance decreased, glomerular filtration rate abnormal, glomerular filtration rate decreased. ALT: alanine aminotransferase; AST: aspartate aminotransferase; URTI: upper respiratory tract infection; UTI: urinary tract infection; n: number.

1304

haematologica | 2018; 103(8)

Long-term second-line bosutinib for CP CML

year) were experienced by patients during year 1 (99.6%) of treatment, with rates somewhat lower in years 2 (74%), 3 (68%), 4 (52%), and 5 (57%) (Figure 4). Common AEs (in >5 patients) newly-occurring in year 3 were cough [5% (n=8)], increased blood creatinine [5% (n=7)], and pyrexia [4% (n=6)]; most events were grade 1/2. Common newly-occurring AEs in year 4 were increased blood creatinine [5% (n=6)] and pleural effusion [5% (n=7)]; 2 events (both grade 4 pleural effusion considered probably not related to bosutinib) resulted in hospitalization. No newly-occurring AEs were reported in more than 5 patients in year 5. Adverse events led to treatment discontinuation in 69 (24%) patients throughout the study, including one who also discontinued due to PD and another who discontinued due to subject request as the primary reason. AEs resulting in treatment discontinuation in 2% or more of patients overall were thrombocytopenia [6% (n=17)], neutropenia [2% (n=6)], and alanine aminotransferase increased [2% (n=6)]. Of these 69 patients, 28 (41%) discontinued treatment without attempting a dose reduction to less than 500 mg/day. The majority (86%) of discontinuations due to AEs occurred during the first two years of treatment (Online Supplementary Table S2). AEs led to treatment discontinuation in 7 patients in years 3-5: coronary artery disease, scleroderma, and renal failure in year 3; ascites and serositis (same patient), increased blood creatinine, and pulmonary hypertension in year 4; and thrombocytopenia in year 5. Although diarrhea was the most common AE [IM-R, 86% (n=167); IM-I, 85% (n=76)], in most instances this was grade 1 or 2 [IM-R, 76% (n=149); IM-I, 75% (n=67)] (Online Supplementary Table S6). Only 3 (2%) IM-R and one (1%) IM-I patient discontinued bosutinib treatment primarily because of diarrhea, all within the first two years. Diarrhea (any grade) occurred most frequently within year 1 (84%) with a median (range) time to first occurrence of 2 (1-1330) days; only 4 patients experienced diarrhea for the first time during years 2-5 (Figure 4). Cardiac AEs occurred in 37 (13%) patients [IM-R, 13% (n=25); IM-I, 14% (n=12)], 12 (32%) of whom had a medical history of these events; maximum grade 3, 4, and 5 cardiac AEs occurred in 11 (4%), 5 (2%), and 2 (1%) patients, respectively (Online Supplementary Table S6). The most common cardiac AEs (occurring in â&#x2030;Ľ5 patients) were pericardial effusion [3% (grade 3/4, 1%)], congestive cardiac failure [2% (grade 3/4, 2%)], atrial fibrillation [2% (grade 3/4, 1%)], bradycardia [2% (grade 3/4, 1%)], and cardiac failure [2% (grade 3/4, 1%)]. Twelve (4%) patients [IM-R, 3% (n=6); IM-I, 7% (n=6)] experienced cardiac AEs considered by the investigator to be treatment-related, only 3 of whom experienced grade 3/4 events. The median (range) time to first cardiac event was 184 (1-2563) days with the incidence of newly-occurring cardiac AEs decreasing after year 1 (Figure 4). Cardiac AEs led to treatment discontinuation in one patient (cardiac failure in year 2) and death in 2 patients (both congestive heart failure unrelated to bosutinib, occurring in years 3 and 7). Twenty-two (8%) patients [IM-R, 7% (n=13); IM-I, 10% (n=9)] had vascular AEs including 8 (36%) who had a medical history of vascular events; 11 (4%) patients had grade 3/4 vascular AEs (Online Supplementary Table S6). In 4 (1%) patients [IM-R, 2% (n=3); IM-I, 1% (n=1)], vascular AEs were considered by the investigator to be treatmentrelated, only one of whom experienced a grade 3/4 vascuhaematologica | 2018; 103(8)

lar event. Most vascular AEs initially occurred within two years with a median (range) time to onset of 548 (47-2452) days. Only one patient discontinued treatment due to vascular AEs (coronary artery disease in year 3). No patients died because of vascular AEs. Twenty-six (9%) patients [IM-R, 10% (n=19); IM-I, 8% (n=7)] experienced hypertension-related AEs, 10 (38%) of whom had a history of vascular events. Events were of low severity in the majority of these patients (maximum grade 1/2, n=18; grade 3, n=8; no grade â&#x2030;Ľ4) and 5 (2%) experienced events considered by the investigator to be related to treatment; however, no patients discontinued due to hypertension-related AEs. As with vascular AEs, the incidence of newly-occurring hypertension-related AEs did not increase over time (Figure 4). Renal AEs occurred in 37 (13%) patients [IM-R, 14% (n=27); IM-I, 11% (n=10)], 7 (19%) of whom had a medical history of renal events (Online Supplementary Table S6). Six (2%) patients had maximum grade 3/4 events and 14 (all grade; grade 3/4, n=1) had events considered by the investigator to be treatment related. The median (range) time to first renal AE was 673 (8-2695) days. Renal AEs led to treatment discontinuation in 3 patients (1 each in years 1, 2, and 3) and death in one patient (acute kidney injury in year 1 related to PD and unrelated to bosutinib).

Cross-intolerance Eighty-nine patients were intolerant to prior imatinib (Online Supplementary Table S7). Of 85 patients with a specific AE reported as the reason for discontinuation of imatinib, 52 (61%) experienced the same AE with bosutinib that led to imatinib discontinuation, most commonly hematologic AEs (thrombocytopenia, n=12; neutropenia, n=5; anemia, n=5) or gastrointestinal AEs (diarrhea, n=6; nausea, n=4); 14 (16%) had cross-intolerance (defined as having discontinued bosutinib due to the same AE that led to prior imatinib discontinuation). Twenty-five (29%) patients experienced the same grade 3/4 AE while on bosutinib. No patient receiving bosutinib died due to the same AE that led to intolerance to prior imatinib.

Discussion After five years of follow up, the final results of this phase I/II study demonstrated durable efficacy and acceptable long-term safety for second-line bosutinib in patients with CP CML resistant or intolerant to imatinib. The estimated probabilities of responders maintaining an MCyR or CCyR at year 5 (71%, 69%) decreased modestly from the estimated probabilities at year 2 (76%, 78%). Resistance and intolerance to prior imatinib did not appear to result in differences in response durability, as rates observed at years 2 and 5 were similar for both IM-R and IM-I patients. Additionally, late disease progression was uncommon, supporting the observed response durability [although 38 (13%) patients discontinued after year 3, potentially biasing the interpretation of subsequent outcomes]. Cumulative response rates at years 5 and 2 were similar (year 5: MCyR, 60% and CCyR, 50%; year 2: MCyR, 58% and CCyR, 46%). However, it should be noted that results reported here are based on a finalized database resulting in slight differences from previously published data.10 The response rates achieved in this study are compara1305

C. Gambacorti-Passerini et al.

ble to those observed in studies of second-line nilotinib and dasatinib. With similar follow-up durations, CCyR rates of 37% and 49% were reported with nilotinib and dasatinib, respectively, compared with 47% with bosutinib in the present study.4,8,13 Estimated rates of on-treatment PD/death (19%) and transformation to AP/BP CML (5%) remain low with bosutinib; only 2 IM-R patients had on-treatment transformation to AP after year 2, although there is a potential bias from patients lost to follow up. Similar rates of transformation were observed with second-line dasatinib (5%).5,14 The estimated OS rate at 5 years is high, with a modest decrease from the 2-year OS rate (84% vs. 91%). This 5-year rate is also comparable to those reported for dasatinib (91%), nilotinib (87%), and ponatinib (81%) in CP CML patients after prior TKI failure.8 Responses were observed in all but 2 (T315I and M244V) of the 26 patients with newly-emerging BCRABL1 mutations. All but one of 14 patients with newlyemerging mutations that are highly resistant to bosutinib15 had a best response of at least CHR; 5 (36%) had a best response of at least PCyR. Effects of dose reductions on response were limited as most patients who dose reduced dose attained/maintained an MCyR. Only 4% and 2% of patients who reduced dose to 400 mg/day and 300 mg/day, respectively, lost their previously achieved MCyR. Gastrointestinal toxicities remained the most commonly reported AEs overall at the 5-year follow up (diarrhea, 86%; nausea, 46%; vomiting, 37%). Initial events occurred early, with incidences through year 2 of 84% for diarrhea, 45% for nausea, and 37% for vomiting.11 Although diarrhea was common, grade 3 events occurred in only 10% of patients (no grade 4), and only 4 patients discontinued because of this AE, all within two years of initiating bosutinib. Grade 3/4 hematologic AEs, such as thrombocytopenia (25%) and neutropenia (10%), occurred at rates similar to or lower than those observed with second-line dasatinib (24% and 36%), nilotinib (30% and 31%), and ponatinib (35% and 23%).4,5,16 Rates of cross-intolerance between bosutinib and prior imatinib were low, suggesting that most patients intolerant to imatinib therapy may be successfully treated with bosutinib. Given the long-term nature of TKI therapy, late-emerging toxicities are of concern, particularly cardiac and vascular events. In a study of bosutinib versus imatinib as firstline treatment for CP CML, the incidence of cardiac and vascular AEs with bosutinib was low and similar to that of imatinib.17,18 In the present study, the incidence of newly-

References 1. Gambacorti-Passerini C, Antolini L, Mahon FX, et al. Multicenter independent assessment of outcomes in chronic myeloid leukemia patients treated with imatinib. J Natl Cancer Inst. 2011;103(7):553-561. 2. Vigano I, Di Giacomo N, Bozzani S, Antolini L, Piazza R, Gambacorti Passerini C. First-line treatment of 102 chronic myeloid leukemia patients with imatinib: a long-term single institution analysis. Am J Hematol. 2014;89(10):E184-187. 3. GLEEVEC® (imatinib mesylate). Full

1306

4. 5. 6. 7.

occurring cardiac and vascular AEs with second-line bosutinib remained low after year 2. However, most (85%) discontinuations due to AEs as the primary reason occurred within the first two years; thus, patients remaining on treatment after year 2 may have a lower risk of experiencing these events. The incidence of renal AEs, while low, remained similar in years 3-5. Bosutinib has been associated with a decrease in glomerular filtration rate that is typically modest and potentially reversible (similar to what has been reported with imatinib).19,20 Dose adjustments are recommended in patients with baseline and treatmentemergent renal impairment.6,19 Careful monitoring, supportive care, and prompt management of toxicities may allow patients to continue treatment long term. Most baseline and on-treatment factors examined appeared not to be predictive of response duration, OS, or PFS. Baseline Ph+ ratio ≤35% (vs. ≥95) was associated with all 3 types of long-term outcomes (MCyR duration but not CCyR duration). Lower percentage of peripheral blood blasts at baseline and MCyR by week 12 were associated with both improved OS and PFS. Having a baseline BCRABL1 mutation, regardless of sensitivity to bosutinib, was predictive of decreased OS and, interestingly, having an abnormal LFT on-treatment was predictive of increased OS. This unexpected result may be due to increased bosutinib exposure levels resulting from the underlying cause of the abnormal LFT, leading to an increase in efficacy; however, population pharmacokinetics modeling from this study has found no relationship between baseline LFTs and bosutinib pharmacokinetics. Notably, prior response or resistance to IM did not predict any long-term outcomes. Because P-values were not adjusted for multiple comparisons, marginally significant P-values should be interpreted with caution. The potent and durable activity and distinct toxicity profile of bosutinib confirm it is an important option for treating CML patients in the second-line setting, as demonstrated by its long-term efficacy and safety in these patients; a 10-year follow up is planned for patients enrolled in an ongoing extension study. Acknowledgments The authors would like to acknowledge Dr. H. Jean Khoury for his extraordinary contributions to the research and treatment of hematologic malignancies. This study was sponsored by Pfizer Inc. Medical writing support was provided by Johna Van Stelten, PhD, of Complete Healthcare Communications, LLC, and was funded by Pfizer Inc. Dr. Jorge Cortes’ participation in this study was supported in part by NCI grants CA016672 and CA049639.

Prescribing Information. Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA, 2016. TASIGNA® (nilotinib). Full Prescribing Information. Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA, 2016. SPRYCEL® (dasatinib). Full Prescribing Information. Bristol-Myers Squibb Company, Princeton, NJ, USA, 2016. BOSULIF® (bosutinib). Full Prescribing Information. Pfizer Labs, New York, NY, USA, 2017. Ramirez P, DiPersio JF. Therapy options in imatinib failures. Oncologist. 2008;13(4): 424-434.

8. Jabbour E, Kantarjian H. Chronic myeloid leukemia: 2016 update on diagnosis, therapy, and monitoring. Am J Hematol. 2016;91(2):252-265. 9. Jabbour E, Kantarjian H, Cortes J. Use of second- and third-generation tyrosine kinase inhibitors in the treatment of chronic myeloid leukemia: an evolving treatment paradigm. Clin Lymphoma Myeloma Leuk. 2015;15(6):323-334. 10. Cortes JE, Kantarjian HM, Brummendorf TH, et al. Safety and efficacy of bosutinib (SKI-606) in chronic phase Philadelphia chromosome-positive chronic myeloid leukemia patients with resistance or intol-

haematologica | 2018; 103(8)

Long-term second-line bosutinib for CP CML

erance to imatinib. Blood. 2011;118(17): 4567-4576. 11. Gambacorti-Passerini C, Brummendorf TH, Kim DW, et al. Bosutinib efficacy and safety in chronic phase chronic myeloid leukemia after imatinib resistance or intolerance: minimum 24-month follow-up. Am J Hematol. 2014;89(7):732-742. 12. Brummendorf TH, Cortes JE, Khoury HJ, et al. Factors influencing long-term efficacy and tolerability of bosutinib in chronic phase chronic myeloid leukaemia resistant or intolerant to imatinib. Br J Haematol. 2016; 172(1):97-110. 13. Shah NP, Kim DW, Kantarjian H, et al. Potent, transient inhibition of BCR-ABL with dasatinib 100 mg daily achieves rapid and durable cytogenetic responses and high transformation-free survival rates in chronic phase chronic myeloid leukemia patients

haematologica | 2018; 103(8)

14.

15.

16. 17.

with resistance, suboptimal response or intolerance to imatinib. Haematologica. 2010;95(2):232-240. Shah NP, Kantarjian H, Kim D-W, et al. Sixyear (yr) follow-up of patients (pts) with imatinib-resistant or -intolerant chronicphase chronic myeloid leukemia (CML-CP) receiving dasatinib. J Clin Oncol. 2012;30 (15_suppl):6506. Redaelli S, Piazza R, Rostagno R, et al. Activity of bosutinib, dasatinib, and nilotinib against 18 imatinib-resistant BCR/ABL mutants. J Clin Oncol. 2009;27 (3):469-471. ICLUSIGÂŽ (ponatinib). Full Prescribing Information. ARIAD Pharmaceuticals Inc., Cambridge, MA, USA, 2016. Cortes JE, Jean Khoury H, Kantarjian H, et al. Long-term evaluation of cardiac and vascular toxicity in patients with Philadelphia

chromosome-positive leukemias treated with bosutinib. Am J Hematol. 2016;91(6): 606-616. 18. Brummendorf TH, Cortes JE, de Souza CA, et al. Bosutinib versus imatinib in newly diagnosed chronic-phase chronic myeloid leukaemia: results from the 24-month follow-up of the BELA trial. Br J Haematol. 2015;168(1):69-81. 19. Cortes JE, Gambacorti-Passerini C, Kim DW, et al. Effects of Bosutinib Treatment on Renal Function in Patients With Philadelphia Chromosome-Positive Leukemias. Clin Lymphoma Myeloma Leuk. 2017;17(10): 684-695. 20. Marcolino MS, Boersma E, Clementino NC, et al. Imatinib treatment duration is related to decreased estimated glomerular filtration rate in chronic myeloid leukemia patients. Ann Oncol. 2011;22(9):2073-2079.

1307

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Phase I trial of plerixafor combined with decitabine in newly diagnosed older patients with acute myeloid leukemia Gail J. Roboz,1 Ellen K. Ritchie,1 Yulia Dault,1 Linda Lam,1 Danielle C. Marshall,1 Nicole M. Cruz,1 Hsiao-Ting C. Hsu,1 Duane C. Hassane,1 Paul J. Christos,2 Cindy Ippoliti,1 Joseph M. Scandura1 and Monica L. Guzman1

Division of Hematology and Medical Oncology, Leukemia Program, Weill Cornell Medicine/New York-Presbyterian Hospital and 2Division of Biostatistics and Epidemiology, Weill Cornell Medicine, New York, NY, USA

Haematologica 2018 Volume 103(8):1308-1316

ABSTRACT

Correspondence: gar2001@med.cornell.edu

Received: October 26, 2017. Accepted: April 27, 2018. Pre-published: May 3, 2018.

cute myeloid leukemia carries a dismal prognosis in older patients. The objective of this study was to investigate the safety and efficacy of decitabine combined with the CXCR4 antagonist plerixafor in newly diagnosed older patients with acute myeloid leukemia and to evaluate the effects of plerixafor on leukemia stem cells. Patients were treated with monthly cycles of decitabine 20 mg/m2 days 1-10 and escalating doses of plerixafor (320-810 mcg/kg) days 1-5. Sixty-nine patients were treated, with an overall response rate of 43%. Adverse karyotype did not predict response (P=0.31). Prior hypomethylating agent treatment was the strongest independent predictor of adverse overall survival (hazard ratio 3.1; 95%CI: 1.3-7.3; P=0.008) and response (14% in previously treated patients, 46% in treatment naïve; P=0.002). As expected, the most common toxicities were myelosuppression and infection. Plerixafor induced mobilization of leukemia stem and progenitor cells, but did not cause clinically significant hyperleukocytosis. Reduction in leukemia stem cells appeared to correlate with duration of response. Plerixafor can be safely added to decitabine in poor-prognosis, elderly acute myeloid leukemia patients. The maximum tolerated dose of the combination was 810 mcg/kg. While mobilization of leukemia stem cells was observed in some patients, the clinical benefit of adding plerixafor was uncertain. This trial was registered at clinicaltrials.gov identifier: 01352650.

doi:10.3324/haematol.2017.183418 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1308 ©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.

1308

Introduction Acute myeloid leukemia (AML) carries a dismal prognosis in older patients, especially those with adverse cytogenetics and/or poor performance status. The median age at diagnosis is 67 years, with 5-year survival less than 10%.1 Standard induction chemotherapy with cytarabine and an anthracycline can achieve remission in selected older AML patients, but this regimen is often not feasible due to toxicity and poor tolerability. Furthermore, long-term survival after initial intensive chemotherapy is rare in these patients, even in the setting of initial complete remission (CR). Decitabine (5-aza-2’-deoxycytidine), a DNA methyltransferase inhibitor, has shown efficacy with an acceptable extramedullary toxicity profile in newly diagnosed older AML patients, with 25% CR, 7% 30-day mortality, and a median overall survival of 7.7 months when administered using a schedule of 20 mg/m2 over one hour (h) daily for five days. Of note, patients in this multicenter study received a median of 3 cycles of treatment (range 1-25 cycles).2 Blum et al. treated 53 AML patients with a median age of 74 years (range 60-85 years) with decitabine 20 mg/m2 daily for ten days every four weeks.3 The CR rate was 47% after a median of 3 cycles of therapy, with 2% 30-day and 15% 8-week induction mortality, mostly due to disease progression. Median overall and disease-free survival were 55 haematologica | 2018; 103(8)

Decitabine and plerixafor in elderly AML

and 46 weeks, respectively.3 Similar results have been achieved in other single-center trials using single-agent decitabine in a 10-day schedule, including at our own center.4,5 Still, since most newly-diagnosed AML patients treated with decitabine generally relapse within 6-18 months, and overall median survival is less than one year, a variety of potentially synergistic agents are under investigation. Acute myeloid leukemia originates from a rare population of leukemia stem cells (LSCs) that are capable of selfrenewal, proliferation, and differentiation into malignant blasts. LSCs and leukemic blasts can persist after treatment and contribute to disease relapse.6 Drugs that release LSCs and blasts from their protective microenvironment may leave them more vulnerable to therapy, as they are strongly dependent on the bone marrow niche for proliferation and survival.7 CXCR4, the chemokine receptor for stromal cellderived factor 1 (CXCL12/SDF-1), is a critical component of the bone marrow niche and is expressed on both normal stem cells and AML blasts. The CXCR4/SDF-1 axis in AML promotes leukemic cell homing to the marrow, as well as in vivo growth.8,9 Plerixafor, a small molecule antagonist of CXCR4, is commercially available as a stem cell mobilizing agent and is approved by the US Food and Drug Administration (FDA) for use in combination with granulocyte-colony stimulating factor (G-CSF) for patients with multiple myeloma or non-Hodgkin lymphoma undergoing autologous stem cell transplantation (Mozobil, SanofiAventis). Plerixafor blocks CXCR4-mediated signaling and significantly decreases the survival of AML cells in vitro. Plerixafor has been safely combined with cytotoxic chemotherapy in several studies of patients with relapsed/refractory AML.10-12 While mobilization of leukemic blasts was achieved, these trials were not randomized and, thus, the impact of plerixafor on clinical outcomes was unclear. The objective of this investigator-initiated clinical trial was to investigate the safety and efficacy of adding plerixafor to decitabine in newly diagnosed older patients with AML. Extensive correlative scientific studies were performed to determine the effects of plerixafor on the mobilization of LSCs and leukemic progenitor cells.

Methods This trial was registered at clinicaltrials.gov identifier: 01352650 and was approved by the Institutional Review Board of Weill Cornell Medical College. The study was performed in accordance with the Declaration of Helsinki, and all subjects provided written informed consent.

Patient selection and study design

The study population included patients ≥60 years old with newly diagnosed, pathologically confirmed AML, as defined by World Health Organization criteria.13 Patients with an antecedent hematologic disorder or therapy-related myeloid neoplasm were included, but those with acute promyelocytic leukemia or favorable risk cytogenetics according to the European LeukemiaNet (ELN) criteria were excluded from participation.14 Patients with a history of prior treatment with either decitabine or plerixafor, and those undergoing active treatment for a concomitant malignancy were also excluded. There were no mandatory requirements for organ system function or performance status, but patients with a calculated CrCl of ≤50 mL/min using the Cockcroft-Gault formula had a dose reduction of plerixafor by one-third during that cycle, haematologica | 2018; 103(8)

as per the FDA package insert for plerixafor. The trial was designed as an open-label, phase I feasibility study to optimize mobilization of leukemia stem and progenitor cells using a fixed dose and schedule of decitabine combined with escalating doses of plerixafor. Based on previous data, it was expected that patients would require between 1-4 10-day cycles of decitabine to achieve clinical response. Plerixafor was administered during alternating treatment cycles, which allowed each patient to serve as his/her own “control” for measurements of mobilization and other correlative scientific studies. Half of the patients received plerixafor during odd-numbered treatment cycles, and half during even-numbered cycles, as the optimal timing of plerixafor administration was unknown.

Treatment schedule Patients were treated according to the schedule in Figure 1. Ninety-three patients were screened and 69 were enrolled onto the trial. Prior to protocol treatment, patients were treated with hydroxyurea to reduce the total white blood cell count to <30 x109/L. Up to 4 induction cycles of decitabine, with or without the addition of plerixafor, were permitted, with 28-56 days between the starting days of each cycle. Decitabine was administered as an intravenous infusion of 20 mg/m2 over 1 hour (h) on days 1-10 of every treatment cycle. Plerixafor was administered 4 h prior to decitabine, during alternating treatment cycles: schedule A patients received plerixafor during even-numbered cycles and schedule B patients received plerixafor during odd-numbered cycles. There were three dosing cohorts of plerixafor. Cohorts 1, 2 and 3 received 320, 540, and 810 μg/kg of plerixafor intravenously on days 1-5, respectively, during alternating treatment cycles. All patients in all groups were treated with decitabine at the same dose and schedule. Patients with evidence of clinical benefit from treatment, including improved blood counts, reduced transfusion requirements, and/or improved performance status were eligible for treatment with ongoing monthly maintenance cycles of five days of decitabine, with plerixafor administered during alternate cycles according to the same dose and schedule as during induction. Patients were treated with antibiotics, transfusions, and other supportive care measures as per institutional guidelines. The use of erythropoietic growth factors was not permitted. GSCF was permitted at the discretion of the investigator, but could not be administered on the same days as plerixafor. Plerixafor was provided by Genzyme Inc., which was later acquired by Sanofi Oncology.

Safety assessments Patients were hospitalized for daily laboratory and clinical monitoring, as per institutional practice. Adverse events were reported using the National Cancer Institute (NCI) Common Terminology Criteria (CTCAE) v.4.0. A data and safety monitoring board (DSMB) was established as per the guidelines of Weill Cornell Medical College, and assessments of dose-limiting toxicity (DLT) were made in conjunction with the Data and Safety Monitoring Board.

Response assessments Responses were determined using the International Working Group criteria.15 Complete remission (CR) was defined as a decrease in bone marrow blasts to less than 5% and absence of blasts in the peripheral blood, coupled with recovery of the absolute neutrophil count (ANC) to ≥1.0x106/mL and platelet count to ≥100x106/mL. Patients who met all criteria for CR except ANC or platelet recovery were defined as CR with incomplete peripheral blood count recovery (CRi). Partial response (PR) was defined as a ≥50% reduction in bone marrow blasts. 1309

G.J. Roboz et al.

Figure 1. Treatment schema.

Statistical analysis

Table 1. Patients' baseline characteristics.

The primary study end point was to determine the safety and toxicity of this novel treatment regimen combining decitabine and plerixafor. The secondary clinical end points included overall response rate and overall survival. Overall survival was measured in months from cohort assignment to date of death or last followup date. The overall response rate (ORR) calculation included CR, CRi and PR. Descriptive statistics (i.e. median, range, frequency, and percent) were calculated to characterize the study population. Overall survival (OS) was assessed by Kaplan-Meier survival analysis, and univariate associations between demographic/clinical variables of interest and OS were assessed by the log-rank test. The independent effect of demographic/clinical predictors of interest on OS was assessed by multivariable Cox proportional hazards regression analysis. Adjusted hazard ratios (HR) were computed and 95% confidence intervals (95%CI) for hazard ratios and median OS time estimates are presented to assess the precision of the obtained estimates. Median follow-up time for the study group was computed based on survivors. Associations between demographic/clinical variables of interest and overall response were evaluated by the χ2 test or Fisher’s exact test, as appropriate. All P-values are two-sided with statistical significance evaluated at the 0.05 alpha level. All analyses were performed in SPSS v.24.0 (SPSS Inc., Chicago, IL, USA) and Stata v.14.0 (StataCorp, College Station, TX, USA).

Characteristic or demographic

Correlative scientific studies Details of correlative scientific studies and mutational profiling can be found in Online Supplementary Appendix 1.

Results Patients' characteristics Ninety-five newly diagnosed AML patients were screened for eligibility and 69 patients were enrolled onto the study between June 2011 and January 2013. Of the 26 patients ineligible for study participation, 18 were reclassified as having myelodysplastic syndrome and 8 patients chose to receive standard induction chemotherapy. Baseline characteristics are presented in Table 1. The baseline mutational profile of the cohort is presented in Figure 2. The most frequently observed mutations were in RUNX1 (38%), TET2 (31%), ASXL1 (30%), SRSF2 (30%), DNMT3A (28%), BCOR (18%), and TP53 (18%). The median age at diagnosis was 73 years (range 56-87 years); 1310

N. of patients

Age at diagnosis, years Median Range Sex Female Male ECOG performance status 0 1 2 CALGB cytogenetic risk classification20 Intermediate Adverse Antecedent hematologic disorder Present Not present Treament-related AML Yes No Prior hypomethylating agent Yes No Bone marrow blasts Median Range Ejection fraction Median Range Creatinine clearance ≤50 mL/min Bilirubin >1.5 x ULN White blood cell count >30x109/L

73 56-87 31 38

44.9 55.1

12 44 13

17.4 63.8 18.8

39 30

56.5 43.5

30 39

43.5 56.5

31 38

44.9 55.1

14 55

20.3 79.7

54 6-99 62 30-76 28 3 10

40.6 4.4 14.5

N: number; ECOG: Easter Cooperative Oncology Group; CALGB: Cancer and Leukemia Group B; AML: acute myeloid leukemia; ULN: upper limit of normal. Based on Byrd et al. CALGB cytogenetic risk criteria.20

55% were male. Approximately 80% of the patients had Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0-1 and the remainder had ECOG PS 2. Forty-four percent of patients had adverse cytogenetics, 44% had an antecedent hematologic disorder, and 45% had therapy-related AML. The median bone marrow haematologica | 2018; 103(8)

Decitabine and plerixafor in elderly AML

Figure 2. Mutational profile of patients studied. Each column indicates a patient, while each row indicates a gene tested (right label) and the percent of patients mutated for each gene (left label). Each mutation is colored according to the mutation type(s) present. Bar plots show the number of mutations per patient (top) and the total number of mutations per gene (right).

blasts was 54% (range 6-99%) and 15% of patients had a baseline white blood cell count greater than 30x109/L. Forty-five percent of patients had impaired hepatic or renal function at study entry. Twenty percent of patients had received prior treatment with azacitidine. Patients were evenly distributed among the three plerixafor dosing cohorts and received a median of 3 cycles of decitabine and one cycle of plerixafor. Thirty-two percent (n=21) of patients received more than 3 cycles of decitabine and more than one cycle of plerixafor.

Safety Grade 3 and 4 adverse events were as expected for older AML patients and included myelosupression in all patients, febrile neutropenia (65%), bacteremia (30%), and respiratory infections (23%). While myelosuppression was common, there were no episodes of unexpectedly prolonged myelosuppression in the absence of residual AML. Grade 1 and 2 adverse events observed in more than 20% of patients included, in ascending order of frequency: depression (23%), anxiety (26%), limb-related pain (29%), anorexia (30%), fever (30%), peripheral edema (33%), infection (36%), elevated creatinine (46%), nausea (48%), fatigue (51%), constipation (52%), skin disorders (53%), diarrhea (58%), and elevated bilirubin (83%) (Table 2). Grade 1 and 2 insomnia was unique to plerixafor-containing cycles and was experienced by 56.1% of patients. There was no clinically significant hyperleukocytosis caused by plerixafor. There was one episode of dose-limhaematologica | 2018; 103(8)

Table 2. Grade 1/2 adverse events observed in over 20% of subjects, regardless of drug attribution. Depression Anxiety Limb-related pain Anorexia Fever Peripheral edema Infection Elevated creatinine Nausea Fatigue Constipation Skin disorders Insomnia Diarrhea Elevated bilirubin

N (%) 16 (23) 18 (26) 20 (29) 21 (30) 21 (30) 23 (33) 25 (36) 32 (46) 33 (48) 35 (51) 36 (52) 37 (53) 39 (56) 40 (58) 57 (83)

N: number.

iting toxicity, renal insufficiency, in the 810 Îźg/kg cohort, but this was not clearly related to plerixafor. Still, this event, combined with the clinical impression of increased insomnia and increased non-serious gastrointestinal complaints in the 810 Îźg/kg cohort, led us to recommend 675 1311

G.J. Roboz et al. Îźg/kg as the maximum tolerated dose of plerixafor for this regimen. The 30-day induction mortality was 5.8% (n=4) and 60-day induction mortality was 13.0% (n=9).

Efficacy Of 69 evaluable patients, the overall response rate was 43% (35% CR, 7% CRi, 1% PR), with a median time to best response of 1.9 months (2 cycles, range 0.9-7.9 months). The median response duration was 4.5 months, with a median follow-up time for the entire study group (based on survivors) of 9.9 months (range 5.4-24.8 months). Median OS was 11.2 months (95%CI: 8.5 months, 13.9 months) (Figure 3A). As expected, the median OS for responders was significantly longer (18 months; 95%CI: 10.5-25.4 months) than for non-responders (5.0 months; 95%CI: 1.4-8.6 months) (P<0.0001). Prior treatment with azacitidine was the strongest independent predictor of OS (adjusted hazard ratio 3.1, 95%CI: 1.3-7.3; P=0.008) (Figure 3B). The median survival of patients previously treated with azacitidine was also much shorter than for previously untreated patients (2.5 months, 95%CI: 1.1-3.8 months vs. 12.6 months, 95%CI: 9.5-15.7 months; P=0.001). In addition, whereas 52% of HMA treatment-naĂŻve patients responded to treatment (46%/6% CR/CRi), only 14% of patients previously treated with an HMA achieved a response (0%/14% CR/CRi; P=0.002). Finally, adverse karyotype did not predict overall response (P=0.31) and 53% of patients with adverse cytogenetics achieved responses (43% CR, 7% CRi, 3% PR). Median OS was 10.9 months in 59 patients without a TP53 mutation and 18.1 months in the 10 patients with TP53 mutation, but this result was not statistically significant, probably due to the small sample size. In multivariate analysis, however, adverse karyotype, as well as baseline bone marrow blasts more than 54%, were significant predictors of poor OS. Neither therapy-related AML nor the presence of an antecedent hematologic disorder was a significant predictor of response; 10 patients with therapy-related AML (32%) and 14 patients (45%) with an antecedent hematologic disorder achieved CR/CRi. There were no significant differences in ORR or OS based on plerixafor dose level (P=0.55 and P=0.19, respectively) or treatment schedule (A

vs. B; P=0.71 and P=0.53, respectively), but the study was not powered for these comparisons. After discontinuation of study treatment, 28 patients (42%) received further anti-leukemia therapies. Of these 28 patients, 8 achieved first or second remission with standard cytarabine/daunorubicin induction (n=5), additional decitabine (n=2), or elacytarabine (n=1). Thirteen patients (20%) underwent hematopoietic stem cell transplantation, of whom 9 were in remission, one was in partial remission, and 3 were not in remission at the time of transplant.

Correlative studies The primary goal of the correlative studies was to assess the ability of plerixafor to induce mobilization. As shown in Figure 4, treatment with plerixafor resulted in significant mobilization of leukemic stem and progenitor cells, but not as robustly and consistently as predicted. The mobilizing effect of plerixafor was highly significant (P=0.0221) among clinical responders (n=21) versus nonresponders (n=19) (Figure 4B). Patients who received plerixafor combined with decitabine starting in their first cycle of treatment (B cohorts) were observed to have their most significant mobilizations during plerixafor-containing cycles (n=26; P=0.0318) (Figure 4C and D). As plerixafor is a CXCR4 antagonist that is expected to induce cell cycle entry of LSCs via loss of CXCR4-SDF1 interaction, we evaluated CXCR4 expression prior to plerixafor (Figure 5A), as well as cell cycle status before and after exposure to plerixafor (Figure 5C). As expected, we found that, within the responder group, plerixafor was more likely to mobilize CXCR4+ cells (Figure 5A and B). Interestingly, no significant differences were found for the non-responder group when evaluating mobilization with respect to CXCR4 expression (Online Supplementary Figure S1A). In addition, we found that plerixafor was more likely to increase the cycling of stem/progenitor cells, as measured by Ki-67 staining (Figure 5C and D). However, even though we observed that mobilizers tended to have increased proportions of cycling cells, we did not find differences in the duration of response or time to relapse between patients with and without increased cycling (Online Supplementary Figure S1B).

Figure 3. Overall survival (OS) for all evaluable patients in the study. (A) Kaplan-Meier OS curves for all patients (n=69) enrolled on the study. Median OS 11.2 months (95% CI: 8.6 months, 13.8 months). (B) OS for patients stratified by prior hypomethylating agent (HMA) therapy. No prior HMA (n=55), median OS 12.6 months (95%CI: 9.2, 15.9 months); prior hypomethylating agent (n=14), median OS 2.5 months (95%CI: 1.1, 3.8 months); P=0.0008 by log-rank test. PD: plerixafor + decitabine; D: decitabine.

1312

haematologica | 2018; 103(8)

Decitabine and plerixafor in elderly AML

Phenotypically defined LSCs were evaluated in the bone marrow during treatment. It is clear that LSCs were shown to persist even after multiple cycles of decitabine (Figure 6). An increase in LSC frequency was noted after cycle 3, when relapses were beginning to become evident (Figure 6A). Interestingly, among non-mobilizers, there was a continuous increase in LSCs (Figure 6B), suggesting the inability of decitabine to eliminate these cells. Furthermore, among responders, the absence of an increased population of LSCs after cycle 3 correlated with prolonged remission, with relapse in most patients only after 8 cycles of therapy. In contrast, all patients with a greater than two-fold increase in LSCs at cycle 3 relative to the previous bone marrow time point replapsed before completing 7 cycles of therapy (Figure 6C).

Discussion This investigator-initiated clinical trial was the first time a mobilizing agent has been combined with a DNA methyltransferase inhibitor in newly diagnosed patients with AML. The study was based on the hypothesis that

blockade of CXCR4/SDF-1 signaling with plerixafor would mobilize leukemic stem and progenitor cells out of their bone marrow microenvironment and make them more susceptible to the effects of decitabine. We found that adding plerixafor to decitabine was safe and tolerable. The toxicity profile was as expected for older patients with newly diagnosed AML treated with 10-day cycles of decitabine and included myelosuppression, febrile neutropenia, and infections. Importantly, the addition of plerixafor did not induce clinically significant hyperleukocytosis or tumor lysis syndrome. Grade 1 and 2 insomnia and gastrointestinal disturbances were more frequent with plerixafor, but there were no major, unexpected events. We established the maximum tolerated dose of plerixafor as 810 Îźg/kg and the recommended treatment dose for combination with decitabine as 675 Îźg/kg, based on an event of renal insufficiency, combined with increased episodes of gastrointestinal complaints and insomnia. We designed extensive correlative scientific studies in an effort to understand the effects of plerixafor on leukemic stem and progenitor populations. Treatment with plerixafor resulted in significant mobilization of leukemic stem and progenitor cells, but not as effectively

haematologica | 2018; 103(8)

Figure 4. Plerixafor increases mobilization of stem/progenitor cells. (A) Violin plot representing the fold change, evaluated at 4 hours after infusion, in stem/progenitor cells (CD34+CD38-) for all patients comparing cycles containing plerixafor (DP) to cycles without plerixafor (P). Scatter plots representing the average fold change in stem/progenitor cells + (CD34 CD38 ) comparing (B) responders and non-reponders in all cohorts, (C) cohorts A and B, and (D) all cohorts. Each symbol represents a patient, horizontal bar represents the mean, error bars represent the Standard Error of Mean. Paired t-tests between plerixafor and non-plerixafor cycles.

1313

G.J. Roboz et al. A

Figure 5. Plerixafor is more likely to mobilize CXCR4+ cells and increase the cycling of stem/progenitor cells. (A) Representative flow cytometry histograms for CXCR4 expression in diagnostic samples from a mobilizer and a non-mobilizer. Blue filled histogram represents control, red represents CXCR4-stained. (B) Scatter plots for the expression of CXCR4 at diagnosis comparing mobilizers and non-mobilizers within the responder group. (C) Representative flow cytometry dot plots to determine cell cycle status. Red squares show quiescent cells. (D) Scatter plots representing Ki-67+ stem/progenitor cells in mobilizers and non-mobilizers during plerixafor-containing cycles (PD) and non-plerixafor cycles (D). Each symbol represents a patient, horizontal bar represents the mean, error bars represent the Standard Error of Mean.

as predicted. There are several possible reasons for this, including: a) suboptimal dose and/or schedule of plerixafor, leading to inadequate CXCR4 blockade; b) suboptimal CXCR4 blockade despite adequate dosing; c) adequate CXCR4 blockade, but presence of additional factors influencing the dependency of LSCs on their microenvironment, such as VLA-4 and/or E-selectin. Also, CXCR4expressing LSCs were more effectively mobilized than those without CXCR4 expression and plerixafor effectively increased the cycling of stem/progenitor cells, but we did not measure the effect of increased cycling on the ability of decitabine to incorporate into these cells. Our data show that leukemia stem and progenitor cells persist after treatment with decitabine. Persistence of LSCs has been associated with disease progression in larger AML studies using chemotherapy,16 but the significance of phenotypically-defined LSCs in disease progression or clinical out1314

comes for HMA-based therapies requires further investigation in larger studies. We did not observe consistent data indicating that CXCR4 blockade with plerixafor sensitizes LSCs and progenitors to decitabine, but depletion of LSC populations by cycle 3 among responders seemed to be predictive of longer remission duration. While this trial was designed to optimize acquisition of clinically relevant, correlative scientific data, the design may not have optimized the antileukemic effects of the plerixafor/decitabine combination, since patients received plerixafor only every other cycle and for only half of the decitabine doses (5 of 10 doses). At the trial outset, we were uncertain as to the optimal dose and schedule of plerixafor, and concerned about its potential for inducing leukocytosis. These issues led to a treatment design in which plerixafor was administered either during even numbered or odd-numbered cycles for each dosing haematologica | 2018; 103(8)

Decitabine and plerixafor in elderly AML

Figure 6. Leukemia stem/progenitor cells persist after treatment with decitabine; absence of increase in leukemia stem cells (LSCs) at cycle 3 in responders correlates with longer remissions. (A) Scatter plots representing the fold change in LSCs (CD34+CD38-CD123+CD90-) in the bone marrow at the end of the cycle indicated, relative to the previous time point. (B) Fold change in LSCs in the bone marrow for a subset of responders from cohorts 3A and 3B comparing mobilizers and non-mobilizers. (C) Fold change in LSCs in the bone marrow at cycle 3 comparing patients who relapsed before cycle 7 (C <7) to those who relapsed after cycle 8 (C >8); patients who were transplanted or off trial are shown. Each symbol represents a patient, horizontal bar represents the mean, error bars represent the Standard Error of Mean.

cohort, with the objective of using each patient as his/her own control. Mobilization may have been enhanced by additional days of dosing, or by concomitant administration of G-CSF, as recommended for plerixaforâ&#x20AC;&#x2122;s FDAapproved indication. The overall response rate of 43% and overall survival of 11 months are consistent with, but not significantly better, than the results from single-agent, 10-day schedules of decitabine reported by our center and others. Responses were seen in poor-prognosis, older patients with AML, including in those with particularly aggressive biological features, such as adverse karyotype, TP53 mutation, or therapy-related disease. Of interest, as has been previously reported with decitabine,17 OS was longer in patients with TP53 mutations, though the small sample size precluded statistical significance. While our data did not confirm a robust clinical benefit for the addition of plerixafor to decitabine in the doses and schedules investigated, the strong scientific rationale for the combination, the sugges-

haematologica | 2018; 103(8)

tive correlative scientific data, and the safety and feasibility of the approach argue that the concept of CXCR4 blockade in AML therapy should not be abandoned. Strategies for further investigation could include mobilization with alternative CXCR4 antagonists, such as novel small molecule inhibitors and antibodies, as well as combination strategies with G-CSF, E-selectin antagonists, and/or BCL2-inhibitors. Acknowledgments The authors would like to acknowledge Genzyme/Sanofi Oncology for research support and providing plerixafor, the investigational product for this trial. The authors are grateful to Wen Xie for expert technical assistance. The authors also gratefully acknowledge contributors to Leukemia Fighters. MG and GR were partially supported by LLS 6427-13, DP2 OD007399, R01 CA102031. Dr. Paul Christos was partially supported by the following grant: Clinical and Translational Science Center at Weill Cornell Medical College (UL1-TR000457-06).

1315

G.J. Roboz et al.

References 1. Institute NC. Cancer Stat Facts: Acute Myeloid Leukemia (AML). 2014. Available from: http://seer.cancer.gov/statfacts/html/ amyl.html 2. Cashen AF, Schiller GJ, O'Donnell MR, DiPersio JF. Multicenter, phase II study of decitabine for the first-line treatment of older patients with acute myeloid leukemia. J Clin Oncol. 2010;28(4):556-561. 3. Blum W, Garzon R, Klisovic RB, et al. Clinical response and miR-29b predictive significance in older AML patients treated with a 10-day schedule of decitabine. Proc Natl Acad Sci USA. 2010;107(16):7473-7478. 4. Bhatnagar B, Duong VH, Gourdin TS, et al. Ten-day decitabine as initial therapy for newly diagnosed patients with acute myeloid leukemia unfit for intensive chemotherapy. Leuk Lymphoma. 2014;55 (7):1533-1537. 5. Ritchie EK, Feldman EJ, Christos PJ, et al. Decitabine in patients with newly diagnosed and relapsed acute myeloid leukemia. Leuk Lymphoma. 2013;54(9):2003-2007. 6. Roboz GJ, Guzman M. Acute myeloid leukemia stem cells: seek and destroy. Expert Rev Hematol. 2009;2(6):663-672. 7. Konopleva MY, Jordan CT. Leukemia stem cells and microenvironment: biology and therapeutic targeting. J Clin Oncol.

1316

2011;29(5):591-599. 8. Sugiyama T, Kohara H, Noda M, Nagasawa T. Maintenance of the hematopoietic stem cell pool by CXCL12-CXCR4 chemokine signaling in bone marrow stromal cell niches. Immunity. 2006;25(6):977-988. 9. Konopleva M, Tabe Y, Zeng Z, Andreeff M. Therapeutic targeting of microenvironmental interactions in leukemia: mechanisms and approaches. Drug Resist Updat. 2009;12(4-5):103-113. 10. Cooper TM, Sison EAR, Baker SD, et al. A phase 1 study of the CXCR4 antagonist plerixafor in combination with high-dose cytarabine and etoposide in children with relapsed or refractory acute leukemias or myelodysplastic syndrome: A Pediatric Oncology Experimental Therapeutics Investigators' Consortium study (POE 1003). Pediatr Blood Cancer. 2017;64(8). 11. Martinez-Cuadron D, Boluda B, Martinez P, et al. A phase I-II study of plerixafor in combination with fludarabine, idarubicin, cytarabine, and G-CSF (PLERIFLAG regimen) for the treatment of patients with the first early-relapsed or refractory acute myeloid leukemia. Ann Hematol. 2018;97(5):763-772. 12. Uy GL, Rettig MP, Motabi IH, et al. A phase 1/2 study of chemosensitization with the CXCR4 antagonist plerixafor in relapsed or refractory acute myeloid leukemia. Blood.

2012;119(17):3917-3924. 13. Vardiman JW, Thiele J, Arber DA, et al. The 2008 revision of the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009;114(5):937-951. 14. Dohner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. 15. Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. 2003;21 (24):4642-4649. 16. van Rhenen A, Moshaver B, Kelder A, et al. Aberrant marker expression patterns on the CD34+CD38- stem cell compartment in acute myeloid leukemia allows to distinguish the malignant from the normal stem cell compartment both at diagnosis and in remission. Leukemia. 2007;21(8):1700-1707. 17. Welch JS, Petti AA, Miller CA, et al. TP53 and Decitabine in Acute Myeloid Leukemia and Myelodysplastic Syndromes. N Engl J Med. 2016;375(21):2023-2036.

haematologica | 2018; 103(8)

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1317-1328

Dalila Salvatore,1,2 Myriam Labopin,1,3,4 Annalisa Ruggeri,1,5 Giorgia Battipaglia,1,2 Ardeshir Ghavamzadeh,6 Fabio Ciceri,7 Didier Blaise,8 William Arcese,9 Gerard Sociè,10 Jean Henri Bourhis,11 Maria Teresa Van Lint,12 Benedetto Bruno,13 Anne Huynh,14 Stella Santarone,15 Eric Deconinck,16 Mohamad Mohty1,3,4 and Arnon Nagler4,17

Service d'Hématologie et Thérapie Cellulaire Hôpital Saint Antoine, Paris, France; Hematology Department, Federico II University, Naples, Italy; 3Hospital Saint-Antoine, Paris University UPMC, France; 4Acute Leukemia Working Party of EBMT, Paris, France; 5 Department of Pediatric Hematology and Oncology, IRCCS Bambino Gesù Children's Hospital, Roma, Italy; 6Shariati Hospital, Hematology-Oncology and BMT Research, Teheran, Iran; 7Haematology and BMT Unit, IRCCS Ospedale San Raffaele, Milano, Italy; 8 Programme de Transplantation &Therapie Cellulaire, Centre de Recherche en Cancérologie de Marseille, Institut Paoli Calmettes, France; 9Stem Cell Transplant Unit, Policlinico Universitario Tor Vergata, Rome, Italy; 10Hopital St. Louis, Dept.of Hematology – BMT, Paris, France; 11Hematology department, Institut Gustave Roussy, Villejuif, France; 12Ospedale San Martino, Department of Haematology II Genova; 13A.O.U. Città della Salute e della Scienza, Torino, Italy; 14Institut Universitaire du Cancer Toulouse, Oncopole, France; 15Unità Terapia Intensiva Ematologica per il Trapianto Emopoietico, Ospedale Civile, Pescara, Italia; 16Hopital Jean Minjoz, Service d'Hématologie, Besançon, France and 17Chaim Sheba Medical Center, Tel-Hashomer, Israel 1 2

ABSTRACT

llogeneic hematopoietic stem cell transplantation is the optimal care for patients with high-risk or intermediate - acute myeloid leukemia. In patients lacking matched sibling donor, haploidentical donors are an option. We compared outcomes of unmanipulated (Haplo) to matched sibling donor transplant in acute myeloid leukemia patients in first complete remission. Included were intermediate and high-risk acute myeloid leukemia in first complete remission undergoing Haplo and matched sibling donor transplant from 2007-2015, and reported to the ALWP of the EBMT. A propensity score technique was used to confirm results of main analysis: 2 matched sibling donors were matched with 1 Haplo. We identified 2654 pts (Haplo =185; matched sibling donor =2469), 2010 with intermediate acute myeloid leukemia (Haplo=122; matched sibling donor =1888) and 644 with high-risk acute myeloid leukemia (Haplo =63; matched sibling donor =581). Median follow up was 30 (range 1-116) months. In multivariate analysis, in intermediate - acute myeloid leukemia patients, Haplo resulted in lower leukemia-free survival (Hazard Ratio 1.74; P<0.01), overall-survival (HR 1.80; P<0.01) and GvHD-free-relapse-free survival (Hazard Ratio 1.32; P<0.05) and higher graft-versus-host disease (GvHD) non-relapse mortality (Hazard Ratio 3.03; P<0.01) as compared to matched sibling donor. In high-risk acute myeloid leukemia, no differences were found in leukemia-free survival, overall-survival, and GvHD-free- relapse-free survival according to donor type. Higher haematologica | 2018; 103(8)

Correspondence: annalisaruggeri80@hotmail.com

Received: January 22, 2018. Accepted: May 10, 2018. Pre-published: May 10, 2018. doi:10.3324/haematol.2018.189258 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1317 ©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.

1317

D. Salvatore et al.

grade II-IV acute GvHD was observed for Haplo in both high-risk (Hazard Ratio 2.20; P<0.01) and intermediate risk (Hazard Ratio 1.84; P<0.01). A trend for a lower Relapse-Incidence was observed in Haplo among high-risk acute myeloid leukemia (Hazard Ratio 0.56; P=0.06). The propensity score analysis confirmed results. Our results underline that matched sibling donor is the first choice for acute myeloid leukemia patients in first complete remission. On the other hand, results of Haplo transplants are similar to matched sibling donor transplants in acute myeloid leukemia patients with high risk cytogenetics.

Introduction

Methods

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is a potentially curative treatment for patients with acute myeloid leukemia (AML).1 However, a human leukocyte antigen (HLA)-identical sibling2,3 is available in only 25-35% of the patients.4 For patients lacking a full matched sibling donor (MSD), other stem cell sources are available such as unrelated donors,5 umbilical cord blood units,6 or HLA-mismatched family donors (Haplo).7,8 The advantage of the latter is the rapid availability of the donors both for the transplant procedure and for subsequent adoptive immunotherapies. Initial concerns with the Haplo-HSCT were the high rate of graft failure, of severe graft-versus-host disease (GvHD) due to the multiple class I and II HLA disparities between donor and recipient, and the high non-relapse mortality (NRM).9,10,11 Advances in HLA typing, optimization of GvHD prophylaxis and other transplantation techniques allowed outcome improvements,8 such as the use of non T-cell depleted (TCD) unmanipulated grafts with new strategies to modulate donor T-cell alloreactivity. In particular, the use of post-transplant high-dose cyclophosphamide (PTCY) or the addition of anti thymocyte globulin (ATG) to standard GvHD prophylaxis ensured higher rates of engraftment while keeping an acceptable incidence of GvHD.12,13,14 This contributed to the increase in the number of unmanipulated Haplo-HSCT performed in recent years.15 Single center or registry-based studies have reported similar outcomes between Haplo-HSCT and unrelated or cord blood allo-HSCT for patients with hematological malignancies.14,16,17,18 Data comparing Haplo -HSCT to MSD-HSCT in AML patients are limited. In a recent prospective multicenter non-randomized study from China, Wang et al.19 showed in a very young cohort of AML patients (median age of 28 years in the Haplo group) similar outcomes for Haplo and MSD-HSCT in AML patients in first complete remission (CR1). Similarly, Yoon et al.20 analyzed long-term outcomes of 561 patients with intermediate (n=417) or poor risk (n=144) AML that underwent HSCT in CR1 from various donors including from MSD and Haplo. In poor risk AML, the authors observed a 5-year disease-free survival (DFS) of 47% versus 60% (P<0.01) for MSD and Haplo, respectively; while in intermediate risk AML, DFS was 66% and 68% (P=0.08) for MSD and Haplo, respectively. Herein, we conducted a registry-based study of adult patients undergoing either an unmanipulated Haplo or a MSD allo-HSCT for high or intermediate risk AML in CR1, reported to the Acute Leukemia Working Party (ALWP) of EBMT.

We retrospectively analyzed adult patients (≥18 years) diagnosed with AML with intermediate or unfavorable cytogenetics who underwent their first allo-HSCT in CR1 between 2007 and 2015, from either a MSD or Haplo donor, and whose data were reported to the ALWP of the EBMT. The EBMT is a voluntary working group of more than 500 transplant centers that are required to report all consecutive stem cell transplantations and follow up once a year. Audits are routinely performed to determine the accuracy of the data. This study was approved by the ALWP of the EBMT institutional review board and conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. Patients were stratified according to cytogenetic status at diagnosis in intermediate or high risk, according to the previous definition from Grimwade et al.21 Of note, included in the Haplo group, were only patients receiving an unmanipulated graft with the use of in vivo TCD or PTCY. Ex vivo graft manipulation was an exclusion criteria. Conditioning regimen was defined myeloablative (MAC) when containing total body irradiation (TBI) with a dose >6 Gray or a total dose of busulfan (Bu) >8 mg/kg or >6.4 mg/kg when administered orally or intravenously, respectively. All other regimens were defined as RIC.22 Primary end-point was leukemia-free survival (LFS), defined as the probability of being alive and disease-free at any time point. Both death and relapse were considered events. Patients alive and in CR were censored at their last follow up. Overall survival (OS) was defined as the probability of being alive at any time point. Other secondary endpoints were engraftment, acute (aGvHD) and chronic (cGvHD) GvHD, relapse incidence (RI), non-relapse mortality (NRM) and refined graft-versus-host/relapse free survival (GRFS),23 defined as being alive with neither grade III-IV aGvHD, severe cGvHD nor disease relapse at any time point. Modified Glucksberg criteria and revised Seattle criteria were used to grade aGvHD24 and cGvHD,25 respectively. Engraftment was defined as achieving an absolute neutrophil count greater than or equal to 0.5×109/L for three consecutive days. NRM was defined as death from any cause without previous relapse or progression. Median values and ranges were used for continuous variables and percentages for categorical variables. Patient-, disease- and transplantrelated variables were compared using χ2 or Fischer exact test for categorical variables, and Mann–Whitney test for continuous variables. Probabilities of OS, LFS and GRFS were calculated using Kaplan-Meier method.26 Cumulative incidence functions (CIF) were used to estimate RI and NRM in a competing risks setting. To study GvHD, death and relapse were considered as competing events. Univariate analyses were performed using the log rank test for OS, LFS and GRFS, while the Gray test was used for CIF. Multivariate analyses adjusted for differences between the groups were performed using Cox proportional hazards regression model.27

1318

haematologica | 2018; 103(8)

Haplo versus MSD in AML in CR1

All interactions between donor type and other covariates were tested; a significant interaction according to cytogenetics has been found, thus a stratification (intermediate or high cytogenetics risk AML) with two separate analysis was made. Propensity score matching was also performed to reduce or eliminate confounding effects. Two MSD were matched with each Haplo using the nearest neighbor or exact matching.28 Matching was done without replacement. Included in the propensity score model were: age, year of allo-HSCT, time from diagnosis to allo-HSCT, conditioning regimen (RIC), source of stem cells, cytogenetics, patient and donor CMV serology status. All tests were two-sided and P values < 0.05 were considered statistically significant. Analyses were performed using the R statistical software version 3.2.3 (available online at http://www.Rproject.org), and propensity score analysis was performed using the ‘MatchIt’.29

Results Patients, disease and transplant characteristics Patients and transplant characteristics are summarized in Table 1. Median follow up was 22 (range 3-96) months and 31 (range 1-116) months for Haplo and MSD, respectively (P<0.01). We identified a total of 2654 patients (Haplo=185; MSD=2469), including 2010 intermediate AML (Haplo=1122; MSD=1888) and 644 high risk-AML (Haplo=163; MSD=581) transplanted in 227 EBMT centers. Median age at allo-HSCT was 50 (range 18-74) years for both Haplo and MSD (P=0.63). There were some differences between the two groups: Haplo underwent alloHSCT more recently compared to MSD recipients (2014 versus 2010; P<0.01) and had a longer time from diagnosis to allo-HSCT (6 versus 5 months, P<0.01); furthermore, in

Table 1. Patient, disease and transplant characteristics.

Characteristic (%) Median age, years (range) Median year of allo-HSCT (range) Interval from diagnosis to allo-HSCT, months (range) Cytogenetics Intermediate High risk Patient’s sex Male Female Donor’s sex Male Female Patient CMV serostatus Negative Positive Donor CMV serostatus negative positive Missing Conditioning regimen MAC RIC Stem cell source BM PBSC GVHD prophylaxis CsA alone CsA + MMF Csa + MTX PT-CY Other Missing In vivo TCD Median follow-up, months (range)

Haplo (n=185)

MSD (n=2469)

50 (18-74) 2014 (2007-2015) 6 (1-17)

50 (18-75) 2010 (2007-2015) 5 (1-18)

0.63 <0.01 <0.01

122 (66) 63 (34)

1888 (76) 581 (24)

<0.01

103 (56) 82 (44)

1296 (53) 1172 (47)

0.41

96 (52) 89 (48)

1322 (54) 1140 (46)

0.43

28 (15) 155 (85)

777 (32) 1660 (68)

<0.01

51 (28%) 132 (72%) 2

927 (38%) 1492 (62%) 50

<0.01

93 (50) 92 (50)

1302 (53) 1167 (47)

0.52

92 (50) 93 (50)

473 (19) 1996 (81)

<0.01 <0.01

4 (2) 4 (2) 7 (4) 137 (74) 33 (18) 0 54 (31) 22 (3-96)

470 (19) 487 (20) 1273 (51) 36 (2) 182 (7) 21 (1) 863 (35) 31 (1- 116)

0.30 <0.01

Haplo: haploidentical family donor; MSD: matched sibling donor; allo-HSCT: allogeneic hematopoietic stem cell transplantation; CMV: cytomegalovirus; MAC: myeloablative conditioning regimen; RIC: reduced intensity conditioning regimen; BM: bone marrow; PBSC: peripheral blood stem cells; CSA: cyclosporine; MMF: mycophenolate mofetil; MTX: methotrexate; PT-CY: post-transplant cyclophosphamide; TCD: in vivo T-cell depletion.

haematologica | 2018; 103(8)

1319

D. Salvatore et al.

the Haplo group there was a higher proportion of highrisk AML (34% versus 24% in MSD, P<0.01), bone marrow (BM) as stem cell source (50% versus 19% in MSD; P<0.01) and CMV positive donors (72% versus 62% in MSD; P<0.01). Conditioning regimen was MAC in approximately 50% of cases in both Haplo and MSD (P=0.52). In the Haplo group, the most frequently used MAC contained Thiotepa-Busulfan-Fludarabine, while the most frequent RIC contained cyclophosphamide and 2 or 4 Gy TBI. In the MSD group, the most frequently used MAC and RIC regimen were BusulfanCyclophosphamide and Busulfan-Fludarabine, respectively. Details on conditioning regimens are reported in the Online Supplementary Table. Among Haplo recipients, 137 (74%) received PTCY and 54 (31%) received ATG as GvHD prophylaxis.

Univariate analysis for the whole population The results of univariate analysis are summarized in Table 2A. A higher engraftment rate was observed in MSD recipients (99% versus 96%, P<0.01), with a shorter median time to engraftment in this group (16 versus 18 days in Haplo, P<0.01) Higher incidence of grade II-IV aGvHD was found in Haplo (21% versus 31%, P<0.01) while cGvHD was lower as compared to MSD (33% versus 35%, P=0.05). Main causes of death were disease recurrence (in 30% versus 59%), GvHD in 16% versus 18% and infections in 33% versus 12% of Haplo and MSD, respectively. At 2 years, CI of relapse was 19% versus 24% (P=0.10) and NRM was 23% versus 10% (P<0.01) in Haplo and MSD recipients, respectively. The probability of LFS and OS were 58% versus 67% (P<0.01) and 68% versus 76% (P<0.01), in Haplo and MSD, respectively. Probability of GRFS was 47% versus 50% (P=0.25), respectively.

Multivariate analysis for the whole population In a multivariate analysis adjusted on the main differ-

ences between the two groups (Table 3A), Haplo was associated with a higher risk of grade II-IV aGvHD (HR=1.94; 95% CI: 1.38-2.73; P<0.01), a higher NRM (HR=2.56; 95% CI:1.73-3.77; P<0.01), a lower LFS (HR=1.33; 95% CI: 1.03-1.71; P<0.04) and a lower OS (HR=1.34; 95% CI: 1.03-1.75; P<0.04). Moreover, due to a significant interaction between donor type and cytogenetic risk on LFS (P<0.01), all further analyses were stratified on cytogenetic group.

Outcomes according to cytogenetics: intermediate and high-risk AML Intermediate risk AML The results of univariate analysis in this group are summarized in Table 2B. Grade II-IV aGvHD was 29% versus 20% (P<0.03) for Haplo and MSD recipients, respectively. At 2 years, CI of cGvHD was 30% versus 36% (P<0.02) for Haplo and MSD recipients, respectively. The probability of LFS and OS were 56 % versus 70% (P<0.01) and 68% versus 79% (P<0.01) in Haplo and MSD, respectively. Probability of GRFS was 45% versus 54% (P<0.05), in Haplo and MSD, respectively. CI of relapse was 18% versus 20% (P=0.52) and NRM was 26% versus 10% (P<0.01) in Haplo and MSD recipients, respectively. In multivariate analysis, Haplo was associated with a higher risk of grade II-IV aGvHD (HR 1.84; 95% CI 1.202.82; P<0.01), higher NRM (HR 3.03; 95% CI 1.98-4.62; P<0.01), lower LFS (HR 1.74; 95% CI 1.30-2.32; P<0.01), OS (HR 1.80; 95% CI 1.32-2.45; P<0.01) and GRFS (HR 1.32; 95% CI 1.01-1.72; P<0.05). No significant differences were found for cGvHD and RI. Results of multivariate analysis for donor type and other factors associated with the main outcomes are reported in table 3B.

High risk AML The results of univariate analysis are summarized in table 2C For Haplo and MSD recipients, grade II-IV aGvHD was 36% versus 24% (P<0.04) and cGvHD was

Table 2. Results of univariate analysis for main outcomes at 2 years after allo-HSCT according to donor type (A) in patients with intermediate (B) and high risk (C) AML.

A) Outcome

RI % ±s.d.

NRM % ±s.d.

LFS % ±s.d.

OS % ±s.d.

Gr. II-IV aGvHD % ±s.d.

cGvHD% ±s.d.

GRFS% ±s.d.

Haplo MSD P

19±6 24±2 0.10

23±6 10±2 <0.01

58±6 67±4 <0.01

68±6 76±2 <0.01

31±7 21±2 <0.01

33±6 35±2 0.05

47±8 50±2 0.25

B) Outcome

RI % ±s.d.

NRM % ±s.d.

LFS % ±s.d.

OS % ±s.d.

Gr. II-IV aGvHD % ±s.d.

cGvHD% ±s.d.

GRFS% ±s.d.

Haplo MSD P

18±7 20±3 0.52

26±8 10±2 <0.01

56±9 70±2 <0.01

68±8 79±2 <0.01

29±8 20±2 <0.03

30±9 36±2 <0.02

45±9 54±2 <0.05

C) 2-years outcome

RI% ±s.d

NRM% ±s.d

LFS% ±s.d

OS % ±s.d

aGvHD gr II-IV % ±s.d

cGvHD % ±s.d

GRFS % ±s.d

Haplo MSD P

21±9 36±4 <0.02

18±9 10±3 0.16

61±13 55±4 0.14

67±12 66±4 0.26

36±12 24±3 <0.04

39±12 33±4 0.79

49±13 40±4 0.17

RI: relapse incidence; s.d.: standard deviation; NRM: non-relapse mortality; LFS: leukemia-free survival; OS: overall survival; GRFS: refined graft-versus-host disease/relapse free survival; aGvHD: acute graft-versus-host disease; cGvHD: chronic-graft-versus host disease; Haplo: haploidentical donor; MSD: matched sibling donor.

1320

haematologica | 2018; 103(8)

Haplo versus MSD in AML in CR1

39% versus 33% (P=0.79). The probability of LFS and OS were 61 % versus 55% (P=0.14) and 67% versus 66% (P=0.26) in Haplo and MSD, respectively. Probability of GRFS was 49% versus 40% (P=0.17), in Haplo and MSD, respectively. CI of relapse was 21% versus 36% (P<0.02) and NRM was 18% versus 10% (P=0.16) for Haplo and MSD, respectively.

In multivariate analysis, Haplo was associated with a higher risk of grade II-IV aGvHD (HR 2.20; 95% CI 1.293.74; P<0.01) and a trend for a lower RI (HR 0.56; 95% CI 0.31-1.01; P=0.06). No significant differences were found for other outcomes. Results of multivariate analysis for donor type and other factors associated with the main outcomes are reported in table 3C.

Figure 1. Outcomes at two years according to pair-matched analysis in patients with intermediate-risk acute myeloid leukemia. (A) Relapse-incidence. (B) Nonrelapse mortality. (C Leukemia-free survival. (D) Overall survival.

Figure 2. Outcomes at two years according to pair-matched analysis in patients with high-risk acute myeloid leukemia. (A) Relapse-incidence. (B) Non-relapse mortality. (C) Leukemia-free survival. (D) Overall survival.

haematologica | 2018; 103(8)

1321

D. Salvatore et al. Table 3. Results of multivariate analysis of main outcomes after HSCT in the entire population (A) and in patients with intermediate (B) or high risk (c) AML.

A) Variable RI Haplo versus. MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagnosis to allo-HSCT > median NRM Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagnosis to allo-HSCT > median LFS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagnosis to allo-HSCT > median OS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagnosis to allo-HSCT > median GRFS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM

HR (95% CI)

0.86 (0.60-1.22) 1.03 (0.97-1.11) 1.00 (0.97-1.04) 1.12 (0.93-1.34) 0.95 (0.78-1.17) 0.91 (0.76-1.09) 1.07 (0.90-1.27) 0.90 (0.76-1.05) 1.90 (1.62-2.22) 0.94 (0.90-0.98)

0.41 0.27 0.62 0.21 0.66 0.33 0.43 0.20 <0.01 <0.01

2.56 (1.73-3.77) 1.24 (1.11-1.37) 0.96 (0.92-1.01) 0.79 (0.61-1.03) 0.98 (0.73-1.31) 1.33 (1.05-1.67) 1.22 (0.93-1.59) 1.28 (1.00-1.64) 1.03 (0.79-1.34) 1.00 (0.96-1.05)

<0.01 <0.01 0.17 0.08 0.89 0.01 0.13 0.04 0.79 0.76

1.33 (1.03-1.71) 1.09 (1.03-1.16) 0.99 (0.97-1.02) 1.00 (0.86-1.16) 0.96 (0.81-1.13) 1.04 (0.91-1.20) 1.11 (0.96-1.29) 1.00 (0.87-1.15) 1.59 (1.39-1.81) 0.97 (0.94-1.00)

<0.04 <0.01 0.76 0.93 0.65 0.50 0.13 0.94 <0.01 0.08

1.34 (1.03-1.75) 1.15 (1.08-1.22) 0.99 (0.96-1.02) 0.94 (0.81-1.09) 1.04 (0.88-1.23) 1.09 (0.94-1.27) 1.15 (0.99-1.34) 1.01 (0.87-1.16) 1.71 (1.49-1.97) 0.97 (0.94-1.01)

<0.04 <0.01 0.71 0.45 0.60 0.21 0.06 0.87 <0.01 0.17

1.17 (0.92-1.48) 1.07 (1.01-1.12) 0.98 (0.96-1.00) 0.90 (0.79-1.03) 1.07 (0.92-1.24)

0.18 <0.01 0.12 0.14 0.36 continued on the next page

1322

haematologica | 2018; 103(8)

Haplo versus MSD in AML in CR1 continued from the previous page

Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagnosis to allo-HSCT > median aGvHD II-IV Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagn to allo-HSCT > median cGvHD Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Poor cytogenetics versus other Time from diagnosis to allo-HSCT > median

B) RI Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median NRM Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median LFS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT

1.20 (1.07-1.36) 1.04 (0.92-1.18) 1.03 (0.92-1.16) 1.40 (1.24-1.58) 0.98 (0.95-1.00)

<0.01 0.47 0.55 <0.01 0.17

1.94 (1.38-2.73) 1.00 (0.92-1.08) 0.96 (0.92-0.99) 0.69 (0.55-0.86) 0.98 (0.77-1.24) 1.31 (1.08-1.59) 0.83 (0.67-1.02) 1.17 (0.96-1.43) 1.23 (1.01-1.50) 0.95 (0.91-1.00)

<0.01 0.98 0.02 <0.01 0.87 <0.01 0.09 0.11 0.03 0.04

0.80 (0.57-1.13) 1.09 (1.02-1.16) 0.99 (0.96-1.02) 0.77 (0.65-0.92) 1.15 (0.94-1.40) 1.42 (1.22-1.65) 0.92 (0.78-1.07) 1.05 (0.91-1.23) 1.04 (0.88-1.22) 0.99 (0.96-1.02)

0.21 <0.01 0.88 <0.01 0.16 <0.01 0.31 0.46 0.62 0.76

Variable

HR (95% CI) P

1.12 (0.74-1.71) 0.99 (0.91-1.08) 1.01 (0.98-1.05) 1.25 (1.01-1.57) 0.98 (0.78-1.24) 0.87 (0.70-1.08) 0.99 (0.80-1.22) 1.06 (0.87-1.30) 0.94 (0.89-0.98)

0.58 0.80 0.47 <0.05 0.86 0.20 0.93 0.55 <0.01

3.03 (1.98-4.62) 1.27 (1.12-1.43) 0.99 (0.94-1.05) 0.86 (0.64-1.15) 1.11 (0.80-1.53) 1.29 (1.00-1.67) 1.15 (0.86-1.55) 1.41 (1.06-1.87) 0.99 (0.94-1.04)

<0.01 <0.01 0.85 0.30 0.54 0.06 0.35 <0.02 0.69

1.74 (1.30-2.32) 1.08 (1.01-1.16) 1.01 (0.98-1.04)

<0.01 <0.04 0.59 continued on the next page

haematologica | 2018; 103(8)

1323

D. Salvatore et al. continued from the previous page

RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median OS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median GRFS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median aGvHD II-IV Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagn to allo-HSCT > median cGvHD Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median

c) RI Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT

1.09 (0.92-1.30) 1.02 (0.85-1.23) 1.02 (0.87-1.20) 1.04 (0.88-1.24) 1.17 (0.99-1.38) 0.96 (0.93-1.00)

0.32 0.82 0.80 0.63 0.06 <0.04

1.80 (1.32-2.45) 1.14 (1.06-1.23) 1.01 (0.97-1.04) 1.01 (0.84-1.22) 1.13 (0.92-1.38) 1.09 (0.92-1.30) 1.10 (0.92-1.32) 1.15 (0.96-1.37) 0.94 (0.93-1.01)

<0.01 <0.01 0.72 0.91 0.25 0.33 0.30 0.13 0.13

1.32 (1.01-1.72) 1.07 (1.01-1.13) 0.99 (0.96-1.02) 0.97 (0.83-1.13) 1.02 (0.87-1.20) 1.22 (1.06-1.40) 1.06 (0.91-1.22) 1.16 (1.01-1.33) 0.97 (0.95-1.01)

<0.05 <0.03 0.44 0.68 0.77 <0.01 0.46 <0.04 0.11

1.84 (1.20-2.82) 1.05 (0.95-1.16) 0.96 (0.92-1.00) 0.65 (0.50-0.85) 0.90 (0.68-1.18) 1.47 (1.18-1.84) 0.81 (0.64-1.04) 1.32 (1.03-1.69) 0.96 (0.91-1.01)

<0.01 0.32 0.06 <0.01 0.44 <0.01 0.10 <0.03 0.12

0.75 (0.50-1.13) 1.10 (1.02-1.19) 1.00 (0.97-1.03) 0.73 (0.60-0.89) 1.18 (0.94-1.48) 1.48 (1.25-1.74) 0.94 (0.78-1.12) 1.07 (0.90-1.27) 1.00 (0.97-1.04)

0.17 <0.02 0.99 <0.01 0.15 <0.01 0.47 0.45 0.7

Variable 0.56 (0.31-1.01) 1.16 (1.03-1.29) 1.00 (0.95-1.05)

HR (95% CI) P 0.06 <0.02 0.94 continued on the next page

1324

haematologica | 2018; 103(8)

Haplo versus MSD in AML in CR1 continued from the previous page

RIC versus MAC PBSC versus BM Female to male recipient vs. other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median NRM Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median LFS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median OS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median GRFS Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median aGvHD II-IV Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other haematologica | 2018; 103(8)

0.88 (0.66-1.18) 0.85 (0.60-1.20) 0.99 (0.74-1.34) 1.19 (0.90-1.59) 0.66 (0.51-0.86) 0.95 (0.88-1.03)

0.41 0.36 0.97 0.22 <0.01 0.20

1.40 (0.62-3.13) 1.17 (0.95-1.43) 0.88 (0.79-0.97) 0.57 (0.33-0.97) 0.70 (0.39-1.24) 1.49 (0.90-2.46) 1.56 (0.87-2.77) 0.95 (0.57-1.56) 1.08 (0.98-1.19)

0.41 0.12 0.01 0.04 0.22 0.12 0.13 0.83 0.11

0.73 (0.46-1.17) 1.16 (1.05-1.28) 0.97 (0.93-1.02) 0.80 (0.62-1.03) 0.81 (0.60-1.09) 1.11 (0.86-1.43) 1.26 (0.98-1.63) 0.72 (0.57-0.91) 0.99 (0.93-1.05)

0.19 <0.01 0.21 0.08 0.16 0.42 0.07 <0.01 0.80

0.73 (0.44-1.20) 1.19 (1.07-1.32) 0.98 (0.93-1.03) 0.79 (0.60-1.02) 0.82 (0.60-1.12) 1.15 (0.88-1.51) 1.24 (0.95-1.62) 0.77 (0.60-0.98) 0.99 (0.93-1.06)

0.21 <0.01 0.38 0.07 0.21 0.30 0.11 0.03 0.84

0.88 (0.58-1.34) 1.09 (0.99-1.19) 0.96 (0.92-1.01) 0.78 (0.62-0.99) 1.13 (0.85-1.51) 1.19 (0.94-1.50) 1.05 (0.84-1.32) 0.80 (0.65-0.99) 0.97 (0.92-1.03)

0.56 0.07 0.09 0.04 0.39 0.14 0.67 0.04 0.28

2.20 (1.29-3.74) 0.90 (0.78-1.03) 0.96 (0.90-1.02) 0.83 (0.56-1.21) 1.30 (0.83-2.04) 1.07 (0.73-1.55)

<0.01 0.12 0.22 0.32 0.25 0.74

continued on the next page

1325

D. Salvatore et al. Patient CMV seropositivity Donor CMV seropositivity Time from diagn to allo-HSCT > median cGvHD Haplo versus MSD Age (incremental age of 10 years) Year of allo-HSCT RIC versus MAC PBSC versus BM Female to male recipient versus other Patient CMV seropositivity Donor CMV seropositivity Time from diagnosis to allo-HSCT > median

0.95 (0.65-1.39) 0.94 (0.66-1.33) 0.95 (0.87-1.04)

0.80 0.72 0.24

1.02 (0.58-1.78) 1.07 (0.94-1.21) 1.00 (0.94-1.06) 0.96 (0.68-1.36) 1.03 (0.68-1.56) 1.23 (0.88-1.72) 0.87 (0.62-1.20) 1.04 (0.76-1.42) 0.94 (0.87-1.03)

0.95 0.33 0.96 0.83 0.89 0.23 0.39 0.79 0.18

HR: hazard ratio; CI: confidence interval; RI: relapse incidence; NRM: non-relapse mortality; LFS: leukemia free survival; OS: overall survival; GRFS: refined graft-versus-host disease/relapse-free survival; aGvHD: acute graft-versus-host disease; cGvHD: chronic graft-versus-host disease; Haplo: haploidentical donor; MSD: matched sibling donor; alloHSCT: allogeneic hematopoietic stem cell transplantation; RIC: reduced intensity conditioning regimen; MAC: myeloablative conditioning regimen; PBSC: peripheral blood stem cells; BM: bone marrow; CMV: cytomegalovirus.

Propensity score matching analysis We were able to pair-match 183 Haplo with 364 MSD. The results of propensity score analysis are summarized in Table 4. In the group of patients presenting an intermediate risk cytogenetics, Haplo was associated with a higher risk of NRM (HR 2.59. 95% CI: 1.59-4.20. P<0.01), lower LFS (HR 1.60; 95% CI: 1.15- 2.22; P<0.01) and OS (HR 1.61; 95% CI: 1.12-2.30; P<0.01). There was no significant association between Haplo grade II-IV aGvHD, cGvHD and GRFS. In the group of patients presenting cytogenetics classified as high risk, Haplo was associated to higher risk of acute GvHD grade II-IV (HR 2.06; 95% CI: 1.14-3.75; P=0.02) and a trend for a lower risk of relapse (HR 0.53; 95% CI: 0.28-1.01; P=0.053). There was no significant association between Haplo and other main outcomes. Survival curves according to the results of pair-matched analysis in each cytogenetic group are shown in Figure 1 and 2.

Discussion Allogeneic HSCT might be a curative option in patients diagnosed with AML and achieving CR, especially in those with unfavorable cytogenetics for which prognosis is very poor with chemotherapy alone. Use of HSCT in patients with intermediate risk cytogenetics is sometimes debated, according to the different transplant center policies. Subsequently, for these two cytogenetic risk categories, a donor search might be immediately launched at time of diagnosis.32 In the absence of a MSD, Haplo may represent a valid alternative, despite initial concerns being raised due to the high risk of graft failure and NRM in this setting.9 The aim of the current study was to compare the outcomes of patients transplanted either from a MSD or Haplo donor in patients with AML in first CR. According to cytogenetic at time of diagnosis, AML was classified as intermediate or high risk. Moreover, due to a significant interaction according to cytogenetic risk, intermediate and high-risk AML were then analyzed separately. According to donor type, higher risk of grade II-IV aGvHD was reported in Haplo recipients. Furthermore, donor CMV 1326

positive serology was found as a risk factor for aGvHD, as already shown by others.33,34 In agreement with previous reports,35,36 among AML with intermediate cytogenetic risk, the intensity of the conditioning regimen was associated with higher risk of aGvHD, as well as female to male donor, while in AML with high cytogenetic risk, the only factor associated with higher risk of aGvHD was the type of donor. Furthermore, stem cell sources were not influential for acute GvHD, as previously described.20 No significant differences in the CI of cGvHD were found according to donor type. This could also be related to the higher proportion of BM in the Haplo group. Our results are in some part different to those reported by Luznik et al.12 Importantly, the experience reported by the Baltimore group is mainly in non myeloablative conditioning regimen and BM as stem cell source and this could in part explain the difference among our results. Also, being a registry study, we reported data from several transplant centers including different immunosuppressive protocols according to different Centers and as compared to previous reports19 and therefore no direct comparison could be performed. Compared with MSD recipients, Haplo recipients had a longer time to neutrophils recovery with a median time to engraftment of 2 days longer than MSD, in line with previous studies;17,18 this is probably due to the higher proportion of patients receiving bone marrow graft among Haplos and the myelosuppression from PT-CY. NRM was worse in the Haplo recipients in univariate and multivariate analysis. When looking at cytogenetics groups, this result was confirmed in intermediate risk, but not in high risk, where Haplo and MSD had similar NRM, in line with previous reports.17,19,20 Furthermore, female donor to male recipient was associated to a higher NRM in intermediate AML and not in high risk AML. Therefore, one can speculate that the impact of female to male mismatch could depend on the risk of the underlying disease, as previously shown.36 However, a possible explanation to the results in the high-risk group might be related to the low number of patients, preventing us to make definitive conclusions. Death from infections was more common in Haplo transplants than MSD maybe due to a slower immune haematologica | 2018; 103(8)

Haplo versus MSD in AML in CR1 Table 4. Propensity score analysis for main outcomes after allo-HSCT according to donor type in patients with intermediate (a) and high risk (b) AML.

a) Outcome A) Outcome

RI % RI % ±s.d.

NRM NRM % % ±s.d.

Haplo MSD HR (95% CI) P

18±6 21±5 1.04 (0.65-1.66) 0.86

26±8 10±4 2.59 (1.59-4.20) <0.01

B) Outcome

RI % ±s.d.

NRM % ±s.d.

Haplo MSD HR (95% CI) P

22±11 39±10 0.53 (0.28-1.00) 0.05

17±10 13±7 1.07 (0.45-2.51) 0.87

LFS % LFS % ±s.d.

OS % OS % ±s.d.

56±8 68±9 69±6 79±5 1.60 (1.15-2.22) 1.60 (1.12-2.29) <0.01 <0.01

LFS % ±s.d.

OS % ±s.d.

61±13 67±13 48±10 57±9 0.68 (0.40-1.13) 0.68 (0.39-1.19) 0.14 0.18

Gr.II-IV aGvHD % Gr. II-IV aGvHD % ±s.d.

cGvHD% cGvHD% ±s.d.

GRFS% GRFS% ±s.d.

29±7 21±5 1.49 (0.95-2.31) 0.07

30±9 35±6 0.82 (0.54-1.24) 0.37

45±10 53±7 1.27 (0.94-1.71) 0.11

Gr. II-IV aGvHD % ±s.d.

cGvHD% ±s.d.

GRFS% ±s.d.

37±12 21±7 2.06 (1.13-3.74) 0.01

37±13 31±10 0.98 (0.54-1.77) 0.95

51±13 41±10 0.82 (0.52-1.28) 0.39

RI: relapse incidence; NRM: non-relapse mortality; LFS: leukemia-free survival; OS: overall survival; GRFS: refined graft-versus-host-free relapse free survival; Gr. II-IV aGvHD: grade IIIV acute graft-versus-host disease; cGvHD: chronic graft-versus-host disease; HAPLO: haploidentical donor; MSD: matched sibling donor; HR: hazard ratio; CI: confidence interval.

reconstitution in Haplo setting, also favored by the use of additional high doses of immunosuppressive agents as compared to MSD. However, as ours is a registry-based study, details on type of infections were not available. Importantly, the type of donor did not influence the risk of relapse in intermediate AML. Recently, Ringden et al.37 published no difference in leukemic relapse between MSD and Haplo. On the other hand, in high-risk AML, we found a trend for higher RI in MSD recipients; this could reflect a lower immunogenicity of MSD transplant in AML with more biological aggressive characteristics. Our results should be taken with caution as there are important factors that we were not able to take into account, such as molecular biology data, important for disease stratification. Risk group was, indeed, defined according to cytogenetics at diagnosis. In intermediate AML, a RIC regimen was associated to higher risk of relapse as previously described,35 while in high-risk AML, the type of conditioning regimen affected neither relapse nor GvHD incidence In this setting, CMV serology and incremental age were the only factors affecting risk of relapse, while the type of donor was the only related to risk of GvHD. The probability of LFS was lower in Haplo, in line with previous reports.37 In a retrospective study from a single center, Bashey et al.17 reported outcomes of 475 patients receiving unmanipulated Haplo transplant using PT-CY in comparison to MSD or 10/10 matched unrelated donors. This series on patients with lymphoid and myeloid malignancies included 170 patients with AML. In line with our results, OS was superior in MSD as compared to Haplo recipients. Of

References 1. Kanakry CG, de Lima MJ, Luznik L. Alternative donor allogeneic hematopoietic cell transplantation for acute myeloid leukemia. Semin Hematol. 2015; 52(3):232242. 2. Clift RA, Hansen JA, Thomas ED, et al. Marrow transplantation from donors other

haematologica | 2018; 103(8)

note, they also found higher incidence of grade II-IV aGvHD, without differences in cGvHD, and higher NRM in the Haplo setting. In our study, as in Bashey population, the time from diagnosis to transplant was longer for Haplo than MSD and this could have negatively affected outcomes of transplant. In multivariate analysis, incremental age produced effects on LFS and OS, regardless of cytogenetics, in line with others.40 Our data were analyzed using the propensity score analysis in order to balance characteristics of the two populations. The matched pair analysis confirmed the results of higher aGvHD incidence in Haplo compared to MSD, and confirmed the main outcome results that we found in standard analysis, for both intermediate and high risk AML. Given the main finding of our study, outcomes of transplantation from Haplo versus MSD depend on the leukemic cytogenetics risk. Intermediate AML outcomes were better in the MSD setting as compared to Haplo with no significant differences in RI among the two types of donor. Whilst in high-risk AML, there were no significant differences in the main transplantation outcomes between Haplo and MSD, except for the lower risk of relapse in the Haplo group. However, we acknowledge that the number of patients with high risk cytogenetics in our study was low and, consequently, the statistical power was too. In conclusion, our results underline that matched sibling donor remain the first donor choice for AML patients in first CR when available. It should be of interest to further investigate the role of Haplo in this setting with welldesigned prospective studies.

than HLA-identical siblings. Transplantation. 1979;28(3):235-242. 3. Powles RL, Morgenstern GR, Kay HE, et al. Mismatched family donors for bone-marrow transplantation as treatment for acute leukaemia. Lancet. 1983;1(8325):612-615. 4. Gragert L, Eapen M, Williams E, et al. HLA match likelihoods for hematopoietic stemcell grafts in the U.S. Registry. N Engl J Med. 2014;371(4):339-348.

5. Lown RN, Shaw BE. Beating the odds: factors implicated in the speed and availability of unrelated haematopoietic cell donor provision. Bone Marrow Transplant. 2013; 48(2):210-219. 6. Scaradavou A, Brunstein CG, Eapen M, et al. Double-unit grafts successfully extend the application of umbilical cord blood transplantation in adults with acute leukemia. Blood. 2013;121(5):752-758.

1327

D. Salvatore et al. 7. Aversa F. Haploidentical haematopoietic stem cell transplantation for acute leukaemia in adults: experience in Europe and the United States. Bone Marrow Transplant. 2008;41(5):473-481. 8. Raiola AM , Dominietto A , di Grazia C, et al. Unmanipulated haploidentical transplants compared with other alternative donors and matched sibling grafts. Biol Blood Marrow Transplant. 2014;20 (10):1573-1579. 9. Ciceri F, Labopin M, Aversa F, et al. A survey of fully haploidentical hematopoietic stem cell transplantation in adults with high-risk acute leukemia: a risk factor analysis of outcomes for patients in remission at transplantation. Blood. 2008; 112(9):3574-3581. 10. Mancusi A, Ruggeri L, Velardi A. Haploidentical hematopoietic transplantation for the cure of leukemia: from its biology to clinical translation. Blood. 2016; 128(23):2616-2623. 11. Mehta J, Singhal S, Gee AP, et al. Bone marrow transplantation from partially HLAmismatched family donors for acute leukemia: single-center experience of 201 patients. Bone Marrow Transplant. 2004; 33(4):389–396. 12. Luznik L, O'Donnell PV, Symons HJ, et al. HLA-haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant. 2008;14(6):641-650. 13. Lu DP, Dong L, Wu T, et al. Conditioning including antithymocyte globulin followed by unmanipulated HLA-mismatched/haploidentical blood and marrow transplantation can achieve comparable outcomes with HLA-identical sibling transplantation. Blood. 2006;107(8):3065-3073. 14. Ciurea SO, Zhang MJ, Bacigalupo A, et al. Haploidentical transplant with posttransplant cyclophosphamide vs matched unrelated donor transplant for acute myeloid leukemia. Blood. 2015;126(8):1033-1040. 15. Passweg JR, Baldomero H, Bader P, et al. Use of haploidentical stem cell transplantation continues to increase: the 2015 European Society for Blood and Marrow Transplant activity survey report. Bone Marrow Transplant. 2017;52(6):811-817. 16. 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. 17. Bashey A, Zhang X, Jackson K, et al. Comparison of outcomes of hematopoietic cell transplants from T-replete haploidenti-

1328

18.

19.

20.

21.

22.

23.

24.

25. 26. 27.

28.

29.

cal donors using post-transplantation Cyclophosphamide with 10 of 10 HLA-A, B, -C, -DRB1, and -DQB1 allele-matched unrelated donors and HLA-identical sibling donors: a multivariable analysis including disease risk index. Biol Blood Marrow Transplant. 2016;22(1):125-133. Di Stasi A, Milton DR, Poon LM, et al. Similar transplant outcomes for AML/MDS patients with haploidentical versus 10/10 HLA matched unrelated and related donors. Biol Blood Marrow Transplant. 2014;20(12):1975–1981. Wang Y, Liu Q, Xu L, et al. Haploidentical vs identical-sibling transplant for AML in remission: a multicenter, prospective study. Blood. 2015;125(25):3956-3962. Yoon JH, Kim HJ, Park SS, et al. Long term clinical outcomes of hematopoietic cell transplantation for intermediate to poor risk acute myeloid leukemia during first remission according to available donor types. Oncotarget. 2017;8(25):4159041604. Grimwade D, Hills RK, Moorman V, et al. Refinement of cytogenetic classification in acute myeloid leukaemia: Determination of prognostic significance of rarer recurring chromosomal abnormalities amongst 5,876 younger adult patients treated in the UK Medical Research Council trials. Blood. 2010;116(3):354-365. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant. 2009; 15(12):16281633. Ruggeri A, Labopin M, Ciceri F, et al. Definition of GvHD-free, relapse-free survival for registry-based studies: an ALWP– EBMT analysis on patients with AML in remission. Bone Marrow Transplant. 2016;51(4):610-611. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus conference on AGvHD grading. Bone Marrow Transplant. 1995;15(6):825-828. Lee SJ, Vogelsang G, Flowers ME. Chronic graft versus host disease. Biol Blood Marrow Transplant. 2003;9(4):215-233. Kaplan EL, Mayer P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 2008;53(282):457-481. Fine JP, Gray RJ. A Proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 2009; 446(94):496-509. Ho D, Imai K, King G, et al. Matching as non-parametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal. 2007;15(3):199-236. Ho ED, Imai K, King G, Stuart EA. MatchIt:

30. 31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

Nonparametric Preprocessing for Parametric Causal Inference. J Stat Softw. 2011;42(8):1-28. Ringdén O, Karlsson H, Olsson R, et al. The allogeneic graft-versus-cancer effect. Br J Haematol. 2009;147(5):614-633. Weiden PL, Sullivan KM, Flournoy N, et al. Antileukemic effect of chronic graft-versushost disease: contribution to improved survival afterallogeneic marrow transplantation. N Engl J Med. 1981;304(25):15291533. Cornelissen JJ, Gratwohl A, Schlenk RF, et al. The European LeukemiaNet AML Working Party consensus statement on allogeneic HSCT for patients with AML in remission: an integrated risk adapted approach. Nat Rev Clin Oncol. 2012; 9(10):579-590. Miller W, Flynn P, McCullough J, et al. Cytomegalovirus infection after bone marrow transplantation: an association with acute graft-versus-host disease. Blood. 1986;67(4):1162-1167. Ljungman P, Perez-Bercoff L, Jonsson J, et al. Risk factors for the development of cytomegalovirus disease after allogeneic stem cell transplantation. Haematologica. 2006;91(1):78-83. Abdul Wahid SF, Ismail NA, Mohd-Idris MR, et al. Comparison of reduced-intensity and myeloablative conditioning regimens for allogeneic hematopoietic stem cell transplantation in patients with acute myeloid Leukemia and acute lymphoblastic Leukemia: a meta-analysis. Stem Cells Dev. 2014; 23(21):2535-2352. Nannya Y, Kataoka K, Hangaishi A, et al. The negative impact of female donor/male recipient combination in allogeneic hematopoietic stem cell transplantation depends on disease risk. Transpl Int. 2011; 24(5):469-476. Ringdén O, Labopin M, Ciceri F, et al. Is there a stronger graft-versus-leukemia effect using HLA-haploidentical donors compared with HLA-identical sibling? Leukemia. 2016;30(2):447-455. Marmont AM, Horowitz MM, Gale RP, et al. T-cell depletion of HLA-identical transplants in leukemia. Blood. 1991;78(8):21202130. Ringden O, Horowitz MM, Sondel P, et al. Methotrexate, cyclosporine, or both to prevent graft-versus-host disease after HLAidentical sibling bone marrow transplants for early leukemia? Blood. 1993;81(4):10941101. Yanada M, Emi N, Naoe T, et al. Allogeneic myeloablative transplantation for patients aged 50 years and over. Bone Marrow Transplant. 2004;34(1):29-35.

haematologica | 2018; 103(8)

ARTICLE

Non-Hodgkin Lymphoma

Novel GPR34 and CCR6 mutation and distinct genetic profiles in MALT lymphomas of different sites

Ferrata Storti Foundation

Sarah Moody,1 Joe Sneath Thompson,1 Shih-Sung Chuang,2 Hongxiang Liu,3 Markus Raderer,4 George Vassiliou,5 Iwona Wlodarska,6 Fangtian Wu,1 Sergio Cogliatti,7 Alistair Robson,8 Margaret Ashton-Key,9 Yingwen Bi,10 John Goodlad11 and Ming-Qing Du1,3,12

Division of Cellular and Molecular Pathology, Department of Pathology, University of Cambridge, UK; 2Department of Pathology, Chi-Mei Medical Centre, Tainan, Taiwan; 3 Molecular Malignancy Laboratory, Addenbrooke's Hospital, Cambridge University Hospitals NHS Foundation Trust, UK; 4Department of Medicine I, Clinical Division of Oncology, Medical University of Vienna, Austria; 5The Wellcome Trust Sanger Institute, Wellcome Genome Campus, Hinxton, Cambridge, UK; 6Center for Human Genetics, KU Leuven, Belgium; 7Institute of Pathology, State Hospital St. Gallen, Switzerland; 8 Department of Dermatopathology, St John's Institute of Dermatology, London, UK; 9 Department of Cellular Pathology, Southampton University Hospitals National Health Service Trust, UK; 10Department of Pathology, Eye & ENT Hospital, Fudan University, Shanghai, PR China; 11Department of Pathology, Western General Hospital, NHS Lothian University Hospitals Trust, Edinburgh, UK and 12Department of Histopathology, Addenbrooke's Hospital, Cambridge University Hospitals NHS Foundation Trust, UK 1

Haematologica 2018 Volume 103(8):1329-1336

ABSTRACT

ucosa-associated lymphoid tissue (MALT) lymphoma originates from a background of diverse chronic inflammatory disorders at various anatomic sites. The genetics underlying its development, particularly in those associated with autoimmune disorders, is poorly characterized. By whole exome sequencing of 21 cases of MALT lymphomas of the salivary gland and thyroid, we have identified recurrent somatic mutations in 2 G-protein coupled receptors (GPR34 and CCR6) not previously reported in human malignancies, 3 genes (PIK3CD, TET2, TNFRSF14) not previously implicated in MALT lymphoma, and a further 2 genes (TBL1XR1, NOTCH1) recently described in MALT lymphoma. The majority of mutations in GPR34 and CCR6 were nonsense and frameshift changes clustered in the C-terminal cytoplasmic tail, and would result in truncated proteins that lack the phosphorylation motif important for b-arrestin-mediated receptor desensitization and internalization. Screening of these newly identified mutations, together with previously defined genetic changes, revealed distinct mutation profiles in MALT lymphoma of various sites, with those of salivary gland characterized by frequent TBL1XR1 and GPR34 mutations, thyroid by frequent TET2, TNFRSF14 and PIK3CD mutations, and ocular adnexa by frequent TNFAIP3 mutation. Interestingly, in MALT lymphoma of the salivary gland, there was a significant positive association between TBL1XR1 mutation and GPR34 mutation/translocation (P=0.0002). In those of ocular adnexa, TBL1XR1 mutation was mutually exclusive from TNFAIP3 mutation (P=0.049), but significantly associated with IGHV3-23 usage (P=0.03) and PIK3CD mutation (P=0.009). These findings unravel novel insights into the molecular mechanisms of MALT lymphoma and provide further evidence for potential oncogenic co-operation between receptor signaling and genetic changes.

haematologica | 2018; 103(8)

Correspondence: mqd20@cam.ac.uk

Received: February 19, 2018. Accepted: April 18, 2018. Pre-published: April 19, 2018.

doi:10.3324/haematol.2018.191601 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1329 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.

1329

S. Moody et al.

Introduction

Methods

Extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma) originates from acquired MALT at diverse anatomic sites. Accordingly, its development is closely associated with distinct inflammatory disorders. For example, MALT lymphomas of the stomach and ocular adnexa are associated with chronic infection by Helicobacter pylori and Chlamydia psittaci, respectively, while those of the salivary gland and thyroid are associated with Sjögren syndrome and Hashimoto thyroiditis, respectively.1 In addition to its diverse etiology, MALT lymphomas of various sites also show distinct biased usage of IGHV. For instance, up to one-third of ocular adnexal MALT lymphomas have IGHV4-34 BCR that is most likely auto-reactive, binding to the carbohydrate I/i antigens, while more than 50% of salivary gland MALT lymphomas bear IGHV1-69 BCR with features of rheumatoid factors.2-9 These findings suggest that various chronic inflammatory disorders may generate immune responses, which preferentially expand B cells with certain properties, such as autoreactive BCR, consequently leading to the development of MALT lymphoma.1,10 There are a number of genetic changes described in MALT lymphomas, but many of these changes occur at considerable variable frequencies at different sites despite the fact that they commonly affect the NF-κB pathway.1 For example, t(11;18)(q21;q21)/API2-MALT1 and to a lesser extent t(1;14)(p22;q32)/BCL10-IGH, are frequently seen in MALT lymphoma of the lung and stomach, but are rarely or not seen in those of the salivary gland and thyroid.11-13 In contrast, TNFAIP3 mutation and/or deletion, which encodes a NF-κB negative regulator, is frequent in MALT lymphoma of the ocular adnexa, but occurs at low frequencies in those of other sites.14-16 The reason for such dramatic differences in the genetics of MALT lymphomas of different sites is unclear, and it remains to be investigated whether the occurrence of these genetic changes is influenced by the distinct background disorders or vice versa. The development of MALT lymphoma is the result of oncogenic co-operation between immunological drive and acquired genetic changes as neither the above genetic changes nor the immunological drive alone is sufficient for malignant transformation. Nevertheless, the exact mechanisms of oncogenic co-operation in MALT lymphoma of various sites remain largely elusive. In a recent study, we found a significant association between TNFAIP3 inactivation and biased IGHV4-34 usage in ocular adnexal MALT lymphoma, arguing for their co-operation in sustaining chronic BCR signaling, and thus NF-κB activation.17 Interestingly, a very similar finding, namely a significant association between biased usage of IGHV1-2 and inactivating mutation of KLF2 (a negative regulator of NF-κB), was previously reported in splenic marginal zone lymphoma.18 Taken together, such co-operation between antigenic stimulation and genetic abnormalities may represent a common feature in marginal zone B-cell lymphoma. Nonetheless, this has not been fully investigated in marginal zone lymphoma of various sites, particularly in those of the salivary gland and thyroid, where very few genetic features have been identified. In the present study, we first investigated the mutation profile of 21 MALT lymphomas of the salivary gland and thyroid by whole exome sequencing (WES), then validated the recurrent mutations in MALT lymphoma of various sites by targeted sequencing, and finally investigated their association with IGHV usage.

Patients' samples and DNA exaction

1330

A total of 249 cases of MALT lymphoma were included in this study, originating from the ocular adnexa (n=115), salivary gland (n=58), stomach (n=36), thyroid (n=13), lung (n=13), and other sites (n=14); 179 of these cases had been the subject of a previous study for somatic mutation in 17 genes.17 Genomic DNA was extracted from tumor rich areas (>40%) of formalin fixed paraffin embedded (FFPE) lymphoma biopsies in 217 cases, and where possible from non-neoplastic cells by microdissection using the QIAamp DNA micro kit (Qiagen, the Netherlands). Additionally, high molecular weight DNA was available in 32 cases of MALT lymphoma. DNA quality was assessed by PCR amplification of variably sized genomic fragments, with those amenable to PCR of ≥300bp genomic fragment used for mutation analyses by targeted sequencing.19 Among the cases included in this study, clinical information such as site involvement and treatment details were available in 98 cases of ocular adnexal MALT lymphomas, and their correlation with genetic changes was examined. Local ethical guidelines for the use of archival tissues for research were adopted with the approval of the ethics committees of the institutions involved.

Whole exome sequencing A total of 21 MALT lymphomas (14 from salivary gland, 7 from thyroid) together with one matched non-neoplastic biopsy were investigated by whole exome sequencing (WES) (Online Supplementary Table S1). The initial 2 cases (one thyroid and one salivary gland MALT lymphoma) were carried out on high molecular weight tumor DNA at the Eastern Sequence and Informatics Hub (EASIH). Genomic libraries were prepared using the Illumina Truseq DNA sample preparation v.2 kit (Illumina, CA, USA), captured with the Nimblegen SeqCap EZ Exome v.3 (Roche, Germany) and sequenced on an Illumina HiSeq platform. The remaining 19 cases (6 thyroid and 13 salivary gland MALT lymphoma) together with one matched non-neoplastic sample were performed on FFPE tissue DNA at the Wellcome Trust Sanger Institute. Genomic libraries were generated using the Illumina Paired End Sample Prep Kit, enriched using the Agilent SureSelect Human All Exon 50Mb kit (Agilent, CA, USA), and sequenced on an Illumina HiSeq platform using a 75bp paired end protocol. Sequencing reads were aligned to the hg19 reference genome using BWA with default settings, with single nucleotide variants called by CaVEMan (Cancer Variants through Expectation Maximization), and insertions and deletions by PINDEL. Data were filtered against 300 unmatched normal controls available from the Cancer Genome Project in addition to the 1000 genomes project and Ensembl variation databases to remove known single nucleotide polymorphisms (SNP).

Somatic variant validation by fluidigm access array PCR and Illumina MiSeq sequencing Where indicated, the novel and shortlisted variants identified by WES were confirmed by PCR and Sanger Sequencing, and their somatic origin ascertained by PCR and Sanger sequencing analysis of DNA samples from microdissected non-neoplastic cells. Mutations in CCR6, FGFR3, FOXO1, GPR34, IKBKB, PIK3CD, NOTCH1, TBL1XR1, TET2 and TNFRSF14 were then screened using the Fluidigm Access Array PCR and Illumina MiSeq protocol, as previously described.19 Primer sequences and PCR conditions are shown in Online Supplementary Table S2. In addition, mutation in a further 17 genes including BCR signaling (CD79A, CD79B, CARD11), NF-κB (TNFRSF11A, TNFAIP3, TRAF3), TLR signaling (MYD88), plasma cell differentiation (PRDM1), histone modifiers haematologica | 2018; 103(8)

Novel GPCR mutation in MALT lymphoma

(CREBBP, EP300, EZH2, MLL2, MEF2B, KDM2B), antigen presentation (B2M, CD58), and apoptosis (TP53) were similarly screened in the cases (n=70) that were not investigated previously.17 Briefly, DNA samples were sequenced in duplicate and analyzed using an in-house variant calling protocol optimized against a large panel of known mutations.19 Variants that appeared in both replicates at â&#x2030;Ľ10% AAF (alternative allele frequency) were further validated by inspection of their bam files, and then considered as true genetic changes.19 Where possible, variants were confirmed as somatic by PCR and Sanger sequencing of non-neoplastic DNA or assumed to be somatic if previously reported as somatic in the COSMIC database or other samples in the cohort.

Fluorescence in situ hybridization analysis Fluorescence in situ hybridization (FISH) analysis was performed on FFPE tissue sections with Vysis LSI IGH Dual Colour Break Apart Probe (Abbott Molecular, IL, USA) and an in house GPR34 Dual Colour Break Apart probe, as previously described.16,20 To generate the in-house GPR34 break-apart probe, BAC clones RP11-643E21 and RP11-524P6 centromeric to GPR34, and RP11360E17 and CTD-3202J9 telemeric to GPR34 were amplified using the Templiphi kit (GE Healthcare, IL, USA) and then labeled with SpectrumOrange and SpectrumGreen respectively using nick translation with random priming (Abbott Molecular, IL, USA).16,20

Sequencing analysis of the rearranged immunoglobulin heavy chain genes The sequence of clonally rearranged immunoglobulin heavy chain genes was available in 94 cases from a previous study.17 Additional PCR and sequencing of the rearranged immunoglobulin heavy chain genes were performed for the cases investigated by WES using the BIOMED-2 VH FR1-JH or FR2-JH primers, as described previuosly.17 Sequences were analyzed using IMGT/VQuest software (www.imgt.org/IMGT_vquest/ vquest), and successfully annotated in a total of 101 cases (Online Supplementary Table S3).

Statistical analysis Fishers exact test was used to test for associations between categorical variables.

Results Novel mutations in MALT lymphomas identified by WES Whole exome sequencing was successful in a total of 21 MALT lymphomas (14 from salivary gland, 7 from thyroid) together with one matched non-neoplastic DNA (Online Supplementary Table S4). After filtering known SNPs, synonymous changes and benign variants by Polyphen2, 72 variants were seen in the case with matched germline DNA, while an average of 111 variants (range 46-264/case) were observed in the remaining cases without matched germline DNA (Online Supplementary Figure S4). Based on their frequencies, potential functional impact, involvement in cancer and lymphocyte biology and hits against relevant GO terms, novel variants in 10 genes were selected for validation and their somatic origin was confirmed by Sanger sequencing of corresponding non-neoplastic DNA (Online Supplementary Table S5). These shortlisted genes included 2 G protein-coupled receptors (GPR34, CCR6) not yet reported as a mutation target in human malignancies, 6 (FGFR3, FOXO1, IKBKB, PIK3CD, TET2, TNFRSF14) not yet implicated in MALT lymphoma, and 2 (TBL1XR1, NOTCH1) recently described in MALT lymphoma (Online haematologica | 2018; 103(8)

Supplementary Table S4 and Online Supplementary Figure S2).21 All of these mutations identified by WES were potentially pathogenic (see section below). To further investigate their mutation frequency and characteristics in MALT lymphoma, we screened a large cohort from various anatomic sites for mutations by Fluidigm Access Array PCR and Illumina MiSeq sequencing.

Variable involvement of newly identified mutations in MALT lymphoma of different sites A total of 249 cases of MALT lymphoma from the ocular adnexa (n=115), salivary gland (n=58), stomach (n=36), thyroid (n=13), lung (n=13), and other sites (n=14) were investigated for somatic mutations in 27 genes with the 10 genes mentioned above investigated exclusively by the present study, and 17 genes studied by previous (n=149) and present (n=70) studies, respectively (Online Supplementary Tables S6 and S7).17 Recurrent genetic changes are shown in Figure 1 (Online Supplementary Figure S3). Interestingly, the findings showed remarkable variations in their involvement among MALT lymphoma of different sites. GPR34 mutation was almost exclusively found in MALT lymphoma of salivary gland (9 of 56, 16%), with the exception of a single case from ocular adnexa (Figure 1). The majority of GPR34 mutations were nonsense changes (n=6) and frameshift indel (n=1) that were clustered in the C-terminal regulatory regions, resulting in truncated proteins (Figure 2). The remaining 3 mutations were missense changes, namely R84H, D151A and Y327N. GPR34 is a G-protein coupled receptor (GPCR) and is deregulated by t(X:14)(p11.4;q32) in MALT lymphoma of salivary gland and lung with a history of SjĂśgren syndrome.20,22,23 In view of this finding, we screened salivary gland MALT lymphoma for t(X;14)(p11;q32) by interphase FISH, and identified the translocation in 2 of 58 (3%) cases (Online Supplementary Figure S4). Interestingly, GPR34 mutation and translocation were mutually exclusive, together seen in 11 of 58 (19%) cases of salivary gland MALT lymphoma (Figure 1 and Online Supplementary Figure S4). TBL1XR1 mutation was most frequent in MALT lymphoma of salivary gland (14 of 58, 24%), occurring less frequently in MALT lymphomas of the stomach (3 of 36, 8%), lung (1 of 13, 8%), ocular adnexa (7 of 115, 6%), skin (1 of 8, 13%) and in a single case from tonsil, but not in the thyroid (Figure 1). A total of 30 mutations were found in 27 cases, with 3 cases affected by two mutations. The majority of these mutations (20 missense changes and 5 inframe deletions) were within the WD40 domains and frequently affected regions or residues of structural importance, which are likely critical for interaction with NCoR (Figure 3). The remaining 5 mutations included 3 substitution changes at exon 11 splice site, one frameshift deletion and one nonsense change, most likely resulting in truncated proteins with potential increased binding to UbcH5.24 TET2 mutation was found in the majority of MALT lymphomas of the thyroid (8 of 13, 62%), and also recurrently seen in those of salivary gland (5 of 58, 9%), ocular adnexa (5 of 115, 4%), stomach (3 of 3, 8%), and lung (1 of 13, 8%) (Figure 1). A total of 31 TET2 mutations were detected in 23 cases, with 8 cases showing multiple mutations (Figure 2). Interestingly, 7 of 8 of the cases with multiple TET2 mutations were MALT lymphomas of the thyroid. The vast majority of these mutations were deleterious changes including 11 frameshift indels, 4 nonsense, one splice site, 1331

S. Moody et al.

and 15 missense mutations. Of these, the majority of frameshift (8 of 11) and nonsense (3 of 4) mutations were found in MALT lymphomas of the thyroid. To our surprise, TNFRSF14 mutation was highly frequent in MALT lymphoma of thyroid (6 of 13, 46%), albeit infrequent in those of other sites (Figure 1). A total of 21 TNFRSF14 mutations were detected in 18 cases, and all mutations were deleterious changes comprising 6 nonsense, 2 frameshift indel, one inframe deletion, 11 missense changes including 2 in the first codon thus abolishing the translation start site, and a single splice site mutation (Figure 2). As TNFRSF14 mutation is common in pediatric-type follicular lymphoma and the mutations identified are of a similar nature,25 we carefully reviewed the histopathology and immunophenotype of our cases with TNFRSF14 mutations, where necessary performing further immunohistochemistry for CD10 and BCL6, and also FISH for BCL2 and BCL6 translocation. These additional analyses confirmed the original diagnosis of MALT lymphoma in these cases. PIK3CD was another gene frequently mutated in MALT lymphoma of thyroid (3 of 13, 23%), followed by salivary gland (5 of 58, 9%), but rarely in those of other sites (Figure 1). A total of 11 PIK3CD mutations were identified in 11 MALT lymphomas, and the vast majority of these mutations were missense changes including N334K and E1021K, previously reported in patients with activated PI3KÎ´ syndrome (hyper IgM syndrome) and in DLBCL (Figure 2).26-29 Mutation in the remaining genes was relevantly infre-

quent, without any apparent bias among the sites examined. Mutation in CCR6 was recurrent, and seen in a total of 7 MALT lymphomas, with 7 mutations including 5 deleterious changes (3 nonsense, 3 frameshift indels) and 2 missense changes, largely clustered in the C-terminal regulatory region (Figures 1 and 2). Interestingly, mutations in GPR34 and CCR6, both members of the GPCR family, were mutually exclusive.

Distinct association of genetic changes in MALT lymphoma of various sites Correlation analysis revealed several distinct associations among genetic changes in MALT lymphoma of different sites. In salivary gland MALT lymphoma, there was a significant association between TBL1XR1 mutation and GPR34 mutation/translocation (P=0.0002), with TBL1XR1 mutation seen in 8 of 11 (73%) cases with GPR34 mutation/translocation, but only in 6 of 47 (13%) cases without GPR34 genetic abnormalities (Figure 4 and Online Supplementary Figure S5). Both GPR34 and TBL1XR1 genetic changes were more frequently seen in the cases with nonIGHV1-69 than those with IGHV1-69 rearrangement (29% vs. 9%; 57% vs. 18%, respectively) although this does not reach statistical significance. In ocular adnexal MALT lymphoma, TBL1XR1 mutation was mutually exclusive from TNFAIP3 mutation (P=0.049), but significantly associated with PIK3CD mutation (P=0.009), with TBL1XR1 mutation seen in 2 of 3 (67%)

Figure 1. Distinct mutation profiles in mucosa-associated lymphoid tissue (MALT) lymphoma of various sites. A total of 27 genes including the 10 MALT lymphomaassociated genes indentified in the present study were screened for mutation by targeted sequencing, but only those showing recurrent changes are presented here. API2-MALT1 denotes t(11;18)(q21;q21) from previous studies.11,12 GPR34 results in MALT lymphoma of the salivary gland include both mutation and t(X;14)(p11;q32)/GPR34-IGH. MALT lymphomas of the salivary gland are featured by frequent TBL1XR1 and GPR34 mutations, while those of the thyroid by frequent TET2, TNFRSF14 and PIK3CD mutations, and ocular adnexa by frequent TNFAIP3 mutation.

1332

haematologica | 2018; 103(8)

Novel GPCR mutation in MALT lymphoma

cases with PIK3CD mutation, but in only 5 of 112 (4%) cases without PIK3CD mutation (Figure 4 and Online Supplementary Figure S5). TBL1XR1 mutation was also mutually exclusive from IGHV4-34 rearrangement, but significantly associated with IGHV3-23 usage (P=0.03), with TBL1XR1 mutation seen in 3 of 11 (27%) cases with IGHV3-23 rearrangement, but in only 2 of 52 (4%) cases with other rearrangements (Figure 4). MYD88 mutation also appeared to be more frequent in TBL1XR1 mutated cases (2 of 7, 29%) compared to non-mutated cases (6 of 108, 6%), although this did not reach significance (P=0.07). In gastric MALT lymphoma, t(11;18) API2-MALT1 was mutually exclusive from other genetic changes (P=0.025) (Online Supplementary Figure S5).

Correlation between genetic changes and clinicopathological parameters This was carried out in 98 cases of ocular adnexal MALT lymphoma where clinicopathological data were available (Online Supplementary Table S8). TBL1XR1 mutation and IGHV3-23 usage were significantly associated with involvement of the conjunctiva (P=0.002 in each, respectively), while TNFRSF14 mutation was significantly associated with involvement of the orbit (P=0.04). MYD88 and TP53 mutations were much higher in cases involving both orbit and conjunctiva than those involving only a single site but

only the former showing a statistical significance (4 of 23, 17% vs. 2 of 75, 3%; P=0.03). With the exception of TNFAIP3 deletion/mutation that was significantly associated with higher radiation dosages to achieve complete remission as reported previously,16 none of the other genetic changes showed any significant association with radiation dosages.

Discussion By WES of 21 cases of MALT lymphomas of the salivary gland and thyroid, we have identified several novel genetic changes in MALT lymphoma, including GPR34 and CCR6 mutations not yet previously reported in human malignancies. Screening these newly identified mutations revealed distinct mutation profiles in MALT lymphoma of various sites, and provides further evidence of potential oncogenic co-operation between receptor signaling and genetic changes.

GPR34 and CCR6 mutations in MALT lymphoma Both GPR34 and CCR6 are members of class A G protein-coupled receptor (GPCR) superfamily, which transduce extracellular stimulation into intracellular signaling through G protein and b-arrestin. By interacting with G protein,

Figure 2. Nature and distribution of mutations in newly identified mucosa-associated lymphoid tissue (MALT) lymphoma-associated genes. Site of MALT lymphoma in which the mutation was identified is indicated by color (blue: salivary gland; red: thyroid; green: ocular adnexa; orange: stomach; purple: lung; gray: other sites). TM: transmembrane domain; ABD: adaptor binding domain; RBD: ras binding domain; DSBH: double-stranded b-helix.

haematologica | 2018; 103(8)

1333

S. Moody et al.

GPCR can activate diverse signaling pathways critical for cellular function. This receptor signaling is tightly regulated through a process known as desensitization that is mediated by GPCR phosphorylation, allowing its binding to barrestin which triggers receptor internalization, consequently dampening intracellular signaling.30 A common phosphorylation code (motif) has been identified in the Cterminal cytoplasmic tail of most GPCRs, and its phosphorylation enables electrostatic interactions with b-arrestin.31 GPR34 contains 7 transmembrane domains, followed by a C-terminal cytoplasmic tail. The majority of GPR34 mutations identified in MALT lymphoma are nonsense changes or frameshift indels that are clustered in its C-terminal region, resulting in truncated proteins. A phosphorylation code (motif) (S351S353T356) is seen in the C-terminal cytoplasmic tail of GPR34 according to Zhou et al.,31 and all the above nonsense and frameshift changes would eliminate or impair these phosphorylation sites, thus potentially deregulating the receptor desensitization process. The remaining GPR34 mutations are missense changes including R84H, D151A and Y327N. R84H affects the tribasic motif (RKR) in the first intracellular loop, which is the key topogenic signal determining the orientation of the first transmembrane domain.32 D151A affects the highly conserved E/DRY motif and is predicted to cause the receptor constitutive activation.33 Y327N is close to the interface of the seventh TM domain and the cytoplasmic tails, and is predicted to be highly damaging by PolyPhen-2, although its potential functional impact is unclear. Similarly, CCR6 also contains 7 transmembrane domains, followed by a C-terminal cytoplasmic tail, and the majority of CCR6 mutations are nonsense or frameshift changes that are clustered in its C-terminal region. A phosphorylation code (motif) (S357T360T363) is present in the C-terminal cytoplasmic tail of CCR6,31 and all the nonsense and frameshift changes identified would eliminate these phosphorylation sites, potentially impairing the receptor desensitization process. The functional impact of the two CCR6 missense

mutations (R159S and Y352C) is unclear. The nature and distribution of mutations in GPR34 and CCR6 identified in this study are very similar to those seen in CXCR4 in WaldenstrĂśm macroglobulinemia and WHIM syndrome, as well as those found in CCR4 in adult T-cell leukemia/lymphoma, which have been shown to be gainof-function changes.34-36 Despite their similar mutation pattern, GPR34 and CCR6 have distinct ligands, and thus respond to different environmental cues, potentially explaining their differential involvement in MALT lymphoma of different sites. GPR34 has been shown to be triggered by lyso-phosphatidylserine, while CCR6 is activated by CCL20.37-39 Overexpression of wild-type GPR34 in vitro resulted in ERK, PI3K/AKT and PKC signaling, inducing AP1 and NF-ÎşB-mediated gene transcription.20,40 It still remains to be investigated whether the above GPR34 and CCR6 mutations impact similar downstream signaling pathways, and if they would enhance these intracellular signaling pathways through impaired receptor desensitization and internalization or via independent mechanisms.

Distinct genetic profile in MALT lymphoma of various sites The present study provides further evidence showing distinct genetic profiles among MALT lymphoma of various sites (Figure 1). MALT lymphoma of the salivary gland features frequent mutation of TBL1XR1 and GPR34, while those of the thyroid are characterized by frequent mutations in TET2, TNFRSF14 and PIK3CD. MALT lymphoma of the ocular adnexa is noted for frequent TNFAIP3 mutation. In contrast, MALT lymphomas of the stomach and lung are distinguished by a high prevalence of t(11;18)(q21;q21). The present study also unravels several distinct associations among genetic changes in MALT lymphoma. In MALT lymphoma of the salivary gland, there is a significant positive association between TBL1XR1 mutation and GPR34 mutation/translocation. The effect of TBL1XR1 mutations has not been fully characterized, but a previous

Figure 3. Nature and distribution of TBL1XR1 mutations in mucosa-associated lymphoid tissue (MALT) lymphoma, and their potential functional impact. (A) TBL1XR1 mutations identified in this study. The site of MALT lymphoma in which mutation was identified is indicated by color: red: salivary gland; red: thyroid; green: ocular adnexa; orange: stomach; purple: lung; gray: other sites. (B) Detailed analyses of TBL1XR1 mutations within the WD40 domains. TBL1XR1 mutations identified in the present and a previous study by Jung et al. are included in the analyses.41 The majority of TBL1XR1 mutations are localized in the regions of structural importance, such as HSDW tetrad indicated by arrows and the residues predicted to be top facing (WDSP predictor, indicated in red text). The approximate position in the HMM logo for WD40 repeats (Prosite PS00678) is shown below the sequence.

1334

haematologica | 2018; 103(8)

Novel GPCR mutation in MALT lymphoma

Figure 4. Significant association among genetic changes in mucosa-associated lymphoid tissue (MALT) lymphoma of the salivary gland and ocular adnexa. In both cohorts, samples lacking the described changes are not included. P-value in red: positive correlation; P-value in black: negative correlation; significant values are in bold; X: inactivating frameshift and nonsense mutations; D: deletion; s: splice site mutations; T: translocation.

report suggests that mutations enhance TBL1XR1 binding to NCoR and facilitate its degradation, consequently promoting NF-κB and JUN target gene expression.41 Thus, TBL1XR1 mutation may sustain GPR34 mediated NF-κB and AP1 activation via this mechanism. Why GPR34 mutation and translocation are preferentially associated with MALT lymphoma of the salivary gland is unclear. GPR34 has been shown to be activated by lysophosphatidylserine, which could be generated by hydrolysis of the exposed membrane lipids in apoptotic cells through phospholipase A.37,38 Among MALT lymphoma of various sites, the lymphoepithelial lesions are most prominent in those of salivary gland. It remains to be investigated whether such lymphoepithelial lesions may facilitate the production of lyso-phosphatidylserine, the ligand for GPR34, thereby facilitating the expansion and localization of B cells carrying GPR34 genetic changes surrounding the lymphoepithelial lesion. In MALT lymphoma of the ocular adnexa, TBL1XR1 mutation is mutually exclusive from TNFAIP3 mutation and IGHV4-34 usage, but significantly associated with IGHV323 usage and PIK3CD mutation, albeit with limited numbers of cases involved. Like IGHV4-34 rearrangements, the IGHV3-23 rearrangement seen in ocular adnexal MALT lymphoma also encodes autoreactive BCR as shown by recombinant antibody studies.5 In this context, the signifihaematologica | 2018; 103(8)

cant association between TBL1XR1 mutation and IGHV323 rearrangement is very much analogous to that between TNFAIP3 mutation and IGHV4-34 rearrangement reported previously,17 hence expanding the evidence supporting oncogenic co-operation between antigenic drive and acquired genetic changes in MALT lymphoma. Similarly, the significant association between TBL1XR1 and PIK3CD mutation may also signify oncogenic co-operation between the two genetic events. In summary, our study reveals several novel genetic changes including GPR34 and CCR6 mutations not yet reported in human malignancies, and provides further evidence of distinct mutation profiles in MALT lymphoma of various sites. These novel findings offer exciting prospects for further characterization of the molecular pathogenesis of MALT lymphoma. Acknowledgments The authors would like to thank David Withers, Cambridge Genomics Services and Graeme Clark (Medical Genetics) for technical assistance with sequencing. The research was supported by grants from Bloodwise (13006, 15002, 15019), UK, and the Kay Kendal Leukaemia Fund (KKL582), UK. SM was initially supported by a PhD studentship from the Medical Research Council, Department of Pathology, University of Cambridge and Addenbrooke’s Charitable Trust, UK. 1335

S. Moody et al.

References 1. Du MQ. MALT lymphoma: A paradigm of NF-kappaB dysregulation. Semin Cancer Biol. 2016;39:49-60. 2. van Maldegem F, Wormhoudt TA, Mulder MM, et al. Chlamydia psittaci-negative ocular adnexal marginal zone B-cell lymphomas have biased VH4-34 immunoglobulin gene expression and proliferate in a distinct inflammatory environment. Leukemia. 2012;26(7):1647-1653. 3. Mannami T, Yoshino T, Oshima K, et al. Clinical, histopathological, and immunogenetic analysis of ocular adnexal lymphoproliferative disorders: characterization of malt lymphoma and reactive lymphoid hyperplasia. Mod Pathol. 2001;14(7):641649. 4. Bahler DW, Szankasi P, Kulkarni S, et al. Use of similar immunoglobulin VH gene segments by MALT lymphomas of the ocular adnexa. Mod Pathol. 2009;22(6):833-838. 5. Zhu D, Bhatt S, Lu X, et al. Chlamydophila psittaci-negative ocular adnexal marginal zone lymphomas express self polyreactive B-cell receptors. Leukemia. 2015;29(7):15871599. 6. Dagklis A, Ponzoni M, Govi S, et al. Immunoglobulin gene repertoire in ocular adnexal lymphomas: hints on the nature of the antigenic stimulation. Leukemia. 2012;26(4):814-821. 7. Bende RJ, Aarts WM, Riedl RG, et al. Among B cell non-Hodgkin's lymphomas, MALT lymphomas express a unique antibody repertoire with frequent rheumatoid factor reactivity. J Exp Med. 2005;201 (8):1229-1241. 8. Bahler DW, Miklos JA, Swerdlow SH. Ongoing Ig gene hypermutation in salivary gland mucosa- associated lymphoid tissuetype lymphomas. Blood. 1997;89(9):33353344. 9. Miklos JA, Swerdlow SH, Bahler DW. Salivary gland mucosa-associated lymphoid tissue lymphoma immunoglobulin V(H) genes show frequent use of V1-69 with distinctive CDR3 features. Blood. 2000;95(12): 3878-3884. 10. Hamoudi RA, Appert A, Ye H, et al. Differential expression of NF-kappaB target genes in MALT lymphoma with and without chromosome translocation: insights into molecular mechanism. Leukemia. 2010;24(8):1487-1497. 11. Ye H, Liu H, Attygalle A, et al. Variable frequencies of t(11;18)(q21;q21) in MALT lymphomas of different sites: significant association with CagA strains of H pylori in gastric MALT lymphoma. Blood. 2003;102(3): 1012-1018. 12. Ye H, Gong L, Liu H, et al. MALT lymphoma with t(14;18)(q32;q21)/IGH-MALT1 is characterized by strong cytoplasmic MALT1 and BCL10 expression. J Pathol. 2005;205(3):293-301. 13. Streubel B, Simonitsch-Klupp I, Mullauer L, et al. Variable frequencies of MALT lymphoma-associated genetic aberrations in MALT lymphomas of different sites.

1336

Leukemia. 2004;18(10):1722-1726. 14. Chanudet E, Ye H, Ferry J, et al. A20 deletion is associated with copy number gain at the TNFA/B/C locus and occurs preferentially in translocation-negative MALT lymphoma of the ocular adnexa and salivary glands. J Pathol. 2009;217(3):420-430. 15. Honma K, Tsuzuki S, Nakagawa M, et al. TNFAIP3/A20 functions as a novel tumor suppressor gene in several subtypes of nonHodgkin lymphomas. Blood. 2009;114(12):2467-2475. 16. Bi Y, Zeng N, Chanudet E, et al. A20 inactivation in ocular adnexal MALT lymphoma. Haematologica. 2012;97(6):926-930. 17. Moody S, Escudero-Ibarz L, Wang M, et al. Significant association between TNFAIP3 inactivation and biased immunoglobulin heavy chain variable region 4-34 usage in mucosa-associated lymphoid tissue lymphoma. J Pathol. 2017;243(1):3-8. 18. Clipson A, Wang M, de Leval L, et al. KLF2 mutation is the most frequent somatic change in splenic marginal zone lymphoma and identifies a subset with distinct genotype. Leukemia. 2015;29(5):1177-1185. 19. Wang M, Escudero-Ibarz L, Moody S, et al. Somatic Mutation Screening Using Archival Formalin-Fixed, Paraffin-Embedded Tissues by Fluidigm Multiplex PCR and Illumina Sequencing. J Mol Diagn. 2015;17(5):521532. 20. Ansell SM, Akasaka T, McPhail E, et al. t(X;14)(p11;q32) in MALT lymphoma involving GPR34 reveals a role for GPR34 in tumor cell growth. Blood. 2012;120(19):3949-3957. 21. Johansson P, Klein-Hitpass L, Grabellus F, et al. Recurrent mutations in NF-kappaB pathway components, KMT2D, and NOTCH1/2 in ocular adnexal MALT-type marginal zone lymphomas. Oncotarget. 2016;7(38):62627-62639. 22. Baens M, Finalet FJ, Tousseyn T, et al. t(X;14)(p11.4;q32.33) is recurrent in marginal zone lymphoma and up-regulates GPR34. Haematologica. 2012;97(2):184-188. 23. Akasaka T, Lee YF, Novak AJ, et al. Clinical, histopathological, and molecular features of mucosa-associated lymphoid tissue (MALT) lymphoma carrying the t(X;14) (p11;q32)/GPR34-immunoglobulin heavy chain gene. Leuk Lymphoma. 2017;58(9):14. 24. Perissi V, Aggarwal A, Glass CK, Rose DW, Rosenfeld MG. A corepressor/coactivator exchange complex required for transcriptional activation by nuclear receptors and other regulated transcription factors. Cell. 2004;116(4):511-526. 25. Schmidt J, Gong S, Marafioti T, et al. Genome-wide analysis of pediatric-type follicular lymphoma reveals low genetic complexity and recurrent alterations of TNFRSF14 gene. Blood. 2016;128(8):11011111. 26. Shibata D, Weiss LM. Epstein-Barr virusassociated gastric adenocarcinoma. Am J Pathol. 1992;140:769-774. 27. Elgizouli M, Lowe DM, Speckmann C, et al. Activating PI3Kdelta mutations in a cohort of 669 patients with primary immunodefi-

28.

29.

30. 31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

ciency. Clin Exp Immunol. 2016;183(2):221229. Zhang J, Grubor V, Love CL, et al. Genetic heterogeneity of diffuse large B-cell lymphoma. Proc Natl Acad Sci USA. 2013;110(4):1398-1403. Crank MC, Grossman JK, Moir S, et al. Mutations in PIK3CD can cause hyper IgM syndrome (HIGM) associated with increased cancer susceptibility. J Clin Immunol. 2014;34(3):272-276. Rajagopal S, Shenoy SK. GPCR desensitization: Acute and prolonged phases. Cell Signal. 2018;41:9-16. Zhou XE, He Y, de Waal PW, et al. Identification of Phosphorylation Codes for Arrestin Recruitment by G Protein-Coupled Receptors. Cell. 2017;170(3):457-469. Hasegawa H, Patel N, Ettehadieh E, Li P, Lim AC. Topogenesis and cell surface trafficking of GPR34 are facilitated by positive-inside rule that effects through a tri-basic motif in the first intracellular loop. Biochim Biophys Acta. 2016;1863(7 Pt A):1534-1551. Fanelli F, De Benedetti PG, Raimondi F, Seeber M. Computational modeling of intramolecular and intermolecular communication in GPCRs. Curr Protein Pept Sci. 2009;10(2):173-185. Lagane B, Chow KY, Balabanian K, et al. CXCR4 dimerization and beta-arrestinmediated signaling account for the enhanced chemotaxis to CXCL12 in WHIM syndrome. Blood. 2008;112(1):34-44. Cao Y, Hunter ZR, Liu X, et al. The WHIMlike CXCR4(S338X) somatic mutation activates AKT and ERK, and promotes resistance to ibrutinib and other agents used in the treatment of Waldenstrom's Macroglobulinemia. Leukemia. 2015;29(1): 169-176. Nakagawa M, Schmitz R, Xiao W, et al. Gain-of-function CCR4 mutations in adult T cell leukemia/lymphoma. J Exp Med. 2014;211(13):2497-2505. Ikubo M, Inoue A, Nakamura S, et al. Structure-activity relationships of lysophosphatidylserine analogs as agonists of G-protein-coupled receptors GPR34, P2Y10, and GPR174. J Med Chem. 2015;58(10):42044219. Aoki J, Nagai Y, Hosono H, Inoue K, Arai H. Structure and function of phosphatidylserine-specific phospholipase A1. Biochim Biophys Acta. 2002;1582(1-3):26-32. Lee AY, Phan TK, Hulett MD, Korner H. The relationship between CCR6 and its binding partners: does the CCR6-CCL20 axis have to be extended? Cytokine. 2015;72(1):97-101. Zuo B, Li M, Liu Y, et al. G-protein coupled receptor 34 activates Erk and phosphatidylinositol 3-kinase/Akt pathways and functions as alternative pathway to mediate p185Bcr-Abl-induced transformation and leukemogenesis. Leuk Lymphoma. 2015;56(7):2170-2181. Jung H, Yoo HY, Lee SH, et al. The mutational landscape of ocular marginal zone lymphoma identifies frequent alterations in TNFAIP3 followed by mutations in TBL1XR1 and CREBBP. Oncotarget. 2017;8(10):17038-17049.

haematologica | 2018; 103(8)

ARTICLE

Non-Hodgkin Lymphoma

End-of-treatment and serial PET imaging in primary mediastinal B-cell lymphoma following dose-adjusted EPOCH-R: a paradigm shift in clinical decision making Christopher Melani,1 Ranjana Advani,2 Mark Roschewski,1 Kelsey M. Walters,2 Clara C. Chen,3 Lucia Baratto,4 Mark A. Ahlman,3 Milos D. Miljkovic,1 Seth M. Steinberg,5 Jessica Lam,2 Margaret Shovlin,1 Kieron Dunleavy,6 Stefania Pittaluga,7 Elaine S. Jaffe7 and Wyndham H. Wilson1

Lymphoid Malignancies Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD; 2Stanford Cancer Institute, Stanford University, CA; 3Nuclear Medicine Division, Radiology and Imaging Sciences, Clinical Center, National Institutes of Health, Bethesda, MD; 4Nuclear Medicine and Molecular Imaging Division, Stanford University, CA; 5Biostatistics and Data Management Section, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD; 6George Washington University Cancer Center, DC and 7Laboratory of Pathology, Clinical Center, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1337-1344

CM and RA contributed equally to this work.

ABSTRACT

ose-adjusted-EPOCH-R obviates the need for radiotherapy in most patients with primary mediastinal B-cell lymphoma. Endof-treatment PET, however, does not accurately identify patients at risk of treatment failure, thereby confounding clinical decision making. To define the role of PET in primary mediastinal B-cell lymphoma following dose-adjusted-EPOCH-R, we extended enrollment and follow up on our published phase II trial and independent series. Ninety-three patients received dose-adjusted-EPOCH-R without radiotherapy. Endof-treatment PET was performed in 80 patients, of whom 57 received 144 serial scans. One nuclear medicine physician from each institution blindly reviewed all scans from their respective institution. End-of-treatment PET was negative (Deauville 1-3) in 55 (69%) patients with one treatment failure (8-year event-free and overall survival of 96.0% and 97.7%). Among 25 (31%) patients with a positive (Deauville 4-5) end-oftreatment PET, there were 5 (20%) treatment failures (8-year event-free and overall survival of 71.1% and 84.3%). Linear regression analysis of serial scans showed a significant decrease in SUVmax in positive end-oftreatment PET non-progressors compared to an increase in treatment failures. Among 6 treatment failures, the median end-of-treatment SUVmax was 15.4 (range, 1.9-21.3), and 4 achieved long-term remission with salvage therapy. Virtually all patients with a negative end-of-treatment PET following dose-adjusted-EPOCH-R achieved durable remissions and should not receive radiotherapy. Among patients with a positive end-of-treatment PET, only 5/25 (20%) had treatment-failure. Serial PET imaging distinguished end-of-treatment PET positive patients without treatment failure, thereby reducing unnecessary radiotherapy by 80%, and should be considered in all patients with an initial positive PET following dose-adjusted-EPOCH-R (clinicaltrials.gov identifier 00001337).

Introduction Primary mediastinal B-cell lymphoma (PMBCL) is a subtype of diffuse large B-cell lymphoma that is clinically and biologically related to nodular sclerosis Hodgkin lymphoma (nsHL).1,2 As such, it primarily presents as a bulky mediastinal mass in adolescents and young adults and is more common in females.3-8 R-CHOP is comhaematologica | 2018; 103(8)

Correspondence: wilsonw@mail.nih.gov

Received: March 5, 2018. Accepted: May 10, 2018. Pre-published: May 10, 2018. doi:10.3324/haematol.2018.192492 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1337 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.

1337

C. Melani et al.

monly used to treat PMBCL but retrospective studies indicate that this therapy alone is inadequate for many patients,9 resulting in the frequent use of consolidative mediastinal radiotherapy, as part of combined modality treatment.6,8,10 It is well documented, however, that mediastinal radiotherapy is associated with significant late toxicity including premature death due to cardiovascular complications and second malignancies,11-14 which has led to efforts to minimize its use in mediastinal lymphomas.1518 In an effort to reduce mediastinal radiotherapy in PMBCL, we conducted a prospective study of DAEPOCH-R based on hypothesis-generating evidence that dose-intensive regimens may be more effective and showed DA-EPOCH-R obviated the need for radiotherapy in most patients.4 An important, albeit preliminary observation from this study, was that most patients with a positive end-of-treatment (EOT) 18F-fluorodeoxyglucose-positron-emission tomography (FDG-PET) scan achieved durable remissions without further therapy, calling into question the positive predictive value (PPV) of EOT FDG-PET following DAEPOCH-R.4 This is in line with several other retrospective studies as well as the prospective IELSG-26 study that have eluded to the low PPV and high false-positive rate of EOT FDG-PET imaging in PMBCL; however, a variety of induction chemoimmunotherapy regimens were utilized with most EOT FDG-PET positive patients going on to receive salvage radiotherapy or high-dose chemotherapy with autologous stem-cell transplantation, making the results inapplicable to DA-EPOCH-R or chemotherapy alone.19-22 While it is routine clinical practice to consider a positive (Deauville 4-5) EOT FDG-PET scan indicative of persistent disease and the need for radiotherapy,23,24 our findings raise a potential paradigm shift whereby singular EOT FDG-PET scans are inadequate following DAEPOCH-R. Indeed, even the significance of a negative EOT FDG-PET following front-line chemoimmunotherapy remains an open question and the subject of a randomized phase III study of post-treatment radiotherapy versus observation (clinicaltrials.gov identifier 01599559). To fully characterize the role of EOT and serial FDGPET imaging on clinical decision making and to provide further data on the clinical outcome of DA-EPOCH-R in PMBCL, we significantly extended enrollment on our phase II trial and independent clinical series. Herein, we provide an in-depth analysis of single EOT and serial FDG-PET scans and long-term patient outcome following DA-EPOCH-R for previously untreated PMBCL.

Response Assessment EOT response assessment was performed using CT in all patients and FDG-PET beginning in September 2002. Published guidelines recommend EOT FDG-PET a minimum of 3 weeks, preferably 6-8 weeks, following completion of chemotherapy.24 All patients with an EOT FDG-PET following the last dose of chemotherapy up to 8 weeks post-therapy (11 weeks post day 1 of the final cycle) were included for analysis. EOT FDG-PET was performed a median 3 weeks (range, 1-10) from day 1 of the final cycle of therapy. Scans were retrospectively scored per the 5-point Deauville scale26 with scores 1-3 negative and 4-5 positive.23,24 Thirty-five of 55 (64%) and 22 of 25 (88%) patients with negative and positive EOT scans, respectively, underwent serial FDG-PET imaging. Tumor biopsy and salvage therapy was implemented per investigator discretion. Surveillance CT scans were performed for up to 5 years post-therapy. One nuclear medicine physician from each institution reviewed and scored all FDG-PET scans from their respective institution without knowledge of clinical outcome. Calculation of metabolic tumor volume (MTV) and total lesion glycolysis (TLG = MTV x SUVmean) was performed on all NCI FDG-PET scans using Osirix version 8.50 (Pixmeo SARL, Bernex, Switzerland).

Statistical Analysis Overall survival (OS) and event-free survival (EFS) was calculated from the on-study date until date of death or last follow up or date of death, relapse, progression, second lymphoma treatment, or last follow up, respectively. Treatment failure was defined as relapse, progression, or residual disease following therapy. Probabilities of OS/EFS were calculated using the Kaplan-Meier (KM) method,27 with the significance of the difference between a pair of KM curves determined via an exact log-rank test. Characteristics were compared between patients with and without evaluable EOT FDG-PET scans and between patients by institution. Dichotomous characteristics, ordered characteristics, and continuous parameters were compared using Fisherâ&#x20AC;&#x2122;s exact test, an exact Cochran-Armitage test, and an exact Wilcoxon rank sum test, respectively. Linear regression was used in patients with serial FDG-PET scans to determine the slope of the change in SUVmax over time. Tests of the slopes being 0 within each group, tests of slopes among the 3 groups, and pairwise comparisons between 2 groups at a time were done using a Wilcoxon signed rank test, an exact Kruskal-Wallis test and Wilcoxon rank sum test, respectively. All P-values are two-tailed and not adjusted for multiple comparisons. Median potential follow up was calculated from the date of enrollment through April 2018, the date of the most recent data update.

Results Methods Patient Characteristics Patients/Treatment Ninety-three PMBCL patients received DA-EPOCH-R on the prospective NCI (N=59) and retrospective Stanford (N=34) study from November 1999 through July 2016. This includes 67 patients from the previously published study4 plus an additional 26 patients; 8 NCI and 18 Stanford. All patients received 6-8 cycles of DA-EPOCH-R (dose-adjusted etoposide, cyclophosphamide, and doxorubicin with prednisone, vincristine and rituximab) with G-CSF support as previously described, without consolidation radiotherapy.4,25 The study was approved by the NCI IRB and all patients provided written informed consent in accordance with the Declaration of Helsinki. clinicaltrials.gov identifier 00001337. 1338

Baseline characteristics of the 93 patients from NCI and Stanford were similar aside from a higher proportion of patients with an ECOG of 2-3 (29% vs. 3%, P=0.00058) in the Stanford cohort [Table 1]. Thirteen patients did not have evaluable EOT FDG-PET scans. Reasons included; treatment prior to routine FDG-PET use (N=9), FDG-PET performed prior to the last dose of chemotherapy (N=2), FDG-PET performed later than 8 weeks post completion of chemotherapy (N=1), and extensive brown fat uptake (N=1); exclusion of the 3 patients with FDG-PET scans had no significant impact on the study results or conclusions. The 80 remaining patients with evaluable EOT FDG-PET scans had similar baseline characteristics to the haematologica | 2018; 103(8)

Serial FDG-PET Imaging in PMBCL

Table 1. Baseline Characteristics of the Study Patients.

Characteristic

Female sex- no. (%) Age- yr. Median Range Bulky tumor, > 10 cm Patients- no. (%) Maximal diameter- Median (Range), cm Stage IV disease- no. (%) International prognostic index (IPI)- no. (%) Low (0-1) Low-intermediate (2) Intermediate-high (3) High (4-5) ECOG- no. (%) 0-1 2-3 Elevated LDH- no. (%) Extranodal site- no. (%) 0-1 â&#x2030;Ľ2 Any Pleural effusion- no. (%) Pericardial effusion- no. (%)

Total Cohort (N=93)

Evaluable EOT FDG-PET (N=80)

Prospective NCI Cohort (N=59)

Retrospective Stanford Cohort (N=34)

55 (59)

44 (55)

35 (59)

20 (59)

31 18-68

30 19-54

32.5 18-68

54 (59)a 10.7 (4-18.9) 18 (19)

52 (66)b,d 10.9 (5.5-18.9)e 14 (18)

36 (61) 10.9 (4-18.9) 14 (24)

18 (55)c 10 (4.9-18.3) 4 (12)

60 (65) 22 (24) 8 (9) 3 (3)

53 (66) 18 (23) 7 (9) 2 (3)

37 (63) 15 (25) 6 (10) 1 (2)

23 (68) 7 (21) 2 (6) 2 (6)

81 (87) 12 (13) 68 (74)a

69 (86) 11 (14) 59 (75)b

57 (97) 2 (3)f 46 (78)

24 (71) 10 (29) 22 (65)c

80 (86) 13 (14) 38 (41) 45 (48) 38 (41)

69 (86) 11 (14) 30 (38) 40 (50) 35 (44)

50 (85) 9 (15) 27 (46) 27 (46) 21 (36)

30 (88) 4 (12) 11 (32) 18 (53) 17 (50)

a N = 92 patients; bN = 79 patients; cN = 33 patients; dP=0.0013 comparing patients with and without evaluable EOT FDG-PET scans; eP= 0.0009 comparing patients with and without evaluable EOT FDG-PET scans; fP=0.00058 comparing patients treated at NCI vs. Stanford; ECOG: Eastern Cooperative Oncology Group performance status; LDH: lactate dehydrogenase; EOT FDG-PE: end-of-treatment; 18F-fluorodeoxyglucose-positron-emission tomography; NCI: National Cancer Institute.

13 patients without evaluable scans other than significantly more bulky tumors > 10 cm (66% vs. 15%, P=0.0013).

Clinical Outcome With a median potential follow up of 8.4 years (range, 1.7-18.4), EFS and OS at 8-years is 90.6% (95% confidence interval [CI]; 81.8-95.2) and 94.7% (95% CI; 86.3-98.0), respectively [Figure 1 A-B]. The NCI and Stanford cohorts had similar outcome with an 8-year EFS of 90.6% vs. 91.0% (P=0.71) and OS of 95.6% vs. 93.8% (P=0.30), respectively [Figure 1 C-D]. The outcome of the 13 patients without evaluable EOT FDG-PET scans was not statistically different from the 80 patients with evaluable scans; 8-year EFS 100% vs. 89.0% (P=0.17) and OS 100% vs. 93.8% (P=0.24), respectively.

EOT FDG-PET and CT Response Eighty (86%) patients had evaluable EOT FDG-PET scans following DA-EPOCH-R. Fifty-five (69%) patients had a negative (Deauville 1-3) and 25 (31%) patients had a positive (Deauville 4-5) EOT FDG-PET [Table 2]. Treatment failure occurred in 1 of 55 (2%) patients with a negative EOT FDG-PET and in 5 of 25 (20%) patients with a positive EOT FDG-PET scan. All 5 treatment failures in patients with a positive EOT FDG-PET occurred at or immediately following the EOT FDG-PET scan, and the one treatment failure in the patient with a negative EOT haematologica | 2018; 103(8)

FDG-PET occurred at day 320. One of 17 (6%) Deauville 4 patients and 4 of 8 (50%) Deauville 5 patients had treatment failure following front-line therapy. Four of 6 (67%) treatment failures were successfully salvaged with radiotherapy alone in 2 (both Deauville 5), resection alone in 1 (Deauville 4), and chemotherapy/transplantation/radiotherapy in 1 (Deauville 2) with a median remission duration of 6.4 years (range, 2-11.3). Two patients (both Deauville 5) died of progressive disease 7 and 17 months after multiple salvage regimens and 2 patients died without disease. Patients with a negative (Deauville 1-3) EOT FDG-PET had a significantly better 8-year EFS of 96.0% vs. 71.1% (P=0.0010) and OS of 97.7% vs. 84.3% (P=0.0115) compared to patients with positive (Deauville 4-5) scans [Figure 2 A-B]. In an exploratory analysis, patients with Deauville 5 scans had the poorest outcome with an 8-year EFS of 50% vs. 93.3% (P=0.0003) and OS of 75% vs. 95.9% (P=0.029) compared to patients with Deauville 1-4 scans [Figure 2 C-D]. Using conventional groupings of Deauville 1-3 versus 4-5, EOT FDG-PET had a positive predictive value (PPV) of 20% and a negative predictive value (NPV) of 98%. All 89 patients with complete tumor measurements had a reduction in the bi-dimensional product of the largest tumor mass by CT. There was no relationship between EOT tumor reduction and EOT FDG-PET Deauville score 1339

C. Melani et al.

[Figure 3]. Furthermore, there was no difference in tumor reduction when comparing patients with (N=6) and without (N=83) treatment failure; median reduction of 92% (range, 65-99) vs. 93% (range, 62-100), respectively [Figure 3].

Serial FDG-PET Scans Fifty-seven of 80 patients with evaluable EOT FDG-PET scans underwent 144 total serial scans; median of 2 (range, 1-6). Among the 54 patients with a negative EOT FDGPET who are progression-free, 34 had serial scans. Linear regression analysis demonstrated an overall decrease in SUVmax over time with a median change per day in SUVmax of -0.005 (range, -0.134-0.010; P=0.0018) [Online Supplementary Figure S1A]. Among the 20 patients with a positive EOT FDG-PET who are progression-free, 17 had serial scans. SUVmax decreased in these patients as well with linear regression analysis revealing a median change per day in SUVmax of -0.006 (range, -0.070-0.002; P=0.0005) [Figure 4A]. In the 6 treatment failures, the median EOT FDG-PET SUVmax was 15.4 (range, 1.9-21.3) [Figure 4C]. All 6 treatment failures had evidence of disease, which was documented by biopsy in 4 and by standard imaging criteria in 2 patients. One patient without biopsy confirmation

showed progression on CT with an EOT SUVmax of 14.5 and received salvage radiotherapy. A second patient without biopsy showed progression on treatment with increases in SUVmax from 10.2 to 21.3, and appearance of a new lesion, and received radiotherapy. Serial scans in 5 treatment failures all revealed progressive increases in SUVmax, which normalized in 3 patients following radiotherapy, resection, and chemotherapy/transplantation/ radiotherapy, respectively. Two patients had continued progression of SUVmax despite multiple salvage therapies and both died of progressive disease. Linear regression analysis in the 5 treatment failures with serial scans showed an overall increase in SUVmax per day across serial scans, with a median of 0.023 (range, -0.007-0.267; P=0.13), which was statistically greater than both positive and negative EOT FDG-PET non-progressors (P=0.011 and P=0.0037, respectively). Among 51 non-progressing patients with serial scans, 10 (20%) continued to have positive and 29 (57%) continued to have negative Deauville scores. Seven (14%) patients converted from positive to negative and 5 (10%) converted from negative to positive [Figure 4B; Online Supplementary Figure S1B]. In the 5 patients with treatment failure and serial scans, Deauville score remained stable in 4 (80%) and increased in 1 (20%) [Figure 4D].

P=0.71

P=0.30

Figure 1. Kaplanâ&#x20AC;&#x201C;Meier estimates of event-free and overall survival of all patients and by study group. DA-EPOCH-R was administered to a total of 93 patients; 59 treated on the NCI prospective study and 34 treated on the retrospective Stanford study. (A). Event-free survival 90.6% (95% CI, 81.8-95.2) at 8-years for the total cohort. (B). Overall survival 94.7% (95% CI, 86.3-98.0) at 8-years for the total cohort. (C). Event-free survival 90.6% (95% CI, 78.8-96.0) for the NCI cohort and 91.0% (95% CI, 74.6-97.0) for the Stanford cohort (P=0.71) at 8-years. (D). Overall survival 95.6% (95% CI, 83.5-98.8) for the NCI cohort and 93.8% (95% CI, 77.5-98.4) for the Stanford cohort (P=0.30) at 8-years.

1340

haematologica | 2018; 103(8)

Serial FDG-PET Imaging in PMBCL

Changes in MTV and TLG across serial FDG-PET scans generally mimicked that of SUVmax with greater variability in value between patients within each EOT FDG-PET subgroup [Online Supplementary Figures S2-3].

Discussion These extended results from our initial study4 show that DA-EPOCH-R in untreated PMBCL patients results in an 8year EFS and OS of 90.6% and 94.7%, respectively, while obviating the need for radiotherapy in all but 5 (5%) patients. In contrast, retrospective studies suggest R-CHOP alone is inadequate for many PMBCL patients due to an unacceptable rate of primary induction failure up to 21% in one series,9 necessitating the frequent use of post-treatment radiotherapy, as part of combined modality treatment.510,15,19,21 A recent multicenter, retrospective study comparing

R-CHOP to DA-EPOCH-R as front-line therapy for PMBCL showed no significant difference in 2-year PFS or OS between the two treatments; however, this was achieved through significantly greater radiotherapy use with RCHOP (59% vs. 13%, P<0.001).28 Although excellent outcomes can be achieved via combined modality treatment, routine mediastinal radiotherapy use significantly increases the risk of late toxicity, including premature death from cardiovascular disease and second cancers.11-14 Unfortunately, due to the absence of prospective studies of R-CHOP in PMBCL, an accurate assessment cannot be made of its curative potential and who requires post-treatment radiotherapy. Nonetheless, it is presently accepted that patients with a positive EOT FDG-PET scan following R-CHOP require consolidation radiotherapy, and it remains uncertain if patients with a negative EOT FDG-PET benefit from radiotherapy, which is the endpoint of the IELSG-37 phase III randomized study (clinicaltrials.gov identifier 01599559).

Event-free Survival (Deauville 1-3 vs. 4-5)

P=0.0010

Overall Survival (Deauville 1-3 vs. 4-5)

P=0.0115

Event-free Survival (Deauville 1-4 vs. 5)

P=0.0003

Overall Survival (Deauville 1-4 vs. 5)

P=0.029

Figure 2. Kaplanâ&#x20AC;&#x201C;Meier estimates of event-free and overall survival by Deauville group. Event-free survival and overall survival according to EOT FDG-PET Deauville group. (A). Event-free survival 96.0% (95% CI, 84.8-99.0) vs. 71.1% (95% CI, 43.6-86.9) (P=0.0010) for Deauville 1-3 (blue curve) and Deauville 4-5 (red curve), respectively, at 8-years. (B). Overall survival 97.7% (95% CI, 84.6-99.7) vs. 84.3% (95% CI, 56.5-95.0) (P=0.0115) for Deauville 1-3 (blue curve) and Deauville 4-5 (red curve), respectively, at 8-years. (C). Event-free survival 93.3% (95% CI, 82.8-97.5) vs. 50.0% (95% CI, 15.2-77.5) (P=0.0003) for Deauville 1-4 (blue curve) and Deauville 5 (red curve), respectively, at 8-years. (D). Overall survival 95.9% (95% CI, 84.5-99.0) vs. 75.0% (95% CI, 31.5-93.1) (P=0.029) for Deauville 1-4 (blue curve) and Deauville 5 (red curve), respectively, at 8-years.

Table 2. EOT FDG-PET Response Following DA-EPOCH-R Therapy.

Lymphoma Status (N=80 total with EOT FDG-PET)

No treatment failure- no. patients Treatment failure- no. patients

Deauville Score 1

Negative (55/80, 69%) 2

Positive (25/80, 31%) 3

(30%) 24* 0

(24%) 18 1

(15%) 12 0

(21%) 16* 1

(10%) 4 4

*Indicates 1 patient death without evidence of disease recurrence; EOT FDG-PET end-of-treatment 18F-fluorodeoxyglucose-positron-emission tomography.

haematologica | 2018; 103(8)

1341

C. Melani et al.

Figure 3. Tumor reduction by end-of-treatment CT. Reduction of the bi-dimensional product of the largest mediastinal mass for the 89 patients with complete tumor measurements by EOT CT. All patients had reduction in tumor bi-dimensional product and there was no relationship between EOT tumor reduction and EOT FDGPET Deauville score. No difference in tumor reduction was demonstrated between patients with (N=6, red arrows) and without (N=83) treatment failure; Median reduction 92% (range, 65-99) vs. 93% (range, 62-100), respectively.

Our results indicate that very few patients require posttreatment radiotherapy following DA-EPOCH-R, irrespective of their EOT FDG-PET scans. These findings provide substantial evidence that patients with negative EOT FDG-PET scans rarely recur and are unlikely to benefit from additional mediastinal radiotherapy. Furthermore, they provide evidence for our initial observation that most patients with a positive EOT FDG-PET achieve long-term remission following DA-EPOCH-R and, as a group, would not benefit from empirical consolidation radiotherapy. Indeed, the discrepancy between our findings that routine consolidation radiotherapy is unnecessary following DA-EPOCH-R, and the accepted need for post-treatment radiotherapy in patients with positive EOT FDG-PET scans following R-CHOP has led to uncertainty. Unfortunately, it is not uncommon for patients with a positive EOT FDG-PET following DAEPOCH-R to receive post-treatment radiotherapy. Such an approach in our study would have resulted in 31% (25/80) of patients receiving radiotherapy, most of whom (80%) were already cured with DA-EPOCH-R alone. A clinically important aspect of our study is distinguishing treatment failures following DA-EPOCH-R. Given the worse outcome of PMBCL compared to DLBCL with salvage therapy,29 early recognition of patients with persistent disease is critical to optimize the curative potential of radiotherapy while averting its use in patients already cured with DA-EPOCH-R. We first looked at tumor mass reduction based on EOT CT, and observed no predictive value on outcome or any relationship with EOT FDG-PET. We also assessed the ability of single EOT and serial FDGPET imaging to detect treatment failure. Following DAEPOCH-R, 69% of patients had a negative EOT FDG-PET. Notably, 98% of these patients never progressed, indicat1342

ing such patients rarely require radiotherapy. Among the 31% of patients with positive EOT scans, only 5 ultimately had treatment failure of which 4 occurred in patients with Deauville 5 scans. These results are consistent with the prospective IELSG-26 study, which revealed a significantly worse outcome in patients with Deauville 4-5 EOT FDG-PET scans (5-yr. PFS 68% vs. 99%, P<0.0001; 5-yr. OS 83% vs. 100%, P=0.003), with the greatest number of treatment failures in Deauville 5 patients.22 In contrast to our study, however, variable chemoimmunotherapy was used and most patients (89%) received consolidation radiotherapy. We found serial FDG-PET imaging to be a highly effective strategy to distinguish persistent disease from posttreatment inflammatory changes. Linear regression analysis in 17 non-progressing patients with a positive EOT FDG-PET and serial imaging showed an overall decrease in SUVmax across serial scans. In contrast, serial FDG-PET imaging in 5 treatment failures with serial scans showed an increase in SUVmax that was statistically greater than patients who never progressed, regardless of EOT FDGPET response (P=0.011 and P=0.0037 for positive and negative EOT FDG-PET non-progressors, respectively). Overall, use of serial FDG-PET imaging effectively reduced radiotherapy from a potential 31% (25/80) of patients with a positive EOT FDG-PET scan to only 5 (5%) patients with confirmed treatment failure. We also explored the use of quantitative FDG-PET parameters (i.e., MTV and TLG) to assess if they improved upon SUVmax in identification of treatment failures. These methods were limited by the overall low volume of disease following therapy as well as inability to exclude nonmalignant causes of FDG uptake, resulting in a wide variability in value between patients. Although these paramhaematologica | 2018; 103(8)

Serial FDG-PET Imaging in PMBCL

Figure 4. Evolution of serial FDG-PET imaging. Heatmap depiction of (A). SUVmax, and (B). Deauville score, over time in the 20 non-progressing patients with a positive EOT FDG-PET scan. Heatmap depiction of (C). SUVmax, and (D). Deauville score, over time in the 6 patients with treatment failure. FDG-PET scans performed prior to the EOT FDG-PET are listed as negative numbers with those following the EOT FDG-PET listed as positive numbers. The EOT FDG-PET scan is bordered by black dashed lines. FDG-PET scans performed following salvage intervention are shaded in black.

eters were not superior to monitoring SUVmax in our study, other recent reports indicate these quantitative parameters may be beneficial for baseline prognostication as well as when combined with EOT Deauville score.30,31 Our findings are supported by a recent retrospective multi-center analysis of 156 PMBCL patients treated with DA-EPOCH-R which reported a 3-year EFS and OS of 85.9% and 95.4%, respectively.32 Overall, 14.9% of patients received post-treatment radiotherapy, which was administered at the discretion of the treating physician. In that study, 75% of patients achieved a negative EOT FDG-PET and 95.4% remained progression-free, consistent with our findings that consolidation radiotherapy is virtually never indicated in this patient group. Less clear are their results in patients with positive EOT FDG-PET scans. Among the 31 patients with positive EOT scans, 19 received no further treatment with 68% progression-free at a median follow up of 17 months, indicating that a substantial subset of these patients are likely cured with DAEPOCH-R alone.32 Twelve patients with a positive EOT FDG-PET received post-treatment radiotherapy and 33.3% remain progression-free at 2 years. It is important to note that serial FDG-PET was not a prospective endpoint of our trial and decisions regarding which patients should receive serial scans and the timing of those scans was left to the discretion of the treating haematologica | 2018; 103(8)

physician. Indeed, the aim of this study was to provide a descriptive look at EOT and serial PET imaging in PMBCL following DA-EPOCH-R as it occurs in the real-world clinical setting, where decisions are often left to clinical judgement. The notion, however, that physician discretion influenced these observational findings is obviated by the extended follow up, which showed who did and did not recur and by the absence of late recurrences. In conclusion, our results indicate that a negative EOT FDG-PET following DA-EPOCH-R in PMBCL is highly predictive of cure and radiotherapy in these patients is unnecessary. The unique biology of PMBCL results in a high rate of false-positive EOT FDG-PET scans indicating the need for a paradigm shift in clinical decision making for this group of patients when receiving DA-EPOCH-R. A singular EOT FDG-PET did not accurately identify treatment failure but serial FDG-PET imaging effectively discriminated residual disease from post-treatment inflammatory changes. Serial FDG-PET imaging should be considered in all patients with an initial positive EOT FDG-PET to identify treatment failures that require radiotherapy. Funding Research support was provided through the intramural program of the National Cancer Institute, National Institutes of Health. 1343

C. Melani et al.

References 12. 1. Rosenwald A, Wright G, Leroy K, et al. Molecular diagnosis of primary mediastinal B cell lymphoma identifies a clinically favorable subgroup of diffuse large B cell lymphoma related to Hodgkin lymphoma. J Exp Med. 2003;198(6):851-862. 2. Savage KJ, Monti S, Kutok JL, et al. The molecular signature of mediastinal large Bcell lymphoma differs from that of other diffuse large B-cell lymphomas and shares features with classical Hodgkin lymphoma. Blood. 2003;102(12):3871-3879. 3. Bishop PC, Wilson WH, Pearson D, Janik J, Jaffe ES, Elwood PC. CNS involvement in primary mediastinal large B-cell lymphoma. J Clin Oncol. 1999;17(8):24792485. 4. Dunleavy K, Pittaluga S, Maeda LS, et al. Dose-adjusted EPOCH-rituximab therapy in primary mediastinal B-cell lymphoma. N Engl J Med. 2013;368(15):1408-1416. 5. Rieger M, Osterborg A, Pettengell R, et al. Primary mediastinal B-cell lymphoma treated with CHOP-like chemotherapy with or without rituximab: results of the Mabthera International Trial Group study. Ann Oncol. 2011;22(3):664-670. 6. Xu LM, Fang H, Wang WH, et al. Prognostic significance of rituximab and radiotherapy for patients with primary mediastinal large B-cell lymphoma receiving doxorubicin-containing chemotherapy. Leuk Lymphoma. 2013;54(8):1684-1690. 7. Vassilakopoulos TP, Pangalis GA, Katsigiannis A, et al. Rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone with or without radiotherapy in primary mediastinal large B-cell lymphoma: the emerging standard of care. Oncologist. 2012;17(2):239-249. 8. Zinzani PL, Martelli M, Bertini M, et al. Induction chemotherapy strategies for primary mediastinal large B-cell lymphoma with sclerosis: a retrospective multinational study on 426 previously untreated patients. Haematologica. 2002;87(12):1258-1264. 9. Soumerai JD, Hellmann MD, Feng Y, et al. Treatment of primary mediastinal B-cell lymphoma with rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone is associated with a high rate of primary refractory disease. Leuk Lymphoma. 2014;55(3):538-543. 10. Binkley MS, Hiniker SM, Wu S, et al. A single-institution retrospective analysis of outcomes for stage I-II primary mediastinal large B-cell lymphoma treated with immunochemotherapy with or without radiotherapy. Leuk Lymphoma. 2016;57(3): 604-608. 11. van Leeuwen FE, Ng AK. Long-term risk of second malignancy and cardiovascular disease after Hodgkin lymphoma treatment.

1344

13.

14.

15.

16.

17.

18.

19.

20.

21.

Hematology Am Soc Hematol Educ Program. 2016;2016(1):323-330. Bhakta N, Liu Q, Yeo F, et al. Cumulative burden of cardiovascular morbidity in paediatric, adolescent, and young adult survivors of Hodgkin's lymphoma: an analysis from the St Jude Lifetime Cohort Study. Lancet Oncol. 2016;17(9):1325-1334. Sud A, Thomsen H, Sundquist K, Houlston RS, Hemminki K. Risk of Second Cancer in Hodgkin Lymphoma Survivors and Influence of Family History. J Clin Oncol. 2017;35(14):1584-1590. Schaapveld M, Aleman BM, van Eggermond AM, et al. Second Cancer Risk Up to 40 Years after Treatment for Hodgkin's Lymphoma. N Engl J Med. 2015;373(26):2499-2511. Pinnix CC, Dabaja B, Ahmed MA, et al. Single-institution experience in the treatment of primary mediastinal B cell lymphoma treated with immunochemotherapy in the setting of response assessment by 18fluorodeoxyglucose positron emission tomography. Int J Radiat Oncol Biol Phys. 2015;92(1):113-121. Russo F, Corazzelli G, Frigeri F, et al. A phase II study of dose-dense and doseintense ABVD (ABVDDD-DI ) without consolidation radiotherapy in patients with advanced Hodgkin lymphoma. Br J Haematol. 2014;166(1):118-129. Raemaekers JM, Andre MP, Federico M, et al. Omitting radiotherapy in early positron emission tomography-negative stage I/II Hodgkin lymphoma is associated with an increased risk of early relapse: Clinical results of the preplanned interim analysis of the randomized EORTC/LYSA/FIL H10 trial. J Clin Oncol. 2014;32(12):1188-1194. Wolden SL, Chen L, Kelly KM, et al. Longterm results of CCG 5942: a randomized comparison of chemotherapy with and without radiotherapy for children with Hodgkin's lymphoma--a report from the Children's Oncology Group. J Clin Oncol. 2012;30(26):3174-3180. Cheah CY, Hofman MS, Seymour JF, et al. The utility and limitations of (18)F-fluorodeoxyglucose positron emission tomography with computed tomography in patients with primary mediastinal B-cell lymphoma: single institution experience and literature review. Leuk Lymphoma. 2015;56(1):49-56. Filippi AR, Piva C, Giunta F, et al. Radiation therapy in primary mediastinal B-cell lymphoma with positron emission tomography positivity after rituximab chemotherapy. Int J Radiat Oncol Biol Phys. 2013;87(2):311-316. Vassilakopoulos TP, Pangalis GA, Chatziioannou S, et al. PET/CT in primary mediastinal large B-cell lymphoma responding to rituximab-CHOP: An analysis of 106 patients regarding prognostic sig-

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

nificance and implications for subsequent radiotherapy. Leukemia. 2016;30(1):238242. Martelli M, Ceriani L, Zucca E, et al. [18F]fluorodeoxyglucose positron emission tomography predicts survival after chemoimmunotherapy for primary mediastinal large B-cell lymphoma: results of the International Extranodal Lymphoma Study Group IELSG-26 Study. J Clin Oncol. 2014;32(17):1769-1775. Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014;32(27):3059-3068. Barrington SF, Mikhaeel NG, Kostakoglu L, et al. Role of imaging in the staging and response assessment of lymphoma: consensus of the International Conference on Malignant Lymphomas Imaging Working Group. J Clin Oncol. 2014;32(27):3048-3058. Wilson WH, Grossbard ML, Pittaluga S, et al. Dose-adjusted EPOCH chemotherapy for untreated large B-cell lymphomas: a pharmacodynamic approach with high efficacy. Blood. 2002;99(8):2685-2693. Meignan M, Gallamini A, Haioun C. Report on the First International Workshop on Interim-PET-Scan in Lymphoma. Leuk Lymphoma. 2009;50(8):1257-1260. Kaplan EL, Meier P. Nonparametric Estimation from Incomplete Observations. Journal of the American Statistical Association. 1958;53(282):457-481. Shah NN, Szabo A, Huntington SF, et al. RCHOP versus dose-adjusted R-EPOCH in frontline management of primary mediastinal B-cell lymphoma: a multi-centre analysis. Br J Haematol. 2018;180(4):534-544. Kuruvilla J, Pintilie M, Tsang R, Nagy T, Keating A, Crump M. Salvage chemotherapy and autologous stem cell transplantation are inferior for relapsed or refractory primary mediastinal large B-cell lymphoma compared with diffuse large B-cell lymphoma. Leuk Lymphoma. 2008;49(7):13291336. Ceriani L, Martelli M, Zinzani PL, et al. Utility of baseline 18FDG-PET/CT functional parameters in defining prognosis of primary mediastinal (thymic) large B-cell lymphoma. Blood. 2015;126(8):950-956. Ceriani L, Martelli M, Conconi A, et al. Prognostic models for primary mediastinal (thymic) B-cell lymphoma derived from 18FDG PET/CT quantitative parameters in the International Extranodal Lymphoma Study Group (IELSG) 26 study. Br J Haematol. 2017;178(4):588-591. Giulino-Roth L, O'Donohue T, Chen Z, et al. Outcomes of adults and children with primary mediastinal B-cell lymphoma treated with dose-adjusted EPOCH-R. Br J Haematol. 2017;179(5):739-747.

haematologica | 2018; 103(8)

ARTICLE

Non-Hodgkin Lymphoma

Sergio Storti,1 Michele Spina,2 Emanuela Anna Pesce,3 Flavia Salvi,4 Michele Merli,5 Alessia Ruffini,6 Giuseppina Cabras,7 Annalisa Chiappella,8 Emanuele Angelucci,9 Alberto Fabbri,10 Anna Marina Liberati,11 Monica Tani,12 Gerardo Musuraca,13 Annalia Molinari,14 Maria Pia Petrilli,1 Carmela Palladino,4 Rosanna Ciancia,2 Andrea Ferrario,5 Cristiana Gasbarrino,1 Federico Monaco,4 Vincenzo Fraticelli,1 Annalisa De Vellis,1 Francesco Merli15 and Stefano Luminari15,16

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1345-1350

1 Department of Hematology, Universita' Cattolica Sacro Cuore Campobasso; 2Division of Oncology A, National Cancer Institute Aviano; 3Fondazione Italiana Linfomi Onlus; 4 Hematology Unit, Antonio e Biagio e Cesare Arrigo Hospital, Alessandria; 5Department of Hematology, Ospedale di Circolo e Fondazione Macchi - ASST Sette Laghi, Varese; 6GRADE Gruppo Amici Dell’Ematologia; 7Unit of Hematology Ospedale Businco, Cagliari; 8 Department of Hematology, Città della Salute Hospital and University, Torino; 9Ospedale Policlinico San Martino, Genova; 10Azienda Ospedaliera Universitaria Senese, U.O.C. Ematologia, Siena; 11Università degli Studi di Perugia, A.O.S. Maria, Terni; 12Department of Hematology, S. Maria delle Croci Hospital, Ravenna; 13Department of Hematology, IRCCS Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (I.R.S.T.), Meldola; 14 Department of Hematology, Infermi Hospital, Rimini; 15Department of Hematology, Azienda Unità Sanitaria Locale IRCCS di Reggio Emilia and 16Department of Clinical Diagnostic and Public Health Medicine, Università degli Studi di Modena e Reggio Emilia, Modena, Italy

ABSTRACT

e conducted a phase II study to assess activity and safety profile of bendamustine and rituximab in elderly patients with untreated diffuse large B-cell lymphoma (DLBCL) who were prospectively defined as frail using a simplified version of the Comprehensive Geriatric Assessment (CGA). Patients had to be over 70 years of age, with histologically confirmed DLBCL. Frail patients were those younger than 80 years with a frail profile at CGA or older than 80 years with an unfit profile. Treatment consisted of 4-6 courses of bendamustine [90 mg/m2 days (d)1-2] and rituximab (375 mg/m2 d1) administered every 28 days. Other main study end points were complete remission rate and the rate of extra-hematologic adverse events. Forty-nine patients were enrolled of whom 45 were confirmed eligible. Overall, 24 patients achieved a complete remission (53%; 95%CI: 38-68%) and the overall response rate was 62% (95%CI: 47-76%). The most frequent grade 3-4 adverse event was neutropenia (37.8%). Grade 3-4 extra-hematologic adverse events were observed in 7 patients (15.6%; 95%CI: 6.529.5%); the most frequent was grade 3 infection in 2 patients. With a median follow up of 33 months (range 1-52), the median progression-free survival was ten months (95%CI: 7-25). The study shows promising activity and manageable toxicity profile of BR combination as first-line therapy for patients with DLBCL who are prospectively defined as frail according to a simplified CGA, as adopted in this trial (clinicaltrials.gov identifier: 01990144).

haematologica | 2018; 103(8)

Correspondence: stefano.luminari@ausl.re.it

Received: December 20, 2017. Accepted: May 3, 2018. Pre-published: May 10, 2018.

doi:10.3324/haematol.2017.186569 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1345 ©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.

1345

S. Storti et al.

Introduction Diffuse large B-cell lymphoma (DLBCL) is the most frequent non-Hodgkin lymphoma (NHL) and typically affects elderly patients; approximately 50% of the patients are older than 65 years and 15% are older than 80 years.1 More importantly, life expectancy has markedly improved over the past century and it is expected that the number of people older than 75 years will triple by the year 2030.2 As a consequence, the burden of age-related diseases, including DLBCL, is expected to increase in the near future. R-CHOP is the undisputed standard for the treatment of elderly patients up to 80 years of age.3 A remaining unmet need, however, is related to the treatment of elderly patients who, due to age or comorbidity, cannot be treated with full-dose standard treatment. In these subjects, the availability of less toxic regimens is strongly warranted but the decision-making process concerning therapeutic intervention should also include an accurate and objective patient selection.4 Bendamustine is an alkylating agent with properties of a purine analog and is approved for the treatment of chronic lymphocytic leukemia (CLL) and indolent NHL. Bendamustine was also studied in aggressive lymphomas; several phase II studies of BR (bendamustine in association with rituximab) in patients with relapsed refractory DLBCL showed promising efficacy with overall good tolerance of this regimen in the salvage setting.5-7 The activity of BR was also tested with promising results in small phase II studies on untreated patients who were generally considered not eligible for standard R-CHOP.8,9 In 2014, the Fondazione Italiana Linfomi (FIL) started a phase II study to investigate the activity and the safety profile of a combination regimen of BR for the initial therapy of patients with DLBCL who were not eligible to receive standard anthracycline-based therapy. In contrast to other studies, we included patients who were prospectively classified as frail according to the Comprehensive Geriatric Assessment (CGA),10 and we evaluated a modified BR schedule. Based on available evidence, there is still no standard treatment for this subset of patients, complete response to therapy is approximately 15%,11 and, even if rituximab is used, the median OS is approximately 20 months.12

patients aged between 70 and 80 years, ADL<4 or IADL<5 or 1 grade 3 comorbidity or >8 grade 2 comorbidities were required; in patients older than 80 years, ADL>5 or IADL>6 or 5-8 grade 2 comorbidities were required10 (Table 1). A full list of inclusion criteria and study procedures is available in the Online Supplementary Appendix.

Treatment Patients received bendamustine (90 mg/m2, d1-2) combined with rituximab (375 mg/m2, d1) every 28 days. Patients with age-adjusted International Prognostic Index (aaIPI) equal to 0 and non-bulky disease received 4 cycles of BR followed by 2 cycles of rituximab. All other patients received 6 cycles of BR followed by 2 cycles of rituximab. Bendamustine was supplied for free by Mundipharma. The use of consolidation radiotherapy was allowed. Prophylaxis with valacyclovir and cotrimoxazole was mandatory. The use of granulocyte-colony stimulating factor (G-CSF) and erythropoietin was recommended.

Statistical analysis The main study end point was the complete remission rate (CRR) that was calculated on the efficacy population (EP; i.e. patients receiving at least 2 courses of BR) using Cheson 1999 criteria.13 A further end point was the rate of grade 3-4 extrahematologic adverse events (eeAEs) calculated on the safety population (SP; i.e. all patients who have received at least one dose of study medication) using CTCAE v.4.0 (Common Terminology Criteria for Adverse Events). Secondary end points were progression-free survival (PFS) and overall survival (OS).14 The study was conducted according to a Simon 2-stage design. The null hypothesis (p0) for CRR was set to 0.15.11 With a type I error of 0.05 and a type II error of 0.10, and considering the alternative hypothesis (p1) at 0.35, at least 4 CRs were required after the enrollment of the first 19 cases. With the full enrollment of the 44 cases, at least 11 patients in CR were required to confirm activity of the combination to be promising. Considering a 10% drop-out rate, the final study enrollment was set at 49 patients. Maximum tolerated toxicity rate for eeAEs was set at 30%. Considering an alpha error of 0.05, and according to the 2-stage Simon design, the study had to be stopped if grade 3-4 events were observed in 12 or more patients out of the first 19 enrolled. With the enrollment of the planned 44 patients, a maximum number of 20 patients with grade 3-4 eeAEs were allowed.

Results Methods Patients Study design and objectives This is a phase II open-label, non-randomized study to investigate activity and safety of BR combination therapy in elderly patients with DLBCL, prospectively defined as frail according to the CGA. The study was approved by the local ethics committees.

From February 2012 to February 2014, 49 patients were enrolled into the study by 24 Italian centers. Three patients

Table 1. Comprehensive Geriatric Assessment criteria for definition of fit, unfit and frail patients.

Patient eligibility Eligible patients were elderly frail patients aged 70 years or over with a newly diagnosed, histologically proven DLBCL. All patients were prospectively evaluated by the CGA, as originally reported as part of pre-therapy assessment. Briefly, the CGA used was based on the assessment of Activities of Daily Living (ADL), Instrumental Activities of Daily Living (IADL), and Cumulative Illness Rating Scale-Geriatric (CIRS-G). Patients were classified as frail if the following criteria were met: in 1346

ADL IADL CIRS** Age

FIT

UNFIT

6 8 0 score=3-4 < 5 score=2

5* 7-6* 0 score=3-4 5-8 score=2 ≥ 80 fit

FRAIL

≤4 ≤5 1 score=3-4 > 8 score=2 ≥ 80 unfit

ADL: Activity of Daily Living; IADL: Instrumental Activities of Daily Living; CIRS: Cumulative Illness Rating Scale. *Residual function. **Scores are referred to severity of assessed comorbidities (as reported by Tucci et al.10)

haematologica | 2018; 103(8)

BR in frail elderly patients with DLBCL

were considered ineligible due to violation of inclusion criteria (one patient was not confirmed as frail, one did not undergo CGA evaluation, one case was HCV positive). One additional patient was excluded from the study analysis due to death before treatment start. The clinical characteristics of the remaining 45 patients are shown in Table 2. A full list of comorbidities with observed rates is provided in Online Supplementary Table S1.

Feasibility and Efficacy Treatment was started in 45 eligible patients. Nine patients were at low risk (aaIPI 0 and non-bulky disease) and had to receive 4 BR cycles followed by two doses of rituximab. All the other patients (n=36) were at high risk and had to receive all 6 BR cycles followed by two doses of rituximab. Overall, 25 patients received all planned bendamustine doses (55%); 7 of 9 at low risk, and 18 of 36 at high risk. Treatment was discontinued due to progressive disease (12 patients), eeAEs (8 patients) and physician's decision (7 patients). Four non-bulky patients (3 with stage I-II, 1 with stage IV) received consolidation radiotherapy. In 4 high-risk cases, treatment was interrupted before administration of the 2 consolidation doses of rituximab. Three additional patients interrupted treatment after cycle 4 (n=1) and 5 (n=2) due to physician's decision. The calculated administered dose intensity for rituximab and bendamustine was 0.990 (25-75 percentiles: 0.901-1.016), and 0.996 (25-75 percentiles: 0.877-1.017). The efficacy analysis was based on all 45 cases. After the enrollment of the first 19 patients, 5 CRs were observed and the accrual to study stage 2 opened. Overall, 24 patients achieved a CR at the end of treatment (53%; 95%CI: 38-68%), 4 patients achieved a partial remission (PR), one patient had a stable disease (SD), 13 patients had progressive disease (PD), and in 3 patients response was not evaluable (due to AEs: pelvic fracture, pneumonia, heart attack). The overall response rate (ORR) was 62% (95%CI: 47-76%). The observed CRR was higher than that initially required for efficacy assessment by the study design (Table 3).

Safety The safety analysis was available for all 45 eligible patients and for 244 cycles. Thirty-five grade 3-4 AEs were reported in 23 patients (51.1%; 95CI: 35.8-66.3%); the most frequent grade 3-4 AE was neutropenia with 17 events (37.8%) (Table 4). Though not mandatory, G-CSF was used in 26 patients (58%; 95CI: 42-72%). The rate of grade 3-4 eeAE was monitored during stage 1 and stage 2 of the study. Overall, grade 3-4 eeAEs were reported in 7 patients (15.6%; 95CI: 6.5-29.5%) and grade 3-4 hematologic AEs were reported in 21 cases (46.7%; 95CI: 31.7-62.1%). The most frequent grade 3-4 eeAE was infection (n=2 patients, 4.4%); no grade 4 infections were reported. A detail of all reported AEs is provided in Table 4. Both for study stage I and stage II, the rate of grade 3-4 eeAEs never fell beyond the maximum rate allowed.

Survival analysis The median follow up was 33 months (range 1-52). Thirty-two patients experienced PD, including 13 PD at the end of induction therapy, 12 relapses, and 7 deaths for causes unrelated to lymphoma (pneumonia, heart failure, hepatocarcinoma, neurological disorder, respiratory disorder, cachexia, unknown). The 2-year PFS was 38% haematologica | 2018; 103(8)

(95%CI: 24-51%) and the median PFS was 10 months (95%CI: 7-25%) (Figure 1). Overall, 24 patients died: 8 during treatment and 16 during follow up. Cause of death was lymphoma progression in 14 patients (58%), and unrelated causes in 8 patients (the 7 deaths reported above, plus one patient who died due to secondary acute myeloyd leukemia after lymphoma progression). Two-year OS was 51% (95%CI: 3565%) and median OS was 30 months (95%CI: 10-NA). The results of the CGA scales used during patientsâ&#x20AC;&#x2122; baseline assessment were correlated with OS and PFS in univariate analysis, and no association was found (Table 5). In addition, no association was found between results of CGA scales and safety results (data not shown).

Discussion We report the results of a phase II study to investigate the activity and toxicity of a combination regimen of rituximab and bendamustine for the initial treatment of patients with DLBCL who were classified as frail based on

Table 2. Baseline characteristics of patients eligible for the study (n=45). Sex Age (years) Hb (g/dL) Stage

ECOG PS ENS LDH IPI CGA LVEF (%)

Status

Missing

N (%)

Male Median (range) Median (range) I II III IV >1 >1 >ULN 3-5 Unfit with age â&#x2030;Ľ80 years frail Median (range)

26 (58) 81 (71-89) 12.9 (7.8-16.1) 7 (16) 10 (23) 6 (14) 22 (48) 16 (36) 11 (24) 16 (36) 25 (57) 35 (78) 10 (22) 60 (43-70)

1 1 4

Hb: hemoglobin; ECOG PS: Eastern Cooperative Oncology Group Performance Status; ENS: extra nodal site; LDH: lactate dehydrogenase; ULN:upper limit of normal; IPI: International Prognostic Index; CGA: comprehensive geriatric assessment; LVEF: left ventricular ejection fraction.

Table 3. Response after planned rituximab plus bendamustine treatment. Status Missing N (%)

Response

R8+B6 (n=36) N (%)

CR 19 (53) PR 4 (11) ORR SD 1 (3) PD 9 (25) Not assessed 1 (3) Death in treatment * 2 (6)

R6+B4 (n=9) N (%)

Total (n=45) N (%, %95CI)

5 (56) -

24 (53; 38-68) 4 (9; 2-21) 28 (62; 47-76) 1 (2; 0-12) 13 (29; 16-44) 1 (2; 0-12) 2 (4; 1-15)

4 (44) -

*Dead for heart failure and pneumonia. R8+B6: rituximab for 8 cycles and bendamustine for 6 cycles; R6+B4: Rituximab for 6 cycles and bendamustine for 4 cycles; n/N:: number; CI: Confidence Interval; CR: complete remission; PR: partial remission; ORR: overall response rate; SD: stable disease; PD: progression disease.

1347

S. Storti et al.

a CGA. With the observed 53% CR rate, and with the 15.6% rate of grade 3-4 eeAEs, the study met its primary objectives and shows the promising activity of the combination in a difficult-to-treat patient population Besides the analysis of the main study end points (response rate and safety), we acknowledge that the small sample size of this study does not allow any firm conclusions to be drawn on the interpretation of secondary end points, and in particular on PFS and OS data. Moreover, the lack of a centralized histology review and cell of origin analysis does not allow us to present any hypothesis on the differential activity of bendamustine and rituximab combination among DLBCL subtypes. Since the activity of bendamustine in DLBCL was first documented by Weidmann et al.,15 several phase II and retrospective studies have been published to assess the activity and the safety profile of this combination. In the first prospective studies on relapsed refractory patients, bendamustine was used at higher doses (120 mg/m2/d) and with a shorter interval between cycles.5-7 In these trials, the ORR ranged between 46% and 63% (CR 15-37%) and the median PFS between 3.6 and 6.7 months (Table 6). Park et al.8 recently published the results of a small phase II trial of bendamustine at 120 mg/m2 in combination with rituximab in untreated older patients (>65 years) who were poor candidates for R-CHOP. Among the 23 enrolled patients, the median age was 80 years, the ORR and the CRR were 78% and 52%, respectively, but the median PFS and OS were only 5.4 and 10 months, respectively. In our study, in consideration of the patientsâ&#x20AC;&#x2122; age and frail status, we opted for a regimen with bendamustine at a lower dose of 90 mg/m2 combined with standard rituximab doses, and administered every four weeks. Looking at our results, this choice did not seem to significantly reduce treatment activity compared to more intense published BR regimens. Conversely, the adoption of a less intense BR combination was well tolerated and was a good choice for patients who were prospectively identi-

fied as frail. Comparing our data with those from prior phase II reports, it should also be acknowledged that Park et al. did not include stage I disease and had more poor performance status patients, while Weidmann et al. had more early stage patients, and a higher median age (85 years). Our data should also be compared with a previous analysis of 99 elderly frail patients with DLBCL who were analyzed in a prospective observational study.12 The

Figure 1. Estimated progression-free survival (PFS) with 95% confidence interval (gray area).

Table 4. Overall toxicities according to CTCAE v.4.0 categories with grade.

All grades Anemia Leucopenia Neutropenia Thrombocytopenia Febrile neutropenia Infections Fever Cardiac disorders Gastrointestinal disorders General disorders and administration site conditions* Hepatobiliary disorders Metabolism and nutrition disorders Nervous system disorders Renal and urinary disorders Skin and subcutaneous tissue disorders Vascular disorders Other (specify) **

Grade 3

Grade 4

20 19 29 20 3 9 2 4 14

44.4 42.2 64.4 44.4 6.7 20.0 4.4 8.9 31.1

1 3 8 4 0 2 0 1 0

2.2 6.7 17.8 8.9 0.0 4.4 0.0 2.2 0.0

0 2 9 0 1 0 0 1 0

0.0 4.4 20.0 0.0 2.2 0 0.0 2.2 0.0

5 2 3 4 4 9 2 4

11.1 4.4 6.7 8.9 8.9 20.0 4.4 8.9

1 0 1 0 1 0 0 0

2.2 0.0 2.2 0.0 2.2 0.0 0.0 0.0

0 0 0 0 0 0 0 0

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

CTCAE: Common Terminology Criteria for Adverse Events; n: number. *Asthenia; laboratory abnormalities; fever. **Other - Grade 1: epistaxis and flu; Grade 2: accidental fall and cough.

1348

haematologica | 2018; 103(8)

BR in frail elderly patients with DLBCL

majority of frail subjects of this study were treated with rituximab-containing therapy (39%) and 31 of them also received an anthracycline-containing regimen. Considering only the patients who were treated with chemo-immunotherapy, the 2-year and median OS were 48% and 20 months, respectively. These data compare favorably with those observed in the current study that used a more strict definition of frailty (i.e. unfit patients with intermediate profile at CGA were not included) and were obtained without the use of doxorubicin. When this trial was started, data from a large observational study on 173 consecutive elderly patients with DLBCL were already available. These suggested that the use of chemotherapy regimens with curative intent were not able to improve patients’ survival compared to palliative therapies for the subgroup of patients who were classified as frail according to the same CGA that we adopted in this trial.10 Based on this observation, we considered frail those patients ineligible to receive anthracycline-containing regimens, also if administered at lower doses, and identified an unmet need in the search of active therapies in this subset of patients. Unfortunately, lacking a randomized comparison, we cannot draw any conclusion on the relative efficacy of the BR regimen in comparison with other immunochemotherapy options. Due to its favorable safety profile, however, the use of a BR combination seems an excellent option and should be set as a reference to identify more effective treatment strategies in future trials. Regarding safety, the recent experience with BR in follicular lymphoma reported an increase in toxicities; with the limitations of a small number of patients, in our population, BR was manageable and safe; the two deaths due to second malignancies is not an unexpected finding in a very elderly population. In order to further improve these results, additional efforts should be made to attempt to increase the response rate and to prolong the short duration of response. New drugs such as lenalidomide and ibrutinib have shown activity against DLBCL, are both associated with an excellent safety profile,16,17 and could be used to improve the results with an acceptable toxicity. A phase II trial of rituximab in association with a lenalidomide combination in elderly frail patients is currently being conducted by our group and actively recruiting patients (clinicaltrials.gov identifier: 02955823). In addition, the published results of the REMARC trial has demonstrated, for the first time after several unsuccessful attempts with other drugs,18 that a maintenance therapy with lenalidomide in elderly patients with DLBCL who responded to initial immunochemother-

apy is associated with a reduced risk of disease progression compared to observation.19 Finally, our study is part of a larger project for elderly patients with DLBCL for whom a preliminary CGA is required to define patient fitness status and to adapt treatment goals accordingly. Patients prospectively enrolled in this so called “elderly project” are evaluated by a simplified version of the CGA20 and are categorized into one of three groups: fit, unfit and frail (clinicaltrials.gov identifier: 02364050). Fit patients are then offered a standard R-CHOP treatment with curative intent, and unfit patients are considered better candidates for adapted RCHOP regimens with reduced drug doses to achieve a cure with reduced toxicity. Finally, no clear benefit was observed for frail patients treated with curative intent compared with those treated with palliative intent, and there was no standard or reference regimen.21 To the best of our knowledge, the elderly project is the first attempt to try to objectify the treatment approach to elderly patients with lymphoma and to promote clinical research in this population. With our study, and with the adoption of a prospective definition of patient fitness, we have been able to show that, also in the population of frail patients, the use of a low toxicity regimen allows a cure of the lymphoma to be achieved in a good proportion of patients. These data, along with the adopted CGA evaluation, help

Table 5. Association of age, geriatric scales and IPI with OS and PFS (univariate analysis).

Variable Age ≤80 >80 ADL 6 5 0-4 IADL 8 6-7 0-5 IPI 1-2 3-5

1yr-OS % (95CI)

Log-rank P

1yr PFS % (95%CI)

0.694 68 (29-88) 62 (44-76)

Log-rank P 0.685

60 (25-83) 46 (29-61) 0.335

70 (48-85) 50 (21-74) 63 (23-86)

0.536 56 (35-73) 25 (6-50) 62 (23-86)

0.350 33 (5-68) 71 (49-85) 62 (32-82)

0.020 33 (5-68) 56 (35-73) 43 (18-66)

0.531 77 (50-91) 52 (31-69)

0.581 58 (33-76) 40 (21-58)

yr: year; CI: Confidence Interval; ADL: Activities of Daily Living; IADL: Instrumental Activities of Daily Living; IPI: International Prognostic Index; OS: overall survival; PFS: progression-free survival.

Table 6. Summary of prospective studies of bendamustine in association with rituximab in diffuse large B-cell lymphoma.

Author Ohmachi et al.6 Vacirca et al.7 Weidmann et al.9 Park et al.8 Current report

Year (medAge)

Patients

Phase

Benda dose mg/m2

Days

CRR%

ORR%

mPFS months

2013 2014 2011 2016 2016

59 (65) 48 (74) 13 (85) 23 (80) 45 (81)

R/R R/R Untr. Untr. Untr.

120 90/120* 120 120 90

21 28 21 21 28

37.7 15.3 54 52 53

62.7 45.8 69 78 62

6.7 3.6 7.7 5.4 10

*Bendamustine (Benda) dosage 90 mg/m2 was administered for the first 2 patients on study. Following the US FDA approval, the dosage was amended to 120 mg/m2. R/R: relapsed/refractory; medAge: median age; Untr: untreated; CRR: complete remission rate; ORR: overall response rate; mPFS: median progression-free survival.

haematologica | 2018; 103(8)

1349

S. Storti et al.

to define new reference points for future therapeutic development in an otherwise highly heterogeneous and difficult-to-treat population. In conclusion, considering our results, and the other available data discussed above, we believe that the combination of bendamustine and rituximab, even if not curative, is a good option for the treatment of elderly frail patients with DLBCL, also when used at reduced doses and administered at 4-week intervals as in our study. Treatment of elderly frail patients remains a challenge for the clinician, and the choice of treatment should be indi-

References 1. Bellera C, Praud D, Petit-Moneger A, McKelvie-Sebileau P, Soubeyran P, Mathoulin-Pelissier S. Barriers to inclusion of older adults in randomised controlled clinical trials on Non-Hodgkin's lymphoma: a systematic review. Cancer Treat Rev. 2013;39(7):812-817. 2. Jemal A, Siegel R, Ward E, Murray T, Xu J, Thun MJ. Cancer statistics, 2007. CA Cancer J Clin. 2007;57(1):43-66. 3. Coiffier B, Lepage E, Briere J, et al. CHOP chemotherapy plus rituximab compared with CHOP alone in elderly patients with diffuse large-B-cell lymphoma. N Engl J Med. 2002;346(4):235-242. 4. Peyrade F, Jardin F, Thieblemont C, et al. Attenuated immunochemotherapy regimen (R-miniCHOP) in elderly patients older than 80 years with diffuse large B-cell lymphoma: a multicentre, single-arm, phase 2 trial. Lancet Oncol. 2011;12(5):460468. 5. Ogura M, Ando K, Taniwaki M, et al. Feasibility and pharmacokinetic study of bendamustine hydrochloride in combination with rituximab in relapsed or refractory aggressive B cell non-Hodgkin's lymphoma. Cancer Sci. 2011;102(9):1687-1692. 6. Ohmachi K, Niitsu N, Uchida T, et al. Multicenter phase II study of bendamustine plus rituximab in patients with relapsed or refractory diffuse large B-cell lymphoma. J Clin Oncol. 2013;31(17):21032109. 7. Vacirca JL, Acs PI, Tabbara IA, Rosen PJ, Lee P, Lynam E. Bendamustine combined with rituximab for patients with relapsed or refractory diffuse large B cell lymphoma. Ann Hematol. 2014;93(3):403-409.

1350

vidualized considering all available options and through an accurate assessment of the risk-benefit ratio for each single patient. At least as important as the choice of treatment, a concerted effort should be made to adopt validated tools to assess patient fitness status before treatment start and to adapt treatment goals accordingly. In this context, we recommend the use of the simplified version of the CGA as used in this study. Funding This work was supported by Mundipharma Pharmaceuticals.

8. Park SI, Grover NS, Olajide O, et al. A phase II trial of bendamustine in combination with rituximab in older patients with previously untreated diffuse large B-cell lymphoma. Br J Haematol. 2016;175(2): 281-289. 9. Weidmann E, Neumann A, Fauth F, et al. Phase II study of bendamustine in combination with rituximab as first-line treatment in patients 80 years or older with aggressive B-cell lymphomas. Ann Oncol. 2011;22(8):1839-1844. 10. Tucci A, Martelli M, Rigacci L, et al. Comprehensive geriatric assessment is an essential tool to support treatment decisions in elderly patients with diffuse large B-cell lymphoma: a prospective multicenter evaluation in 173 patients by the Lymphoma Italian Foundation (FIL). Leuk Lymphoma. 2015;56(4):921-926. 11. Monfardini S, Aversa SM, Zoli V, et al. Vinorelbine and prednisone in frail elderly patients with intermediate-high grade nonHodgkin's lymphomas. Ann Oncol. 2005;16(8):1352-1358. 12. Merli F, Luminari S, Rossi G, et al. Outcome of frail elderly patients with diffuse large Bcell lymphoma prospectively identified by Comprehensive Geriatric Assessment: results from a study of the Fondazione Italiana Linfomi. Leuk Lymphoma. 2014;55(1):38-43. 13. Cheson BD, Horning SJ, Coiffier B, et al. Report of an international workshop to standardize response criteria for nonHodgkin's lymphomas. NCI Sponsored International Working Group. J Clin Oncol. 1999;17(4):1244. 14. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579586.

15. Weidmann E, Kim SZ, Rost A, et al. Bendamustine is effective in relapsed or refractory aggressive non-Hodgkin's lymphoma. Ann Oncol. 2002;13(8):1285-1289. 16. Hitz F, Zucca E, Pabst T, et al. Rituximab, bendamustine and lenalidomide in patients with aggressive B-cell lymphoma not eligible for anthracycline-based therapy or intensive salvage chemotherapy - SAKK 38/08. Br J Haematol. 2016;174(2):255-263. 17. Maddocks K, Christian B, et al. A phase 1/1b study of rituximab, bendamustine, and ibrutinib in patients with untreated and relapsed/refractory non-Hodgkin lymphoma. Blood. 2015;125(2):242-248. 18. Habermann TM, Weller EA, Morrison VA, et al. Rituximab-CHOP versus CHOP alone or with maintenance rituximab in older patients with diffuse large B-cell lymphoma. J Clin Oncol. 2006;24(19):31213127. 19. Thieblemont C, Tilly H, Gomes da Silva M, et al. Lenalidomide Maintenance Compared With Placebo in Responding Elderly Patients With Diffuse Large B-Cell Lymphoma Treated With First-Line Rituximab Plus Cyclophosphamide, Doxorubicin, Vincristine, and Prednisone. J Clin Oncol. 2017;35(22):2473-2481. 20. Spina M, Balzarotti M, Uziel L, et al. Modulated chemotherapy according to modified comprehensive geriatric assessment in 100 consecutive elderly patients with diffuse large B-cell lymphoma. Oncologist. 2012;17(6):838-846. 21. Tucci A, Ferrari S, Bottelli C, Borlenghi E, Drera M, Rossi G. A comprehensive geriatric assessment is more effective than clinical judgment to identify elderly diffuse large cell lymphoma patients who benefit from aggressive therapy. Cancer. 2009;115(19):4547-4553.

haematologica | 2018; 103(8)

ARTICLE

Non-Hodgkin Lymphoma

Marek Trnĕný,1 Gregor Verhoef,2 Martin JS Dyer,3 Dina Ben Yehuda,4 Caterina Patti,5 Miguel Canales,6 Andrés Lopez,7 Farrukh T Awan,8 Paul G Montgomery,9 Andrea Janikova,10 Anna M Barbui,11 Kazimierz Sulek,12 Maria J Terol,13 John Radford,14 Anna Guidetti,15,16 Massimo Di Nicola,15 Laure Siraudin,17 Laurence Hatteville,18 Sandrine Schwab,19 Corina Oprea18 and Alessandro M Gianni15,16

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1351-1358

Charles University, General Hospital, Prague, Czech Republic; 2Department of Hematology, University Hospital, Leuven, Belgium; 3Ernest and Helen Scott Haematological Research Institute, University of Leicester, UK; 4Hadassah Medical Center, Jerusalem, Israel; 5 PA Cervello EMAT, Palermo, Italy; 6Hospital la Paz, Madrid, Spain; 7Vall d’Hebron Research Institute, Barcelona, Spain; 8Ohio State University, Columbus, OH, USA; 9Boise VA Medical Center, Boise, ID, USA; 10Department of Hematology and Oncology, University Hospital Brno, Czech Republic; 11Azienda Ospedaliera Papa Giovanni XXIII, Bergamo, Italy; 12Wojskowy Instytut Medyczny, Warsaw, Poland; 13Hospital Clínico Universitario de Valencia, Health Research Institute INCLIVA, Spain; 14University of Manchester and The Christie NHS Foundation Trust, Manchester Academic Health Science Centre, UK; 15Fondazione Istituto Nazionale Tumori, Milan, Italy; 16University of Milan, Italy; 17Lincoln, Paris, France; 18Sanofi R&D, Vitry sur Seine, France and 19Sanofi R&D, Chilly-Mazarin, France 1

ABSTRACT

his phase II, single-arm, multicenter study examined the efficacy and safety of coltuximab ravtansine (an anti-CD19 antibody drug conjugate) in 61 patients with histologically documented (de novo or transformed) relapsed or refractory diffuse large B-cell lymphoma who had previously received rituximab-containing immuno-chemotherapy. Patients had received a median of 2.0 (range 0-9) prior treatment regimens for diffuse large B-cell lymphoma and almost half (45.9%) had bulky disease (≥1 lesion >5 cm) at trial entry. Patients received coltuximab ravtansine (55 mg/m2) in 4 weekly and 4 biweekly administrations until disease progression or unacceptable toxicity. Forty-one patients were eligible for inclusion in the per protocol population. Overall response rate (International Working Group criteria) in the per protocol population, the primary end point, was 18/41 [43.9%; 90% confidence interval (CI:) 30.6-57.9%]. Median duration of response, progressionfree survival, and overall survival (all treated patients) were 4.7 (range 0.0-8.8) months, 4.4 (90%CI: 3.02-5.78) months, and 9.2 (90%CI: 6.5712.09) months, respectively. Common non-hematologic adverse events included asthenia/fatigue (30%), nausea (23%), and diarrhea (20%). Grade 3-4 adverse events were reported in 23 patients (38%), the most frequent being hepatotoxicity (3%) and abdominal pain (3%). Eye disorders occurred in 15 patients (25%); all were grade 1-2 and none required a dose modification. Coltuximab ravtansine monotherapy was well tolerated and resulted in moderate clinical responses in pre-treated patients with relapsed/refractory diffuse large B-cell lymphoma. (Registered at: clinicaltrials.gov identifier: 01472887)

haematologica | 2018; 103(8)

Correspondence: trneny@cesnet.cz

Received: April 10, 2017. Accepted: May 3, 2018. Pre-published: May 10, 2018.

doi:10.3324/haematol.2017.168401 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1351 ©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.

1351

M. Trnĕný et al.

Introduction Diffuse large B-cell lymphoma (DLBCL) is the most frequent form of non-Hodgkin lymphoma, representing approximately 30-58% of cases.1 The majority of cases of DLBCL occur de novo, although some develop from indolent lymphoma.2 DLBCL is subclassified as germinal center B-cell-like (GCB) or activated B-cell-like (ABC) subtypes based on gene expression profiling. The ABC subtype has a worse prognosis than the GCB subtype.3 In addition, concurrent deregulation of MYC and BCL2 has been associated with poor outcomes,4,5 however the prognostic significance of these rearrangements remains controversial.6-8 Standard first-line therapy for DLBCL is cyclophosphamide, hydroxydaunorubicin, vincristine, and prednisone, combined with rituximab (R-CHOP). Five-year overall survival (OS) in patients treated with this regimen is over 70%.9,10 Dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab (DA-EPOCH-R), showed promise as an alternative firstline regimen to R-CHOP in a phase II study,11 but failed to demonstrate superior event-free survival or OS in a phase III trial which directly compared the two regimens.12 The majority of patients in the phase III study had good prognostic features, and therefore it is possible that DAEPOCH-R may provide an advantage in patients with an adverse prognosis (such as MYC/BCL2 double-hit lymphoma) or rare subtypes (such as primary mediastinal lymphoma). However, the phase III study was not designed to answer this question, and R-CHOP remains the standard of care for the majority of unselected patients with DLBCL.12-15 Salvage treatment with autologous stem cell transplantation (ASCT) is the most effective approach at first relapse. However, it can only be offered to young, fit patients, and long-term survival is only 40%.16 There are limited treatment options with unsatisfying results for patients relapsing after, or ineligible for, ASCT.17 New therapeutic strategies are essential for these patients. Coltuximab ravtansine (SAR3419) is an anti-CD19 monoclonal antibody conjugated to a potent cytotoxic maytansinoid, DM4, via an optimized, hindered, disulfide bond. The antibody selectively binds to the CD19 antigen present on the majority of B cells, resulting in internalization of the receptor-drug complex and intracellular release of DM4. DM4 is a potent inhibitor of tubulin polymerization and microtubule assembly, functioning by similar mechanisms to vincristine and vindesine.18,19 Coltuximab ravtansine has been evaluated in patients with relapsed/refractory (R/R) B-cell non-Hodgkin lymphoma. A first-in-human phase I study examined several dose levels in 3-weekly administrations. At the maximum tolerated dose (160 mg/m2) few clinical responses and high levels of treatment-related ocular toxicity were observed.20 A further phase I, dose-escalation study examined onceweekly dosing and a modified schedule consisting of 4 weekly doses followed by 4 doses given once every 2 weeks. Both schedules showed anti-lymphoma activity in approximately 30% of patients with either indolent or aggressive disease. The maximum tolerated dose was 55 mg/m2, and the modified dosing schedule was found to limit drug accumulation, reduce toxicity, and improve response rates.19 To confirm the clinical benefit observed in the phase I setting in a population with aggressive lymphoma, we 1352

conducted a phase II, open-label, multicenter study evaluating coltuximab ravtansine monotherapy in transplantineligible patients with CD19-positive, R/R DLBCL.

Methods Study design In this phase II, open-label, single-arm study patients received 4 weekly doses of intravenous (iv) coltuximab ravtansine 55 mg/m2, followed by a 1-week rest period, then biweekly doses until disease progression (PD), unacceptable toxicity, or discontinuation of treatment. One cycle was 4 weeks, except for cycle 1 (5 weeks). At the investigator’s discretion, patients received premedication consisting of iv diphenhydramine 50 mg and oral acetaminophen 650 mg 30-45 minutes before each infusion. Dose reductions were permitted (see Online Supplementary Methods).

Patients Adult patients with de novo or transformed histologically confirmed DLBCL and more than 30% of cells expressing CD19 (local assessment) were enrolled. Patients had relapsed (progression ≥6 months after completion of last line of therapy) or refractory (progression during or within 6 months of a prior therapy) disease and had previously received standard chemotherapy (including rituximab). Patients with primary refractory disease (refractory to first-line therapy) were ineligible. However, some primary refractory patients were wrongly enrolled (see Results section). Full inclusion and exclusion criteria are included in the Online Supplementary Appendix. All patients provided written informed consent. The protocol and subsequent amendments were approved by independent ethics committees and/or institutional review boards at each center. The study was conducted according to the Declaration of Helsinki.

Outcomes The primary end point was overall response rate [ORR; proportion of patients achieving a partial response (PR) or complete response (CR) (International Working Group criteria21)]. Secondary end points included duration of response (DOR; time from first PR or CR until PD or death), progression-free survival (PFS; time from first study treatment until PD or death), OS (time from first study treatment until death), and safety. Assessment of biomarkers was an exploratory end point.

Assessments Assessment of clinical response involved physical examination, bone marrow biopsy, and computerized tomography (CT) every 12 weeks until PD or treatment discontinuation. Positron-emission tomography (PET) was performed at baseline and, if positive, repeated to confirm a CR. Patients with a negative CT but positive PET were classified as PR. Adverse events (AEs) were classified using National Cancer Institute Common Terminology Criteria for Adverse Events (v.4.03). Pre-specified AEs of special interest were eye disorders, neuropathy, and infusion-related reactions (all drug hypersensitivity reactions and treatment-related AEs occurring on the day of infusion). Details of biomarker assessments are included in the Online Supplementary Methods.

Statistical analysis The predicted beneficial ORR was 40% or over. Assuming haematologica | 2018; 103(8)

Coltuximab ravtansine monotherapy in DLBCL

44 patients were evaluable for response, the study had 90% power to reject the null hypothesis of an ORR of 20% with a onesided a=0.05. An ORR of less than 20% was considered clinically uninteresting based on available observations from coltuximab ravtansine and new agents in relapsed/refractory non-Hodgkin lymphoma and/or DLBCL, for which activity ranged between 15% and 30% in phase II studies.22-28 The primary end point (ORR), was assessed in the per protocol (PP) population (all treated patients who had an evaluable response assessment during or at the end of treatment or who died due to PD before response assessment, without any important protocol deviations affecting efficacy at study entry). ORR was also assessed in the biomarkerevaluable population (all patients with results of biomarker analysis from a fresh or archival sample). DOR and PFS were assessed in the PP population, and OS and safety were assessed in all treated patients (safety population). Statistical analyses of biomarkers are detailed in the Online Supplementary Appendix.

Results Overall, 61 patients were enrolled (January 20, 2012 to July 23, 2013) and received at least one dose of study drug (safety population). Twenty patients were excluded from the PP population (Figure 1), of whom 16 were wrongly enrolled in the study due to misclassification of their prior treatment history. Of these 16 patients, primary refracto-

ry disease was the sole important deviation at study entry in 14. The primary end point (ORR) was analyzed separately in this subgroup. Baseline characteristics of the safety population are summarized in Table 1. Most patients (50 of 61 patients; 82.0%) presented with DLBCL at initial diagnosis. Of those patients with transformed lymphoma (n=11), 7 were initially diagnosed with follicular lymphoma, and 9 had received prior anticancer therapy for non-DLBCL lymphoma (6 patients received â&#x2030;Ľ1 prior anti-CD20-containing regimen). Almost half of the patients (45.9%) had bulky disease (defined as longest diameter of the lesion >5 cm for at least one location). Patients had received a median of 2.0 (range 0-9) prior treatment regimens for DLBCL, with 18 patients (29.5%) having received 3 or more prior regimens. Patients received a median of 3 (range 1-10) cycles of therapy [median duration of treatment 13.3 (range 5-41) weeks]. Thirty-nine of 61 treated patients (63.9%) received 3 or more treatment cycles, including 16 patients who received 6 cycles or more. Overall, 56 patients discontinued treatment due to: PD (n=47), AEs (n=6), or investigatorâ&#x20AC;&#x2122;s decision (n=3). At the time of analysis (May 6, 2014), 5 patients were continuing on therapy. The ORR (primary end point), analyzed in the PP population (n=41), was 43.9% (18 of 41; 90%CI: 30.6-57.9%); therefore, the null hypothesis was rejected (P<0.0001). Among the 18 responders, 6 achieved CR (PET negative) and 12 achieved PR [PET positive (n=8) or not examined

Figure 1. Inclusion of patients in the per protocol (PP) efficacy analysis. Patients were recruited at 28 sites in the USA, Belgium, Czech Republic, Israel, Italy, Poland, Spain, Turkey, and UK. The PP population consisted of all treated patients who had an evaluable response assessment during or at the end of the treatment protocol or who died due to progressive disease before response assessment, without any important protocol deviations affecting efficacy at study entry. CT: computed tomography; DLBCL: diffuse large B-cell lymphoma. *Some patients met multiple exclusion criteria. â&#x20AC; Fourteen patients had primary refractory disease as their only protocol deviation.

haematologica | 2018; 103(8)

1353

M. Trnĕný et al.

(n=4)] (Table 2). Seven patients (7 of 41; 17.1%) had stable disease, and the remaining patients (16 of 41; 39.0%) had progressive disease. Higher response rates were observed among patients with relapsed DLBCL (14 of 26; 53.8%, 90%CI: 36.2-70.8%) compared with patients refractory to their last regimen (4 of 15; 26.7%, 90%CI: 9.7-51.1%). A higher ORR (56.3%) was also observed in patients who received only one prior therapeutic regimen (n=16). At the time of analysis, 6 patients with relapsed disease were still responding to therapy (3 CRs and 3 PRs). Table 1. Baseline characteristics (safety population; n=61).

Variable

Value, n (%)

Median age (range), years Age group, years <65 65–75 >75 Sex Male Female Histology (investigator determined) De novo DLBCL Transformed DLBCL Cell of origin classification* ABC GCB Unclassified ECOG performance status† 0 1 2 Ann Arbor stage I II III IV International Prognostic Index score Low Low intermediate High intermediate High Lactate dehydrogenase >ULN† Extranodal involvement Bulky disease‡ Prior transplant for DLBCL Disease status at study entry§ Primary refractory Refractory to last regimen Relapsed Number of prior regimens for DLBCL 0 1 2 3 >3 Prior regimen for non-DLBCL lymphoma¶

69 (30–88) 17 (27.9%) 26 (42.6%) 18 (29.5%) 31 (50.8%) 30 (49.2%) 50 (82.0%) 11 (18.0%) 16 (43.2%) 17 (45.9%) 4 (10.8%) 27 (45.0%) 26 (43.3%) 7 (11.7%) 4 (6.6%) 11 (18.0%) 15 (24.6%) 31 (50.8%) 12 (19.7%) 11 (18.0%) 25 (41.0%) 13 (21.3%) 41 (68.3%) 36 (59.0%) 28 (45.9%) 12 (19.7%) 16 (26.7%) 16 (26.7%) 28 (46.7%) 1 (1.6%) 25 (41.0%) 17 (27.9%) 9 (14.8%) 9 (14.8%) 9 (81.8%)

Data are number (n) (%) unless otherwise stated. ABC: activated B-cell-like; DLBCL: diffuse large B-cell lymphoma; ECOG: Eastern Cooperative Oncology Group; GCB: germinal center B-cell-like; ULN: upper limit of normal. *N=37. †N=60. ‡Longest diameter of lesion >5 cm for at least 1 location. §N=60 (1 patient had received no prior regimen for DLBCL). ¶N=11 (patients with transformed DLBCL).

1354

Overall response rate was also assessed in 14 patients with primary refractory disease (sole important deviation affecting efficacy) who were excluded from the PP population. Among these patients, the ORR was 21.4% (3 of 14; 90%CI: 6.1-46.6%), with the majority having PD (9 of 14; 64.3%) and only one patient achieving CR. Figure 2 shows the DOR in individual patients in the PP population according to initial responses. The median DOR was 4.7 (range 0-8.8) months. Of 18 patients who responded to coltuximab ravtansine treatment (PR or better), 4 achieved a DOR of >6 months (one of 4 patients with refractory disease and 3 of 14 patients with relapsed disease). At the time of analysis, 34 of 41 patients (82.9%) in the PP population had experienced PD and the median PFS was 4.4 (90%CI: 3.02-5.78) months. Forty-one of the 61 patients in the safety population had died at the analysis cut-off date. Estimated median OS was 9.2 (90%CI: 6.57-12.09) months (Figure 3). CD19 was locally assessed in all patients (n=41) during enrollment, and centrally assessed in 37 of 41 PP patients (90.2%) during biomarker analysis. Overall, 35 patients had 30% or more CD19-positive cells (range 30-100%). Variable levels of expression were recorded, with 11, 16, and 8 samples having a mean intensity of 1+, 2+, and 3+, respectively. The median H-score (see Online Supplementary Methods) was 162 (range 0-270). There was no relationship between levels of expression of CD19 and response; some patients with high CD19 expression had PD as their best response whereas some patients with lower expression experienced a PR (Online Supplementary Figure S1). Two patients with absent CD19 staining had progressive disease. For each measure of CD19 expression, the receiver operating characteristic curve AUC values varied between 0.42 and 0.65, indicating that none of the CD19 expression measures showed good predictive accuracy for distinguishing between responders and nonresponders (Online Supplementary Table S1). No significant optimal cut-off point for CD19 expression was identified. In addition, there was no apparent correlation between cell of origin classification or MYC/BCL2 expression and response rate (data not shown). All 61 patients in the safety population (Table 3) experienced at least one AE, including 33 of 61 patients (54%) who experienced at least one treatment-related AE. Grade 3-4 AEs were reported in 23 of 61 patients (38%), the most frequent being hepatotoxicity (2 of 61, 3%) and abdominal pain (2 of 61, 3%). Serious AEs (SAEs) were reported in 24 of 61 patients (39%). Six SAEs (occurring in Table 2. Summary of best response to treatment by subgroup based on International Working Group criteria.

Response, n (%) ORR 90% CI* CR PR SD PD

All (n=41)

Refractory to last regimen (n=15)

Relapsed (n=26)

Primary refractory (n=14)

18 (43.9%) 30.6–57.9 6 (14.6%) 12 (29.3%) 7 (17.1%) 16 (39.0%)

4 (26.7%) 9.7–51.1 1 (6.7%) 3 (20.0%) 3 (20.0%) 8 (53.3%)

14 (53.8%) 36.2–70.8 5 (19.2%) 9 (34.6%) 4 (15.4%) 8 (30.8%)

3 (21.4%) 6.1–46.6 1 (7.1%) 2 (14.3%) 2 (14.3%) 9 (64.3%)

n: number; CI: confidence interval; CR: complete response; ORR: overall response rate; PD: progressive disease; PR: partial response; SD: stable disease. *Estimated by Clopper–Pearson exact method.

haematologica | 2018; 103(8)

Coltuximab ravtansine monotherapy in DLBCL

3 patients) were considered related to treatment: hepatotoxicity (n=2), pneumonia, abdominal pain, nausea, and grade 5 febrile neutropenia (n=1). The most common grade 3-4 hematologic laboratory abnormalities were neutropenia (25%), lymphopenia (21%), and leukopenia (15%) (Table 3). Grade 3-4 nonhematologic laboratory abnormalities were rare, with elevated levels of aspartate aminotransferase, alkaline phosphatase, alanine aminotransferase, and creatinine each occurring in 2 patients. Grade 3-4 febrile neutropenia was also observed in one patient, but this did not require growth factor administration.

Eye disorders occurred in 15 patients (25%); all were grade 1-2 and none required a dose modification. Nineteen extracorneal eye disorders were observed in 13 patients (21.3%), with the first occurrence during cycle 1 (6 patients), cycle 2 (n=2), cycle 3 (n=3), cycle 7 (n=1), and cycle 9 (n=1). Fourteen of these events had resolved at the time of data cut-off, with a median recovery time of 12.5 days (range 1-47). One patient experienced a corneal event (grade 2 keratitis during cycle 4), which resolved within 9 days. A further 2 patients experienced dry eyes, occurring during cycle 1 and resolving after 13 and 17 days, respectively. Neuropathy was observed in

Figure 2. Duration of response for individual patients in the per protocol population. Patients with a duration of response of 0.03 months were censored to the first documentation of the response, in the absence of another evaluable assessment before the cut-off date. CR: complete response; DLBCL: diffuse large B-cell lymphoma; PR: partial response.

Figure 3. Kaplanâ&#x20AC;&#x201C;Meier curve of progression-free survival (PFS) (per protocol population) and overall survival (OS) (safety population).

haematologica | 2018; 103(8)

1355

M. Trnĕný et al.

7 patients (11%). Five patients (8%) reported peripheral neuropathy (PN) occurring during cycle 1 (n=3) or cycle 2 (n=2), including one case of grade 3 PN (unrelated to study treatment) in a patient with a history of the condition. Dose modifications were not required in any of the patients with PN, although none of these events had resolved at the time of analysis. A further 2 patients presented with events compatible with optic neuropathy (grade 1); this diagnosis was not confirmed, but could not be confidently excluded. Neither of these patients required a dose modification and both events resolved within a median of 9 days (range 4-14). Overall, infusionrelated reactions occurred in 10 patients (16%), and were most commonly gastrointestinal in nature (nausea 10%, vomiting 3%). Drug hypersensitivity was observed in one patient. Dose modifications (dose omission, interruption, or cycle delay) due to AEs were required in 17 patients (28%), including 9 patients (15%) who experienced a grade 3-4 AE. Nine patients (15%) had at least one cycle delayed by >3 days, and 9 patients (15%) had one dose omitted. One patient (2%) required a dose interruption due to grade 1 hypotension, which was considered to be unrelated to treatment. Of 8 patients (13%) who experienced AEs leading to death, 7 were due to PD. The other patient who died developed febrile neutropenia 34 days after the last dose of coltuximab ravtansine while receiving further anticancer therapy (gemcitabine-cisplatin); the investigator could not exclude the possibility that the event was due to a delayed effect of coltuximab ravtansine treatment.

Discussion The results of this phase II trial indicate that treatment with coltuximab ravtansine as monotherapy is associated with moderate clinical responses in a proportion of DLBCL patients previously treated with rituximab-based chemotherapy, and has a favorable toxicity profile. The responses described here are numerically higher than those reported in a phase II study of coltuximab ravtansine in combination with rituximab [ORR 44% (90%CI: 30.6-57.9%) vs. 31% (90%CI: 22.0-41.6%), respectively].29 However, patients enrolled in the combination study were limited to 3 cycles of treatment, whereas in the current study patients continued on therapy until disease progression or discontinuation due to an AE or investigator’s decision. Additionally, the patients in the combination therapy study could be described as a more refractory population (60% of patients had primary refractory disease), whereas the primary analysis population for the current study excluded patients with primary refractory disease. It should be noted that some patients with primary refractory disease were wrongly included in this study due to a misclassification of their prior treatment history. The response rates described here are in line with other antibody-drug conjugates, when tested as monotherapy (44-56%)30,31 or in combination with rituximab (29-54%).32,33 Interestingly, the anti-CD30 antibodydrug conjugate brentuximab vedotin achieved an ORR of 44% among patients with R/R DLBCL, most of whom were refractory to their first (76%) and last (82%) line of therapy.30 The response rates were also similar to antiCD19 monoclonal antibodies, such as MEDI-551, 1356

Table 3. Adverse events (AEs) occurring in ≥10% of patients (safety population; n=61).

AE, n (%)

All grades

Grade 3-4

Grade 5

Any AE Serious AEs AE leading to dose modification* AE leading to discontinuation Non-hematologic AEs Asthenia/fatigue Nausea Diarrhea Cough Vomiting Decreased appetite Disease progression Back pain Abdominal pain Dyspnea Constipation Peripheral edema Laboratory abnormalities Hematologic AEs† Anemia Lymphopenia Leukopenia Thrombocytopenia Neutropenia Hepatic and renal abnormalities AST Alkaline phosphatase‡ ALT Creatinine Bilirubin

61 (100%) 24 (39%) 17 (28%) 4 (7%)

23 (38%) 14 (23%) 9 (15%) 0

8 (13%) 8 (13%) – –

18 (30%) 14 (23%) 12 (20%) 11 (18%) 8 (13%) 8 (13%) 8 (13%) 7 (11%) 7 (11%) 6 (10%) 6 (10%) 6 (10%)

1 (2%) 1 (2%) 0 0 0 0 3 (5%) 1 (2%) 2 (3%) 1 (2%) 0 0

0 0 0 0 0 0 5 (8%) 0 0 0 0 0

53 (87%) 41 (67%) 39 (64%) 35 (57%) 32 (52%)

4 (7%) 13 (21%) 9 (15%) 6 (10%) 15 (25%)

– – – – –

37 (61%) 26 (45%) 27 (44%) 19 (31%) 9 (15%)

2 (3%) 2 (3%) 2 (3%) 2 (3%) 1 (2%)

– – – – –

ALT: alanine aminotransferase; AST: aspartate aminotransferase. *Including dose omission, interruption, and cycle delays. †Laboratory evaluations. ‡N=58.

MOR208, and blinatumomab, currently in phase II development for DLBCL.34-36 In comparison, in a recent multicenter, randomized study of the aza-anthracenedione pixantrone in patients with aggressive B-cell lymphoma (DLBCL, transformed indolent lymphoma, or follicular lymphoma),37 the ORR was 26% (CR, 15%), with a median PFS of 5.7 months (95%CI: 2.4-6.5). The response rates among patients refractory to their first or last line of therapy were numerically lower than those observed among the relapsed patients included in the study (21.4% and 26.7% vs. 53.8%, respectively). However, given the limited numbers of patients in each group it is difficult to draw firm conclusions. Biomarker analysis revealed no apparent correlation between cell of origin classification or MYC/BCL2 expression and clinical response. In addition, none of the CD19 expression measures analyzed showed good predictive accuracy for distinguishing responders and non-responders, and no significant optimal cut-off point for CD19 expression could be identified. This lack of correlation between CD19 expression and efficacy is counterintuitive, but may represent an effect of coltuximab ravtansine on the tumor microenvironment that is important for lymphoma cell growth and survival.38 Additional ad hoc analyses would be required to investigate this further. Interestingly, pre-clinical studies have also demonstrated haematologica | 2018; 103(8)

Coltuximab ravtansine monotherapy in DLBCL

that low levels of CD22 or CD79B expression on target cells does not reduce the antitumor activities of pinatuzumab vedotin or polatuzumab vedotin, respectively.39 Similar findings have also been reported in a brentuximab vedotin phase II study in DLBCL, in which responses were not dependent on CD30 expression.30 Overall, coltuximab ravtansine exhibited a favorable safety profile, with the majority of the most common AEs reported at grade 1-2. The most frequent grade 3-4 AEs were hematologic or gastrointestinal in nature. No studyonset occurrences of grade 3-4 PN or ocular toxicity were observed, and grade 1-2 toxicities were reversible and manageable. In addition, the majority of these events occurred during cycles 1-2 suggesting that they may result from the more intensive dosing of coltuximab ravtansine during the first cycle of the study, rather than drug accu-

References 1. Sant M, Allemani C, Tereanu C, et al. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood. 2010;116(19):3724-3734. 2. Montoto S, Fitzgibbon J. Transformation of indolent B-cell lymphomas. J Clin Oncol. 2011;29(14):1827-1834. 3. Dunleavy K, Wilson WH. Appropriate management of molecular subtypes of diffuse large B-cell lymphoma. Oncology (Williston Park). 2014;28(4):326-334. 4. 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. 5. 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. 6. Tilly H, Gomes da SM, Vitolo U, et al. Diffuse large B-cell lymphoma (DLBCL): ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2015;26 Suppl 5:v116-v125. 7. 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. 8. Petrella T, Copie-Bergman C, Briere J, et al. BCL2 expression but not MYC and BCL2 coexpression predicts survival in elderly patients with diffuse large B-cell lymphoma independently of cell of origin in the phase 3 LNH03-6B trial. Ann Oncol. 2017;28(5): 1042-1049. 9. Sehn LH, Berry B, Chhanabhai M, et al. The revised International Prognostic Index (R-IPI) is a better predictor of outcome than the standard IPI for patients with diffuse large B-cell lymphoma treated with RCHOP. Blood. 2007;109(5):1857-1861. 10. Zhou Z, Sehn LH, Rademaker AW, et al. An

haematologica | 2018; 103(8)

11.

12.

13.

14.

15.

16.

17.

18.

19.

mulation. Indeed Ribrag et al.19 demonstrated a reduced incidence of ocular toxicities and PN with the optimized schedule used here versus a weekly dosing schedule. Dose modifications were required in 28% of patients due to AEs, approximately half of which were grade 3-4. No dose reductions were required during the study. SAEs considered related to study treatment were uncommon. In conclusion, the results of this phase II study indicate that the optimized dosing regimen of coltuximab ravtansine may have some efficacy in patients with relapsed or refractory DLBCL, previously treated with rituximab. Acknowledgments The authors would like to thank Amy-Leigh Johnson, PhD, of Adelphi Communications Ltd. (Bollington, UK) for editorial support. This study and the editorial support were funded by Sanofi.

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. Dunleavy K, Fanale M, Lacasce A, et al. Preliminary report of a multicenter prospective phase II study of DA-EPOCH-R in MYC-rearranged aggressive B-cell lymphoma. 56th American Society of Hematology (ASH) Annual Meeting and Exposition, San Francisco, CA, USA, December 6-9, 2014. Wilson WH, Jung SH, Pitcher BN, et al. Phase III randomized Study of R-CHOP versus DA-EPOCH-R and molecular analysis of untreated diffuse large B-cell lymphoma: CALGB/Alliance 50303. Blood. 2016;128 (22):469. Vitolo U, Trneny M, Belada D, et al. Obinutuzumab or rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone in previously untreated diffuse large B-cell lymphoma. J Clin Oncol. 2017:JCO2017733402. Arakaki H, Nakazato T, Osada Y, Ito C, Aisa Y, Mori T. Comparison of R-CVP with R-CHOP for very elderly patients aged 80 or over with diffuse large B cell lymphoma. Ann Hematol. 2017;96(7):1225-1226. Nolasco-Medina D, Reynoso-Noveron N, Mohar-Betancourt A, Aviles-Salas A, Garcia-Perez O, Candelaria M. Comparison of three chemotherapy regimens in elderly patients with diffuse large B cell lymphoma: experience at a single national reference center in Mexico. Biomed Res Int. 2016;2016: 9817606. 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. Van Den Neste E, Schmitz N, Mounier N, et al. Outcomes of diffuse large B-cell lymphoma patients relapsing after autologous stem cell transplantation: an analysis of patients included in the CORAL study. Bone Marrow Transplant. 2017;52(2):216221. Raufi A, Ebrahim AS, Al-Katib A. Targeting CD19 in B-cell lymphoma: emerging role of SAR3419. Cancer Manag Res. 2013;5:225233. Ribrag V, Dupuis J, Tilly H, et al. A doseescalation study of SAR3419, an anti-CD19

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

antibody maytansinoid conjugate, administered by intravenous infusion once weekly in patients with relapsed/refractory B-cell non-Hodgkin lymphoma. Clin Cancer Res. 2014;20(1):213-220. Younes A, Kim S, Romaguera J, et al. Phase I multidose-escalation study of the antiCD19 maytansinoid immunoconjugate SAR3419 administered by intravenous infusion every 3 weeks to patients with relapsed/refractory B-cell lymphoma. J Clin Oncol. 2012;30(22):2776-2782. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579586. Advani A, Coiffier B, Czuczman MS, et al. Safety, pharmacokinetics, and preliminary clinical activity of inotuzumab ozogamicin, a novel immunoconjugate for the treatment of B-cell non-Hodgkin’s lymphoma: results of a phase I study. J Clin Oncol. 2010;28(12):2085-2093. Wiernik PH, Lossos IS, Tuscano JM, et al. Lenalidomide monotherapy in relapsed or refractory aggressive non-Hodgkin’s lymphoma. J Clin Oncol. 2008;26(30):49524957. Witzig TE, Reeder CB, LaPlant BR, et al. A phase II trial of the oral mTOR inhibitor everolimus in relapsed aggressive lymphoma. Leukemia. 2011;25(2):341-347. Witzig TE, Vose JM, Zinzani PL, et al. An international phase II trial of single-agent lenalidomide for relapsed or refractory aggressive B-cell non-Hodgkin’s lymphoma. Ann Oncol. 2011;22(7):1622-1627. Hernandez-Ilizaliturri FJ, Deeb G, Zinzani PL, et al. Higher response to lenalidomide in relapsed/refractory diffuse large B-cell lymphoma in nongerminal center B-celllike than in germinal center B-cell-like phenotype. Cancer. 2011;117(22):5058-5066. Friedberg JW, Sharman J, Sweetenham J, et al. Inhibition of Syk with fostamatinib disodium has significant clinical activity in non-Hodgkin lymphoma and chronic lymphocytic leukemia. Blood. 2010;115(13): 2578-2585. Dunleavy K, Pittaluga S, Czuczman MS, et al. Differential efficacy of bortezomib plus chemotherapy within molecular subtypes of diffuse large B-cell lymphoma. Blood. 2009;113(24):6069-6076. Coiffier B, Thieblemont C, de GS, et al. A

1357

M. Trnĕný et al. phase II, single-arm, multicentre study of coltuximab ravtansine (SAR3419) and rituximab in patients with relapsed or refractory diffuse large B-cell lymphoma. Br J Haematol. 2016;173(5):722-730. 30. Jacobsen ED, Sharman JP, Oki Y, et al. Brentuximab vedotin demonstrates objective responses in a phase 2 study of relapsed/refractory DLBCL with variable CD30 expression. Blood. 2015;125(9):13941402. 31. Palanca-Wessels MC, Czuczman M, Salles G, et al. Safety and activity of the antiCD79B antibody-drug conjugate polatuzumab vedotin in relapsed or refractory B-cell non-Hodgkin lymphoma and chronic lymphocytic leukaemia: a phase 1 study. Lancet Oncol. 2015;16(6):704-715. 32. Morschhauser F, Flinn I, Advani RH, et al. Preliminary results of a phase II randomized study (ROMULUS) of polatuzumab vedotin (PoV) or pinatuzumab vedotin (PiV) plus rituximab (RTX) in patients (Pts) with relapsed/refractory (R/R) nonHodgkin lymphoma (NHL). American

1358

Society of Clinical Oncology (ASCO), Chicago, IL, USA, May 30-June 3, 2014. Abstract 8519. 33. Wagner-Johnston ND, Goy A, Rodriguez MA, et al. A phase 2 study of inotuzumab ozogamicin and rituximab, followed by autologous stem cell transplant in patients with relapsed/refractory diffuse large B-cell lymphoma. Leuk Lymphoma. 2015;56(10): 2863-2869. 34. Viardot A, Goebeler M, Hess G, et al. Treatment of relapsed/refractory diffuse large B-cell lymphoma with the bispecific Tcell engager (TiTE) antibody construct blinatumomab: primary analysis results from an open-label, Phase 2 study. 56th Amercian Society of Hematology (ASH) Annual Meeting and Exposition, San Francisco, CA, USA, December 6-9, 2014. Abstract 4460. 35. Goswami T, Forero A, Hamadani M, et al. Phase I/II study of MEDI-551, a humanized monoclonal antibody targeting CD19, in subjects with relapsed or refractory advanced B-cell malignancies. American Society of Clinical Oncology (ASCO),

36.

37.

38.

39.

Chicago, IL, USA, June 1-5, 2012. Abstract 8065. Jurczak W, Zinzani PL, Goy A, et al. Phase IIa study of single-agents MOR208 in patients with relapsed or refractory B-cell non-Hodgkin’s lymphoma (NHL). ASCO Annual Meeting, Chicago, IL, USA, May 29-June 2, 2015. Abstract 8500. Pettengell R, Coiffier B, Narayanan G, et al. Pixantrone dimaleate versus other chemotherapeutic agents as a single-agent salvage treatment in patients with relapsed or refractory aggressive non-Hodgkin lymphoma: a phase 3, multicentre, open-label, randomised trial. Lancet Oncol. 2012;13(7): 696-706. Shain KH, Dalton WS, Tao J. The tumor microenvironment shapes hallmarks of mature B-cell malignancies. Oncogene. 2015;34(36):4673-4682. Pfeifer M, Zheng B, Erdmann T, et al. AntiCD22 and anti-CD79B antibody drug conjugates are active in different molecular diffuse large B-cell lymphoma subtypes. Leukemia. 2015;29(7):1578-1586.

haematologica | 2018; 103(8)

ARTICLE

Plasma Cell DIsorders

Maternal embryonic leucine zipper kinase inhibitor OTSSP167 has preclinical activity in multiple myeloma bone disease Joséphine Muller,1,a Arnold Bolomsky,2,a Sophie Dubois,1 Elodie Duray,1 Kathrin Stangelberger,2 Erwan Plougonven,3 Margaux Lejeune,1 Angélique Léonard,3 Caroline Marty,4 Ute Hempel,5 Frédéric Baron,1,6 Yves Beguin,1,6 Martine Cohen-Solal,4 Heinz Ludwig,2 Roy Heusschen1,b and Jo Caers1,6,b

Laboratory of Hematology, GIGA-I3, University of Liège, Belgium; Wilhelminen Cancer Research Institute, Department of Medicine I, Wilhelminenspital, Vienna, Austria; 3PEPs (Products, Environments, Processes), Chemical Engineering, Liège, Belgium; 4INSERMUMR-1132, Université Paris Diderot, France; 5Institute of Physiological Chemistry, Faculty of Medicine Carl Gustav Carus, Technische Universität Dresden, Germany and 6Department of Hematology, CHU de Liège, Belgium 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1359-1368

Joséphine Muller and Arnold Bolomsky are co-first authors. bRoy Heusschen and Jo Caers are co-senior authors.

ABSTRACT

ultiple myeloma bone disease is characterized by an uncoupling of bone remodeling in the multiple myeloma microenvironment, resulting in the development of lytic bone lesions. Most myeloma patients suffer from these bone lesions, which not only cause morbidity but also negatively impact survival. The development of novel therapies, ideally with a combined anti-resorptive and bone-anabolic effect, is of great interest because lesions persist with the current standard of care, even in patients in complete remission. We have previously shown that MELK plays a central role in proliferation-associated high-risk multiple myeloma and its inhibition with OTSSP167 resulted in decreased tumor load. MELK inhibition in bone cells has not yet been explored, although some reports suggest that factors downstream of MELK stimulate osteoclast activity and inhibit osteoblast activity, which makes MELK inhibition a promising therapeutic approach. Therefore, we assessed the effect of OTSSP167 on bone cell activity and the development of myeloma-induced bone disease. OTSSP167 inhibited osteoclast activity in vitro by decreasing progenitor viability as well as via a direct anti-resorptive effect on mature osteoclasts. In addition, OTSSP167 stimulated matrix deposition and mineralization by osteoblasts in vitro. This combined anti-resorptive and osteoblast-stimulating effect of OTSSP167 resulted in the complete prevention of lytic lesions and bone loss in myeloma-bearing mice. Immunohistomorphometric analyses corroborated our in vitro findings. In conclusion, we show that OTSSP167 has a direct effect on myeloma-induced bone disease in addition to its antimultiple myeloma effect, which warrants further clinical development of MELK inhibition in multiple myeloma.

Introduction The development of lytic bone lesions due to multiple myeloma bone disease (MMBD) is a hallmark of multiple myeloma (MM).1 MMBD occurs in more than 80% of MM patients2 and is caused by an uncoupling of bone remodeling. MMBD not only results in morbidity but also directly stimulates MM tumor growth through multiple mechanisms, resulting in a vicious cycle of bone destruction and MM growth.3,4 Although novel therapies continue to increase the life expectancy for MM patients, lytic bone lesions in these patients rarely heal.4 Bisphosphonates are the current standard of care for MMBD but can be responhaematologica | 2018; 103(8)

Correspondence: jo.caers@chu.ulg.ac.be

Received: November 26, 2017. Accepted: May 3, 2018. Pre-published: May 10, 2018. doi:10.3324/haematol.2017.185397 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1359 ©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.

1359

J. Muller et al.

sible for side effects such as osteonecrosis of the jaw, renal impairment, atypical fractures and hypocalcemia.5,6 These limitations highlight the need for new therapeutic strategies that ideally have a combined anti-MM and anti-MMBD effect. We recently reported that maternal embryonic leucine zipper kinase (MELK) expression is strongly associated with proliferative high-risk MM, and that MELK inhibition with a small molecule inhibitor, OTSSP167, reduces tumor load in a murine MM model.7 Overexpression of MELK as well as an inverse correlation between MELK expression and survival has been reported for multiple malignancies.8-10 MELK promotes cell cycle progression and interacts with M-phase inducer phosphatase 2 (CDC25B) and co-localizes with key cell cycle regulators such as cyclin B1 and cyclin-dependent kinase 1 (CDK1).11 Downstream targets of MELK include the transcription factor forkhead box protein M1 (FOXM1)12 and the histone-methyltransferase enhancer of zeste homolog 2 (EZH2).13 Of note, FOXM1 can also directly regulate MELK expression,10 presumably resulting in a positive feedback loop, and has been identified as a therapeutic target for high-risk MM.14 The role of MELK and FOXM1 in osteoclasts and osteoblasts has not yet been explored. Regarding EZH2, Fang et al. showed that EZH2 promotes osteoclastogenesis by reducing the transcription factor IRF8 and subsequent upregulation of NFATC1.15 In addition, EZH2 prevents osteogenic differentiation of mesenchymal cells and suppression of runt-related transcription factor 2 (RUNX2) has been implicated in this process.16 Inhibition of EZH2 has been shown to mitigate bone loss in ovariectomized mice17 and to reverse MM-induced suppression of osteoblast differentiation in vitro.18 Because the described roles of downstream targets of MELK suggest that its inhibition could concurrently block osteoclast function and stimulate osteoblast function, we examined the potential of OTSSP167 as a novel therapeutic agent for MMBD.

Methods

Cell viability assay and cell cycle analysis RAW264.7 and PBMC viability were assessed with the cell proliferation kit I (Roche). BMSC-TERT viability was assessed with the Cell Counting Kit 8 (Sigma-Aldrich). For cell cycle analysis, cells were stained using PI/RNase staining buffer (BD Biosciences), followed by FACS analysis on a FACSCalibur (BD Biosciences).

Osteoclast differentiation and in vitro bone matrix resorption PBMCs were seeded at a density of 750,000 cells/cm2 in alphaMEM (Lonza) supplemented with 10% FCS, 2 mM L-glutamine and 1% P/S. Cells were left to adhere for 4 hours. Next, the medium was refreshed and supplemented with 25 ng/ml human MCSF and 50 ng/ml human sRANKL (Peprotech). The culture medium was refreshed twice per week and cultures were stopped on day 14. RAW264.7-derived osteoclast cultures were established as described previously.19 TRAP activity in osteoclast cultures was detected using the Leukocyte TRAP kit (Sigma-Aldrich). Alternatively, cultures were lysed for RNA or protein extraction. Bone resorption by osteoclasts was assessed in Osteo Assay 96well plates (Corning) as described previously.19 Actin ring formation was assessed by staining cultures with phalloidin-FITC (Sigma-Aldrich), followed by analysis on an A1R confocal fluorescent microscope (Nikon).

Quantification of reactive oxygen species Reactive oxygen species (ROS) were detected using the Cellular Reactive Oxygen Species Detection Assay kit (Abcam). In short, cells were stained with DCDFA for 30 minutes at 37 ºC and fluorescence was measured (Ex/Em = 485/535 nm) on an Infinite M200 Pro plate reader (Tecan).

Osteoblast differentiation and functional analyses BMSC-TERT were seeded at a density of 25,000 cells/cm2 and grown to 70–80 % confluence. Osteoblast differentiation was initiated by changing the medium to alpha-MEM supplemented with 10 % FCS, 2 mM L-glutamine, 1% P/S, 100 nM dexamethasone, 50 μg/ml ascorbic acid and 3 mM b-glycerophosphate (Sigma-Aldrich). Osteogenic medium was refreshed twice per week. Collagen secretion and matrix mineralization were assessed by Sirius Red and Alizarin Red staining, respectively, as described previously.19,20

Reagents OTSSP167 (Biorbyt) was dissolved in DMSO and stored at 20°C. For in vivo studies, OTSSP167 was dissolved in 0.5% methylcellulose (Sigma-Aldrich) and stored at -20°C. The following antibodies were used: anti-FOXM1 (SC-502, Santa Cruz), anti-EZH2 (#4905, Cell Signaling Technology) anti-MELK (GTX111958, GeneTex and 2274S, Cell Signaling Technology), anti-a-tubulin (T6074, Sigma), anti-GAPDH (2118, Cell Signaling Technology), anti-rabbit-HRP (P0217, Agilent) and anti-mouse HRP (P0260, Agilent).

Real-time PCR Real-time PCR was performed as described previously19 using 250 nmol/L of the appropriate primers (IDT, Online Supplementary Table S1) or pre-designed Taqman gene expression assays (Applied Biosystems). Gene expression was normalized to b-actin and b2microglobubulin expression (osteoclasts) or RPLP0 (osteoblasts). Measurements were performed at least in triplicate and relative expression levels were determined using the ΔCt method.

Western blotting Cells and culture conditions Human peripheral blood mononuclear cells (PBMCs) were obtained after Ficoll (GE Healthcare) separation of whole blood. RAW264.7 cells and 5TGM.1GFP+ cells were cultured in DMEM (Lonza) supplemented with 10% fetal bovine serum (FBS)(Sigma-Aldrich), 2mM L-glutamine (Lonza) and 1% penicillin/streptomycin (P/S) (Lonza). TERT+ bone marrow mesenchymal stromal cells (BMSC-TERT) (kindly provided by Dr. D Campana, St. Jude Children’s Research Hospital, Memphis, TN, USA) were cultured in RPMI-1640 (Gibco) supplemented with 10% FCS, 2mM L-glutamine and 1% P/S. 1360

Cells were lysed in RIPA Lysis and Extraction buffer (Thermo Scientific) supplemented with cOmplete Protease Inhibitor Cocktail (Roche). Twenty μg of protein were separated by gel electrophoresis on a 10% SDS-polyacrylamide gel and transferred onto PVDF membranes (BioRad). Membranes were blocked with 5% BSA/PBS/Tween20 and incubated overnight at 4°C with primary antibodies (MELK: 1:1000, FOXM1: 1:100, EZH2: 1:1000, a-tubulin: 1:5000, GAPDH: 1:2000). The next day, blots were incubated with a HRP-conjugated secondary antibody (1:5000), followed by visualization on an ImageQuant LAS4000 (GE Healthcare). haematologica | 2018; 103(8)

OTSSP167 has activity in myeloma bone disease

Treatment of myeloma-bearing mice with OTSSP167 We used the C57BL/KaLwRij 5TGM.1 mouse model21 to assess the effect of OTSSP167 on the development of MMBD. Nineweek-old female mice were injected i.v. with 5.0 x 105 5TGM.1GFP+ cells and OTSSP167 or vehicle solution was administered by oral gavage at different dose levels. For every cohort, mice were randomly divided into a naive group (no MM, n=5), a vehicle group (MM + vehicle, n=10) and a treated group (MM + OTSSP167, n=10). When mice showed signs of active myeloma, i.e., paraplegia, all mice within a cohort were sacrificed. Bone marrow plasmocytosis was determined by flushing the bones of one leg followed by flow cytometry on a FACSCanto flow cytometer (BD Biosciences). The femur and tibia of the opposite leg were

defleshed and stored in 70% EtOH. Ethical approval was obtained for all mouse experiments (ULg license no. 1336).

Micro-computed tomography Micro-computed tomography (ÎźCT) was performed on distal femurs with the Skyscan 1172 system (Bruker), as described previously.19 3D models of bones were generated using CTVol software (Bruker). The number of cortical perforations > 50 Îźm in diameter was counted blinded on reconstructed images.

Bone histomorphometry Femurs were embedded in methylmethacrylate. All parameters were recorded as recommended by the American Society for Bone

Figure 1. OTSSP167 hampers osteoclast differentiation. A) mRNA levels of MELK, EZH2, FOXM1 and relevant transcription factors during osteoclastogenesis in murine cultures, MO: monocytes; OC: osteoclasts. B) The effect of OTSSP167 treatment on MELK protein expression in RAW264.7 cultures. C) MTT assay on PBMCs and RAW264.7 cells incubated with a range of OTSSP167 concentrations. D+E) Representative images of TRAP-stained human (huOC) and RAW264.7-derived osteoclast cultures continuously treated with a range of OTSSP167 concentrations. F) Quantification of osteoclast numbers per field of view (N.OC/FOV) in huOC cultures. G) Number of nuclei per osteoclast in huOC cultures. H) Quantification of N.OC/FOV in RAW264.7-derived cultures. I) Number of nuclei per osteoclast in RAW264.7derived cultures. All data are represented as mean +/- standard error. *: P<0.05, **: P<0.01, ***: P<0.001. For clarity, only significant differences with vehicletreated cultures are shown.

haematologica | 2018; 103(8)

1361

J. Muller et al.

D F

Figure 2. OTSSP167 inhibits bone matrix resorption. A+B) Representative images of Von Kossa-stained bone matrix resorption assays of huOC and RAW264.7derived osteoclast cultures continuously treated with a range of OTSSP167 concentrations. C+D) Quantification of the matrix resorption area in huOC and RAW264.7derived osteoclast cultures. E) Real-time PCR analysis of MELK, IRF8, DC-STAMP and NFATc1 following OTSSP167 treatment of RAW264.7-derived osteoclast cultures. F) Representative confocal microscopy images of phalloidin-FITC-stained RAW264.7-derived osteoclast cultures. All data are represented as mean +/- standard error. *: P<0.05, **: P<0.01, ***: P<0.001. For clarity, only significant differences with vehicle-treated cultures are shown.

and Mineral Research Histomorphometry Nomenclature Committee.22 Osteoblast surface and osteoid surface were measured on sections stained with toluidine blue and Masson’s trichrome (Sigma-Aldrich). Osteoclasts were detected by TRAP staining (Sigma-Aldrich) and osteoclast surface was determined. Briefly, sections were stained for acid phosphatase using naphthol ASTR phosphate as substrate in the presence of 50 mM tartrate with hexazotised pararosaniline, and counterstained with methyl green. Measurements were performed in NDP.View 2.6.13.

Statistical analyses All experiments were performed at least in triplicate. Results are shown as means +/- standard error and representative images are shown. For comparisons of 2 means, a Student t-test was used. For comparisons of multiple means, a one-way ANOVA was used, followed by a Dunnett’s post-hoc test or Tukey’s post-hoc test. All statistical analyses were performed with Prism 5 (Graphpad software). P-values below 0.05 were considered significant and P-values are represented as follows: *=P<0.05, **=P<0.01, ***=P<0.001.

cells and on the differentiation into osteoclasts. Treatment of RAW264.7 cells with 25 nM OTSSP167 for 24 hours resulted in decreased MELK protein levels (Figure 1B). Because of the described role of MELK in cell cycle progression, we assessed the effect of continuous OTSSP167 treatment on RAW264.7 and human PBMC viability, and found that the viability of both osteoclast progenitor populations decreased (IC50: 12.8 nM and 43.2 nM, respectively) (Figure 1C). This coincided with an induction of G2/M cell cycle arrest (Online Supplementary Figure S1A). The decrease in progenitor viability corresponded with a decrease in both human (Figure 1D and 1F) and murine (Figure 1E and 1H) osteoclast differentiation following continuous OTSSP167 treatment. Although the number of osteoclasts decreased, osteoclast size was markedly increased in human cultures treated with 10 nM OTSSP167, with a corresponding increase in the number of nuclei per osteoclasts (Figure 1G). This was not the case for RAW264.7 osteoclast cultures (Figure 1I).

OTSSP167 inhibits bone matrix resorption Results OTSSP167 hampers osteoclast differentiation During normal osteoclastogenesis, MELK levels increased in osteoclasts cultured from murine RAW264.7 cells (Figure 1A) or from primary human mononuclear cells (results not shown) when compared to monocyte cultures (Figure 1A). We could confirm the earlier described15 osteoclast activation pathway (EZH2-IRF8-NFATC1) by showing significant increases in EZH2, NFATC1, TRAP and CTSK and a decrease in IRF8 mRNA (Figure 1A). We subsequently studied the effects of the MELK inhibitor OTSSP167 on the proliferation of osteoclast progenitor 1362

Bone matrix resorption decreased following OTSSP167 treatment of both primary human (Figure 2A and 2C) and murine (Figure 2B and 2D) osteoclast cultures. Of note, this effect was apparent in human osteoclasts treated with 10 nM OTSSP167, suggesting that the enlarged osteoclasts that are present at this concentration are not efficiently resorbing matrix. Similarly, murine osteoclast bone resorption was inhibited at 1 nM OTSSP167, a concentration at which no effect on osteoclast differentiation was observed. The blocking effect of OTSSP167 on osteoclast differentiation was confirmed at mRNA level, where we observed an increased expression of the negative regulator IRF8 and decreased expression of NFATC1 and DCSTAMP (Figure 2E). On the haematologica | 2018; 103(8)

OTSSP167 has activity in myeloma bone disease

Figure 3. Established osteoclast cultures are sensitive to OTSSP167. A) Representative images of TRAP-stained RAW264.7-derived osteoclast cultures treated with a range of OTSSP167 concentrations. Treatment of these cultures was initiated after the final day of osteoclast differentiation (day 5). B) Representative images of Von Kossa-stained bone matrix resorption assays of huOC and RAW264.7-derived osteoclast cultures treated with a range of OTSSP167 concentrations. Treatment of these cultures was initiated when mature osteoclasts appeared (early: day 5, late: day 7). C) Quantification of osteoclast numbers per field of view (N.OC/FOV) in RAW264.7-derived cultures treated with OTSSP167 starting at day 5. D) Quantification of the matrix resorption area in RAW264.7-derived osteoclast cultures treated with OTSSP167 starting at day 5 or day 7. E) Real-time PCR analysis of osteoclast differentiation markers following OTSSP167 treatment starting at day 5 of RAW264.7-derived osteoclast cultures. F) ROS generation by RAW264.7-derived osteoclasts treated with OTSSP167 starting on day 5. All data are represented as mean +/- standard error. *: P<0.05, **: P<0.01, ***: P<0.001.

other hand, we detected an increase in EZH2 mRNA and protein levels, while the phosphorylation level of FOXM1 decreased (Online Supplementary Figure S1B). Finally, we observed no difference in actin ring formation in mature osteoclasts following OTSSP167 treatment (Figure 2F).

Established osteoclast cultures are sensitive to OTSSP167 In order to confirm that the effects of OTSSP167 on osteoclast function are not solely due to a decrease in monocyte viability, we initiated treatment of mature RAW264.7 osteoclasts after the final day of differentiation (day 5). In this setting, treatment with up to 10 nM OTSSP167 had no effect on the number of osteoclasts (Figure 3A and 3C). Initiation of OTSSP167 treatment on day 5 or day 7 of in vitro matrix resorption assays markedly reduced bone matrix resorption (Figure 3B and 3D). Mature osteoclasts appeared at day 5 in these cultures and the presence of mature osteoclasts at the end of both DMSO- and OTSSP167-treated cultures was confirmed (not shown). Twenty-four hour treatment of mature osteoclasts had no impact on NFATc1 or TRAP mRNA levels (Figure 3E). However, we found OTSSP167 greatly reduced CTSK expression and greatly increased calcitonin receptor (CALCR) expression by mature osteoclasts (Figure 3E). Finally, as multiple reports indicate that both osteoclast differentiation and bone resorption by osteoclasts depend on the generation of reactive oxygen species (ROS),23,24 we assessed whether OTSSP167 could hamper the generation of ROS by mature osteoclasts and this was indeed the case (Figure 3F).

OTSSP167 stimulates osteoblast function Contrary to osteoclasts, MELK, EZH2 and FOXM1 expression decreased during OB formation indicating an haematologica | 2018; 103(8)

inhibitory role in OB development (Figure 4A). These data suggest an opposing role for these molecules in OB formation which is in line with the results obtained with OTSSP167. Treatment of BMSC-TERT with OTSSP167 for 24 hours resulted in decreased MELK, EZH2 and FOXM1 protein levels (Figure 4B). Although monocyte viability was considerably reduced following OTSSP167 treatment in the low nanomolar range, bone marrow stromal cell viability was unaffected at these concentrations (IC50: 800.2 nM) (Figure 4C). Accordingly, no G2/M cell cycle arrest was detected (Online Supplementary Figure S1A). Functionally, OTSSP167 increased collagen deposition by osteoblasts (Figure 4D and 4E) and induced a marked increase in matrix mineralization (Figure 4D and 4F). Finally, we determined the effect of OTSSP167 on osteoblast marker gene expression and found that osterix (OSX) expression was increased. Conversely RUNX2, osteopontin (OPN) and interleukin-6 (IL-6) expression were decreased (Figure 4G).

Preclinical activity of OTSSP167 in multiple myeloma bone disease Our in vitro data indicate that OTSSP167 blocks osteoclast activity while stimulating osteoblast activity. Therefore, we assessed the potential of OTSSP167 as a novel therapy for MMBD in vivo in the 5TGM.1 MM model. The effects of OTSSP167 on myeloma development are given and discussed in detail in a separate publication.7 OTSSP167 treatment prevented the development of MMBD and this effect was similar at all dosing schedules tested (Figure 5A). OTSSP167 reduced the number of cortical perforations in MM-bearing mice (Figure 5B), without affecting cortical thickness (Ct.Th) (Online Supplementary Figure S1E). The loss of trabecular bone volume (Tb.BV/TV) that occurs in MM-bearing mice com1363

J. Muller et al.

pared to healthy controls was completely prevented following treatment with OTSSP167 (Figure 5C). This was due to an increase in the number of trabeculae (Tb.N) (Figure 5D) and a decrease in trabecular separation (Tb.Sp) (Online Supplementary Figure S1F). As a result, trabecular connectivity density (Conn.Dn) was similar to healthy mice in OTSSP167-treated MM-bearing mice (Online

Supplementary Figure S1G). Trabecular thickness was not affected in MM-bearing mice (Online Supplementary Figure S1H). Importantly, the observed prevention of MMBD in these mice does not appear to be solely due to a decreased tumor load following OTSSP167 treatment, as MMBD was prevented at a dose (7.5 mg/kg/2d) which had no effect on tumor load (Figure 5E).

Figure 4. OTSSP167 stimulates osteoblast function. A) MELK, EZH2 and FOXM1 mRNA levels during osteoblast differentiation of BMSC-TERT cells. B) The effect of OTSSP167 treatment on MELK, EZH2 and FOXM1 protein levels in BMSC-TERT osteoblast cultures. C) MTT assay on BMSC-TERT cells incubated with a range of OTSSP167 concentrations. D) Representative images of sirius red (top panels) and alizarin red (bottom panels) stainings of BMSC-TERT osteoblast cultures treated with a range of OTSSP167 concentrations. E) Quantification of collagen deposition by BMSC-TERT osteoblasts following OTSSP167 treatment. F) Quantification of matrix mineralization by BMSC-TERT osteoblasts following OTSSP167 treatment. G) Real-time PCR analysis of osteoblast differentiation marker expression by BMSCTERT osteoblasts following OTSSP167 treatment. All data are represented as mean +/- standard error. *:P<0.05, **:P<0.01, ***:P<0.001. Except for panel A, only significant differences with vehicle-treated cultures are shown for clarity.

1364

haematologica | 2018; 103(8)

OTSSP167 has activity in myeloma bone disease

OTSSP167 decreases osteoclast activity and restores osteoblast activity in multiple myeloma-bearing mice Because of the lack of an anti-myeloma effect with an OTSSP167 dose of 7.5 mg/kg/2d, we performed immunohistomorphometry on the bones of mice from this cohort. The increase in osteoclast surface observed in MM-bearing mice was completely prevented after OTSSP167 treatment (Figure 6A and 6D). Together with an increase in osteoclast activity, osteoblast surface was

suppressed in MM-bearing mice and OTSSP167 restored osteoblast surface levels to those observed in naive mice (Figure 6B and 6E). Finally, given the effect of OTSSP167 on osteoblast collagen deposition and mineralizing activity in vitro, we analyzed osteoid and found that OTSSP167 increased osteoid surface in MM-bearing mice (Figure 6C and 6F). In addition, osteoid thickness was reduced in OTSSP167-treated MM-bearing mice (Figure 6G).

Figure 5. Preclinical activity of OTSSP167 in multiple myeloma bone disease. A) Representative 3D-reconstructions of distal femurs of naive or 5TGM.1 MM-bearing mice treated with vehicle solution or various dosing schedules of OTSSP167. Upper panels show a frontal view. Lower panels show a frontal section of the frontal view. Cortical bone is colored white, trabecular bone is colored green. B) Quantification of the number of cortical perforations with a diameter of at least 50 Îźm. CTAn analysis was performed and C) trabecular bone volume (Tb.BV/TV), D) trabecular number (Tb.N), E) Superimposition of tumor load data (red line, relative to vehicle-treated MM-bearing mice) and Tb.BV/TV (black bars, relative to naive mice) in the different mouse cohorts. All data are represented as mean +/- standard error. All statistical differences are reported versus vehicle-treated mice. No difference between the different dosing schedules was observed. *:P<0.05, **:P<0.01, ***:P<0.001.

haematologica | 2018; 103(8)

1365

J. Muller et al.

Discussion In this preclinical study, we demonstrated that MELK inhibitor OTSSP167 is a promising novel therapeutic agent for MMBD since OTSSP167 treatment blocked osteoclast activity and stimulated osteoblast activity in vitro, and completely prevented the development of

MMBD in MM-bearing mice. Consistent with an osteoclast stimulating activity of factors downstream of MELK, MELK mRNA levels increased during osteoclastogenesis. In addition, we found significant mRNA changes in EZH2 and the negative regulator IRF8 known to influence the crucial differentiation factor NFATC1 which was found to be upregulated during osteo-

1366

Figure 6. OTSSP167 decreases osteoclast activity and restores osteoblast activity in multiple myeloma-bearing mice. A) Representative images of TRAPstained distal femur sections of naive, vehicle- and OTSSP167treated mice from the 7.5 mg/kg/2d cohort, which showed no effect on MM tumor load. 40X amplifications are shown in the inserts. B) Representative images of toluidine blue-stained distal femur sections from the same cohort. C) Representative images of Massonâ&#x20AC;&#x2122;s trichrome stained distal femur sections from the same cohort. D) Quantification of osteoclast surface (Oc.S/BS). E) Quantification of osteoblast surface (Ob.S/BS). F) Quantification of osteoid surface (OS/BS). G) Quantification of osteoid thickness (O.Th). All data are represented as mean +/- standard error. *:P<0.05, **:P<0.01, ***:P<0.001.

haematologica | 2018; 103(8)

OTSSP167 has activity in myeloma bone disease

clast differentiation (illustrated in Online Supplementary Figure S1C). OTSSP167 decreased MELK protein levels in RAW264.7 cells, which agrees with previous studies on various cell types.25,26 Of note, OTSSP167 can have off-target effects, which could partially account for the effects of OTSSP167 on osteoclast and osteoblast function.27,28 In accordance with previous reports implicating MELK in G2/M transition and proliferation, OTSSP167 decreased the viability of monocytes due to G2/M cell cycle arrest.26,27,29,30 The decreased progenitor cell viability resulted in a decreased osteoclast differentiation following OTSSP167 treatment. It is difficult to differentiate between an anti-proliferative effect on progenitor cells and an anti-osteoclast differentiation effect. We observed that 2 pathways, known to interact with MELK, were affected by OTSSP167: i) the transcriptional factor FOXM1 which is implicated in proliferation and ii) the EZH2-IRF8-NFATC1 axis, involved in osteoclast differentiation. Unexpectedly, EZH2 levels increased, which can be explained by the strong link between cell cycle arrest and CDK1/2-dependent EZH2 phosphorylation on different residues (T487), which can either disrupt the binding of EZH2 to other partners of the polycomb repressive complex31 or target it for ubiquitin-mediated degradation.32 EZH2 thus becomes less functional, which results is a decline in H3K27 trimethylation and a de-repression of EZH2 target genes (in our case IRF8).33 When mature osteoclasts were treated with OTSSP167, their numbers remained unaltered but matrix resorption was drastically decreased, further corroborating that OTSSP167 has a direct effect on osteoclast activity. Together, our data indicate that OTSSP167 inhibits osteoclast function by hampering monocytic progenitor viability as well as by directly inhibiting mature osteoclast function. MELK, EZH2 and FOXM1 mRNA levels decreased during osteoblast differentiation, consistent with an inhibitory role of MELK and downstream factors such as EZH2 on osteoblast function.16,17 Contrary to osteoclasts and various malignant cells, BMSC-TERT viability was not affected following OTSSP167 treatment at similar concentrations. OTSSP167 treatment increased collagen deposition and strongly stimulated mineralization activity of osteoblasts in vitro and this coincided with an increase in OSX levels but a decrease in RUNX2, OPN and IL-6 mRNA levels. The pro-mineralization activity of OTSSP167 is likely mediated by EZH2 as treatment with an EZH2 inhibitor showed a similar effect.16 However, the decrease in RUNX2 expression following OTSSP167 treatment does not correspond with the described role of EZH2 as a suppressor of RUNX2 transcription.16 Alternative mechanisms of OTSSP167-induced osteoblast maturation include a marked decrease in the expression of OPN, a non-collagenous bone matrix protein that inhibits matrix mineralization34 and IL-6,34 a potent growth factor for MM cells, but also a negative regulator of osteoblast differentiation35 and inducer of bone resorption.36 Deregulation of these genes in conjunction with the above-mentioned upregulation of OSX, a master regulator of mineralization, likely mediates the pro-osteogenic activity of OTSSP167. The regulation of MELK expression and activity by upstream signaling pathways remains poorly understood including in bone cells. E2F137 and FOXM110 have been shown to regulate MELK gene transcription, the former in osteoblastic MC3T3-E1 cells. Of note, E2F1 has been haematologica | 2018; 103(8)

implicated in increased osteoclastogenesis and osteoblast activity.38,39 MELK both regulates and is regulated by one family of MAP kinases, the c-Jun NH(2)-terminal kinases (JNK2), that acts downstream of the RANK-receptor.40 Interestingly, JNK2 does not seem to be required for osteoclast differentiation, but rather appears to be involved in osteoclast survival.41 Given the promising in vitro data, we assessed whether OTSSP167 would affect the development of MMBD in the murine 5TGM.1 MM model. Although this model reflects human myeloma and associated bone disease in an immunocompetent setting, it should be noted that MM growth in this model progresses rapidly. This cell line allows in vivo studies, however, it also shows a BM-independent growth in vitro and its murine origin may not reflect all the human aspects of myeloma disease (from cytogenetic and molecular point of view). We have previously shown that the dosing schemes we used dosedependently decreased MM tumor load.7 OTSSP167 completely prevented the development of MMBD at all doses, with no difference between the treatment groups. Both the development of lytic cortical lesions and the large loss of trabecular bone were completely prevented in MMbearing mice treated with OTSSP167. Importantly, this effect also occurred at an OTSSP167 concentration which had no effect on MM tumor load (7.5 mg/kg/2d), indicating that OTSSP167 has a direct effect on bone cells and does not solely reduce MMBD by reducing MM tumor load. In fact, given the lower concentration of OTSSP167 needed to achieve an anti-MMBD effect, our data suggest that OTSSP167 could exert its anti-MM effect in part by normalizing bone homeostasis. Indeed, the reduced osteoclast numbers following OTSSP167 treatment likely result in reduced myeloma pro-survival factor levels and increased myeloma anti-proliferative factor levels, respectively. Upon treatment with OTSSP167, we found decreased mRNA expression levels of insulin-like growth factor 1 (IGF-1), osteopontin, a proliferation-inducing ligand (APRIL) and interleukin-10 (IL-10) (Online Supplementary Figure S1D). In conclusion, this study provides a novel approach for the treatment of MMBD. The maintenance of bone anabolic activity by OTSSP167 is promising and warrants further investigation. Reducing MM patient morbidity and mortality via the combined anti-MM and anti-MMBD effect of OTSSP167 holds great clinical promise and our results warrant similar studies in other cancers with bone involvement. Acknowledgments The authors would like to thank the GIGA-imaging platform for their excellent technical assistance. Funding JM and ML are Télévie PhD candidates. The Wilhelminen Cancer Research Institute is supported by the Austrian Forum against Cancer. The laboratory of Hematology was supported by Foundation Against Cancer, the Fonds National de la Recherche Scientifique (F.N.R.S., Belgium) and the Fonds spéciaux de la Recherche (University of Liege). Elodie Duray (research fellow), Erwan Plougonven (post-doctoral researcher) and Frédéric Baron (senior research associate) have a mandate supported by the FNRS. Roy Heusschen is a Télévie postdoctoral research associate. Jo Caers is a post-doctorate clinical specialist funded by the Belgian Foundation against Cancer. 1367

J. Muller et al.

References 1. Rollig C, Knop S, Bornhauser M. Multiple myeloma. Lancet. 2015;385(9983):21972208. 2. Roodman GD. Pathogenesis of myeloma bone disease. J Cell Biochem. 2010;109(2):283-291. 3. Heusschen R, Muller J, Duray E, et al. Molecular mechanisms, current management and next generation therapy in myeloma bone disease. Leuk Lymphoma. 2018;59(1):14-28. 4. Galson DL, Silbermann R, Roodman GD. Mechanisms of multiple myeloma bone disease. Bonekey Rep. 2012;1:135. 5. Kennel KA, Drake MT. Adverse effects of bisphosphonates: implications for osteoporosis management. Mayo Clin Proc. 2009;84(7):632-637. 6. Silbermann R, Roodman GD. Current controversies in the management of myeloma bone disease. J Cell Physiol. 2016;231(11):2374-2379. 7. Bolomsky A, Heusschen R, Schlangen K, et al. Maternal embryonic leucine zipper kinase is a novel target for proliferationassociated high-risk myeloma. Haematologica. 2018;103(2):325-335. 8. 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. 9. 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. 10. Wang Y, Lee YM, Baitsch L, et al. MELK is an oncogenic kinase essential for mitotic progression in basal-like breast cancer cells. Elife. 2014;3:e01763. 11. Ganguly R, Hong CS, Smith LG, 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. 12. 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):10511063. 13. Kim SH, Joshi K, Ezhilarasan R, et al. EZH2 protects glioma stem cells from radiationinduced cell death in a MELK/FOXM1dependent manner. Stem Cell Reports. 2015;4(2):226-238. 14. Gu C, Yang Y, Sompallae R, et al. FOXM1 is a therapeutic target for high-risk multiple myeloma. Leukemia. 2016;30(4):873-882. 15. Fang C, Qiao Y, Mun SH, et al. Cutting edge: EZH2 promotes osteoclastogenesis by epigenetic silencing of the negative regulator IRF8. J Immunol. 2016;196(11):4452-4456. 16. Dudakovic A, Camilleri ET, Xu F, et al. Epigenetic control of skeletal development by the histone methyltransferase Ezh2. J

1368

Biol Chem. 2015;290(46):27604-27617. 17. Dudakovic A, Camilleri ET, Riester SM, et al. Enhancer of zeste homolog 2 inhibition stimulates bone formation and mitigates bone loss caused by ovariectomy in skeletally mature mice. J Biol Chem. 2016; 291(47):24594-24606. 18. Adamik J, Jin S, Sun Q, et al. EZH2 or HDAC1 Inhibition reverses multiple myeloma-induced epigenetic suppression of osteoblast differentiation. Mol Cancer Res. 2017;15(4):405-417. 19. 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. 20. Bolomsky A, Schreder M, Meissner T, et al. Immunomodulatory drugs thalidomide and lenalidomide affect osteoblast differentiation of human bone marrow stromal cells in vitro. Exp Hematol. 2014;42(7):516-525. 21. Binsfeld M, Muller J, Lamour V, et al. Granulocytic myeloid-derived suppressor cells promote angiogenesis in the context of multiple myeloma. Oncotarget. 2016; 7(25):37931-37943. 22. Dempster DW, Compston JE, Drezner MK, et al. Standardized nomenclature, symbols, and units for bone histomorphometry: a 2012 update of the report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res. 2013;28(1):217. 23. Garrett IR, Boyce BF, Oreffo RO, et al. Oxygen-derived free radicals stimulate osteoclastic bone resorption in rodent bone in vitro and in vivo. J Clin Invest. 1990; 85(3):632-639. 24. Lee NK, Choi YG, Baik JY, et al. A crucial role for reactive oxygen species in RANKLinduced osteoclast differentiation. Blood. 2005;106(3):852-859. 25. Bolomsky A, Heusschen R, Schlangen K, et al. Maternal embryonic leucine zipper kinase is a novel target for proliferation associated high-risk myeloma. Haematologica. 2018;103(2):325-335. 26. 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. 27. Simon M, Mesmar F, Helguero L, Williams C. Genome-wide effects of MELK-inhibitor in triple-negative breast cancer cells indicate context-dependent response with p53 as a key determinant. PLoS One. 2017; 12(2):e0172832. 28. Lin A, Giuliano CJ, Sayles NM, Sheltzer JM. CRISPR/Cas9 mutagenesis invalidates a putative cancer dependency targeted in ongoing clinical trials. Elife. 2017;6. 29. Kohler RS, Kettelhack H, KnipprathMeszaros AM, et al. MELK expression in ovarian cancer correlates with poor out-

30.

31.

32.

33. 34.

35.

36.

37.

38.

39.

40.

41.

come and its inhibition by OTSSP167 abrogates proliferation and viability of ovarian cancer cells. Gynecol Oncol. 2017;145(1):159-166. Kato T, Inoue H, Imoto S, et al. Oncogenic roles of TOPK and MELK, and effective growth suppression by small molecular inhibitors in kidney cancer cells. Oncotarget. 2016;7(14):17652-17664. Wei Y, Chen Y-H, Li L-Y, et al. CDK1-dependent phosphorylation of EZH2 suppresses methylation of H3K27 and promotes osteogenic differentiation of human mesenchymal stem cells. Nature Cell Biology. 2010;13:87. Wu SC, Zhang Y. Cyclin-dependent kinase 1 (CDK1)-mediated phosphorylation of enhancer of Zeste 2 (Ezh2) regulates its stability. J Biol Chem. 2011;286(32):2851128519. Lim S, Kaldis P. Cdks, cyclins and CKIs: roles beyond cell cycle regulation. Development. 2013;140(15):3079-3093. Yuan Q, Jiang Y, Zhao X, et al. Increased osteopontin contributes to inhibition of bone mineralization in FGF23-deficient mice. J Bone Miner Res. 2014;29(3):693-704. Kaneshiro S, Ebina K, Shi K, et al. IL-6 negatively regulates osteoblast differentiation through the SHP2/MEK2 and SHP2/Akt2 pathways in vitro. J Bone Miner Metab. 2014;32(4):378-392. Ishimi Y, Miyaura C, Jin CH, et al. IL-6 is produced by osteoblasts and induces bone resorption. J Immunol. 1990;145(10):32973303. Verlinden L, Eelen G, Beullens I, et al. Characterization of the condensin component Cnap1 and protein kinase melk as novel E2F target genes down-regulated by 1,25-dihydroxyvitamin D3. J Biol Chem. 2005;280(45):37319-37330. Murata K, Fang C, Terao C, et al. Hypoxiasensitive COMMD1 integrates signaling and cellular metabolism in human macrophages and suppresses osteoclastogenesis. Immunity. 2017;47(1):66-79.e65. Yu S, Yerges-Armstrong LM, Chu Y, Zmuda JM, Zhang Y. E2F1 effects on osteoblast differentiation and mineralization are mediated through up-regulation of frizzled-1. Bone. 2013;56(2):234-241. Caers J, Van Valckenborgh E, Menu E, Van Camp B, Vanderkerken K. Unraveling the biology of multiple myeloma disease: cancer stem cells, acquired intracellular changes and interactions with the surrounding microenvironment. Bull Cancer. 2008;95(3):301313. Amoui M, Sheng MHC, Chen S-T, Baylink DJ, Lau KHW. A transmembrane osteoclastic protein-tyrosine phosphatase regulates osteoclast activity in part by promoting osteoclast survival through c-Src-dependent activation of NF B and JNK2. Arch Biochem Biophys. 2007;463(1):47-59.

haematologica | 2018; 103(8)

ARTICLE

Plasma Cell DIsorders

Therapeutic effects of the novel subtype-selective histone deacetylase inhibitor chidamide on myeloma-associated bone disease

Ferrata Storti Foundation

Jingsong He,1* Qingxiao Chen,1* Huiyao Gu,1 Jing Chen,1 Enfan Zhang,1 Xing Guo,1 Xi Huang,1 Haimeng Yan,1 DongHua He,1 Yang Yang,1 Yi Zhao,1 Gang Wang,1,2 Huang He,1 Qing Yi3 and Zhen Cai1

Bone Marrow Transplantation Center, Department of Hematology, The First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China; 2Quzhou People’s Hospital, Zhejiang Province, China and 3Department of Cancer Biology, Lerner Research Institute, Cleveland Clinic, OH, USA 1

Haematologica 2018 Volume 103(8):1369-1379

*JH and QC contributed equally to the study.

ABSTRACT

istone deacetylases are promising therapeutic targets in hematological malignancies. In the work herein, we investigated the effect of chidamide, a new subtype-selective histone deacetylase inhibitor that was independently produced in China, on multiple myeloma and its associated bone diseases using different models. The cytotoxicity of chidamide toward myeloma is due to its induction of cell apoptosis and cell cycle arrest by increasing the levels of caspase family proteins p21 and p27, among others. Furthermore, chidamide exhibited significant cytotoxicity against myeloma cells co-cultured with bone mesenchymal stromal cells and chidamide-pretreated osteoclasts. Importantly, chidamide suppressed osteoclast differentiation and resorption in vitro by dephosphorylating p-ERK, p-p38, p-AKT and p-JNK and inhibiting the expression of Cathepsin K, NFATc1 and c-fos. Finally, chidamide not only prevented tumor-associated bone loss in a disseminated murine model by partially decreasing the tumor burden but also prevented rapid receptor activator of nuclear factor κ-b ligand (RANKL)-induced bone loss in a non-tumor-bearing mouse model. Based on our results, chidamide exerted dual anti-myeloma and bone-protective effects in vitro and in vivo. These findings strongly support the potential clinical use of this drug as a treatment for multiple myeloma in the near future.

Correspondence: caiz@zju.edu.cn

Received: November 8, 2017. Accepted: April 27, 2018. Pre-published: May 17, 2018.

Introduction

doi:10.3324/haematol.2017.181172

Multiple myeloma (MM) is an incurable plasma malignancy characterized by the accumulation of monoclonal plasma cells in the bone marrow (BM), the secretion of high levels of monoclonal immunoglobulins and osteolytic bone lesions.1 Myeloma cells interact with different cell types in the BM microenvironment, such as osteoclasts (OCs) and mesenchymal stromal cells, which supports their growth, drug resistance and the development of bone disease through cell–cell adhesion and the release of growth factors, such as interleukin-6 (IL-6) and vascular endothelial growth factor (VEGF).1 Osteolytic bone lesions, which result from increased bone resorption by OCs and reduced osteoblastic bone formation, are the most common complication of MM and often decrease the quality of life and survival time of patients with MM.2,3 New active drugs, including proteasome inhibitors and immunomodulatory agents, have improved the outcomes of patients with MM and alleviated bone damage.1 However, disease progression is still inevitable. Therefore, the identification of new targets and development of new drugs that focus on MM cells and their BM microenvironment are crucial to the development of more effective treatments. Chidamide is a novel oral histone deacetylase inhibitor (HDACi) designed to inhibit the activity of HDAC1, 2, 3 and 10 which is produced independently in China and is now undergoing phase I clinical trials in America and Japan.4,5

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1369

haematologica | 2018; 103(8)

1369

J. He et al.

Chidamide has been shown to induce cell apoptosis and cell cycle arrest in a variety of cancers, such as lung cancer, pancreatic cancer, leukemia and lymphoma.4-7 Moreover, HDAC3 was recently shown to be overexpressed in MM and was a key factor that propelled myeloma cell proliferation.8,9 Additionally, HDAC1 and HDAC2 are important positive regulators of osteoclastogenesis.10,11 Based on these data, we hypothesize that chidamide may exert dual anti-myeloma and bone-protective effects. In the work herein, we investigated the dual antimyeloma and bone-protective effects on myeloma cells and preclinical models. To the best of our knowledge, this compound is the first HDACi that was independently developed in China and shows both anti-myeloma and anti-osteolytic bone disease activity in vitro and in vivo, supporting the future clinical application of this drug as a treatment for MM.

ing 50ng/ml RANKL and 25ng/ml monocyte colony stimulating factor (M-CSF) supplemented with 10% fetal bovine serum and 1% L-glutamine was used as osteoclastogenic medium for cells cultured in the presence or absence of chidamide. PBMCs from healthy donors were cultured as pre-OCs for 14 days and as mature OCs for 21 days. The F-actin ring formation assay, assessment of the resorption ability of OCs and tartrate-resistant acid phosphatase (TRAP)+ staining were performed as described in a previous report.12

Drug treatments in different mouse models Two tumor-bearing murine models and a non-tumor-bearing model were employed to investigate the anti-tumor and bone-protective effects of chidamide.4,13-15 Micro computed tomography (CT) was used to evaluate MM bone disease. Further details are provided in the Online Supplementary Materials.16,17

Statistical analysis Methods Drugs Chidamide was a generous gift from the Chipscreen Company, Shenzhen, China. Additional information regarding drugs and reagents is provided in the Online Supplementary Materials.

Human myeloma cell lines, primary myeloma samples, bone marrow stromal cells (BMSCs), healthy peripheral blood mononuclear cells (PBMCs) and bone marrow-derived mononuclear cells (BMMCs) The human myeloma cell lines RPMI-8226, MM.1s, and U266B1 were purchased from the cell bank of the Chinese Academy of Science; H929, CAG, ARP-1, LP-1, and OPM2 cells were generous gifts from Dr. Qing Yi (Department of Cancer Biology, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA). ANBL-6 cells were a generous gift from Professor Zhiqiang Liu, Tianjin Medical University, China. The luciferaseexpressing myeloma cell line RPMI-8226-luc was constructed by infecting RPMI-8226 cells with a lentivirus encoding the luciferase gene, and infected cells were selected with puromycin. Primary myeloma samples and BMSCs were collected from newly diagnosed patients, and PBMCs and BMMCs were collected from healthy donors. Primary CD138(+) cells were sorted using CD138 microbeads (Miltenyi Biotech, CA, USA).

Cell culture Myeloma cell lines were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum and 1% L-glutamine at 37°C in a 5% CO2 atmosphere. BMSCs were cultured in minimum essential medium a (MEM-a) supplemented with 10% fetal bovine serum and 1% L-glutamine. BMSCs were cultured as premature osteoblasts for seven days and as mature osteoblasts for 21 days in the human mesenchymal stem cell (MSC) osteogenic differentiation basal medium.

All data are expressed as means ± SEM and are representative of at least three experiments with similar results performed in triplicate, unless indicated otherwise. A two-tailed Student’s t-test was used to determine the significance of differences between two groups, and one-way analysis of variance was used to estimate the differences between three or more groups. GraphPad Prism 5.0 software (GraphPad Software, CA, USA) was used for the analysis.

Ethical approval Primary myeloma samples, BMSCs, PBMCs, and BMMCs were collected after obtaining informed consent from the patients and approval from the Ethics Committee of the First Affiliated Hospital of Zhejiang University. The animal experiments were carried out after obtaining approval from the Ethics Committee of the First Affiliated Hospital of Zhejiang University.

Results Chidamide exhibits HDAC inhibitory activity in MM cell lines We first investigated the basal expression levels of HDAC1, 2, 3, and 10, which are chidamide targets, in MM cell lines. As shown in Figure 1A, HDAC1, HDAC2 and HDAC3 were all expressed in the nine myeloma cell lines (H929, LP-1, ARP-1, U266B1, RPMI-8226, ANBL-6, OPM2, CAG, and MM.1s), and both HDAC2 and HDAC3 were expressed at slightly higher levels than HDAC1. HDAC10 was barely detectable in human myeloma cell lines. Next, we measured HDAC activity and the acetylation of lysine residues on histones H3 and H4 to determine the inhibitory effect of chidamide on HDAC. As shown in Figure 1B,C, chidamide inhibited HDAC activity and significantly increased the acetylation of H3K8, H3K18 and H4K8. However, the expression of HDAC1, HDAC2, HDAC3 and HDAC10 was not affected (Figure 1D).

Cell proliferation, cell cycle, and cell apoptosis assays Cell counting kit-8 (CCK-8) assays were used to detect MM cell viability. Flow cytometry was used to assess the cell cycle and apoptosis. A detailed description of each procedure is provided in the Online Supplementary Materials.

In vitro OC differentiation, pit formation and F-actin ring formation Osteoclasts were differentiated from PBMCs in osteoclastogenic medium as previously described.12 Briefly, MEM-a contain1370

Chidamide not only exerts anti-myeloma effects but also overcomes the resistance conferred by the BM microenvironment Subsequently, we evaluated the anti-myeloma effect of chidamide. Cell lines were treated with increasing concentrations of chidamide (0.5-8μM) for different times (24, 48, or 72h). CCK-8 assays revealed a dose-dependent and time-dependent pattern of chidamide cytotoxicity (Online Supplementary Figure S1A). Flow cytometry (Annexin V haematologica | 2018; 103(8)

Therapeutic effects of chidamide on myeloma-associated bone disease A

Figure 1. Chidamide exhibits HDAC inhibitory activity in MM cell lines. (A) Western blotting was used to detect the protein levels of HDAC1, HDAC2, HDAC3 and HDAC10. (B) HDAC activity assay showed that chidamide could significantly inhibit HDAC activity (Both in ARP-1 and RPMI-8226, ***P<0.001). (C) Western blot analysis of H3 acetylation at Lys18 and Lys9 and H4 acetylation at Lys8. H3 and H4 expression levels were used as loading controls. (D) Western blot analysis of HDAC1, HDAC2, HDAC3, and HDAC10 levels. Acetylation of α-tubulin at K40 was also examined, and a-tubulin expression was used as the loading control. HDAC: histone deacetylase.

and propidium iodide [PI] staining) showed that chidamide significantly induced MM cell apoptosis at 48h and 72h (Figure 2A and Online Supplementary Figure S1B). ARP-1, MM.1s and CAG cells were more sensitive to chidamide, with IC50 values of approximately 2-4μM, whereas the IC50 values were approximately 4-6μM for the other six cell lines at 48h. When the incubation time was extended to 72h, the IC50 values of most of the cell lines were 1.5-4μM. Furthermore, incubation of ARP-1 and MM.1s cells with an optimal concentration of chidamide increased the levels of cleaved caspases 3, 7, 8, and 9 and cleaved PARP-1 cleavage (Figure 2B). To evaluate whether apoptosis was induced in a caspase-dependent or caspaseindependent manner, we incubated MM.1s and ARP-1 cells with chidamide in the presence of a pan-caspase inhibitor, Q-VD-Oph, for 48h. After co-incubating chidamide with Q-VD-Oph, the apoptotic cells were significantly reduced (Figure 2C and Online Supplementary Figure 2A), with down-regulated active caspase-3 and caspase-9 expression (Online Supplementary Figure 3). Additionally, the cell cycle arrest effect of chidamide was evaluated by flow cytometry. ARP-1, OPM-2, LP-1, and RPMI-8226 cells were treated with increasing doses of chidamide (0.25-6μM) for 48h; chidamide induced ARP-1 cell cycle arrest in the G0/G1 phase and reduced the percentage of cells in the proliferation phases (S and G2) regardless of co-incubation with Q-VD-Oph (***P<0.001, Figure 2D). OPM-2, RPMI-8226, and LP-1 cells showed similar results after incubation with chidamide (Online Supplementary Figure 4). In addition, as shown in Figure 2E, chidamide induced the apoptosis of CD138-positive cells from patients with MM (n=5, 48h). Moreover, the BMMCs from four healthy haematologica | 2018; 103(8)

donors were incubated with 4μM chidamide for 48h to examine the cytotoxicity of chidamide toward non-tumor cells. Although approximately 10% of monocytes underwent apoptosis, the numbers of apoptosis cells were only slightly higher than the vehicle group and far less than MM cells (Online Supplementary Figure 2B). Furthermore, we investigated the cytotoxicity of chidamide in the presence of the BM microenvironment. For this purpose, H929 cells were co-cultured with BMSCs and OCs in the presence of 6µM chidamide for 48h. OCs showed a significant anti-apoptotic effect, but pretreatment with chidamide during osteoclastogenesis suppressed the protective effect of OCs on myeloma cell apoptosis (Figure 3A, Figure 3C). Chidamide-induced apoptosis in approximately 30% of the MM cells, even after co-culture with BMSCs, which usually exert a protective effect on MM cells (Figure 3B,C).

The molecular mechanisms underlying chidamide activity in myeloma cells When ARP-1 and RPMI-8226 cells were treated with an increasing dose of chidamide (0-6μM) (Figure 4A), the expression of Myc, Mcl-1, and Bcl-xL decreased in a dosedependent manner. Additionally, in the presence of chidamide, levels of cell cycle-related proteins, including p27, p21, Cyclin-D2, CDK4, and CDK6, were all reduced (Figure 4A). Finally, as shown in Figure 4B, SOCS3 levels increased and p-JAK2 and p-STAT3 levels decreased in RPMI-8226 and ARP-1 cells treated with different concentrations of chidamide. Additionally, a pan-caspase inhibitor, Q-VD-Oph, was used to treat MM cells together with chidamide to prevent inhibitory effects of activated caspases on the levels of signaling and apoptosis-related 1371

J. He et al.

proteins. As shown in Figure 4 and Online Supplementary Figure S5, most of these proteins showed the same changes in the presence and absence of Q-VD-Oph.

Chidamide decreases the MM tumor burden and prevents myeloma-associated bone loss in a mouse model of disseminated myeloma Two murine models were employed to examine the dual anti-tumor and bone-protective effects of chidamide. First, as shown in Figure 5A, a significant decrease in tumor volume was observed in chidamide-treated NOD/SCID mice compared with that observed in the vehicle group. Consistent with the Western blot results from the cell lines, immunohistochemical staining revealed decreased expression of Bcl-xL, CDK4, CDK6 and Myc and increased levels of cleaved caspase-3 (Online

Supplementary Figure S6). Next, NCG mice were intravenously injected with RPMI-8226-luc cells to establish the murine model of myeloma-induced bone destruction. Compared with the vehicle group, chidamide clearly controlled the tumor burden, as measured by serum levels of human Ig Îť light chain secreted by RPMI-8226-luc cells (***P<0.001, Figure 5B) and the bioluminescence in vivo imaging (Figure 5B). Then, serum levels of the bone resorption marker Carboxy-terminal telopeptide 1 (CTXI) were measured and were found to be significantly diminished in chidamide-treated mice (**P<0.01, Figure 5C); in addition, a significantly increased serum level of amino-terminal propeptide (PINP), the bone formation marker, was observed (***P<0.001, Figure 5C). Consistent with these findings, micro-CT three-dimensional reconstructed images at the metaphysis of distal femurs

A B

1372

Figure 2. MM cells are vulnerable to chidamide. (A) The cytotoxicity of chidamide toward MM cell lines was examined using flow cytometry (Annexin V/PI), and all data are summarized as the meansÂąSEM (n=3). (B) The expression of apoptosis-associated proteins, including PARP-1 and cleaved caspases 3, 7, 8, and 9, in chidamide-treated ARP-1 and MM.1s cells was detected. (C) The effect of chidamide on apoptosis in MM.1s cells after 48h with or without coincubation with the pan-caspase inhibitor Q-VD-Oph. After co-incubation with Q-VD-Oph, chidamide-induced apoptosis was inhibited. (D) Cell cycle arrest was examined using flow cytometry, and ARP-1 cells underwent G1 phase arrest in response to incubation with chidamide regardless of co-incubation with QVD-Oph. (E) Chidamide could significantly induce + primary CD138 cell apoptosis after incubation for 48h (n=5, *P<0.05). (E) Chidamide could significantly induce primary CD138+ cell apoptosis after incubation for 48h (n=5). PI: propidium iodide; MM: multiple myeloma.

haematologica | 2018; 103(8)

Therapeutic effects of chidamide on myeloma-associated bone disease

revealed greater myeloma-associated bone loss in the vehicle group as compared with the chidamide-treated group (Figure 5D). Moreover, based on the bone morphometric parameters, chidamide increased the trabecular number (**P<0.01) and bone volume density (P=0.0755) and reduced trabecular separation (**P<0.01) compared to the control group (Figure 5E).

Chidamide inhibits osteoclastogenesis and bone resorption in vitro As the in vivo study revealed that chidamide exerts antimyeloma effects, we investigated the effect of chidamide on the formation and function of OCs of human origin. The expression of DUSP1, c-fos, NFATc1 and HDAC10 increased during RANKL-induced OC formation (Figure 6A). We then tested several key factors and signaling path-

ways that mediate osteoclastogenesis in order to identify the mechanisms underlying the aforementioned effects. The levels of p-p38, p-ERK1/2, p-JNK and p-AKT were also reduced, indicating that chidamide suppressed the classical pathways of OC activation. Cathepsin K, c-fos, HDAC10 and NFATc1 expression levels were all downregulated after chidamide treatment in a dose-dependent manner. In addition, acetylation of H3K9, H3K18 and H4K8 was increased after chidamide treatment (Figure 6B). PBMCs cultured in osteoclastogenic medium were incubated with different concentrations of chidamide to evaluate its effects on OC formation. At a low concentration (0-1μM) of the drug, the number of TRAP+ multinucleated cells derived from healthy donors was reduced, but the density of cells was not significantly reduced (**P<0.01, Figure 6C,D). Additionally, F-actin ring forma-

B B

haematologica | 2018; 103(8)

Figure 3. Chidamide overcame the resistance conferred by chidamide-pretreated OCs and BMSCs. (A) and (C) H929 cells were co-cultured with OCs and chidamide-pretreated OCs for 48h or with 6μM chidamide, and cell apoptosis was detected by flow cytometry after staining with Annexin V and PI. (B) and (C) H929 cells were co-cultured with BMSCs or with 6μM chidamide for 48h, and cell apoptosis was detected by flow cytometry after staining with Annexin V and PI (H929+chidamide vs. H929+OCs+chidamide, ***P<0.001; H929+chidamide vs. H929+OCs(chida)+chidamide, P>0.05, non-significant; H929+chidamide vs. H929+BMSCs+chidamide, P>0.05 non-significant). PI: propidium iodide; OC: osteoclasts; BMSC: bone marrow stromal cell.

1373

J. He et al.

tion was inhibited in the presence of chidamide during osteoclastogenesis (Figure 6C). Next, we evaluated functional changes in OCs cultured on calcium substrate-coated slides. Chidamide treatment induced a dose-dependent reduction in the area of resorption pits. With increasing drug doses, the resorption area was substantially reduced (Figure 6C, Figure 6E ***P<0.001).

Effect of chidamide on osteoblasts HDAC1 and HDAC3 are regarded as negative regulators of osteoblastogenesis during bone formation, thus we also evaluated the effect of chidamide on osteoblasts. As shown in Online Supplementary Figure S7, primary BMSCs from patients with MM (n=6) were cultured in human MSC osteogenic differentiation basal medium in the presence of different chidamide concentrations. Alkaline phosphatase (ALP) was measured as a positive marker of osteoblast differentiation, and its expression was slightly increased by chidamide treatment. Additionally, Alizarin red staining (ARS) of calcium deposits showed that chidamide had no promotion or inhibitory effect on osteoblast differentiation (Online Supplementary Figure S7A,B). Finally, the expression of osteocalcin (OCN), Activin A and semaphorins at the messenger ribonucleic acid (mRNA) level was also examined; OCN (day 21) was increased, while Activin A (day 21) was down-regulated (Online Supplementary Figure S7C).

Chidamide exerts bone-protective effects on non-tumor-bearing mice Chidamide was administered via oral gavage to nontumor-bearing C57BL/6 mice at a dose of 25mg/kg for 21 days to establish whether chidamide directly exerts its bone-preserving effect on the bone tissue or exerts an indirect effect by decreasing the tumor burden. As shown in Figure 7A, serum CTX-I levels were not significantly different between the vehicle (n=5) and chidamide (n=5) groups, whereas the serum PINP level was clearly increased after treatment with chidamide (***P<0.001). Thereafter, in vivo intraperitoneal injections of soluble receptor activator of nuclear factor-ÎşB ligand (sRANKL; three doses) were administered within 50h followed by gavage with chidamide for 21 days to mimic OC stimula-

tion by myeloma-derived sRANKL. Serum CTX-I levels increased in both the vehicle group (n=5) and the chidamide group (n=5) after sRANKL injections, but the level in the chidamide-treated group was increased to a lesser extent than the level in the vehicle group (*P<0.05). Serum PINP levels were reduced in both groups; however, chidamide attenuated the reduced PINP level in the chidamide-treated group (***P<0.001). TRAP+ staining of bone showed a reduced number of OCs (Figure 7B,C) in the chidamide-treated group (***P<0.001). Moreover, the bone morphometric parameters evaluated by micro-CT (Figure 7D,E) indicated that chidamide increased the trabecular number (***P<0.001) and reduced trabecular separation compared to the control group (**P<0.01). Although bone volume density over total volume (BV/TV, P>0.05) showed an increasing trend in the chidamide-treated group, the difference between the two groups was not significant. Based on these results, chidamide increased the bone volume of healthy mice to some extent and directly prevented RANKL-induced OC activation.

Discussion HDACs are an important family of enzymes with crucial roles in carcinogenesis through their repressive effects on tumor suppressor gene transcription and are proposed as therapeutic targets in oncology.18,19 HDAC inhibitors induce cancer cell apoptosis, cell cycle arrest and promote differentiation, particularly in hematological malignancies. However, the anti-cancer effects of HDAC inhibitors differ and their pharmacological effects vary, depending on the cancer cell types, HDAC targets and doses. Chidamide, a novel HDACi which is currently being widely used in China to treat patients with T-cell lymphoma, shows good efficacy and tolerability.5 In our investigation herein, we evaluated and explored the efficacy of chidamide in treating myeloma and its associated osteolytic bone disease. Chidamide reduced myeloma cell viability in both primary MM cells and MM cell lines, even in the presence of BMSCs or chidamide-pretreated OCs. As the microenvironment is crucial for myeloma drug resistance and

Figure 4. The molecular mechanisms underlying chidamide activity in myeloma cells. (A) Western blot analysis of Mcl-1, Myc, Bcl-xL, Bcl-2, p21, p27, CDK4, CDK6, and Cyclin-D2 levels; a-tubulin was used as the loading control. (B) SOCS-3, p-JAK2, JAK2, p-STAT3-727, p-STAT3-705, and STAT3 levels were analyzed by Western blotting; a-tubulin was used as the loading control.

1374

haematologica | 2018; 103(8)

Therapeutic effects of chidamide on myeloma-associated bone disease

relapse, our data has shown that chidamide overcomes the drug resistance mediated by the interaction between MM cells and the microenvironment. Moreover, the good anti-myeloma effect of chidamide on the disseminated MM mouse model reveals the efficacy of chidamide in the presence of the BM microenvironment. HDACs regulate a variety of targets involved in differ-

ent cell functions and processes, such as apoptosis, the cell cycle and differentiation.20 Chidamide inhibited HDAC activity and induced cell cycle arrest in G1 phase, accompanied by cell apoptosis. Notably, cell cycle arrest and the alterations in levels of related proteins were not reversed by the pan-caspase inhibitor. HDAC inhibitors cause cell cycle arrest though several pathways, the most important

Figure 5. Dual anti-myeloma and bone-protective effects of chidamide on a disseminated myeloma murine model. (A) Chidamide suppresses the growth of MM xenografts, ***P<0.001. (B) Tumor volumes, serum human IgÎť levels in each group and in vivo bioluminescence imaging were used to compare the tumor burdens between the two groups. (C) Serum levels of CTX-I (**P<0.01) and PINP (***P<0.001) in each group. (D) Micro-CT three-dimensional reconstructed images from the vehicle and chidamide groups. (E) Micro-CT parameters of the bone in the vehicle and chidamide groups. Trabecular numbers, ***P<0.001; trabecular space, **P<0.01; BV/TV, non-significant (ns).

haematologica | 2018; 103(8)

1375

J. He et al.

of which seems to be increasing the expression of cell cycle genes. p21 plays a crucial role in chidamide-induced G1 cell cycle arrest. HDAC1 regulates p21 expression to some extent. HDAC1 interacts with the p21 promoter by competing with the p53 protein to decrease the transcription of p21.21 Based on our results from MM cell lines, in chidamide-treated MM cells HDAC1 was released from

the p21 promotor, subsequently increasing p21 expression. Additionally, chidamide inhibited the expression of the Cyclin-D2 gene, thus inhibiting the functions of CDK4 and CDK6. Aberrant activation of the JAK2/STAT3 pathway is required for the pathogenesis of a variety of cancers.22,23 Inappropriate STAT3 activation protects cells from apoptosis. HDAC3 targets STAT3, and inhibition of HDAC3

Figure 6. Chidamide inhibited osteoclast differentiation in vitro. (A) Western blotting was used to detect the expression levels of key factors (DUSP1, c-fos, NFATc1, and HDAC10) during OC maturation; a-tubulin was used as the loading control. (B) Western blot analysis of the expression levels of pathway proteins including pp38, p38, p-ERK, ERK, p-JNK, JNK p-AKT, and AKT; levels of key factors, including Cathepsin K, c-fos, NFATc1 and HDAC10, in OCs treated with different concentrations of chidamide. GAPDH was used as the loading control. Acetylation of H3K18 and H4K8 was also evaluated; H3 and H4 were used as loading controls. (C) TRAP staining of OCs treated with different concentrations of chidamide. F-actin ring formation. Rhodamine-conjugated phalloidin was used to visualize F-actin (red), and nuclei were stained with DAPI (blue), 200Ă&#x2014;. Pit formation assay of OCs treated with different concentrations of chidamide. (D) The number of TRAP-positive cells was reduced in cultures treated with increasing doses of chidamide, **P<0.01. (E) Calcium resorption was reduced in the presence of chidamide (0-1ÎźM). Resorption in the lacunae area (%) diminished with increasing doses of chidamide, ***P<0.001. OC: osteoclasts; TRAP: tartrate-resistant acid phosphatase.

1376

haematologica | 2018; 103(8)

Therapeutic effects of chidamide on myeloma-associated bone disease A

Figure 7. Chidamide inhibited osteoclast differentiation and the function of chidamide in non-tumor-bearing mice was examined. (A) Serum CTX-I and PINP levels before (CTX-1, P>0.05, non-significant; PINP, ***P<0.001) and after intraperitoneal injections of sRANKL following chidamide treatment（CTX-1, *P<0.05; PINP, ***P<0.001). (B) TRAP staining of mouse femur bone tissues. (C) Chidamide treatment decreased the number of OCs (TRAP-positive cells, dark red staining) in the mouse femur compared with the vehicle group; ***P<0.001. (D) Micro-CT three-dimensional reconstructed images of the vehicle and chidamide groups. (E) MicroCT parameters of the bone in the vehicle and chidamide groups. Trabecular numbers, ***P<0.001; trabecular space, **P<0.01; BV/TV, non-significant (ns). RANKL: Receptor activator of nuclear factor κ-Β ligand; CTX: carboxy-terminal telopeptide; PINP: amino-terminal propeptide.

haematologica | 2018; 103(8)

1377

J. He et al.

induces STAT3 de-phosphorylation, thus reducing MM cell growth.8,9 Based on our results and findings from previous studies, we speculated that chidamide promoted MM cell apoptosis by inhibiting the JAK2/STAT3 signaling pathway. Furthermore, platelet factor 4 up-regulates SOSC3 expression to induce MM cell apoptosis by inhibiting STAT3.24 The expression of SOCS3, which acts as a tumor suppressor gene and is always silenced by epigenetic modulation in hematological malignancies, was up-regulated in chidamide-treated cells, which explains the decrease in p-JAK2 and p-STAT3 levels.22,25 Bcl-2 family members, including BclxL and Myc, are downstream targets of the JAK2/STAT3 signaling pathway.23 Moreover, active caspase-3 is a negative regulator of the Bcl-2 family which represses the expression of Bcl-xL and Mcl-1 and subsequently induces cell apoptosis.26 Chidamide inhibited the constitutive activation of the JAK2/STAT3 pathway and caspase-3 activation, thus down-regulating Bcl-xL and Myc. Myeloma-associated bone disease is a major complication of MM and decreases patientsâ&#x20AC;&#x2122; quality of life.2,27 Interactions between myeloma cells and OCs not only support myeloma survival but also increase osteoclastogenesis and inhibit osteoblastogenesis, leading to osteolytic bone lesions and myeloma progression.28,29 An important finding of our study is that chidamide not only suppresses the formation and function of OCs in vitro, but also prevents myeloma-associated bone disease in vivo. Bone remodeling is a balance between bone resorption and bone formation.30 Active OCs, together with myeloma cells, increase bone resorption and inhibit osteoblast differentiation, leading to myeloma-associated bone disease. Chidamide-induced myeloma cell apoptosis prevented osteoclast maturation and showed no inhibitory effect on osteoblast differentiation. HDACs, which are targets of chidamide, are involved in OC formation.10,11,31 HDAC2 is reported to play an important role in OC maturation by activating the AKT pathway.11 Additionally, HDAC3 knockdown decreases the expression of NFATc1, Cathepsin K and DC-STAMP, thus inhibiting OC formation.31 Based on our data, chidamide repressed the expression of key factors, such as NFATc1, c-fos and Cathepsin K, during OC maturation, suggest-

References 1. Rollig C, Knop S, Bornhauser M. Multiple myeloma. Lancet. 2015;385(9983):21972208. 2. Silbermann R, Roodman GD. Myeloma bone disease: Pathophysiology and management. J Bone Oncol. 2013;2(2):59-69. 3. Christoulas D, Terpos E, Dimopoulos MA. Pathogenesis and management of myeloma bone disease. Expert Rev Hematol. 2009;2(4):385-398. 4. Gong K, Xie J, Yi H, Li W. CS055 (Chidamide/HBI-8000), a novel histone deacetylase inhibitor, induces G1 arrest, ROS-dependent apoptosis and differentiation in human leukaemia cells. Biochem J. 2012;443(3):735-746. 5. Shi Y, Jia B, Xu W, et al. Chidamide in relapsed or refractory peripheral T cell lymphoma: a multicenter real-world study in China. J Hematol Oncol. 2017;10(1):69. 6. He M, Qiao Z, Wang Y, et al. Chidamide inhibits aerobic metabolism to induce pan-

1378

10.

11.

ing that chidamide suppresses OC differentiation by inhibiting the function of its targets (such as HDAC2 and HDAC3). A previous study reported the effect of HDAC10 on osteoclast differentiation. They showed that during OC differentiation, HDAC10 levels gradually increased, which is consistent with our study. However, when they knocked down HDAC10 in monocytes, OC differentiation was promoted, indicating that HDAC10 may have a negative effect on OC differentiation.32 However, in our study, following chidamide treatment and inhibition of OC formation, HDAC10 was downregulated at the protein level. Since chidamide can inhibit other HDACs in addition to HDAC10, its inhibitory effect on OC differentiation may not have been caused by HDAC10 down-regulation. Both MM cells and OCs can suppress the differentiation of osteoblasts. As a previous study reported, OCs inhibited osteoblast differentiation though exosomes containing micro ribonucleic acids (miRNAs) and some cytokines.33 In our study, chidamide could induce MM cell apoptosis and abrupt OCs maturation, indicating that chidamide may attenuate the inhibitory effect of MM cells and OCs on osteoblasts. When BMSCs were treated with chidamide during osteoblast differentiation, chidamide increased the gene expression levels of ALP and OCN while reducing the gene expression level of Activin A, which acted as a negative regulator in osteoblast differentiation via SMAD2mediated DLX5 down-regulation.34 The ARS experiment showed neither promotion nor an inhibitory effect on osteoblast differentiation. These results may explain the direct bone-protective effect of chidamide in the mouse models. Our work reveals the dual anti-myeloma and bone-protective effects of chidamide in vitro and in vivo. These findings strongly support the potential clinical use of this drug as a treatment for MM in the near future. Funding This work was supported by the Funds for the National Natural Science Foundation of China (Grant No. 91742110 and 81471532) and the Funds for the Natural Science Foundation of Zhejiang province, China (LY17H080001).

creatic cancer cell growth arrest by promoting Mcl-1 degradation. PLoS One. 2016; 11(11):e0166896. Hu X, Wang L, Lin L, et al. A phase I trial of an oral subtype-selective histone deacetylase inhibitor, chidamide, in combination with paclitaxel and carboplatin in patients with advanced non-small cell lung cancer. Chin J Cancer Res. 2016;28(4):444451. Minami J, Suzuki R, Mazitschek R, et al. Histone deacetylase 3 as a novel therapeutic target in multiple myeloma. Leukemia. 2014;28(3):680-689. Harada T, Ohguchi H, Grondin Y, et al. HDAC3 regulates DNMT1 expression in multiple myeloma: therapeutic implications. Leukemia. 2017;31(12):2670-2677. Cantley MD, Fairlie DP, Bartold PM, Marino V, Gupta PK, Haynes DR. Inhibiting histone deacetylase 1 suppresses both inflammation and bone loss in arthritis. Rheumatology (Oxford). 2015; 54(9):1713-1723. Dou C, Li N, Ding N, et al. HDAC2 regu-

12.

13.

14.

15.

16.

lates FoxO1 during RANKL-induced osteoclastogenesis. Am J Physiol Cell Physiol. 2016;310(10):C780-787. Garcia-Gomez A, Ocio EM, Crusoe E, et al. Dasatinib as a bone-modifying agent: anabolic and anti-resorptive effects. PLoS One. 2012;7(4):e34914. Tomimori Y, Mori K, Koide M, et al. Evaluation of pharmaceuticals with a novel 50-hour animal model of bone loss. J Bone Miner Res. 2009;24(7):1194-1205. Paton-Hough J, Chantry AD, Lawson MA. A review of current murine models of multiple myeloma used to assess the efficacy of therapeutic agents on tumour growth and bone disease. Bone. 2015;77:57-68. Hurchla MA, Garcia-Gomez A, Hornick MC, et al. The epoxyketone-based proteasome inhibitors carfilzomib and orally bioavailable oprozomib have anti-resorptive and bone-anabolic activity in addition to anti-myeloma effects. Leukemia. 2013; 27(2):430-440. Lane NE, Yao W, Nakamura MC, et al. Mice

haematologica | 2018; 103(8)

Therapeutic effects of chidamide on myeloma-associated bone disease

17.

18. 19.

20.

21.

22.

lacking the integrin beta5 subunit have accelerated osteoclast maturation and increased activity in the estrogen-deficient state. J Bone Miner Res. 2005;20(1):58-66. Paino T, Garcia-Gomez A, GonzalezMendez L, et al. The Novel Pan-PIM Kinase Inhibitor, PIM447, Displays Dual Antimyeloma and Bone-Protective Effects, and Potently Synergizes with Current Standards of Care. Clin Cancer Res. 2017;23(1):225-238. Tang J, Yan H, Zhuang S. Histone deacetylases as targets for treatment of multiple diseases. Clin Sci (Lond). 2013;124(11):651-662. Laubach JP, San-Miguel JF, Hungria V, et al. Deacetylase inhibitors: an advance in myeloma therapy? Expert Rev Hematol. 2017;10(3):229-237. Seto E, Yoshida M. Erasers of histone acetylation: the histone deacetylase enzymes. Cold Spring Harb Perspect Biol. 2014;6(4):a018713. Pavithra L, Mukherjee S, Sreenath K, et al. SMAR1 forms a ternary complex with p53MDM2 and negatively regulates p53-mediated transcription. J Mol Biol. 2009; 388(4):691-702. Lesina M, Kurkowski MU, Ludes K, et al. Stat3/Socs3 activation by IL-6 transsignaling

haematologica | 2018; 103(8)

23.

24.

25.

26.

27.

28.

promotes progression of pancreatic intraepithelial neoplasia and development of pancreatic cancer. Cancer Cell. 2011;19(4):456469. Siveen KS, Sikka S, Surana R, et al. Targeting the STAT3 signaling pathway in cancer: role of synthetic and natural inhibitors. Biochim Biophys Acta. 2014; 1845(2):136-154. Liang P, Cheng SH, Cheng CK, et al. Platelet factor 4 induces cell apoptosis by inhibition of STAT3 via up-regulation of SOCS3 expression in multiple myeloma. Haematologica. 2013;98(2):288-295. Ortega-Molina A, Boss IW, Canela A, et al. The histone lysine methyltransferase KMT2D sustains a gene expression program that represses B cell lymphoma development. Nat Med. 2015;21(10):1199-1208. Hu Y, Benedict MA, Wu D, Inohara N, Nunez G. Bcl-XL interacts with Apaf-1 and inhibits Apaf-1-dependent caspase-9 activation. Proc Natl Acad Sci USA. 1998;95(8):4386-4391. Tan E, Weiss BM, Mena E, Korde N, Choyke PL, Landgren O. Current and future imaging modalities for multiple myeloma and its precursor states. Leuk Lymphoma. 2011;52(9): 1630-1640. Dotterweich J, Schlegelmilch K, Keller A, et

29.

30.

31.

32.

33.

34.

al. Contact of myeloma cells induces a characteristic transcriptome signature in skeletal precursor cells. Implications for myeloma bone disease. Bone. 2016;93:155-166. Kawano Y, Moschetta M, Manier S, et al. Targeting the bone marrow microenvironment in multiple myeloma. Immunol Rev. 2015;263(1):160-172. Cantley MD, Zannettino ACW, Bartold PM, Fairlie DP, Haynes DR. Histone deacetylases (HDAC) in physiological and pathological bone remodelling. Bone. 2017; 95:162-174. Pham L, Kaiser B, Romsa A, et al. HDAC3 and HDAC7 have opposite effects on osteoclast differentiation. J Biol Chem. 2011;286(14):12056-12065. Blixt NC, Faulkner BK, Astleford K, et al. Class II and IV HDACs function as inhibitors of osteoclast differentiation. PLoS One. 2017;12(9):e0185441. Sun W, Zhao C, Li Y, et al. Osteoclastderived microRNA-containing exosomes selectively inhibit osteoblast activity. Cell Discov. 2016;2:16015. Vallet S, Mukherjee S, Vaghela N, et al. Activin A promotes multiple myelomainduced osteolysis and is a promising target for myeloma bone disease. Proc Natl Acad Sci USA. 2010;107(11):5124-5129.

1379

ARTICLE

Quality of Life

Ferrata Storti Foundation

Meaningful changes in end-of-life care among patients with myeloma Oreofe O. Odejide,1,2 Ling Li,1 Angel M. Cronin,1 Anays Murillo,1 Paul G. Richardson,3 Kenneth C. Anderson3 and Gregory A. Abel1,4

Division of Population Sciences; 2Center for Lymphoma; 3Center for Myeloma and Center for Leukemia, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA

1 4

Haematologica 2018 Volume 103(8):1380-1389

ABSTRACT

Correspondence: oreofe_odejide@dfci.harvard.edu

Received: January 4, 2018. Accepted: May 3, 2018. Pre-published: May 10, 2018.

atients with advanced myeloma experience a high symptom burden particularly near the end of life, making timely hospice use crucial. Little is known about the quality and determinants of end-of-life care for this population, including whether potential increases in hospice use are also accompanied by “late” enrollment (≤ 3 days before death). Using the Surveillance, Epidemiology, and End-Results-Medicare database, we identified patients ≥ 65 years diagnosed with myeloma between 2000 and 2013 who died by December 31, 2013. We assessed prevalence and trends in hospice use, including late enrollment. We also examined six established measures of potentially aggressive medical care at the end of life. Independent predictors of late hospice enrollment and aggressive end-of-life care were assessed using multivariable logistic regression analyses. Of 12,686 myeloma decedents, 48.2% enrolled in hospice. Among the 6111 who enrolled, 17.2% spent ≤ 3 days there. There was a significant trend in increasing hospice use, from 28.5% in 2000 to 56.5% by 2013 (Ptrend <0.001), no significant rise in late enrollment (12.2% in 2000 to 16.3% in 2013, Ptrend =0.19), and a slight decrease in aggressive end-of-life care (59.2% in 2000 to 56.7% in 2013, Ptrend =0.01). Patients who were transfusion-dependent, on dialysis, or survived for less than one year were more likely to enroll late in hospice and experience aggressive medical care at the end of life. Gains in hospice use for myeloma decedents were not accompanied by increases in late enrollment or aggressive medical care. These findings suggest meaningful improvements in end-of-life care for this population.

Introduction doi:10.3324/haematol.2018.187609 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1380 ©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.

1380

Multiple myeloma is a hematologic cancer diagnosed in over 30,000 individuals each year in the USA.1 It is predominantly a disease of older adults, with a median age of diagnosis at 69 years.2 Although there has been a rapid adoption of novel treatments leading to improvements in survival, myeloma remains an incurable disease.3-6 Moreover, affected patients experience substantial symptom burden throughout the disease trajectory, which intensifies near the end of life (EOL).7,8 Accordingly, high-quality EOL care is crucial for this population. Hospice is a model of care that has been demonstrated to be effective in alleviating patient suffering and improving quality of life for patients near the EOL, through the provision of expert symptom-directed care.9,10 Although hospice enrollment in the USA typically involves discontinuation of chemotherapy and transfusions, this is not the case for many hospice programs in European countries.11 Despite these differences in hospice care delivery in various locations, the central focus is to improve patient quality of life through expert symptomdirected care. In contrast to hospice, medically aggressive care near the EOL is associated with worse patient quality of life.12 Moreover, bereaved caregivers of patients who receive aggressive medical care close to death are less likely to report that their loved ones received “excellent” care, and are also at heightened risk of poor mental health outcomes.9,13 Timely hospice enrollment and avoidance of aggressive medical care near death (e.g., multiple hospitalizations or intensive care haematologica | 2018; 103(8)

End-of-life care for myeloma

unit [ICU] admission) are thus endorsed as indicators of high quality EOL care.14,15 Although little is known about EOL care for patients with myeloma, several studies have demonstrated that patients with hematologic cancers have low rates of hospice enrollment and high rates of aggressive medical care at the EOL compared to patients with solid malignancies.16-19 Rates of hospice use have increased for patients with blood cancers in the past decade; however, there have also been concomitant rises in “late” enrollment (generally defined as ≤ 3 days before death) and aggressive EOL care.20-22 Growth in hospice use that is largely driven by late enrollment is less meaningful, as patients are being admitted to hospice primarily to manage their death, rather than to obtain palliative benefits.23 These trends may not apply to patients with myeloma. Myeloma shares many characteristics with other blood cancers (e.g., bone marrow failure leading to transfusion dependence), has other features that are similar to advanced solid malignancies (e.g., high prevalence of pain, incurability), and still others that are unique (high likelihood of renal disease and dialysis). In this context, we aimed to characterize EOL care among older patients with myeloma. We hypothesized that there would be an increase in hospice use over time; however, given the traditional palliative needs of this population at the EOL and the known incurability of myeloma, we also hypothesized that we would not see increases in late enrollment.

These measures are well-established indicators of potentially suboptimal EOL care.

Covariates In addition to sociodemographic characteristics, we examined comorbidity using the Deyo adaptation of the Charlson Comorbidity Index26 in the twelve months before death, transfusion-dependence (presence of two or more claims for transfusions in the last 30 days of life),20 and dialysis-dependence (presence of two or more claims for dialysis in the last 30 days of life).

Statistical Analyses

Methods

We assessed univariable associations of patient characteristics with outcomes of late hospice enrollment and experiencing at least one indicator of aggressive care using Chi-square tests. We then fit multivariable logistic regression models to characterize factors independently associated with the two aforementioned outcomes. Only covariates with P<0.05 in univariable analysis were included in the models. Trends in overall hospice use, late enrollment, and receipt of at least one indicator of aggressive EOL care were depicted visually using locally weighted scatterplot smoothing, where the day of death was the unit of analysis for the plot. We assessed significant trends over time using the CochranArmitage test, which tested for a monotonic change (i.e., increase or decrease) across the ordered years of death. In a separate multivariable logistic regression model that included hospice use as a covariate, we examined the relationship between hospice enrollment and medically aggressive EOL care. Two-sided P values <0.05 were considered statistically significant. All analyses were performed using SAS version 9.4 (Cary, NC).

Data Source

Results

We used the National Cancer Institute’s Surveillance, Epidemiology, and End Results cancer registry linked to Medicare healthcare claims (SEER-Medicare). This database provides cancer registry data from 18 geographic areas, representing 28% of the population of the USA, linked to billing claims for Medicare beneficiaries.24 At the time of this analysis, the database included diagnoses through 2013 and billing claims through 2014. The DanaFarber/Harvard Cancer Center Office for Human Research Studies deemed the study exempt from review.

Cohort Assembly

We identified patients ≥65 years diagnosed with myeloma or plasmacytoma between 2000 and 2013, who were deceased by December 31, 2013. We excluded patients who died within 30 days of diagnosis. To ascertain complete claims history, patients had to have been continuously enrolled in Medicare Parts A and B with no health maintenance organization enrollment during the twelve months before death. We excluded patients diagnosed with myeloma at death or autopsy, and those who had end-stage renal disease or disability at diagnosis.25 Figure 1 and the Online Supplementary Methods detail the cohort assembly.

Outcomes Hospice use: was defined as the presence of at least one hospice claim (outpatient or inpatient). We defined “late” enrollment as enrollment ≤ 3 days before death. Aggressive EOL care: was defined as the occurrence of at least one of the following indicators: 1) chemotherapy use within 14 days of death, 2) ≥2 emergency department (ED) visits within 30 days of death, 3) ≥2 hospitalizations within 30 days of death, 4) hospital stay >14 days within 30 days of death, 5) at least one ICU admission within 30 days of death, and 6) death in a hospital.12,21 haematologica | 2018; 103(8)

Patient Characteristics This study cohort included 12,686 myeloma decedents. The median age at diagnosis was 77 years. About half of the cohort was male (49.5%) and most were white (80.6%; Table 1). Of the total cohort, 7.3% were transfusion-dependent in the last 30 days of life, and 10.5% were dialysis-dependent. Median survival was 17.6 months (interquartile range [IQR] 5.2 to 39.1 months).

Hospice Use Among the entire cohort, 6111 (48.2%) received hospice care. The median length of stay in hospice was 13 days (IQR 5 to 45 days). The majority (79.6%) used outpatient/home hospice services, while 19.4% used inpatient hospice services, and 1.0% used both. Among those who enrolled, 17.2% spent ≤ 3 days in hospice. In univariable analysis, patients who were transfusion-dependent were more likely to enroll in hospice ≤ 3 days before death compared to those who were not transfusion-dependent (36.5% vs. 16.0%, P<0.001; Table 2). Dialysis-dependence was also associated with late hospice enrollment (32.3% vs. 16.0%; P<0.001). These findings remained consistent in multivariable analysis: patients who were transfusiondependent were more likely to enroll in hospice late (odds ratio [OR] 3.02, 95% confidence internal [CI] 2.39 – 3.82). Similarly, patients on dialysis were also more likely to enroll late (OR 2.22, 95% CI 1.79 – 2.76). Other factors significantly associated with enrolling late in hospice included male sex, living in urban areas, higher comorbidity scores, and surviving less than a year after myeloma diagnosis (Table 3). There was a significant increase in 1381

O.O. Odejide et al.

hospice use over the study period, with rates rising from 28.5% in 2000 to 56.5% by 2013 (Ptrend <0.001); however, there was no significant increase in late enrollment (12.2% in 2000 to 16.3% in 2013, Ptrend =0.19; Figure 2).

Aggressive EOL Care Slightly over half of the cohort (55.8%) had at least one indicator of aggressive EOL care. Nineteen percent of patients received only one indicator of aggressive medical care, 16.0% received two, and only 0.1% received all six indicators of aggressive EOL care. Univariable associations between patient characteristics and aggressive EOL care are displayed in Table 4. In multivariable logistic regression analysis, year of death was a significant determinant of aggressive EOL care. Specifically, we found significantly

lower odds of experiencing any indicator of aggressive care in more recent years compared to earlier years (Table 5). Patients who were transfusion-dependent (OR 3.40, 95% CI 2.87 â&#x20AC;&#x201C; 4.04) or dialysis-dependent (OR 2.32, 95% CI 2.01 â&#x20AC;&#x201C; 2.68) had significantly higher odds of having at least one indicator of medically aggressive care. We also found that age, sex, race, marital status, geographic region, comorbidity, and survival were significantly associated with having one or more indicators of medically aggressive EOL care (Table 5). In univariable analysis examining the relationship between hospice use and medically aggressive care at the EOL, we found that 35.7% of patients who enrolled in a hospice experienced aggressive care compared to 74.5% among those who did not enroll. In a separate multivari-

Figure 1. Cohort assembly.

1382

haematologica | 2018; 103(8)

End-of-life care for myeloma

able model that assessed the relationship between hospice use and aggressive EOL care, patients who enrolled in hospice were significantly less likely to experience any measure of aggressive care at the EOL (OR 0.20, 95% CI 0.18 – 0.22). There was also a slight decrease in experiencing any aggressive EOL care during the study period (59.2% in 2000 to 56.7% in 2013, Ptrend=0.01).

Discussion In this large cohort of older patients with myeloma, almost half enrolled in hospice, and among these, approx-

imately 17% enrolled within 3 days of death. Although hospice enrollment significantly increased between 2000 and 2013, with rates almost doubling, there was no significant rise in late enrollment, suggesting that the increase in hospice use was meaningful. While slightly more than half of the myeloma decedents experienced at least one indicator of medically aggressive care in the last month of life, there was a significant decline in the overall intensity of healthcare utilization during the study period. Moreover, patients who enrolled in a hospice had substantially lower odds of experiencing medically aggressive care at the end of life. Taken together, these data suggest improvements in EOL care for patients with myeloma, which could be

Table 1. Characteristics of patients diagnosed with myeloma who died between 2000 and 2013 (n=12,686).

Characteristic Sex Age at diagnosis (yrs)

Race Marital status at diagnosis Residency College education (census tract quintile)

Median income (census tract quintile)

Region

Time from diagnosis to death* Modified Charlson comorbidity score Dialysis-dependent Transfusion-dependent Year of death

Male Female 65-69 70-74 75- 79 ≥ 80 White Nonwhite Married Other Urban Rural 1 (lowest) 2 3 4 5 (highest) 1 (lowest) 2 3 4 5 (highest) Northeast South Midwest West < 1 year ≥ 1year 0-1 2+ No Yes No Yes 2000 – 2003 2004 – 2008 2009 – 2013

Number

6275 6411 2215 2727 2862 4882 10225 2461 6390 6296 11271 1413 2471 2463 2796 2477 2479 2459 2472 2796 2478 2481 2629 3426 1695 4936 5198 7488 4240 8446 11,350 1336 11,759 927 2407 5047 5232

49.5 50.5 17.5 21.5 22.5 38.5 80.6 19.4 50.4 49.6 88.9 11.1 19.5 19.4 22.0 19.5 19.6 19.4 19.5 22.0 19.5 19.6 20.7 27.0 13.4 38.9 41.0 59.0 33.4 66.6 89.5 10.5 92.7 7.3 19.0 39.8 41.2

*Median duration of disease (from myeloma diagnosis to death) in the cohort was 17.6 months (interquartile range 5.2 to 39.1 months).

haematologica | 2018; 103(8)

1383

O.O. Odejide et al.

further augmented by promoting timely hospice use. The rise in hospice use among myeloma decedents is consistent with prior studies among patients with various malignancies in the USA.16,22,27 Such trends may reflect a greater awareness of the benefits of hospice care, especially as professional oncology organizations have released statements on the importance of hospice.28,29 Moreover, there has also been a substantial growth in the number of hospice organizations serving various locations in the USA over the past two decades.30 Unlike many other

hematologic cancers,21,22 gains in hospice use for patients with myeloma were not accompanied by increases in late enrollment. Distinct features of myeloma compared to other blood cancers, such as incurability and a high prevalence of pain, may make the need for hospice services at the EOL clearer and thus encourage timely enrollment. Indeed, in a prior study examining symptom burden of patients with hematologic malignancies, those with myeloma had the highest number and severity of symptoms, such as pain, fatigue, and constipation.31 A combination of this population’s severe symptom burden and the

Table 2. Univariable analysis of factors associated with hospice enrollment ≤ 3 days before death among myeloma decedents that enrolled in hospice (n=6111).

Characteristic

Sex Age at diagnosis (yrs)

Race Marital status at diagnosis Residency College education (census tract quintile)

Median income (census tract quintile)

Region

Time from diagnosis to death Modified Charlson comorbidity score Dialysis-dependent Transfusion-dependent Year of death

Male Female 65- 69 70-74 75- 79 ≥ 80 White Nonwhite Married Other Rural Urban 1 (lowest) 2 3 4 5 (highest) 1 (lowest) 2 3 4 5 (highest) Northeast South Midwest West < 1 year ≥ 1 year 0-1 2+ No Yes No Yes 2000 – 2003 2004 – 2008 2009 – 2013

Hospice stay ≤ 3 days (n=1054) n (%)

Hospice stay > 3 days (n=5057) n (%)

551 (20.0) 503 (15.0) 177 (19.0) 234 (19.1) 227 (16.6) 416 (16.1) 906 (17.6) 148 (15.6) 572 (18.8) 482 (15.7) 82 (12.7) 972 (17.8) 166 (15.2) 202 (17.1) 243 (17.6) 221 (17.9) 222 (18.1) 156 (14.6) 196 (16.6) 252 (17.4) 210 (17.5) 240 (19.9) 272 (23.7) 228 (12.9) 184 (19.6) 370 (16.4) 426 (18.7) 628 (16.4) 301 (13.1) 753 (19.8) 904 (16.0) 150 (32.3) 922 (16.0) 132 (36.5) 147 (15.7) 396 (17.2) 511 (17.8)

2214 (80.0) 2843 (85.0) 754 (81.0) 993 (80.9) 1142 (83.4) 2168 (83.9) 4255 (82.4) 802 (84.4) 2470 (81.2) 2587 (84.3) 562 (87.3) 4494 (82.2) 929 (84.8) 977 (82.9) 1137 (82.4) 1012 (82.1) 1002 (81.9) 911 (85.4) 988 (83.4) 1199 (82.6) 993 (82.5) 966 (80.1) 876 (76.3) 1543 (87.1) 757 (80.4) 1881 (83.6) 1848 (81.3) 3209 (83.6) 2001 (86.9) 3056 (80.2) 4743 (84.0) 314 (67.7) 4827 (84.0) 230 (63.5) 788 (84.3) 1913 (82.8) 2356 (82.2)

<0.001 0.05

0.14 0.0014 0.0013 0.34

0.02

<0.001

0.02 <0.001 <0.001 <0.001 0.33

All percentages are row percentages.

1384

haematologica | 2018; 103(8)

End-of-life care for myeloma

fact that the hospice model is known to be especially effective in pain management may promote increased enrollment. Additionally, the known incurability of myeloma may temper prognostic uncertainty and encourage earlier EOL discussions compared to blood cancers that are potentially curable.32 The current analysis allowed us to explore potential unique barriers to timely enrollment, such as transfusion-

and dialysis-dependence. Our finding that patients who were transfusion-dependent had three times the odds of enrolling late is provocative, and suggests that transfusiondependence is not only associated with lack of hospice use for patients with blood cancers,20,22 but also impacts the timeliness of enrollment. Although transfusions are palliative, only very few hospices in the USA provide this resource due to reimbursement constraints.33 In some

Figure 2. Trends in overall hospice use and late enrollment (≤ 3 days before death) from 2000 to 2013. Trends in hospice use for myeloma decedents significatly increased from 2000 to 2013 (Ptrend <0.001).Trends in late hospice enrollment (≤ 3 days before death) for myeloma decedents did not significantly increase from 2000 to 2013 (Ptrend =0.19).

Table 3. Multivariable analysis of factors associated with hospice enrollment ≤ 3 days before death among myeloma decedents that enrolled in hospice.

Characteristic Sex Marital status at diagnosis Residency Median income (census tract quintile)

Region

Time from diagnosis to death Modified Charlson comorbidity score Dialysis-dependent Transfusion-dependent

Odds Ratio Male Female Married Other Rural Urban 1 (lowest) 2 3 4 5 (highest) Northeast South Midwest West < 1 year ≥ 1 year 0-1 2+ No Yes No Yes

Ref 0.75 Ref 0.88 Ref 1.34 Ref 1.06 1.06 1.00 1.04 Ref 0.49 0.80 0.64 Ref 0.76 Ref 1.41 Ref 2.22 Ref 3.02

95% Confidence Interval 0.64 – 0.86 0.76 – 1.02 1.03 – 1.75 0.84 – 1.35 0.84 – 1.33 0.76 – 1.24 0.81 – 1.34 0.40 – 0.62 0.63 – 1.01 0.53 – 0.77 0.66 – 0.87 1.22 – 1.65 1.79 – 2.76 2.39 – 3.82

Only variables with P <0.05 in univariable analysis, specifically sex, marital status, urban/rural residency, median income census tract quintile, region, modified Charlson comorbidity score, time from diagnosis to death, dialysis, and transfusion-dependence, were included in the multivariable logistic regression model to generate odds ratios. Odds ratio >1 indicate increased odds of enrolling in hospice late .

haematologica | 2018; 103(8)

1385

O.O. Odejide et al.

health care systems in Europe, where access to transfusion is available in hospice settings, the relationship we observed between transfusion-dependence and late enrollment may not be present. Modifying the current hospice reimbursement structure in the USA to liberalize the use of palliative transfusions would likely improve timely hospice use for patients with myeloma and other hematologic cancers. Indeed, in a national survey of hematologic oncologists, the majority reported that they would refer more

patients to hospice if transfusions were readily available.34 Although providing transfusions necessitates additional costs, our finding that hospice enrollment was associated with a 38.8% absolute reduction in receiving any highcost medically aggressive care at the EOL suggests that this strategy could be overall financially equivalent, at least from the societal perspective. The lack of access to dialysis services in most hospices may contribute to refusals or delays in enrollment among

Table 4. Univariable analysis of factors associated with receipt of at least one indicator of medically aggressive care at the end of life among entire cohort of myeloma decedents from 2000 to 2013 (n=12,686).

Characteristic

Sex Age at diagnosis (yrs)

Race Marital status at diagnosis Residency College education (census tract quintile)

Median income (census tract quintile)

Region

Modified Charlson comorbidity score Time from diagnosis to death Dialysis-dependent Transfusion-dependent Year of death

Male Female 65-69 70-74 75- 79 ≥ 80 White Nonwhite Married Other Rural Urban 1 (lowest) 2 3 4 5 (highest) 1 (lowest) 2 3 4 5 (highest) Northeast South Midwest West 0-1 2+ < 1 year ≥ 1 year No Yes No Yes 2000 – 2003 2004 – 2008 2009 – 2013

Received any aggressive care (n=7079) n (%)

Did not receive any aggressive care (n=5607) n (%)

3670 (58.5) 3409 (53.2) 1415 (63.9) 1689 (61.9) 1629 (56.9) 2346 (48.1) 5514 (53.9) 1565 (63.6) 3717 (58.2) 3362 (53.4) 778 (55.1) 6301 (55.9) 1486 (60.1) 1356 (55.1) 1528 (54.7) 1368 (55.2) 1341 (54.1) 1460 (59.4) 1370 (55.4) 1450 (51.9) 1371 (55.3) 1428 (57.6) 1627 (61.9) 1869 (54.6) 917 (54.1) 2666 (54.0) 1776 (41.9) 5303 (62.8) 3135 (60.3) 3944 (52.7) 6026 (53.1) 1053 (78.8) 6338 (53.9) 741 (79.9) 1417 (58.9) 2781(55.1) 2881 (55.1)

2605 (41.5) 3002 (46.8) 800 (36.1) 1038 (38.1) 1233 (43.1) 2536 (51.9) 4711 (46.1) 896 (36.4) 2673 (41.8) 2934 (46.6) 635 (44.9) 4970 (44.1) 985 (39.9) 1107 (44.9) 1268 (45.3) 1109 (44.8) 1138 (45.9) 999 (40.6) 1102 (44.6) 1346 (48.1) 1107 (44.7) 1053 (42.4) 1002 (38.1) 1557 (45.4) 778 (45.9) 2270 (46.0) 2464 (58.1) 3143 (37.2) 2063 (39.7) 3544 (47.3) 5324 (46.9) 283 (21.2) 5421 (46.1) 186 (20.1) 990 (41.1) 2266 (44.9) 2351 (44.9)

<0.001 <0.001

<0.001 <0.001 0.55 <0.001

<0.001

<0.001 <0.001 <0.001 <0.001 0.003

All percentages are row percentages.

1386

haematologica | 2018; 103(8)

End-of-life care for myeloma

dialysis-dependent myeloma patients. This may partially explain our finding that myeloma patients who were dialysis-dependent were significantly more likely to enroll late. Unlike transfusions, dialysis itself is unlikely to be palliative.35 Accordingly, rather than incorporating dialysis into hospice care, this group of patients may benefit from bridge programs that provide palliative care services before choosing to discontinue dialysis and transition to hospice. Although the rate of medically aggressive care at the EOL for this myeloma cohort (56%) was substantially lower than a prior analysis that included patients with various types of hematologic cancers (78%),19 it is higher than

that described for patients with solid malignancies in both single-institution and population-based studies (3035%).16,19 This intermediate rate supports the hypothesis that the complex features of myeloma that are similar to solid malignancies (e.g., incurability, pain) may ease the transition from more aggressive medical care toward symptom-directed care as compared to other blood cancers. In a qualitative study of hematologic oncologists, physicians who focused on myeloma noted that the incurability of the disease made it less challenging to transition from disease-directed to symptom-focused therapies near the EOL.36 Moreover, in a population-based study of blood

Table 5. Multivariable analysis of factors associated with receipt of at least one indicator of medically aggressive care at the end of life among entire cohort of myeloma decedents from 2000 to 2013 (n=12,686).

Characteristic Sex Age at diagnosis (yrs)

Race Marital status at diagnosis College education (census tract quintile)

Median income (census tract quintile)

Region

Time from diagnosis to death Modified Charlson comorbidity score Dialysis-dependent Transfusion-dependent Year of death

Odds Ratio Male Female 65-69 70-74 75-79 ≥ 80 White Nonwhite Married Other 1 (lowest) 2 3 4 5 (highest) 1 (lowest) 2 3 4 5 (highest) Northeast South Midwest West < 1 year ≥ 1 year 0-1 2+ No Yes No Yes 2000 – 2003 2004 – 2008 2009 – 2013

Ref 0.90 Ref 0.91 0.76 0.55 Ref 1.36 Ref 0.88 Ref 0.86 0.86 0.85 0.78 Ref 0.97 0.85 1.00 1.09 Ref 0.68 0.71 0.73 Ref 0.62 Ref 2.11 Ref 2.32 Ref 3.40 Ref 0.85 0.81

95% Confidence Interval 0.83 – 0.97 0.81 – 1.03 0.68 – 0.86 0.49 – 0.61 1.23 – 1.50 0.81 – 0.95 0.76 – 0.98 0.75 – 0.98 0.73 – 0.99 0.66 – 0.92 0.86 – 1.10 0.74 – 0.98 0.85 – 1.17 0.91 – 1.31 0.60 – 0.77 0.62 – 0.81 0.66 – 0.82 0.58 – 0.68 1.95 – 2.29 2.01 – 2.68 2.87 – 4.04 0.77 – 0.95 0.73 – 0.90

Only variables with P <0.05 in univariable analysis, specifically sex, age, race, marital status, median income census tract, college education census tract, region, modified Charlson comorbidity score, time from diagnosis to death, dialysis-dependence, transfusion-dependence, and year of death, were included in the multivariable logistic regression model to generate odds ratios. Odds ratio >1 indicate higher odds of receiving at least one indicator of aggressive care at the end of life.

haematologica | 2018; 103(8)

1387

O.O. Odejide et al.

cancer patients in the UK, those with myeloma (n=887) were significantly more likely to be referred to palliative care and less likely to die in acute care settings.37,38 Patients who survived more than a year after their diagnosis were more likely to use hospice in a timely fashion and were also less likely to receive aggressive medical care close to death. This is consistent with prior data showing that survival duration is an important determinant of having a home versus hospital death.38 The relationship between survival time and EOL care may reflect increased patient experience with-and thus the desire to avoid-the burden of additional intensive treatments. Moreover, a longer time between diagnosis and death offers more opportunities to engage in advance care planning. Importantly, clear and consistent discussions regarding prognosis and EOL decision-making early in the disease trajectory are necessary if we are going to improve the quality of EOL care across all survival ranges. We acknowledge limitations to our study. First, our cohort was restricted to patients 65 years and older who were enrolled in Medicare, which may limit the generalizability of our findings. Nonetheless, we are reassured that the median diagnostic age for myeloma is well over 65 years. Second, we relied on claims to assess EOL care, which may have variable sensitivity in capturing outcomes of interest. Third, we did not have access to patientsâ&#x20AC;&#x2122; preferences, which are also a significant determinant of the quality of EOL care received. Next, we did not have access to Revised-International Staging System (RISS) stage for patients in this study, and thus could not determine if any association exists between R-ISS and EOL care. Finally, while each indicator of medically aggressive care near the EOL was equally weighted in our analysis as in previous studies,16,19,39 various stakeholders

References 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin. 2017; 67(1):730. 2. SEER Stat Fact Sheets: Myeloma. 3. Warren JL, Harlan LC, Stevens J, Little RF, Abel GA. Multiple myeloma treatment transformed: a population-based study of changes in initial management approaches in the United States. J Clin Oncol. 2013; 31(16):1984-1989. 4. Mateos M-V, Richardson PG, Schlag R, et al. Bortezomib plus melphalan and prednisone compared with melphalan and prednisone in previously untreated multiple myeloma: updated follow-up and impact of subsequent therapy in the phase III VISTA trial. J Clin Oncol. 2010; 28(13):2259-2266. 5. Zonder JA, Crowley J, Hussein MA, et al. Lenalidomide and high-dose dexamethasone compared with dexamethasone as initial therapy for multiple myeloma: a randomized Southwest Oncology Group trial (S0232). Blood. 2010;116(26):5838-5841. 6. Palumbo A, Bringhen S, Liberati AM, et al. Oral melphalan, prednisone, and thalidomide in elderly patients with multiple myeloma: updated results of a randomized controlled trial. Blood. 2008;112(8):31073114.

1388

(patients, hematologic oncologists, policy makers) may assign different levels of importance to each of the indicators. In conclusion, our data suggest that along with vast improvements in treatment and survival, there has also been meaningful progress in EOL care for patients with myeloma in the USA. These patients are not only enrolling more often in hospice, but the increase in use is not driven by late enrollment. Still, there remains ample opportunity for further improvement, particularly among patients who survive less than one year, are dialysisdependent, or transfusion-dependent. Possible solutions include earlier goals of care discussions, bridge palliative care services, and modification of the hospice model to enable transfusion support. Funding OOO received research support from the National Cancer Institute of the National Institutes of Health (NCI K08CA218295), National Palliative Care Research Center Career Development Award, and Harvard Medical School Office for Diversity Inclusion and Community Partnership Faculty Fellowship. Acknowledgments This study used the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, NCI; the Office of Research, Development and Information, Centers for Medicare & Medicaid Services; Information Management Services, Inc.; and the SEER Program tumor registries in the creation of the SEER-Medicare database.

7. Ramsenthaler C, Kane P, Gao W, et al. Prevalence of symptoms in patients with multiple myeloma: a systematic review and meta-analysis. Eur J Haematol. 2016;97(5): 416-429. 8. Ramsenthaler C, Osborne TR, Gao W, et al. The impact of disease-related symptoms and palliative care concerns on health-related quality of life in multiple myeloma: a multi-centre study. BMC Cancer. 2016; 16:427. 9. Wright AA, Keating NL, Balboni TA, Matulonis UA, Block SD, Prigerson HG. Place of death: correlations with quality of life of patients with cancer and predictors of bereaved caregivers' mental health. J Clin Oncol. 2010;28(29):4457-4464. 10. Teno JM, Clarridge BR, Casey V, et al. Family perspectives on end-of-life care at the last place of care. JAMA. 2004;291(1): 88-93. 11. Radbruch L, Payne S, Bercovitch M, et al. White Paper on standards and norms for hospice and palliative care in Europe: Part 2. Eur J of Palliat Care. 2010;17(1):22-23. 12. Wright AA, Zhang B, Ray A, et al. Associations between end-of-life discussions, patient mental health, medical care near death, and caregiver bereavement adjustment. JAMA. 2008;300(14):1665-1673. 13. Wright AA, Keating NL, Ayanian JZ, et al. Family perspectives on aggressive cancer

14.

15.

16.

17.

18.

19.

care near the end of life. JAMA. 2016;315(3):284-292. ASCO. The quality oncology practice initiative quality measures. [cited 2017 September 28]; Available from: http://www.instituteforquality.org/qopi/me asures National quality forum. 2014 [cited 2017 September 28]; Available from: http://www.qualityforum.org/Publications/ 2012/04/Palliative_Care_and_End-ofLife_Care%E2%80%94A_Consensus_Rep ort.aspx Earle CC, Neville BA, Landrum MB, Ayanian JZ, Block SD, Weeks JC. Trends in the aggressiveness of cancer care near the end of life. J Clin Oncol. 2004;22(2):315321. Ho TH, Barbera L, Saskin R, Lu H, Neville BA, Earle CC. Trends in the aggressiveness of end-of-life cancer care in the universal health care system of Ontario, Canada. J Clin Oncol. 2011;29(12):1587-1591. O'Connor NR, Hu R, Harris PS, Ache K, Casarett DJ. Hospice admissions for cancer in the final days of life: independent predictors and implications for quality measures. J Clin Oncol. 2014;32(28):3184-3189. Hui D, Didwaniya N, Vidal M, et al. Quality of end-of-life care in patients with hematologic malignancies: a retrospective cohort study. Cancer. 2014;120(10):1572-1578.

haematologica | 2018; 103(8)

End-of-life care for myeloma

20. Fletcher SA, Cronin AM, Zeidan AM, et al. Intensity of end-of-life care for patients with myelodysplastic syndromes: Findings from a large national database. Cancer. 2016;122(8):1209-1215. 21. Odejide OO, Cronin AM, Earle CC, LaCasce AS, Abel GA. Hospice use among patients with lymphoma: impact of disease aggressiveness and curability. J Natl Cancer Inst. 2016;108:1. 22. Wang R, Zeidan AM, Halene S, et al. Health care use by older adults with acute myeloid leukemia at the end of life. J Clin Oncol. 2017;35(30):3417-3424. 23. Earle CC, Landrum MB, Souza JM, Neville BA, Weeks JC, Ayanian JZ. Aggressiveness of cancer care near the end of life: is it a quality-of-care issue? J Clin Oncol. 2008;26(23):3860-3866. 24. Surveillance, epidemiology, and end results program national cancer institute. 2018 [cited 2018 March 12, 2018]; Available from: https:// healthcaredelivery.cancer.gov/seermedicare/overview/link ed.html 25. Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEERMedicare data: content, research applications, and generalizability to the United States elderly population. Med Care. 2002;40(8 Suppl):IV-3-18. 26. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin

haematologica | 2018; 103(8)

Epidemiol. 1992;45(6):613-619. 27. Wright AA, Hatfield LA, Earle CC, Keating NL. End-of-life care for older patients with ovarian cancer is intensive despite high rates of hospice use. J Clin Oncol. 2014;32(31):3534-3539. 28. Peppercorn JM, Smith TJ, Helft PR, et al. American Society of Clinical Oncology Statement: toward individualized care for patients with advanced cancer. J Clin Oncol. 2011;29(6):755-760. 29. Partridge AH, Seah DSE, King T, et al. Developing a service model that integrates palliative care throughout cancer care: the time is now. J Clin Oncol. 2014; 32(29):33303336. 30. NHPCO's Report Facts and Figures : Hospice Care in America. 2015 [cited 2018 March 13, 2018]; Available from: https://www.nhpco.org/sites/default/files/p ublic/Statistics_Research/ProviderGrowth.p df 31. Johnsen AT, Tholstrup D, Petersen MA, Pedersen L, Groenvold M. Health related quality of life in a nationally representative sample of haematological patients. Eur J Haematol. 2009;83(2):139-148. 32. Odejide OO, Cronin AM, Condron NB, et al. Barriers to quality end-of-life care for patients with blood cancers. J Clin Oncol. 2016;34(26):3126-3132. 33. Johnson KS, Payne R, Kuchibhatla MN, Tulsky JA. Are hospice admission practices

associated with hospice enrollment for older African Americans and Whites? J Pain Symptom Manage. 2016;51(4):697-705. 34. Odejide OO, Cronin AM, Earle CC, Tulsky JA, Abel GA. Why are patients with blood cancers more likely to die without hospice? Cancer. 2017;123(17):3377-3384. 35. Tamura MK, Cohen LM. Should there be an expanded role for palliative care in end-stage renal disease? Curr Opin Nephrol Hypertens. 2010;19(6):556-560. 36. Odejide OO, Salas Coronado DY, Watts CD, Wright AA, Abel GA. End-of-life care for blood cancers: a series of focus groups with hematologic oncologists. J Oncol Pract. 2014;10(6):e396-403. 37. Howell DA, Wang H-I, Roman E, et al. Variations in specialist palliative care referrals: findings from a population-based patient cohort of acute myeloid leukaemia, diffuse large B-cell lymphoma and myeloma. BMJ Support Palliat Care. 2015;5(5):496502. 38. Howell DA, Wang HI, Smith AG, Howard MR, Patmore RD, Roman E. Place of death in haematological malignancy: variations by disease sub-type and time from diagnosis to death. BMC Palliat Care. 2013;12(1):42. 39. Tang ST, Wu SC, Hung YN, Chen JS, Huang EW, Liu TW. Determinants of aggressive end-of-life care for Taiwanese cancer decedents, 2001 to 2006. J Clin Oncol. 2009;27(27):4613-4618.

1389

ARTICLE

Cell Therapy & Immunotherapy

Ferrata Storti Foundation

Haematologica 2018 Volume 103(8):1390-1402

The early expansion of anergic NKG2Apos/ CD56dim/CD16neg natural killer represents a therapeutic target in haploidentical hematopoietic stem cell transplantation

Alessandra Roberto,1* Clara Di Vito,2* Elisa Zaghi,2 Emilia Maria Cristina Mazza,1,3 Arianna Capucetti,2 Michela Calvi,2 Paolo Tentorio,2 Veronica Zanon,1 Barbara Sarina,4 Jacopo Mariotti,4 Stefania Bramanti,4 Elena Tenedini,3 Enrico Tagliafico,3 Silvio Bicciato,3 Armando Santoro,4 Mario Roederer,5 Emanuela Marcenaro,6 Luca Castagna,4 Enrico Lugli1,7* and Domenico Mavilio2,8*

Laboratory of Translational Immunology, Humanitas Clinical and Research Center, Rozzano, Milan, Italy; 2Unit of Clinical and Experimental Immunology, Humanitas Clinical and Research Center, Rozzano, Milan, Italy; 3Department of Life Sciences, University of Modena and Reggio Emilia, Modena, Italy; 4Bone Marrow Transplant Unit, Humanitas Clinical and Research Center, Rozzano, Milan, Italy; 5ImmunoTechnology Section, Vaccine Research Center, NIAID, NIH, Bethesda, MD, USA; 6Dipartimento di Medicina Sperimentale (DI.ME.S.) and Centro di Eccellenza per le Ricerche Biomediche (CEBR) UniversitĂ degli Studi di Genova, Italy; 7Humanitas Flow Cytometry Core, Humanitas Clinical and Research Center, Rozzano, Milan, Italy and 8Department of Medical Biotechnologies and Translational Medicine (BioMeTra), University of Milan, Italy 1

AR and CDV, EL and DM contributed equally to this work.

ABSTRACT

Correspondence: domenico.mavilio@unimi.it or enrico.lugli@humanitasresearch.it Received: December 16, 2017. Accepted: April 23, 2018. Pre-published: April 26, 2018. doi:10.3324/haematol.2017.186619 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1390 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.

1390

atural killer cells are the first lymphocyte population to reconstitute early after non-myeloablative and T cell-replete haploidentical hematopoietic stem cell transplantation with post-transplant infusion of cyclophosphamide. The study herein characterizes the transient and predominant expansion starting from the second week following haploidentical hematopoietic stem cell transplantation of a donorderived unconventional subset of NKp46neg-low/CD56dim/CD16neg natural killer cells expressing remarkably high levels of CD94/NKG2A. Both transcription and phenotypic profiles indicated that unconventional NKp46neg-low/CD56dim/CD16neg cells are a distinct natural killer cell subpopulation with features of late stage differentiation, yet retaining proliferative capability and functional plasticity to generate conventional NKp46pos/CD56bright/CD16neg-low cells in response to interleukin-15 plus interleukin-18. While present at low frequency in healthy donors, unconventional NKp46neg-low/CD56dim/CD16neg cells are greatly expanded in the seven weeks following haploidentical hematopoietic stem cell transplantation, and express high levels of the activating receptors NKG2D and NKp30 as well as of the lytic granules Granzyme-B and Perforin. Nonetheless, NKp46neg-low/CD56dim/CD16neg cells displayed a markedly defective cytotoxicity that could be reversed by blocking the inhibitory receptor CD94/NKG2A. These data open new and important perspectives to better understand the ontogenesis/homeostasis of human natural killer cells and to develop a novel immune-therapeutic approach that targets the inhibitory NKG2A check-point, thus unleashing natural killer cell alloreactivity early after haploidentical hematopoietic stem cell transplantation.

Introduction The development over recent years of new protocols of allogeneic bone marrow transplant (BMT) arises from the need to rapidly identify a reliable source of hematopoietic stem cells (HSCs) to cure life-threatening hematologic malignancies. Indeed, the possibility of having a donor for nearly every patient requiring a haematologica | 2018; 103(8)

Unconventional CD56dim/CD16neg NK cells in HSCT

BMT pushed the optimization of different haploidentical HSC transplants (hHSCT) that combined different conditioned regimes and immune-modulation therapies.1 Both myeloablative (MAC) and non-MAC (NMAC) T cellreplete (TCRe) hHSCT followed by post-transplant cyclophosphamide (Cy) gave remarkable positive clinical outcomes.2-4 Donor-derived immune-reconstitution (IR) is the most important player ruling out either a positive or negative clinical outcome of allogeneic HSCT.5 Natural Killer (NK) cells are key for the prognosis of allogeneic BMT given their ability to kill viral-infected or tumor-transformed cells in the absence of a prior sensitization to specific antigens.6-8 NK cell recognition of “self” relies on a large family of inhibitory NK cell receptors (iNKRs) including killer cell immunoglobulin-like receptors (KIRs) and Ctype lectins, such as CD94/NKG2A, which specifically bind different alleles of major histocompatibility complex of class I (MHC-I). A decreased expression or lack of self-MHC-I on target cells unleash NK cell killing via the engagement of several activating NK cell receptors (aNKRs) (i.e., missing self hypothesis).9-11 In the context of allogeneic and non-myeloablative BMT, the presence of a “mismatch” between iNKRs and HLA alleles on recipient cells induces a condition of alloreactivity that makes it possible for donor-derived NK cells to: i) eliminate recipient immune cells that survived the conditioning regimens (i.e., prevent graft reject), ii) kill recipient antigen presenting cells (APCs) presenting host antigens to donor T cells (i.e., avoid the onset of graft-versus-host disease [GvHD]), and iii) clear residual malignant cells in the recipient (i.e., induce graft-versus-leukemia [GvL] effect). These NK cell-mediated effector-functions in mismatched settings are being currently used to develop adoptive NK cell transfer therapies to cure solid and hematologic tumors.7,12 Circulating NK cell subsets are normally defined on the basis of CD56 and CD16 surface expression. Conventional CD56bright/CD16neg-low (cCD56bright) NK cells account for up to 10-15% of the total NK cell population, and are able to secrete a high amount of pro-inflammatory cytokines while displaying poor cytotoxicity. The conventional CD56dim/CD16pos (cCD56dim) phenotype identifies another highly cytotoxic subset that comprises up to 90% of NK cells in the bloodstream.13 Other than the pathologic NK subsets expanded in immunological disorders and in response to pathogens,14 an additional population of unconventional CD56dim/CD16neg (uCD56dim) NK cells has been identified recently.15-17 Despite being rarely represented in healthy donors, uCD56dim is highly cytotoxic and able to efficiently kill hematologic tumors in vitro.18 A subset of CD56low/CD16low NK cells resembling the phenotype and functions of uCD56dim NK cells is expanded in the bone marrow (BM) of healthy pediatric donors and leukemic patients who have undergone a/b T cell-depleted hHSCT.18,19 The study herein demonstrates that the NK cell IR in TCRe -NMAC-PT/Cy hHSCT with reduced intensity conditioning (RIC) are characterized by the early expansion of donor-derived and anergic uCD56dim NK cells showing a peculiar NKG2Apos/NKp46neg-low phenotype. We also demonstrate herein that the blocking of the NKG2A inhibitory pathway on this subset represents an immunotherapeutic target to improve NK cell alloreactivity early after hHSCT. haematologica | 2018; 103(8)

Methods Study Design 30 patients were treated according to our published hHSCT protocol approved by the institutional review boards (IRB) of Humanitas Research Hospital.20 Patients and donors signed consent forms in accordance with the Declaration of Helsinki. Patients’ clinical features, enrolment criteria and timing of specimen collections are shown in the Online Supplementary Table S1.21,22 Peripheral blood mononuclear cells (PBMCs) from healthy volunteers or patients were isolated as previously described.22,23

Flow cytometry and cell sorting Cells were thawed in medium (Roswell Park Memorial Institute [RPMI]-1640 supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin and 1% L-glutamine) containing benzonase nuclease (Sigma-Aldrich). Cells were stained for 15 minutes at room temperature (RT) with live/dead fixable dead cell stain kit (Life Technologies) and 20 minutes at RT with fluorescent-conjugated monoclonal antibodies (mAbs). All mAbs are listed in Online Supplementary Table S2. For Ki-67, Perforin and Granzyme (GRZ)-B intracellular staining, cells were fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer's protocol. The gating strategy to identify NK cells within total PBMCs is shown in Online Supplementary Figure S1. Samples were acquired on a LSR Fortessa and LSR II flow cytometers or fluorescence-activated cell sorting (FACS)-sorted with FACS Aria III (BD Biosciences). NK cell absolute counts were obtained by calculating the percentage within the lymphocyte gate. Additional Methods are included in the Online Supplementary Material

Results Preferential expansion of uCD56dim NK cells in the first weeks after hHSCT We identified CD14neg/CD3neg/CD20neg NK cell subsets in peripheral blood (PB) and donor BM by polychromatic flow cytometry on the basis of their CD56 and CD16 surface expression within the lymphocyte gate.13 We did not include CD56neg or NKG2Dneg lymphocytes in our gating strategy, as these cells were contaminants lacking the NK cell surface markers NKp46, NKG2A, Perforin and Granzyme-B (Online Supplementary Figure S1). Our results showed that the absolute numbers of circulating NK cells in hHSCT recipients reach levels similar to those of their healthy donors within four to five weeks after hHSCT with a chimerism that is completely donor-derived after 28 days (Figure 1A,B). These findings confirmed that donor-derived NK cells represent the first lymphoid compartment to immune-reconstitute in allogeneic HSCT prior to T and B cells.6,7,21,22,24,25 We then characterized the dynamic of circulating NK cell subset IR early after hHSCT. Four distinct NK cell populations were detectable in both healthy donors and patients and their frequencies were remarkably different within the several time points examined. First, we observed that the frequency of cCD56bright NK cells was significantly higher in the recipients compared to healthy donors starting from the third week from the transplant, and returned to similar physiologic levels 11 weeks after hHSCT. Conversely, the percentage of the cCD56dim NK cells was very low, if not undetectable, in the recipients within the same period, and returned similar to that of their related donors five months after the transplant. The 1391

A. Roberto et al.

low levels of cCD56dim NK cells early after hHSCT were counterbalanced by a significant expansion the uCD56dim NK cell subset which, instead, was present at low levels in PB under homeostatic conditions. uCD56dim NK cells started to statistically increase in the recipients compared to their related donors in the second week after the transplant, outnumbered all other NK cell subsets in the second and third week, and returned to normal level only eight weeks after hHSCT. Different from Human Leukocyte Antigen (HLA)-matched HSCT,26 CD56bright/CD16pos NK cells did not increase in frequency at any time-point in

hHSCT recipients compared to their related donors (Figure 1C,D). We and others have reported that the infection/re-activation with human cytomegalovirus (HCMV) greatly influences the homeostasis and ontogenesis of NK cell subset and induces the expansion of CD56neg/CD16pos NK cells. This phenomenon is particularly relevant in the immune-compromised recipients receiving allogeneic HSCT.14 Although 23 of the 30 (77%) transplanted patients recruited for this study experienced a HCMV infection/reactivation starting from day 29 post hHSCT up to day 64 following the transplant (Online

D Figure 1. Kinetic of NK cell subset immune reconstitution after haploidentical HSCT. (A) Summary graph showing the absolute counts (cells/μL) of circulating natural killer (NK) cells (mean ± SEM) from hematopoietic stem cell healthy donors (HDs) and their related recipients at different time-points after haploidentical HSCT (hHSCT). (B) Representative example of flow cytometry dot plots showing the complete chimerism of HD-derived HLA-A2neg NK cells reconstituting an HLA-A2pos recipient (upper line) after four and eight weeks from hHSCT (lower line). (C) Representative example of flow cytometry dot plots showing the kinetic of HD-derived NK cell subset distribution in the recipient after two, three, four and five weeks from hHSCT. (D) Summary statistical graph showing the frequency (median ± SEM) of conventional CD56bright/CD16neg-low (cCD56bright), CD56bright/CD16pos, conventional CD56dim/CD16pos (cCD56dim) and unconventional CD56dim/CD16neg (uCD56dim) NK cell subsets in the peripheral blood (PB) and bone marrow (BM) of 30 hematopoietic stem cell HDs compared to their counterparts in the blood of the related recipients up to six months after hHSCT. *P<0.05.

1392

haematologica | 2018; 103(8)

Unconventional CD56dim/CD16neg NK cells in HSCT

Supplementary Table S1), neither the frequency (Online Supplementary Figure S2A) nor the phenotype (data not shown) of uCD56dim NK cells were affected over time by this viral infection. Even though the existence and the functional relevance of uCD56dim NK cells have been previously reported,15-17 a recent study claimed that the CD56dim/CD16neg phenotype is induced by cryopreservation.27 To further validate our experimental results performed on cryopreserved cells, we compared the distribution of NK cell subsets between freshly isolated and thawed PBMCs from healthy donors and transplanted patients. Our results showed that both frequencies and phenotypes of the four NK cell subsets from cryopreserved PBMCs are similar to those of freshly purified ones (Online Supplementary Figure S2B-D).

uCD56dim lymphocytes are bona fide NK cells To confirm that CD14neg/CD3neg/CD20neg uCD56dim lymphocytes are indeed NK cells, polychromatic flow cytometry data from 11 healthy donors and from five patients purified three weeks after hHSCT were labelled with a unique computational barcode, concatenated and analyzed by the t-distributed Stochastic Neighbor Embedding (t-SNE) algorithm.28 We arbitrarily identified 13 different

clusters (from C1 to C13) of non-T and non-B lymphocytes on the basis of population boundaries distinguishable on the t-SNE density plots (Figure 2A). We then determined the frequency of antigen expression in each cluster by manual gating (Online Supplementary Figure S3), and displayed the results using a heat map. Figure 2B shows that all clusters except C13 harbor several NK cell surface markers at different degrees of expression. cCD56bright, cCD56dim and uCD56dim NK cell subsets overlap across the distinct clusters defined in the t-SNE map and show a similar distribution in healthy donors and hHSCT recipients (Figure 2C,D). cCD56bright are composed of C7, C8, C9 and, as expected, are NKp46pos and NKG2Apos while expressing low levels of Granzyme-B and Perforin. cCD56dim NK cells are present at higher frequencies in healthy donors compared to hHCST recipients and belong to C3, C4, C5, and C6 groups. These cells are characterized by constitutive high expressions of Granzyme-B and Perforin. uCD56dim cells are comprised of C10, C11, C12 and are NKp46neg-low, NKG2Dpos, Granzyme-Bpos and Perforinpos (Figure 2E). This high-dimensional approach of polychromatic flow cytometry data analysis, although applied to a small number of hHSCT patients, indicate that uCD56dim cells are bona fide NK cells.

Figure 2. Clustering of uCD56dim NK cells. (A) t-distributed Stochastic Neighbor Embedding (t-SNE) plot of lymphocytes from 11 healthy donors (HDs) and five recipients at three weeks after haploidentical HSCT (hHSCT). CD3pos T (green on the left plot) and CD20pos B (orange on the left plot) cells are grouped within the t-SNE map. Within the CD3neg/CD20neg gate (gray within the left plot), 13 (from C1 to C13) different clusters of lymphocytes were defined based on the population boundaries (right plot). (B) Heatmap showing the degree of expression of CD56, CD16, CD8, NKp46, NKG2A, NKG2D, Granzyme-B (GRM-B) and Perforin on the 13 clusters of non-T and non-B lymphocytes defined in the right t-SNE plot of panel A. (C-D) t-SNE plots showing, within the 13 CD3neg/CD20neg clusters of lymphocytes presented in panel B, the clusters of cCD56bright (blue), cCD56dim (black) and uCD56dim (red) NK cell subsets from HDs and from hHSCT-patients three weeks after HSCT together (C) or separately (D). (E) Graphs showing the frequencies (median Âą SEM) of cCD56bright, cCD56dim and uCD56dim from HDs and patients at three weeks after hHSCT (w3) out of the total cells in each of the 13 clusters of CD3neg/CD20neg lymphocytes.

haematologica | 2018; 103(8)

1393

A. Roberto et al.

uCD56dim NK cells are not NK cell precursors and express low levels of NKp46 Ontogenetically, human NK cell precursors have been divided into three main differentiation stages on the basis of their different expression of several surface markers, including CD34, CD117 and CD127. These precursors give rise first to cCD56bright (stage 4) and then to terminally differ-

entiated cCD56dim (stage 5) NK cell subsets that are characterized by a CD34neg/CD117neg/CD127neg phenotype and express aNKRs (i.e., NKG2D and natural cytotoxicity receptors [NCRs]).29,30 Our data showed that uCD56dim NK cells from both healthy donors and hHSCT patients are NKG2Dpos and NKp30pos, but do not express CD34, CD117 and CD127, thus proving that they are NK cells in later

Figure 3. Phenotype of NK cell subsets in healthy donors and haploidentical HSCT patients. (A) Representative example of flow cytometry dot plots from a healthy donors (HDs) and a patient after three weeks from haploidentical HSCT (hHSCT) showing the surface expression of CD117, CD34, CD127 and NKG2D on cCD56bright (blue), cCD56dim (black) and uCD56dim (red) NK cells. The phenotypes of these representative NK cell subsets are overlaid with those of their viable lymphocytes (gray background) used as positive controls. (B) Summary statistical graph showing the expression of CD34, CD117 and CD127 on cCD56bright (blue), cCD56dim (black) and uCD56dim (red) from three HSC HDs and their recipients at three and five weeks after hHSCT. (C) Summary statistical graph showing the expression of NKG2D, Granzyme-B, Perforin, NKp30, NKp46 on cCD56bright (blue), cCD56dim (black) and uCD56dim (red) from HSC HDs and their recipients after three and four weeks from hHSCT. *P<0.05

1394

haematologica | 2018; 103(8)

Unconventional CD56dim/CD16neg NK cells in HSCT

stages of differentiation rather than NK cell precursors (Figure 3A,B). In line with their previously reported high cytotoxicity,18 we also observed that uCD56dim NK cells express high levels of Perforin and Granzyme-B. Interestingly, we also found that uCD56dim NK cells have low levels of NKp46, thus making this NCR an additional phenotypic marker capable of distinguishing the latter subset from both cCD56bright and cCD56dim NK cells (Figure 3C).

first analyzed our results via the principal component analysis (PCA) algorithm. PCA showed that the transcriptional signatures of NK cell subsets from healthy controls are different from those of hHSCT patients (Figure 4A). Moreover, the gene set enrichment analysis (GSEA) performed on healthy donors and hHSCT patients indicates that the majority of the gene sets enriched in the recipients are associated with cell cycle, DNA repair and ribonucleic acid (RNA) transcription (false discovery rate [FDR] <0.00001) (Online Supplementary Table S3). Hence, donorderived NK cells early after hHSCT are activated and endowed with high proliferative potential. We then analyzed the gene expression of cCD56bright, uCD56dim and cCD56dim NK cells from healthy donors, and determined that 3072 transcripts are differentially expressed between these three NK cell subsets. These different transcriptional profiles were then compared with those of uCD56dim and cCD56bright NK cells from patients. Indeed, these latter two NK cell subsets were the only ones to have expanded and be detectable three weeks

uCD56dim NK cells expanded early after hHSCT have a unique transcriptional profile To gain more insights into the biological and functional relevance of uCD56dim NK cells, we assessed the gene expression profiles of FACS-sorted circulating NK cell subsets from three healthy donors (three replicates for each of the three cCD56bright, cCD56dim and uCD56dim NK cell subsets) and from their three hHSCT recipients (three replicates for each of the two cCD56bright and uCD56dim NK cell subsets) three weeks after the transplant. In order to reduce data dimensionality in three main directions, we

Figure 4. Transcriptional profiles of NK cell subsets from healthy donors and patients three weeks after haploidentical HSCT. (A) Principal component analysis (PCA) showing the gene expression profiles of cCD56bright, cCD56dim and uCD56dim NK subsets from healthy donors (HDs) and of cCD56bright and uCD56dim from patients after three weeks from haploidentical HSCT (hHSCT). (B) Hierarchical clustering of NK cell subsets from healthy donors and patients three weeks after hHSCT. Sample grouping, obtained from the expression levels of 3072 genes that are differentially expressed between cCD56bright, uCD56dim and cCD56dim NK cells from HDs. Yellow and violet colors indicate decreased and increased expression, respectively. (C) Log2 expression fold-change (FC) in uCD56dim from hHSCT patients versus cCD56bright cells (x-axis), and versus cCD56dim cells (y-axis) from HDs. C1 and C2 boxes indicate the genes similarly expressed with cCD56dim from HDs but downregulated (C1) and upregulated (C2) in uCD56dim from hHSCT versus cCD56bright from HDs. C3 box indicates the genes downregulated in uCD56dim from hHSCT versus cCD56dim from HDs and similarly expressed with cCD56bright from HDs. Downmodulated genes highlighted in green and upmodulated genes highlighted in red are associated with NK cell maturation, genes highlighted in violet are associated with NK cell activation or cytotoxicity.

haematologica | 2018; 103(8)

1395

A. Roberto et al.

after the transplant (Figure 1C,D). Supervised hierarchical clustering revealed that in healthy donors uCD56dim NK cells have a gene expression similar to that of cCD56dim cells and different from that of cCD56bright NK cells. These results were somewhat expected considering that both uCD56dim and cCD56dim NK cells share a high degree of cytotoxicity in physiological conditions.2 Although showing distinct gene expression profiles, cCD56bright and uCD56dim NK cells from hHSCT recipients group together and are more similar to cCD56dim than cCD56bright NK cell subsets from healthy donors (Figure 4B). We then compared the fold change (FC) of the 3072 differentially expressed genes between uCD56dim NK cells from hHSCT patients and cCD56bright (FCx) or cCD56dim (FCy) NK cells from healthy donors (Figure 4C). Among those transcripts of uCD56dim NK cells from hHSCT patients similarly expressed in cCD56dim NK cells from healthy donors (FCy (log2) < |1|) but differently modulated in cCD56bright NK cells from healthy donors (FCx (log2) > |1|), we found a downregulation of CCR7, interleukin (IL)7R and TCF7 (Figure 4C1) as well as an upmodulation of CCL3, CCL4, PRDM1 and IFN-γ (Figure 4C2). Furthermore, uCD56dim NK cells from hHSCT patients show an increased gene expression of NKp30, Perforin (PRF), Granzyme (GRZ-B), GRZ-A and GRZ-H compared to cCD56bright NK cells from healthy donors (Figure 4C2). Among those transcripts of uCD56dim NK cells from hHSCT patients similarly expressed in cCD56bright NK cells

from healthy donors (FCx (log2) < |1|) but differently modulated in comparison with cCD56dim NK cells from healthy donors (FCy (log2) > |1|) we found a downregulation of the maturation marker KLRG1 (Figure 4C3), of CD160 and GRZ-M (Figure 4C3). These transcriptional profiles parallel our flow cytometry data and confirmed our working hypothesis postulating that uCD56dim NK cells present at high frequencies early after hHSCT are not NK cell precursors, but rather represent lymphocytes in a later stage of differentiation whose gene expression profile is intermediate between cCD56bright and cCD56dim NK cells.

Highly proliferating NKp46neg-low/uCD56dim cells can generate NKp46pos/cCD56bright NK cells Donor-derived uCD56dim and cCD56bright NK cell subsets expressed high levels of Ki-67 (i.e., proliferating) at three and four weeks after hHSCT, while their counterparts in healthy donors were all Ki-67neg (i.e., quiescent) (Figure 5A). These high rates of NK cell proliferation in hHSCT are associated with the so-called “cytokine storm” which occurs early after allogeneic transplant and induces immune cell activation and differentiation to rapidly recuperate the recipients from the previously induced and lifethreatening condition of immunodeficiency.25,31 The preferential expansion of Ki67pos/uCD56dim NK cells starting from the second week after hHSCT prompted us to hypothesize that they could generate either cCD56bright or cCD56dim

Figure 5. FACS-sorted uCD56dim NK cells generate cCD56br NK cells under IL-15 and IL-18 stimulation. (A) Summary statistical graph showing the expression of Ki67 on cCD56bright (blue), cCD56dim (black) and uCD56dim (red) from HSC healthy donors (HDs) and their recipients after three and four weeks from hHSCT. (B) Representative example from a HD of flow cytometry dot plots showing the purity of fluorescence-activated cell sorting (FACS)-sorted cCD56bright (blue), cCD56dim (black) and uCD56dim (red) (left column) natural killer (NK) cell subsets. Highly pure and FACS-sorted NK cell subsets are overlaid with the phenotype of purified CD3neg/CD20neg NK cells expressing CD56 and CD16 (gray). (C) Summary statistical graphs showing the proliferation index of FACS-sorted cCD56bright (blue) and uCD56dim (red) NK cell subsets from six HDs at four, eight and 14 days of culture with interleukin (IL)-15+ IL-18. (D) Summary statistical graph showing the kinetic of CFSE-diluting (CFSEdil) cCD56bright (blue), cCD56dim (black) and uCD56dim (red) NK cell subsets generated from FACS-sorted cCD56bright (upper panel) and uCD56dim (lower panel) from seven HDs. No data are available for the cCD56dim NK cells, as they were not proliferating in response to IL-15 and IL-18. Data are expressed as means ± S.D. *P<0.05; **P<0.01; ***P<0.001.

1396

haematologica | 2018; 103(8)

Unconventional CD56dim/CD16neg NK cells in HSCT

NK cells, whose frequencies in the recipient PB increase over the following weeks (Figure 1C,D). To validate this working hypothesis, we FACS-sorted uCD56dim, cCD56bright and cCD56dim NK cells from healthy donors (Figure 5B), cultured them with IL-15 plus IL-18 and analyzed the phenotype of proliferating cells diluting carboxyfluorescein succinimidyl ester (CFSE) at several time points. IL-15 was chosen on the basis of its enhanced ability to prime NK cell proliferation and because it is increased significantly in the blood of the recipient receiving allogeneic (including haploidentical) HSCT.25,31,32 Similarly, IL-18 is known to have a deep impact on NK cell activation and effector-functions.33,34 In this regard, our transcriptional data showed that IL-18 receptor-1 (IL-18R1) is significantly upregulated in uCD56dim NK cells from hHSCT patients compared to cCD56dim NK cells from healthy donors (FCy (log2) > |1|) (Online Supplementary Table S4). In line with our transcriptional data (Figure 4), we first observed that both uCD56dim and cCD56bright NK cells highly proliferate in response to IL-15 and IL-18, while terminally differentiated cCD56dim NK cells do not. The proliferation index of cCD56bright NK cells is significantly higher after 14 days of culture compared to those of uCD56dim NK cells (Figure 5C and data not shown for

cCD56dim NK cells). Given the differential expression of NKp46 on uCD56dim and cCD56bright NK cells (Figure 3C), we analyzed the surface levels of this NCR to track the fate of these two proliferating subsets. Although cycling at a higher rate, only a minor and not statistically significant fraction of FACS-sorted NKp46pos/cCD56bright NK cells generated NKp46neg-low/uCD56dim NK cells over time. Indeed, the phenotype of proliferating cCD56bright NK cells was similar (range 75-95%) to that of their resting and parental counterparts at day 0. On the other hand, almost all proliferating NKp46neg-low/uCD56dim NK cells gave rise to NKp46pos/cCD56bright NK cells after 14 days of culture, and only a minor fraction retained the NKp46neg-low/uCD56dim phenotype. Finally, we observed that neither FACS-sorted NKp46pos/cCD56bright NK cells nor FACS-sorted NKp46neglow /uCD56dim NK cells were able to generate cCD56dim NK cells (Figure 5D and Figure 6).

uCD56dim cells expanded early after hHSCT are anergic due to high expression of NKG2A Our phenotypic analyses showed that donor-derived uCD56dim NK cells expanded early after hHSCT patients are fully armed to efficiently kill tumor cell targets, as already demonstrated in healthy donors.18 Surprisingly,

Figure 6. Proliferating NKp46neg-low/uCD56dim NK cells differentiate in NKp46pos/cCD56bright NK cells. A) Representative example from a HD of flow cytometry dot plots showing the expression of CD56 on fluorescence-activated cell sorting (FACS)-sorted and CFSE-diluting (CFSEdil) cCD56bright (blue) and uCD56dim (red) from a HD after eight days in culture with interleukin (IL)-15 + IL-18. (B) Representative example from the same HD shown in panel A of flow cytometry histograms showing the level of NKp46 expression, expressed as mean fluorescence intensity (MFI), after eight days in culture with IL-15 + IL+18 (MFI) on cCD56bright (blue line) and uCD56dim (red line) natural killer (NK) cells derived either from FACS-sorted cCD56bright NK cells (blue) or from FACS-sorted uCD56dim NK cells (red). Overlaid gray histograms represent the level of NKp46 expression on the relative freshly purified and FACS-sorted parental NK cells. (C) Summary statistical graph showing the kinetic of NKp46pos/cCD56bright (blue) and of NKp46neg-low/uCD56dim (red) NK cell subsets generated from FACS-sorted cCD56bright (upper graph) and uCD56dim (lower graph) from seven healthy donors. No data are available in panels B and C for cCD56dim NK cells, as they were not proliferating in response to IL-15 and IL-18. Data are expressed as means Âą S.D. *P<0.05; ***P<0.001.

haematologica | 2018; 103(8)

1397

A. Roberto et al.

we found the degree of CD107a degranulation against K562 cell line of uCD56dim cells purified from hHSCT recipients after six weeks from transplant is significantly lower compared to that of their counterparts from healthy donors (Figure 7A,B).

Considering that the cytolytic activity of NK cells is licensed by the dominant regulation of iNKRs over aNKRs,10,11 we screened the expression of a large panel of inhibitory receptors, over time, that could likely explain the anergy of donor-derived uCD56dim NK cells from

Figure 7. Cytotoxicity of NKp46neg-low/uCD56dim from hHSCT patients expressing transient high levels of NK2A early after hHSCT. A) Representative example from a healthy donor (HD) (lower line) of flow cytometry dot plots showing the expression of CD107a (i.e., cytotoxic natural killer [NK] cells) on cCD56bright (left), cCD56dim (middle) and uCD56dim (right) (left column) either in the absence (upper line) or in the presence (lower line) of K562. (B) Summary statistical graph showing the percentage of CD107apos (median ± SEM) on cCD56bright (blue), cCD56dim (black) and uCD56dim (red) from five HDs and four patients after six weeks from haploidentical HSCT (hHSCT). The background of CD107apos NK cells present in the spontaneous degranulation has been subtracted for the analyses performed in the presence of K562 cell line. (C) Summary statistical graph showing the expression of NKG2A on cCD56bright (blue), cCD56dim (black) and uCD56dim (red) from HSC HDs and their recipients at different time points up to six months after hHSCT (medians ± SEM). (D) Summary statistical graph showing the percentage of CD107apos (median ± SEM) on cCD56bright and uCD56dim NK cells from eight HDs and five patients after six weeks from hHSCT alone (-) or incubated (+) with 721.221G cell line, and either in the absence (-) or in the presence (+) of the masking anti-CD94/NKG2A mAb (Y9). *P<0.05; **P<0.01. ns: not significant.

1398

haematologica | 2018; 103(8)

Unconventional CD56dim/CD16neg NK cells in HSCT

hHSCT recipients. We found that the NK cell expression of CD94/NKG2A is remarkably increased after hHSCT. In particular, while only a fraction (median of 47,2 %Âą2,5 %) of uCD56dim NK cells from healthy donors is NKG2Apos, almost all uCD56dim NK cells expanded after hHSCT express this iNKR. The significant higher frequency of NKG2Apos/uCD56dim NK cells in the PB of recipients compared to their related healthy donors is detectable between the fourth and eleventh week following hHSCT (Figure 7C). To understand if this high and transient expression of CD94/NKG2A on donor-derived uCD56dim NK cells reconstituting the recipients could explain their defective cytotoxicity, we performed masking experiments by blocking this iNKR in the presence of a tumor cell line expressing its putative ligand HLA-E (i.e., 721-221G cell line).35,36 Both cCD56bright and uCD56dim NK cells from healthy donors are able to efficiently degranulate, and the masking with an anti-CD94/NKG2A mAb increased the expression of CD107a only in cCD56bright, in line with their high constitutive expression of NKG2A. In contrast, both cCD56bright and uCD56dim NK cells purified from hHSCT after six

weeks from the transplant are not able to efficiently degranulate against 721-221G, and only the blocking of the CD94/NKG2A inhibitory checkpoint significantly increases the expression of CD107a of these NK cell subsets (Figure 7D).

Discussion The recent development of TCRe-NMAC-PT/Cy hHSCT with RIC represented a revolution in the field of BMT for the cure of hematologic malignancies.1,2,37 The study herein characterizes the transient and early expansion of a functionally exhausted subset of donor-derived uCD56dim NK cells that is detectable at a high frequency from the second week after the transplant. uCD56dim NK cells are by far the largest NK cell population in the first weeks after hHSCT and are also characterized by a significant higher expression of CD94/NKG2A compared to their counterparts in healthy donors. The increased surface levels of this iNKR early after hHSCT greatly impairs the cytotoxicity of uCD56dim NK cells, thus affecting those

Figure 8. NK cell immune-reconstitution after haploidentical HSCT. (A) Natural killer (NK) cell ontogenesis (left): after the infusion of unmanipulated graft in non myeloablative haplo- hematopoietic stem cells transplant (hHSCT), the reconstitution of NK cells in recipients is completely donor-dependent and starts from CD34pos hematopoietic stem cells. Indeed, mature NK cells infused with the graft do not survive the PT infusion of cyclophosphamide.25 The first NK cell subsets to be detected starting from the second week after hHSCT are the cCD56bright and uCD56dim NK cells, with the latter population being by far the largest one, expanded early in the first weeks after the transplant. cCD56bright and uCD56dim cells are not NK cell precursors and are able to exert a bi-directional differentiation following stimulation with IL-15, a pro-inflammatory cytokine present at high levels in the sera of lymphopenic recipients in the first weeks after allogeneic transplant.25,31 (B) Clinical impact (lower right): due to the high constitute expression of the inhibitory receptor NKG2A, uCD56dim NK cells are exhausted in their cytolic potential, and this impairment greatly affects: i) the clearance of residual malignant cells in the recipient that survived the conditioning regimens (i.e., decreased graft-versus-leukemia (GvL) effect), ii) the killing of recipient antigen presenting cells (APCs) presenting host antigens to donor T cells (i.e., increase in the onset of graft-versus-host disease [GvHD]), and iii) the elimination of recipient immune T cells that survived the conditioning regimens (i.e., decrease in engraftment).7 (C) Therapeutic insights (upper right): the blocking of the NKG2A inhibitory checkpoint unleashes donor-derived NK cell cytotoxicity and increases their alloreactive potential. Hence, the PT infusion of the humanized anti-NKG2A monoclonal antibody (i.e., Monalizumab) represents a potential novel therapeutic approach to improve the clinical outcome of hHSCT.

haematologica | 2018; 103(8)

1399

A. Roberto et al.

alloreactive NK cell effector functions that are required for a positive clinical outcome of allogeneic HSCTs (Figure 8). It has recently been reported the PT/Cy in hHSCT kills all mature and high proliferating NK cells infused in the recipient with the graft. Indeed, NK cells do not express aldehyde dehydrogenase, thus implying that the immature CD62Lpos/NKG2Apos/KIRneg NK cells reconstituting in the recipients derive from donor HSCs.25 Nonetheless, the kinetic and the clinical impact of NK cell subset distribution in this transplant setting are unknown. Even though the absolute numbers of donor-derived NK cells are restored in the recipients few weeks after the transplant, we show herein that the distribution of their subset takes much longer to acquire a pattern similar to that observed in healthy donors.13 In particular, hHSCT recipients either lack, or have very low frequencies, of circulating cCD56dim NK cells soon after the transplant. The lack of the main cytolytic population that normally accounts for up to 90% of all circulating NK cells is compensated by the expansion, starting from the second week after hHSCT of uCD56dim NK cells, that appear before cCD56dim NK cells and that are present at a higher frequency compared to cCD56bright NK cells early after the transplant. These findings prompted us to first hypothesize that uCD56dim NK cells might represent an additional stage of NK cell differentiation preceding the appearance of cCD56bright and cCD56dim cell subsets. By focusing our analyses on the different transcriptional profiles of NK cell subsets from healthy donors, we first highlighted the peculiar signature of the uCD56dim NK cell subset. These data made it possible to understand how recipients are repopulated over time by specific subsets of donorderived NK cells that are affected by the peculiar lymphopenic environment early after hHSCT, and follow a kinetic highly impacting the functional outcome of the transplant. Indeed, both our transcriptional profiles and phenotypic analyses in healthy donors and hHSCT recipients showed that uCD56dim lymphocytes are bona fide NK cells and not NK cell precursors, are not an artifact of cryopreservation, and express surface markers of late differentiation such as NKG2D and NKp30 as well as lytic granules indicative of a cytotoxic phenotype. Our results are in line with previous studies showing that uCD56dim NK cells in healthy donors, although present at a very low frequency, represent a distinct subset able to efficiently kill tumor cell targets.15-17 It has also been reported that the lack or decreased expression of CD16 on activated and degranulating cCD56dim NK cells is mediated by the metalloproteinase-17 (ADAM17), thus potentially explaining, at least in part, the origin of uCD56dim NK cells.38,39 Considering that mature and highly proliferating NK cells infused with the graft do not survive to the PT/Cy,25 it is highly unlikely that the action of ADAM17 on activated cCD56dim NK cells could alone explain the high frequencies of uCD56dim NK cells early after hHSCT. Indeed, the expansion of this latter circulating NK cell subset has its peak in the first weeks after hHSCT, when cCD56dim NK cells are either undetectable or present at very low frequencies (Figure 1C,D). Moreover, uCD56dim NK cells are characterized by a remarkably high degree of cellular proliferation in response to cytokine activation, are able to generate cCD56bright NK cells, and express a NKp46neg-low phenotype. These functional and phenotypic features do not belong to terminally differentiated cCD56dim NK cells and are neither induced nor mediated by mechanisms associated with 1400

ADAM17 cleavage properties. As a matter of fact, we did not find any difference in the transcriptional levels of ADAM17 between cCD56dim and uCD56dim NK cells purified early after hHSCT (data not shown). Taken together, these data indicate that ADAM17 likely plays a minor role, if any, in the expansion of uCD56dim NK cells early in hHSCT. In the context of the cytokine storm characterizing the systemic lymphopenic environment early after allogeneic HSCT (including hHSCT), IL-15 certainly plays a key role as it is highly increased in patients’ sera in the first week after the transplant.25,31 Indeed, the incubation in vitro with IL-15 plus IL-18 of FACS-sorted NKp46neg-low/uCD56dim purified from healthy donors induces their proliferation and preferential differentiation into NKp46pos/cCD56bright NK cells over time. Only a minor fraction of proliferating uCD56dim NK cells retains its parental phenotype following activation. Additionally we found that, although to a lesser and not statistically significant extent, a small fraction of highly proliferating FACS-sorted NKp46pos/cCD56bright NK cells generate NKp46neglow /uCD56dim NK cells. While these results demonstrate that NKp46 represents an additional surface marker that distinguishes uCD56dim NK cells from both cCD56bright and cCD56dim NK cells, they leave unanswered the question regarding the origin of uCD56dim NK cells. Nonetheless, they indicate the presence of a bi-directional differentiation between this latter subset and cCD56bright NK cells in an ex vivo human setting mimicking a lymphopenic environment highly enriched with IL-15. We are certainly aware that this methodological approach does not resemble the complex cellular and molecular interactions occurring in the human BM niche during lymphopoiesis. Indeed, neither uCD56dim nor cCD56bright NK cells were able to generate terminally-differentiated cCD56dim NK cells, a process that requires the presence of additional signals delivered by fibroblasts, mesenchymal and stromal cells.4042 However, herein we clearly show that both proliferating uCD56dim and cCD56bright NK cells can generate either themselves or their “neighbor” NK cell subset. In this context, the high serum level of IL-15 soon after hHSCT25 could also be manipulated to boost a more potent antitumor response by cCD56bright NK cells, as recently demonstrated in multiple myeloma.43 Further studies are needed in order to disclose the mechanisms that finely tune the bi-directional transition between uCD56dim and cCD56bright, both under physiologic conditions and in the lymphopenic setting following allogeneic HSCTs. Herein, we also demonstrate that the transcriptional profile of uCD56dim NK cells expanded early after hHSCT is distinct from that of their counterparts in healthy donors. This is not surprising in the context of an allogeneic transplant where different stimuli, such as lymphopenia, alloreactivity, a high serum level of cytokines, antigen stimulation, opportunistic viral infections and acute GvHD highly influences the quantity and the quality of IR.1,24,25 Indeed, this peculiar systemic environment induces the preferential expansion starting from the second week after hHSCT of uCD56dim NK cells which, although showing a cytotoxic phenotype, are highly defective in the clearance of tumor cell targets. This cellular functional exhaustion is associated, at least in part, with the transient expression of CD94/NKG2A on all uCD56dim NK cells that account for the majority of the NK cell population within the first weeks following transhaematologica | 2018; 103(8)

Unconventional CD56dim/CD16neg NK cells in HSCT

plant. Indeed, the blocking of CD94/NKG2A with a masking mAb significantly increased the CD107a degranulation as a marker of lytic activity of NK cells purified early after hHSCT. Although the mechanism(s) inducing the expansion of anergic NKG2Apos/uCD56dim NK cells is unknown, this acquired knowledge now makes it possible to develop a therapeutic approach targeting a specific immune check-point whose inhibition can efficiently increase NK cell alloreactivity within a given time-frame after hHSCT. In this regard, the efficacy of the in vivo administration of an anti-CD94/NKG2A blocking mAb (i.e., Monalizumab) in improving NK cell cytotoxicity against solid tumors and leukemic cells has been already reported both in mice and humans.44,45 (clinicaltrials.gov Identifier: 02459301) (Figure 8). Acknowledgments The authors thank the patients for their generosity and participation in this study and the nurses of the Hematology and Bone Marrow Transplant Unit (Humanitas Clinical and Research

References 1. Patriarca F, Luznik L, Medeot M, et al. Experts' considerations on HLA-haploidentical stem cell transplantation. Eur J Haematology. 2014;93(3):187-197. 2. Luznik L, O'Donnell PV, Symons HJ, et al. HLA-haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant. 2008;14(6):641-650. 3. Brunstein CG, Fuchs EJ, Carter SL, et al. Alternative donor transplantation after reduced intensity conditioning: results of parallel phase 2 trials using partially HLAmismatched related bone marrow or unrelated double umbilical cord blood grafts. Blood. 2011;118(2):282-288. 4. Castagna L, Crocchiolo R, Furst S, et al. Bone marrow compared with peripheral blood stem cells for haploidentical transplantation with a nonmyeloablative conditioning regimen and post-transplantation cyclophosphamide. Biol Blood Marrow Transplant. 2014;20(5):724-729. 5. Imamura M, Tanaka J. Immunoregulatory cells for transplantation tolerance and graftversus-leukemia effect. Int J Hematol. 2003;78(3):188-194. 6. Ruggeri L, Capanni M, Urbani E, et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science. 2002; 295(5562):2097-2100. 7. Moretta L, Locatelli F, Pende D, Marcenaro E, Mingari MC, Moretta A. Killer Ig-like receptor-mediated control of natural killer cell alloreactivity in haploidentical hematopoietic stem cell transplantation. Blood. 2011;117(3):764-771. 8. Vivier E, Raulet DH, Moretta A, et al. Innate or adaptive immunity? The example of natural killer cells. Science. 2011;331(6013):44-49. 9. Ljunggren HG, Karre K. In search of the 'missing self': MHC molecules and NK cell recognition. Immunol Today. 1990; 11(7):237-244.

haematologica | 2018; 103(8)

Center). The present study is dedicated to the memory of Alessandro Moretta, a great mentor and a pillar in the field of NK cell biology. Funding This work was supported by Fondazione Cariplo (Grant per la Ricerca Biomedica 2012/0683 to EL and 2015/0603 to DM), Associazione Italiana per la Ricerca sul Cancro (MFAG 10607 to EL, IG.20312 to EM and IG 14687 to DM), by the Italian Ministry of Health (Bando Giovani Ricercatori GR-201102347324 to EL and GR-2013-02356522 to AR), the intramural program of the National Institutes of Allergy and Infectious Diseases (to MR) and intramural research and clinical funding programs of Humanitas Research Hospital (to DM and LC). AR is a recipient of the Guglielmina Lucatello e Gino Mazzega fellowship from the Fondazione Italiana per la Ricerca sul Cancro. EZ is a recipient of the Nella Orlandini fellowship from the Fondazione Italiana per la Ricerca sul Cancro. CDV and EMCM are recipients of post-doctoral fellowships from the Fondazione Umberto Veronesi.

10. Moretta A, Bottino C, Vitale M, et al. Activating receptors and coreceptors involved in human natural killer cell-mediated cytolysis. Annu Rev Immunol. 2001; 19:197-223. 11. Lanier LL. NK cell recognition. Annu Rev Immunol. 2005;23:225-74. 12. Castagna L, Mavilio D. Re-discovering NK cell allo-reactivity in the therapy of solid tumors. J Immunother Cancer. 2016;4:54. 13. Cooper MA, Fehniger TA, Caligiuri MA. The biology of human natural killer-cell subsets. Trends Immunol. 2001;22(11):633640. 14. Lugli E, Marcenaro E, Mavilio D. NK cell subset redistribution during the course of viral infections. Front Immunol. 2014;5:390. 15. Takahashi E, Kuranaga N, Satoh K, et al. Induction of CD16+ CD56bright NK cells with antitumour cytotoxicity not only from CD16- CD56bright NK Cells but also from CD16- CD56dim NK cells. Scand J Immunol. 2007;65(2):126-138. 16. Fan YY, Yang BY, Wu CY. Phenotypically and functionally distinct subsets of natural killer cells in human PBMCs. Cell Biol Int. 2008;32(2):188-197. 17. Penack O, Gentilini C, Fischer L, et al. CD56dimCD16neg cells are responsible for natural cytotoxicity against tumor targets. Leukemia. 2005;19(5):835-840. 18. Stabile H, Nisti P, Morrone S, et al. Multifunctional human CD56 low CD16 low natural killer cells are the prominent subset in bone marrow of both healthy pediatric donors and leukemic patients. Haematologica. 2015;100(4):489-498. 19. Helena S, Paolo N, Giovanna P, et al. Reconstitution of multifunctional CD56lowCD16low natural killer cell subset in children with acute leukemia given alpha/beta T cell-depleted HLA-haploidentical haematopoietic stem cell transplantation. Oncoimmunology. 2017;9(9):e1342024. 20. Raiola A, Dominietto A, Varaldo R, et al. Unmanipulated haploidentical BMT following non-myeloablative conditioning and post-transplantation CY for advanced

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Hodgkin's lymphoma. Bone Marrow Transplant. 2014;49(2):190-194. Roberto A, Castagna L, Gandolfi S, et al. Bcell reconstitution recapitulates B-cell lymphopoiesis following haploidentical BM transplantation and post-transplant CY. Bone Marrow Transplant. 2015;50(2):317319. Roberto A, Castagna L, Zanon V, et al. Role of naive-derived T memory stem cells in Tcell reconstitution following allogeneic transplantation. Blood. 2015;125(18):28552864. Gupta N, Arthos J, Khazanie P, et al. Targeted lysis of HIV-infected cells by natural killer cells armed and triggered by a recombinant immunoglobulin fusion protein: implications for immunotherapy. Virology. 2005;332(2):491-497. Imamura M, Tsutsumi Y, Miura Y, Toubai T, Tanaka J. Immune reconstitution and tolerance after allogeneic hematopoietic stem cell transplantation. Hematology. 2003; 8(1):19-26. Russo A, Oliveira G, Berglund S, et al. NK cell recovery after haploidentical HSCT with posttransplant cyclophosphamide: dynamics and clinical implications. Blood. 2018;131(2):247-262. Dulphy N, Haas P, Busson M, et al. An unusual CD56(bright) CD16(low) NK cell subset dominates the early posttransplant period following HLA-matched hematopoietic stem cell transplantation. J Immunol. 2008;181(3):2227-2237. Lugthart G, van Ostaijen-ten Dam MM, van Tol MJ, Lankester AC, Schilham MW. CD56(dim)CD16(-) NK cell phenotype can be induced by cryopreservation. Blood. 2015;125(11):1842-1843. Amir el AD, Davis KL, Tadmor MD, et al. viSNE enables visualization of high dimensional single-cell data and reveals phenotypic heterogeneity of leukemia. Nat Biotechnol. 2013;31(6):545-552. Yu J, Freud AG, Caligiuri MA. Location and cellular stages of natural killer cell development. Trends Immunol. 2013;34(12):573582. Scoville SD, Freud AG, Caligiuri MA.

1401

A. Roberto et al.

31.

32.

33.

34.

35.

36.

1402

Modeling human natural killer cell development in the era of innate lymphoid cells. Front Immunol. 2017;8:360. Melenhorst JJ, Tian X, Xu D, et al. Cytopenia and leukocyte recovery shape cytokine fluctuations after myeloablative allogeneic hematopoietic stem cell transplantation. Haematologica. 2012;97(6):867873. Mattiola I, Pesant M, Tentorio PF, et al. Priming of human resting NK cells by autologous M1 macrophages via the engagement of IL-1b, IFN-b, and IL-15 pathways. J Immunol. 2015; 195(6):2818-2828. Novick D, Kim S, Kaplanski G, Dinarello CA. Interleukin-18, more than a Th1 cytokine. Semin Immunol. 2013;25(6):439448. Granzin M, Wagner J, Kohl U, Cerwenka A, Huppert V, Ullrich E. Shaping of natural killer cell antitumor activity by ex vivo cultivation. Front Immunol. 2017;8:458. Lee N, Llano M, Carretero M, et al. HLA-E is a major ligand for the natural killer inhibitory receptor CD94/NKG2A. Proc Natl Acad Sci U A. 1998;95(9):5199-5204. Pesce S, Carlomagno S, Moretta A, Sivori S,

37.

38.

39.

40.

Marcenaro E. Uptake of CCR7 by KIR2DS4(+) NK cells is induced upon recognition of certain HLA-C alleles. J Immunol Res. 2015;2015:754373. Blaise D, Furst S, El Cheikh J, et al. Comparison of haploidentical T-replete HSCT followed with post-transplant high dose cyclophosphamide (PT-HDCy) with matched related (MRD) or unrelated (UD) HSCT in patients in or after the 6TH decade. Biol Blood Marrow Transplant. 2015;21(2):S273-S274. Romee R, Foley B, Lenvik T, et al. NK cell CD16 surface expression and function is regulated by a disintegrin and metalloprotease-17 (ADAM17). Blood. 2013; 121(18):3599-3608. Lajoie L, Congy-Jolivet N, Bolzec A, et al. ADAM17-mediated shedding of FcgammaRIIIA on human NK cells: identification of the cleavage site and relationship with activation. J Immunol. 2014; 192(2):741-751. Chan A, Hong DL, Atzberger A, et al. CD56bright human NK cells differentiate into CD56dim cells: role of contact with peripheral fibroblasts. J Immunol.

2007;179(1):89-94. 41. Boissel L, Tuncer HH, Betancur M, Wolfberg A, Klingemann H. Umbilical cord mesenchymal stem cells increase expansion of cord blood natural killer cells. Biol Blood Marrow Transplant. 2008; 14(9):1031-1038. 42. Hosseini E, Ghasemzadeh M, Kamalizad M, Schwarer AP. Ex vivo expansion of CD3depleted cord blood-MNCs in the presence of bone marrow stromal cells; an appropriate strategy to provide functional NK cells applicable for cellular therapy. Stem Cell Res. 2017;19:148-155. 43. Wagner JA, Rosario M, Romee R, et al. CD56bright NK cells exhibit potent antitumor responses following IL-15 priming. J Clin Invest. 2017;127(11):4042-4058. 44. Ruggeri L, Urbani E, Andre P, et al. Effects of anti-NKG2A antibody administration on leukemia and normal hematopoietic cells. Haematologica. 2016;101(5):626-633. 45. McWilliams EM, Mele JM, Cheney C, et al. Therapeutic CD94/NKG2A blockade improves natural killer cell dysfunction in chronic lymphocytic leukemia. Oncoimmunology. 2016;5(10):e1226720.

haematologica | 2018; 103(8)

ARTICLE

Coagulation & Its Disorders

Dexamethasone promotes durable factor VIII-specific tolerance in hemophilia A mice via thymic mechanisms

Ferrata Storti Foundation

Maria T. Georgescu,1 Paul C. Moorehead,2,3 Alice S. van Velzen,4 Kate Nesbitt,1 Birgit M. Reipert,5 Katharina N. Steinitz,5 Maria Schuster,5 Christine Hough1 and David Lillicrap1 Department of Pathology and Molecular Medicine, Queen’s University, Kingston, ON, Canada; 2Janeway Children’s Health and Rehabilitation Centre, St. John’s, NL, Canada; 3 Faculty of Medicine, Memorial University, St. John’s, NL, Canada; 4Department of Pediatric Hematology, Immunology and Infectious Diseases, Emma Children’s Hospital, Amsterdam, the Netherlands and 5Baxalta Innovations GmbH, Vienna, Austria 1

Haematologica 2018 Volume 103(8):1403-1413

ABSTRACT

he development of inhibitory antibodies to factor VIII is the most serious complication of replacement therapy in hemophilia A. Activation of the innate immune system during exposure to this protein contributes to inhibitor development. However, avoidance of factor VIII exposure during innate immune system activation by external stimuli (e.g., vaccines) has not been consistently shown to prevent inhibitors. We hypothesized that dexamethasone, a drug with potent anti-inflammatory effects, could prevent inhibitors by promoting immunologic tolerance to factor VIII in hemophilia A mice. Transient dexamethasone treatment during ainitial factor VIII exposure reduced the incidence of anti-factor VIII immunoglobulin G in both a conventional hemophilia A mouse model (E16KO, 77% vs. 100%, P=0.048) and a hemophilia A mouse model with a humanized major histocompatibility complex type II transgene (E17KO/hMHC, 6% vs. 33%, P=0.0048). More importantly, among E17KO/hMHC mice that did not develop antifactor VIII immunoglobulin G after initial exposure, dexamethasonetreated mice were less likely to develop a response after re-exposure six (7% vs. 52%, P=0.005) and 16 weeks later (7% vs. 50%, P=0.097). Similar results were obtained even when factor VIII re-exposure occurred in the context of lipopolysaccharide (30% vs. 100%, P=0.069). The ability of these mice to develop immunoglobulin G to human von Willebrand factor, a structurally unrelated antigen, remained unaffected by treatment. Transient dexamethasone administration therefore promotes antigen-specific immunologic tolerance to factor VIII. This effect is associated with an increase in the percentage of thymic regulatory T cells (12.06% vs. 4.73%, P<0.001) and changes in the thymic messenger ribonucleic acid transcription profile.

Introduction Neutralizing antibodies (inhibitors) against factor VIII (FVIII) develop in approximately 30% of treated severe hemophilia A (HA) patients, remaining the major complication of therapy in this disease.1 The gold standard for eliminating inhibitors, immune tolerance induction (ITI), is difficult to administer, incompletely effective2 and expensive.3 Strategies for preventing inhibitors are therefore needed. The risk of developing inhibitors is not completely predicted by known patient-related genetic risk factors (e.g., f8 genotype,4 polymorphisms in Il10, Ctla4, Tnfa, major histocompatibility complex class II [MHCII]5), suggesting that inhibitor risk is modifiable. Inhibitors are high-affinity immunoglobulin (Ig) G antibodies that are the result of cognate interactions between FVIII-specific B cells and follicular T helper cells (Tfhs). Tfhs are derived from naïve CD4+ T cells following interactions with mature dendritic cells (DCs).6 In contrast, the interaction of naïve CD4+ T cells haematologica | 2018; 103(8)

Correspondence: david.lillicrap@queensu.ca

Received: January 30, 2018. Accepted: April 19, 2018. Pre-published: April 19, 2018. doi:10.3324/haematol.2018.189852 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/8/1403 ©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.

1403

M.T. Georgescu et al.

with immature DCs results in differentiation to tolerancepromoting regulatory T cells (Tregs)7 or T-cell anergy.8 DC maturation is induced by pro-inflammatory stimuli (e.g., inflammatory cytokines, engagement of pattern recognition receptors), and as such the “decision” regarding immunologic tolerance to FVIII may depend on whether pro-inflammatory stimuli are present during a patient’s initial exposure to FVIII. Inhibitor risk might be reduced by avoiding pro-inflammatory stimuli during initial exposures to FVIII.9 Patients whose first exposure is in the context of prophylactic rather than on-demand therapy may have a lower inhibitor risk.10,11 However, it is not always possible to choose the conditions of first exposure to FVIII, since bleeding that requires treatment may occur before the initiation of prophylaxis. Avoiding FVIII exposure in the presence of other clinically-defined pro-inflammatory stimuli (e.g., febrile illness, vaccines, tissue injury) has been suggested to reduce inhibitor risk in an observational study,9 but these results have not been reproduced. Furthermore, this approach may be difficult to implement,11 making passive avoidance of innate immune stimulation impractical and ineffective. Active pharmacologic suppression of inflammatory signals during initial FVIII exposure would be a more controllable strategy. However, the pro-inflammatory signals responsible for FVIII immunogenicity in HA have not been conclusively identified and therefore cannot be specifically targeted. Glucocorticoids, which affect both innate and adaptive immunity, may mediate the suppression of a variety of pro-inflammatory signals and their immunological consequences.12,13 Therefore, glucocorticoids such as dexamethasone (Dex), are attractive candidates for the suppression of inflammatory danger signals in the context of HA inhibitor development. To test the ability of Dex to promote immunologic tolerance to FVIII and investigate possible mechanisms of action, we used two murine models of HA. The first model is a severe HA mouse (knockout of exon 17 of the f8 gene) in which the murine MHCII loci were replaced with a single transgene for a chimeric human/murine MHCII allele (E17KO/hMHC). Approximately 30% of these mice develop antibodies to human FVIII after repeated exposure,14 suggesting that tolerance is possible, and perhaps inducible, in this model. The second model is a conventional severe HA mouse (knockout of exon 16 of the f8 gene) in which recombinant human FVIII exposure is immunogenic in 100% of animals (E16KO).15 We first hypothesized that E17KO/hMHC mice treated with Dex during an intense initial exposure to FVIII that did not subsequently develop anti-FVIII IgG would, on reexposure to FVIII, be less likely to develop anti-FVIII IgG than would anti-FVIII IgG-negative mice that were initially treated with FVIII alone. We then sought to determine if our treatment protocol could attenuate the anti-FVIII immune response in E16KO mice and investigate potential cellular mechanisms of action.

background. Male mice aged 10-14 weeks were used.14 E16KO. HA mice on a homogeneous C57Bl6 background. Mice were sex-matched across treatment groups and eight weeks of age.16 All animal procedures were in accordance with the Canadian Council on Animal Care guidelines and approved by the Queen’s University Animal Care Committee.

Treatment dosing and blood sampling Dex (Omega) (75μg/dose, ~3mg/kg) was administered intraperitoneally (IP). Recombinant human FVIII (Advate; Baxalta) (6IU/dose, ~240IU/kg unless stated otherwise), lipopolysaccharide (LPS; InvivoGen) (2μg/dose, ~8mg/kg) and ultra-pure plasmaderived human von Willebrand Factor (VWF; Biotest) (2IU/dose, ~80IU/kg) were administered intravenously (IV), via tail vein. Hank’s balanced salt solution (HBSS) was administered as vehicle control at 100μl IP and 250μl IV. Intermittent and final blood samples were obtained via retroorbital plexus and cardiac puncture respectively, then mixed in a 1:10 ratio with 3.2% buffered citrate. Plasma was separated by centrifugation, then stored at -80°C.

Short-term treatment protocol Initial exposure. At week zero, E17KO/hMHC or E16KO mice received FVIII and Dex (FVIII+Dex group) or FVIII alone (FVIII group) for five consecutive days (Figure 1A,B). At week five, blood samples were collected. Re-exposure. FVIII and FVIII+Dex E17KO/hMHC mice with no evidence of anti-FVIII IgG following initial exposure received FVIII (FVIII/FVIII group and FVIII+Dex/FVIII group), or FVIII and lipopolysaccharide (LPS; FVIII/FVIII+LPS group and FVIII+Dex/FVIII+LPS group) for three consecutive days (week six, Figure 1A). At week nine, blood samples were collected.

Long-term treatment protocol Initial exposure. E17KO/hMHC mice received FVIII and Dex (FVIII+Dex group) or FVIII alone (FVIII group) for five consecutive days (week zero, Figure 4). At week four, all mice were sampled. Intermittent low-dose FVIII exposure and re-exposure. FVIII+Dex mice with no evidence of anti-FVIII IgG were divided into two groups. One group received FVIII for three consecutive days at week 16 (FVIII+Dex/FVIII group). The other group received intermittent exposures to low-dose FVIII (2IU/dose at weeks four, eight and 12) followed by FVIII (6IU/dose) for three consecutive days at week 16 (FVIII+Dex/intFVIII+FVIII group). FVIII mice with no evidence of anti-FVIII IgG received FVIII for three consecutive days at week 16 (FVIII/FVIII group). All mice were sampled before (week 14) and after (week 18) FVIII re-exposure. Human VWF exposure. Mice received human plasma-derived VWF once weekly at weeks 18 to 21. At week 22, blood samples were collected.

Anti-FVIII IgG ELISA Anti-FVIII IgG titers were measured via enzyme-linked immunosorbent assay (ELISA) as previously described.17 An optical density (OD) cutoff of 0.3 above the OD490 of the blank sample was the criterion for positivity, and the titer was determined to be the highest dilution at which a given sample was positive. Samples with an OD490 below the cutoff at a 1:40 dilution were considered to have non-detectable anti-FVIII IgG.

Methods Bethesda assay Animals E17KO/hMHC. HA mice with all murine MHCII alleles knocked out and expressing a single chimeric human/murine transgene of the HLADRB1*1501 allele on a mixed C57Bl6/S129 1404

FVIII inhibitory activity was measured via Bethesda assay as previously described.18 Residual FVIII activity was quantified using an automated coagulometer (STA Compact, Stago). Inhibitory activity was calculated only for samples with a residual FVIII activhaematologica | 2018; 103(8)

Dexamethasone promotes tolerance to factor VIII

ity of between 25% and 75%. The reported inhibitory activity was calculated from the plasma dilution that resulted in a residual FVIII activity closest to 50%. Samples with no evidence of FVIII inhibitory activity or with inhibitory activity <0.4BU/ml when undiluted were considered negative.

positivity was the OD490 of pooled plasma (1:40 dilution) taken from FVIII deficient mice with no previous exogenous VWF exposure. Samples with an OD490 below the cutoff at a dilution of 1:40 were considered to have non-detectable anti-human VWF IgG.

Lymphocyte enumeration studies Anti-human VWF IgG ELISA Ninety-six-well plates (4HBX, Immulon) were coated with FVIII free plasma-derived human VWF (0.1IU/ml, ~1μg/ml) (Biotest) overnight at 4°C, then blocked for two hours at room temperature. Plasma samples were diluted to 1:40 and incubated for two hours at room temperature in duplicate. IgG detection was carried out as described for the anti-FVIII IgG ELISA. The OD cutoff for

E16KO mice received HBSS, Dex, FVIII or FVIII+Dex for five consecutive days. Three days or three weeks after the last injection spleen, thymus and blood were collected. Lymphocyte populations were assessed via flow cytometry (MACSQuant Analyzer, Miltenyi) by staining for CD19, CD4, CD8, CD25 and FoxP3 (eBiosciences) as appropriate for the particular organ of origin. Data was analyzed with FlowJoX (Tree Star).

Figure 1. Short-term treatment protocols. A. E17KO/hMHC mice received FVIII (6IU IV) alone or in combination with Dex (75μg IP) for five consecutive days. At week five blood was collected and plasma anti-FVIII IgG titers were measured. Mice with evidence of anti-FVIII IgG were excluded from the remainder of the study. Mice with no evidence of anti-FVIII IgG were re-exposed to FVIII (6IU IV), alone or in combination with LPS (2μg IV), for three consecutive days. At week nine blood was collected. Plasma anti-FVIII IgG titers and FVIII inhibitory activity were measured. B. E16KO mice received FVIII (6IU IV) alone or in combination with Dex (75μg IP) for five consecutive days. At week five blood was collected and plasma anti-FVIII IgG titers as well as FVIII inhibitory activity were measured. FVIII: factor VIII; Dex: dexamethasone; LPS: lipopolysaccharide; Wk: week. : anti-FVIII IgG negative mice; : anti-FVIII IgG positive mice; : injection; : blood collection.

haematologica | 2018; 103(8)

1405

M.T. Georgescu et al.

Messenger ribonucleic acid (mRNA) expression analysis

Statistics

E16KO mice received HBSS, Dex, FVIII or FVIII+Dex for five consecutive days. Three days after the last injection, spleen and thymus were collected and stabilized in RNA Later (Invitrogen). mRNA was then isolated using a commercial kit (RNeasy Plus Mini Kit, Qiagen) and quantified using the NanoString Mouse Immunology Panel (NanoString). Data was analyzed with nSolver Software (NanoString). Genes that were deemed up/down-regulated due to FVIII+Dex treatment had a transcript count ratio â&#x2030;Ľ2:1 in the same direction when comparing both FVIII+Dex against FVIII and Dex against HBSS.

Anti-FVIII IgG and FVIII inhibitors incidence were compared using Fisherâ&#x20AC;&#x2122;s exact test. Anti-FVIII IgG and Bethesda titers were compared using a Mann-Whitney U test. E17KO/hMHC samples, but not E16KO samples, with titers below the detection limit were excluded from statistical analyses. Percentages of lymphocyte populations were compared using unpaired two-tailed t-tests. Statistical analyses were performed using GraphPad Prism 5.0a (GraphPad Software).

P=0.0048

P=0.0485

P=0.00063

Figure 2. Administration of Dex during initial FVIII exposure reduces the initial anti-FVIII immune response in both E17KO/hMHC and E16KO mice. A. Anti-FVIII IgG incidence and B. anti-FVIII IgG titers in E17KO/hMHC mice five weeks after initial treatment with FVIII or FVIII+Dex. A statistical comparison of the positive anti-FVIII IgG titers across treatment groups could not be performed because fewer than three mice had evidence of antibodies in the FVIII+Dex group. G. Anti-FVIII IgG incidence, D. anti-FVIII IgG titers, E. FVIII inhibitor incidence and F. FVIII inhibitory activity in E16KO mice five weeks after initial treatment with FVIII or FVIII+Dex. ND: not detectable; FVIII: factor VIII; Dex: dexamethasone; IgG: immunoglobulin G.

1406

haematologica | 2018; 103(8)

Dexamethasone promotes tolerance to factor VIII

Results Administration of Dex during initial FVIII exposure reduces the initial anti-FVIII immune response in both E17KO/hMHC and E16KO mice Our first aim was to determine the ability of Dex to prevent the anti-FVIII immune response when administered during initial antigen exposure. FVIII was administered alone or in combination with Dex for five consecutive days (week zero, Figure 1A,B). At week five, 6% of E17KO/hMHC FVIII+Dex mice compared to 33% of

E17KO/hMHC FVIII mice had evidence of plasma anti-FVIII IgG (P=0.0050, Figure 2A). Furthermore, FVIII+Dex mice that developed anti-FVIII IgG had lower antibody titers than FVIII mice (Figure 2B). A similar effect was observed in E16KO mice, with 77% of E16KO FVIII+Dex mice compared to 100% of E16KO FVIII mice showing evidence of plasma anti-FVIII IgG (P=0.0485, Figure 2C), and FVIII+Dex mice having significantly lower anti-FVIII IgG titers than FVIII mice (P=0.0063, Figure 2D). Although not statistically significant, a similar trend was observed when looking at inhibitor incidence and activity in E16KO mice (Figure 2E,F).

P=0.0050

P=0.0132

P=0.0699

P=0.0150

haematologica | 2018; 103(8)

P=0.1536

Figure 3. Administration of Dex during initial FVIII exposure induces tolerance to FVIII in E17KO/hMHC mice, even when co-administered with LPS. A. Anti-FVIII IgG incidence, B. antiFVIII IgG titers, C. FVIII inhibitor incidence and D. FVIII inhibitory activity following re-exposure to FVIII in E17KO/hMHC mice initially exposed to FVIII or FVIII+Dex and with no evidence of anti-FVIII IgG at week five. E. Anti-FVIII IgG incidence, F. AntiFVIII IgG titers, G. FVIII inhibitor incidence and H. FVIII inhibitory activity following re-exposure to FVIII+LPS in E17KO/hMHC mice initially exposed to FVIII or FVIII+Dex and with no evidence of anti-FVIII IgG at week five. Some statistical comparisons could not be carried out because fewer than three mice had evidence of antibodies and/or inhibitors. FVIII: factor VIII; Dex: dexamethasone; IgG: immunoglobulin G; LPS: lipopolysaccharide; ND: not detectable.

1407

M.T. Georgescu et al.

Administration of Dex during initial FVIII exposure induces tolerance to FVIII in E17KO/hMHC mice, even when co-administered with LPS Next, we wanted to determine whether E17KO/hMHC mice that did not develop anti-FVIII IgG after initial exposure would be immunologically tolerant upon re-exposure to FVIII. Anti-FVIII IgG negative mice from both the FVIII and FVIII+Dex groups were re-exposed to FVIII alone or in combination with LPS (week 6, Figure 1A). At week nine, 7% of FVIII+Dex/FVIII mice compared to 52% of FVIII/FVIII mice had evidence of plasma anti-FVIII IgG (P=0.0050, Figure 3A). The single FVIII+Dex/FVIII mouse with evidence of anti-FVIII IgG had a lower titer than most of the FVIII/FVIII mice (Figure 3B). Furthermore, 0% of FVIII+Dex/FVIII mice compared to 35% of FVIII/FVIII mice showed evidence of FVIII inhibitors by Bethesda assay (P=0.0132, Figure 3C,D). At week nine, 100% of FVIII/FVIII+LPS mice compared to 30% of FVIII+Dex/FVIII+LPS mice had evidence of anti-FVIII IgG (P=0.0699, Figure 3E). FVIII+Dex/FVIII+LPS mice also had a trend towards lower anti-FVIII IgG titers than FVIII/FVIII+LPS mice (P=0.1536, Figure 3F). Furthermore, 20% of FVIII+Dex/FVIII+LPS compared to 100% of FVIII/FVIII+LPS mice (P=0.0150, Figure 3G) had evidence of FVIII inhibitors and FVIII+Dex/FVIII+LPS mice had lower FVIII inhibitor levels than FVIII/FVIII+LPS (Figure 3H). These data suggest that Dex can promote persistent tolerance to FVIII in the E17KO/hMHC murine

model of HA, and that this tolerance is robust enough to withstand re-exposure to FVIII when co-delivered with LPS, a potent adjuvant.

Administration of Dex during initial FVIII exposure induces durable, antigen-specific tolerance to FVIII in E17KO/hMHC mice We next sought to investigate the durability of Dexinduced tolerance to FVIII, using the long-term treatment protocol described (Figure 4). At week four, 0% of FVIII+Dex E17KO/hMHC mice compared to 64% of FVIII E17KO/hMHC mice had evidence of anti-FVIII IgG (P=0.0001, Figure 5A,B). Mice with evidence of anti-FVIII IgG at week four were excluded from the remainder of the experiment. At week 14, all remaining mice had maintained their anti-FVIII IgG negative status. At week 18, two weeks after re-exposure to FVIII, 7% of FVIII+Dex/FVIII mice and 27% of FVIII+Dex/intFVIII+FVIII mice had evidence of anti-FVIII IgG compared to 50% of FVIII/FVIII mice (P=0.0970 FVIII/FVIII vs. FVIII+Dex/FVIII; P=0.5573 FVIII/FVIII vs. FVIII+Dex/intFVIII+FVIII; P=0.3295 FVIII+Dex/FVIII vs. FVIII+Dex/intFVIII+FVIII, Figure 5C). No apparent differences were seen between the titers of the few mice from each group positive for anti-FVIII IgG (Figure 5D). These data indicate that administration of Dex during initial FVIII exposure confers tolerance that persists for at least 18 weeks, and that ongoing intermittent FVIII exposure is

Figure 4. Long-term treatment protocol. E17KO/hMHC mice received FVIII (6IU IV) alone or in combination with Dex (75Î¼g IP) for five consecutive days. At week four, blood was collected and plasma anti-FVIII IgG titers were measured. Mice with evidence of anti-FVIII IgG were excluded from the remainder of the study. FVIII+Dex mice with no evidence of anti-FVIII IgG received FVIII (6IU IV) for three consecutive days at week 16, or single intermittent exposures to low-dose FVIII (2IU IV) at weeks four, eight and 12, followed by FVIII (6IU IV) for three consecutive days at week 16. FVIII treated mice with no evidence of anti-FVIII IgG at week four received FVIII for three consecutive days at week 16. All mice were sampled two weeks before and two weeks after FVIII re-exposure. Plasma anti-FVIII IgG titers were determined at these time points. At week 18 mice started receiving weekly human VWF injections for four consecutive weeks. At week 22, blood was collected and plasma FVIII inhibitory activity and the presence of plasma anti-human VWF IgG were determined. VWF: von Willebrand Factor; FVIII: factor VIII; Dex: dexamethasone; Wk: week. : anti-FVIII IgG negative mice; : anti-FVIII IgG positive mice; : injection; : blood collection.

1408

haematologica | 2018; 103(8)

Dexamethasone promotes tolerance to factor VIII

not required to maintain this tolerance. To determine whether the observed effect is antigen specific, we injected all mice with a structurally unrelated antigen (human VWF) (week 18-21). At week 22, of the anti-FVIII IgG negative mice, 100% of FVIII+Dex/FVIII, 89% of FVIII+Dex/intFVIII+FVIII and 100% of FVIII/FVIII mice had evidence of anti-human VWF IgG (Online Supplementary Figure S1). We conclude that Dex treatment during initial FVIII exposure does not result in general immunosuppression but rather promotes antigen-specific tolerance to FVIII, and does not impair immune responses to other antigens. At week 22, we also measured FVIII inhibitory activity. 8% of FVIII+Dex/FVIII, 8% of FVIII+Dex/intFVIII+FVIII and 50% of FVIII/FVIII mice had evidence of FVIII inhibitors (P=0.1206 FVIII/FVIII vs. FVIII+Dex/FVIII; P=0.1357 FVIII/FVIII vs. FVIII+Dex/intFVIII+FVIII; P=1 FVIII+Dex/FVIII vs. FVIII+Dex/intFVIII+FVIII, Figure 5E). No apparent differences were seen between the titers of the few mice from each group positive for inhibitors (Figure 5F).

Administration of Dex during initial FVIII exposure causes early changes in lymphocyte populations of E16KO mice To elucidate possible cellular mechanisms of our treatment protocol, we determined the percentage of key lymphocyte populations in the thymus, spleen and blood via flow cytometry. Three days after treatment, FVIII+Dex mice had a decreased percentage of both splenic (47.22% vs. 53.62%, P=0.0395) and blood (18.27% vs. 29.11%, P=0.0050) B cells (CD19+ lymphocytes) compared to FVIII

mice (Figure 6A). At this time point, no significant changes in the percentages of T cells (CD4+CD8â&#x20AC;&#x201C; lymphocytes) were observed across the three tissues (Figure 6B). However, FVIII+Dex mice showed a significant increase in the percentage of thymic Tregs (CD25 and FoxP3 expressing CD4+CD8â&#x20AC;&#x201C; lymphocytes, 12.06% vs. 4.73%, P<0.0010, Figure 6C). Similar trends were observed when comparing Dex and HBSS mice (Figure 6A-C). This suggests that Dex promotes tolerance to FVIII partly by decreasing the percentage of splenic and circulating B cells as well as skewing the distribution of lymphocytes in the thymus towards a regulatory phenotype early after treatment. Three weeks after treatment, FVIII+Dex mice had no significant differences in thymic and splenic lymphocyte populations when compared to FVIII mice (Figure 6D-F). We did however observe an increase in the percentage of blood B cells (36.33% vs. 21.33%, P=0.028) in FVIII+Dex mice compared to FVIII mice (Figure 6D). No significant differences in thymic, splenic or blood lymphocyte populations were detected when comparing Dex and HBSS mice (Figure 6D-F). This suggests that the effects of Dex likely occur early on and have no major lasting impact on lymphocyte populations despite the long-term FVIII tolerance.

Administration of Dex during initial FVIII exposures alters the thymic but not splenic transcript profile of E16KO mice We also assessed the effects of our treatment protocol on thymic and splenic mRNA transcription profiles of E16KO mice. In the thymus, a total of 54 genes had

P=0.1357

P=0.5573

P=0.0001 P=0.0970

P=0.1206 P=0.3295 P=1

Figure 5. Administration of Dex during initial FVIII exposure induces durable tolerance to FVIII in E17KO/hMHC mice. A. Anti-FVIII IgG incidence and B. anti-FVIII IgG titers following initial exposure with FVIII or FVIII+Dex. C. Anti-FVIII IgG incidence, D. anti-FVIII IgG titers, E. FVIII inhibitor incidence and F. FVIII inhibitory activity following FVIII re-exposure in E17KO/hMHC mice initially treated with FVIII or FVIII+Dex and with no evidence of anti-FVIII IgG at week four. Some statistical comparisons could not be carried out because fewer than three mice had evidence of antibodies and/or inhibitors. ND: not detectable; FVIII: factor VIII; Dex: dexamethasone; IgG: immunoglobulin G.

haematologica | 2018; 103(8)

1409

M.T. Georgescu et al.

altered expression due to FVIII+Dex treatment (Figure 7A, Online Supplementary Table S1 and Table S2). There were no differences between the splenic mRNA transcription profiles of FVIII+Dex and FVIII mice (Figure 7B). There were also no differences between the thymic and splenic mRNA transcription profiles of FVIII and HBSS mice (Online Supplementary Figure S2).

Discussion We sought to determine whether Dex, when administered during initial FVIII exposure, could promote immunologic tolerance to FVIII in HA mice. Our experiments indicate that both E17KO/hMHC and E16KO FVIII+Dex mice were less likely to develop anti-FVIII IgG than FVIII mice after initial exposure to FVIII. Although E17KO/hMHC mice can have inherent tolerance to FVIII, the ability of Dex to also reduce inhibitor development in E16KO mice is encouraging. While any immediate effect might have been due to transient immunosuppression, the reduced incidence of anti-FVIII IgG in E17KO/hMHC FVIII+Dex/FVIII mice after re-exposure to FVIII at six, and especially 16, weeks suggests that long-lasting tolerance to FVIII can be promoted. Furthermore, this tolerance is specific to FVIII since these mice mount a robust response to a structurally unrelated antigen (human VWF) despite remote exposure to Dex. Especially noteworthy is the reduced anti-FVIII immune response after Dex exposure in mice whose re-exposure to FVIII was accompanied by LPS. LPS administration with FVIII has been reported to yield anti-FVIII IgG in 100% of E17KO/hMHC mice,14 an effect also observed in our experiments. Compared to FVIII/FVIII+LPS mice, FVIII+Dex/FVIII+LPS mice demonstrated a markedly reduced anti-FVIII immune response, although this effect did not reach statistical significance due to small numbers. We investigated potential cellular mechanisms of Dexmediated tolerance induction by examining its effect on lymphocyte populations of E16KO mice. Three days after treatment we observed a decrease in the percentage of

splenic and circulating B cells. A subset of splenic B cells has been shown to play a role in the initiation of the antiFVIII immune response.19 B cells also maintain this response as their inhibition has been identified as a potential mechanism of ITI in mice.20 Furthermore, in inhibitor patients who fail conventional ITI, the addition of rituximab to target B cells has been shown to increase ITI efficacy.21 Three days post-treatment we also observed an increase in the percentage of thymic Tregs. This T-cell subset has been repeatedly implicated in tolerance to FVIII in HA mouse models.18,22,23 Tregs simultaneously interact with antigen-presenting cells and effector T cells, resulting in effector T-cell suppression. The changes in lymphocyte populations following Dex treatment were no longer present three weeks post-treatment. Our results are in line with previous studies showing that glucocorticoids can induce apoptosis of B and T cells13 and that Tregs, especially those in the thymus, are preferentially spared from Dex-induced cell death.24 There is some evidence suggesting that repeated antigen exposure is required for the maintenance of Treg populations.25 However, in our experiment, FVIII+Dex/intFVIII mice did not maintain tolerance to FVIII better than FVIII+Dex/FVIII mice. FVIII+Dex mice also had altered thymic gene expression, giving further insight into the mechanism of our treatment protocol. We observed down-regulation of genes involved in T-cell receptor formation and rearrangement (Cd4, Rag1, Rag2), genes coding for cytokines that drive effector T-cell proliferation and maturation (Il9, Il12b, Il13,26 Il16,27 Il2728) and genes responsible for T-cell activation (Cd40lg,29 Lck30). In contrast, we saw up-regulation of genes encoding for scavenger receptors involved in clearance of apoptotic cells (Cd36,31 MARCO32) and genes responsible for thymic involution (Pparg33). We also observed an up-regulation of genes encoding extracellular matrix components (Fn134) and adhesion molecules (Ita2b, Cdh5) that may play a role in regulating thymocyte development and migration.35 The splenic gene expression profile in FVIII and HBSS control mice were almost identical, and thus it appears that intense FVIII exposure alone does

Figure 6. Administration of Dex during initial FVIII exposure causes early changes in lymphocytes populations of E16KO mice. The percentage of A. B cells, B. T cells and C. Tregs in the thymus, spleen and blood three days after treatment with HBSS, Dex, FVIII or FVIII+Dex. The percentage of D. B cells, E. T cells and F. Tregs in the thymus, spleen and blood three weeks after treatment with HBSS, Dex, FVIII or FVIII+Dex. n=3-7 for each condition. *P<0.05, **P<0.01, ***P<0.001. : HBSS; : Dex; : FVIII; : FVIII+Dex.

1410

haematologica | 2018; 103(8)

Dexamethasone promotes tolerance to factor VIII

not result in altered expression of immune-relevant genes. A number of prior studies have examined the ability of immunomodulatory agents to prevent FVIII inhibitors in murine models of HA: anti-CD3 monoclonal antibody,18 anti-CD4 monoclonal antibody with adjuvant,36 rapamycin,22 CD40/CD40-ligand interaction blockade,37 and IL2/ anti-IL2 monoclonal antibody complexes.38 However, some of these strategies only induced transient tolerance to FVIII37 or tolerance that was not tested for long-term durability.18 Furthermore, the effect on immune responses to other antigens was not assessed in some cases.18,37 Importantly, none of these other immunomodulatory agents is commonly used in clinical practice. Dex and other glucocorticoids are widely available, routinely used by hematologists and are known to have good oral bioavailability.39 Glucocorticoids have been used in clinical practice to reduce humoral immune responses to protein therapeutics (e.g., infliximab),40 and have even been successful in hemophilia patients with inhibitors.41 Moreover, because the greatest risk of FVIII inhibitors occurs early (~25 first exposure days), clinical application of this approach might only require Dex coverage of a few early FVIII exposures, until the inhibitor risk is reduced. For these reasons, translation of this approach to the bedside seems feasible. A clinical trial examining the use of Dex during early FVIII exposures expected to span several days would be the most immediate application. This study does have some limitations. As previously mentioned, E17KO/hMHC mice have a mixed genetic background. While this variability between animals may be responsible for some of the observed variability in antiFVIII immune responses, it may also indicate some biological robustness of the tolerance effect. Due to the genetic variability of E17KO/hMHC, experiments identifying the mechanistic basis of the effect observed were carried out using the inbred E16KO mouse model. The ability of our

treatment protocol to diminish the anti-FVIII immune response in this model with high propensity for inhibitor development further confirms the robustness of the effect. Although the E17KO/hMHC mouse model of HA recapitulates the epidemiology of anti-FVIII immune responses among humans with severe HA, this does not imply that the immunological mechanisms of these responses are identical. There are significant differences between these species, such as the absence of the IgG4 isotype in mice, which is the dominant IgG subclass associated with inhibitors in HA patients.42 In addition, this mouse model has only one MHC allele and will therefore have nonphysiologic antigen presentation. The age of the model might also be a limitation, as our mice would be considered “young adults”. In contrast, HA patients who are at the greatest risk for inhibitor development are toddlers. The effects of these age differences on the ability of Dex to prevent inhibitors in humans cannot be predicted. Our dose of Dex (~3mg/kg/day, demonstrated to have anti-inflammatory effects in rodents)43 is higher than that typically used in humans. Small animals require higher per-weight doses of medications than humans to achieve equivalent dosing relative to body surface area.44 Dex is given to children in doses of 0.3 – 0.6mg/kg/day for the treatment of croup45 and asthma.46 In these applications, Dex exerts the anti-inflammatory and anti-lymphocytic effects that may be important for promoting immunologic tolerance to FVIII. Therefore, similar dosing would be reasonable to use in a clinical study addressing the mitigation of FVIII inhibitor development. Although Dex can be immunosuppressive, the risk of invasive infections is low after intermittent exposure in young children.47 We also used FVIII doses (~240units/kg/dose) higher than those used in most clinical applications. However, this is the FVIII dose required to provoke FVIII immune responses in E17KO/hMHC mice. Although lower doses

Figure 7. Administration of Dex during initial FVIII exposure alters the thymic but not splenic mRNA transcript profile of E16KO mice. A. Thymic and B. splenic mRNA transcript counts three days after treatment with FVIII+Dex versus mRNA transcript counts three days after treatment with FVIII alone. Each point corresponds with the average mRNA transcript count of three different tissue samples. Labeled points indicate genes that exhibited a ≥2-fold change as a result of treatment with FVIII+Dex. These genes had a gene count ratio ≥2:1 in the same direction when comparing both FVIII+Dex against FVIII and Dex against HBSS. n=3 for each condition. FVIII: factor VIII; Dex: dexamethasone; mRNA: messenger ribonucleic acid.

haematologica | 2018; 103(8)

1411

M.T. Georgescu et al.

could likely have been used in the E16KO mouse model, we wanted a consistent treatment protocol across the two strains. There is no conceptual reason that lower perweight FVIII doses would result in evasion of the tolerance-promoting mechanisms. The use of high FVIII doses over consecutive treatment days simulated “peak treatment moments” known to be associated with an increased risk of inhibitors.10 However, this intense initial exposure to FVIII was not given because of a hemorrhagic event, which from a clinical point of view is unrealistic. Patients often receive more intense initial exposures to FVIII to treat life-threatening bleeding, when inflammation may be present and could enhance the immune response to FVIII. We cannot infer that Dex would exert the same effect in these scenarios. However, animal models may be of limited utility in studying these more clinically relevant circumstances. For example, rodent studies have not consistently identified an association between hemarthrosis48,49 and anti-FVIII immune responses. Ultimately, clinical studies will be needed to answer the most important questions about the use of Dex in patients with severe HA.

References 1. Hay CRM, Palmer B, Chalmers E, et al. Incidence of factor VIII inhibitors throughout life in severe hemophilia A in the United Kingdom. Blood. 2011; 117(23):6367-6370. 2. DiMichele DM, Kroner BL. The North American Immune Tolerance Registry: practices, outcomes, outcome predictors. Thromb Haemost. 2002;87(1):52-57. 3. Rocino A, Cortesi PA, Scalone L, et al. Immune tolerance induction in patients with haemophilia a and inhibitors: Effectiveness and cost analysis in an European Cohort (The ITER Study). Haemophilia. 2016;22(1):96-102. 4. Gouw SC, van den Berg HM, Oldenburg J, et al. F8 gene mutation type and inhibitor development in patients with severe hemophilia A: systematic review and meta-analysis. Blood. 2012;119(12):2922-2934. 5. Oldenburg J, Pavlova A. Genetic risk factors for inhibitors to factors VIII and IX. Haemophilia. 2006;12(s6):15-22. 6. Goenka R, Barnett LG, Silver JS, et al. Cutting edge: dendritic cell-restricted antigen presentation initiates the follicular helper T cell program but cannot complete ultimate effector differentiation. J Immunol. 2011;187(3):1091-1095. 7. Bacchetta R, Gambineri E, Roncarolo M-G. Role of regulatory T cells and FOXP3 in human diseases. J Allergy Clin Immunol. 2007;120(2):227-235. 8. Appleman LJ, Boussiotis VA. T cell anergy and costimulation. Immunol Rev. 2003;192: 161-180. 9. Kurnik K, Bidlingmaier C, Engl W, Chehadeh H, Reipert B, Auerswald G. New early prophylaxis regimen that avoids immunological danger signals can reduce FVIII inhibitor development. Haemophilia. 2010;16(2):256-262. 10. Gouw SC, van den Berg HM, le Cessie S, van der Bom JG. Treatment characteristics and the risk of inhibitor development: a

1412

11. 12.

13.

14.

15.

16.

17.

18.

19.

Conclusions Our experiments show that Dex, when administered during an intense initial exposure to FVIII, diminishes the anti-FVIII immune response in E17KO/hMHC and E16KO HA mice. In addition, this treatment protocol promotes durable and antigen-specific immunologic tolerance to FVIII in E17KO/hMHC mice. This effect appears to be mediated by alterations in lymphocyte populations and thymic gene expression. Clinical studies are needed to determine if this approach can be translated into clinical practice to prevent the devastating occurrence of FVIII inhibitors for which we currently offer no mitigation strategy, even in high-risk patients. Ready access to Dex and other glucocorticoids, ease of administration, and extensive clinical experience with these drugs will make such clinical studies very feasible. Funding This project was supported in part by Operating and Foundation grants from The Canadian Institutes of Health Research (MOP10912, FDN-154285). DL is the recipient of a Canada Research Chair in Molecular Hemostasis.

multicenter cohort study among previously untreated patients with severe hemophilia A. J Thromb Haemost. 2007; 5(7):13831390. Auerswald G, Kurnik K, Aledort LM, et al. The EPIC study: a lesson to learn. Haemophilia. 2015;21(5):622–628. Franchimont D. Overview of the actions of glucocorticoids on the immune response: a good model to characterize new pathways of immunosuppression for new treatment strategies. Ann NY Acad Sci. 2004;1024: 124-137. Amsterdam A, Sasson R. The anti-inflammatory action of glucocorticoids is mediated by cell type specific regulation of apoptosis. Mol Cell Endocrinol. 2002; 189(1–2):1-9. Steinitz KN, van Helden PM, Binder B, et al. CD4+ T-cell epitopes associated with antibody responses after intravenously and subcutaneously applied human FVIII in humanized hemophilic E17 HLA-DRB1*1501 mice. Blood. 2012; 119(17):4073-4082. Qadura M, Waters B, Burnett E, et al. Immunoglobulin isotypes and functional anti-FVIII antibodies in response to FVIII treatment in Balb/c and C57BL/6 haemophilia A mice. Haemophilia. 2011;17 (2):288-295. Bi L, Lawler AM, Antonarakis SE, High KA, Gearhart JD, Kazazian HH. Targeted disruption of the mouse factor VIII gene produces a model of haemophilia A. Nat Genet. 1995;10(1):119-121. Reipert BM, Ahmad RU, Turecek PL, Schwarz HP. Characterization of antibodies induced by human factor VIII in a murine knockout model of hemophilia A. Thromb Haemost. 2000;84(5):826-832. Waters B, Qadura M, Burnett E, et al. AntiCD3 prevents factor VIII inhibitor development in hemophilia A mice by a regulatory CD4+CD25+-dependent mechanism and by shifting cytokine production to favor a Th1 response. Blood. 2009;113(1):193-203. Zerra PE, Cox C, Baldwin WH, et al.

20.

21.

22.

23.

24.

25. 26. 27. 28. 29. 30.

Marginal zone B cells are critical to factor VIII inhibitor formation in mice with hemophilia A. Blood. 2017; 130(23):2556-2568. Hausl C, Ahmad RU, Sasgary M, et al. Highdose factor VIII inhibits factor VIII – specific memory B cells in hemophilia A with factor VIII inhibitors. Blood. 2005; 106(10):34153422. Carcao M, St Louis J, Poon M-C, et al. Rituximab for congenital haemophiliacs with inhibitors: a Canadian experience. Haemophilia. 2006;12(1):7-18. Moghimi B, Sack BK, Nayak S, Markusic DM, Mah CS, Herzog RW. Induction of tolerance to factor VIII by transient co-administration with rapamycin. J Thromb Haemost. 2011;9(8):1524-1533. Kim YC, Zhang A-H, Su Y, et al. Engineered antigen-specific human regulatory T cells: immunosuppression of FVIII-specific T- and B-cell responses. Blood. 2014; 125(7):11071115. Chen X, Murakami T, Oppenheim JJ, Howard OMZ. Differential response of murine CD4+CD25+ and CD4+CD25- T cells to dexamethasone-induced cell death. Eur J Immunol. 2004;34(3):859-869. Ohkura N, Kitagawa Y, Sakaguchi S. Development and maintenance of regulatory T cells. Immunity. 2013;38(3):414-423. Yarilin AA, Belyakov IM. Cytokines in the thymus: production and biological effects. Curr Med Chem. 2004;11(4):447-464. Cruikshank WW, Kornfeld H, Center DM. Interleukin-16. J Leukoc Biol. 2000; 67(6):757-766. Iwasaki Y, Fujio K, Okamura T, Yamamoto K. Interleukin-27 in T Cell Immunity. Int J Mol Sci. 2015;16(2):2851-2863. Grewal IS, Flavell RA. The role of CD40 ligand in costimulation and T-cell activation. Immunol Rev. 1996;153:85-106. Palacios EH, Weiss A. Function of the Srcfamily kinases, Lck and Fyn, in T-cell development and activation. Oncogene. 2004;23(48):7990-8000.

haematologica | 2018; 103(8)

Dexamethasone promotes tolerance to factor VIII

31. Platt N, da Silva RP, Gordon S. Recognizing death: the phagocytosis of apoptotic cells. Trends Cell Biol. 1998;8(9):365-372. 32. Rogers NJ, Lees MJ, Gabriel L, et al. A defect in marco expression contributes to systemic lupus erythematosus development via failure to clear apoptotic cells. J Immunol. 2009; 182(4):1982-1990. 33. Youm Y-H, Yang H, Amin R, Smith SR, Leff T, Dixit VD. Thiazolidinedione treatment and constitutive-PPAR activation induces ectopic adipogenesis and promotes agerelated thymic involution. Aging Cell. 2010;9(4):478-489. 34. Lannes-Vieira J, Dardenne M, Savino W. Extracellular matrix components of the mouse thymus microenvironment: ontogenetic studies and modulation by glucocorticoid hormones. J Histochem Cytochem. 1991;39(11):1539-1546. 35. Gameiro J, Nagib P, Verinaud L. The thymus microenvironment in regulating thymocyte differentiation. Cell Adh Migr. 2010;4(3):382-390. 36. Oliveira VG, Agua-Doce A, Curotto de Lafaille MA, Lafaille JJ, Graca L. Adjuvant facilitates tolerance induction to factor VIII in hemophilic mice through a Foxp3-independent mechanism that relies on IL-10. Blood. 2013;121(19):3936-3945. 37. Reipert BM, Sasgary M, Ahmad RU, Auer W, Turecek PL, Schwarz HP. Blockade of CD40/CD40 ligand interactions prevents

haematologica | 2018; 103(8)

38.

39.

40.

41.

42.

43.

induction of factor VIII inhibitors in hemophilic mice but does not induce lasting immune tolerance. Thromb Haemost. 2001;86(6):1345-1352. Liu CL, Ye P, Lin J, Djukovic D, Miao CH. Long-term tolerance to factor VIII is achieved by administration of interleukin2/interleukin-2 monoclonal antibody complexes and low dosages of factor VIII. J Thromb Haemost. 2014;12(6):921-931. Duggan DE, Yeh KC, Matalia N, Ditzler CA, McMahon FG. Bioavailability of oral dexamethasone. Clin Pharmacol Ther. 1975;18(2):205-209. Farrell RJ, Alsahli M, Jeen Y-T, Falchuk KR, Peppercorn MA, Michetti P. Intravenous hydrocortisone premedication reduces antibodies to infliximab in Crohn’s disease: a randomized controlled trial. Gastroenterology. 2003;124(4):917-924. Edson JR, McArthur JR, Branda RF, McCullough JJ, Chou SN. Successful management of a subdural hematoma in a hemophiliac with an anti-factor VIII antibody. Blood. 1973;41(1):113-122. Montalvão SAL, Tucunduva AC, Siqueira LH, Sambo ALA, Medina SS, Ozelo MC. A longitudinal evaluation of anti-FVIII antibodies demonstrated IgG4 subclass is mainly correlated with high-titre inhibitor in haemophilia A patients. Haemophilia. 2015;21(5):585-692. Knowles RG, Salter M, Brooks SL MS. Anti-

44.

45.

46.

47.

48.

49.

inflammatory glucocorticoids inhibit the induction by endotoxin of nitric oxide synthase in the lung, liver and aorta of the rat. Biochem Biophys Res Commun. 1990;172 (3):1042-1048. Freireich EJ, Gehan EA, Rall DP, Schmidt LH SH. Quantitative comparison of toxicity of anticancer agents in mouse, rat, hamster, dog, monkey, and man. Cancer Chemother Rep. 1966;50(4):219-244. Garbutt JM, Conlon B, Sterkel R, et al. The comparative effectiveness of prednisolone and dexamethasone for children with croup: a community-based randomized trial. Clin Pediatr (Phila). 2013;52(11):1014-1021. Keeney GE, Gray MP, Morrison AK, et al. Dexamethasone for acute asthma exacerbations in children: a meta-analysis. Pediatrics. 2014;133(3):493-499. Aljebab F, Alanazi M, Choonara I, Conroy S. Tolerability of prednisolone and dexamethasone in Saudi Arabia. Arch Dis Child. 2016;101(9):e2. Peyron I, Dimitrov JD, Delignat S, et al. Haemarthrosis and arthropathy do not favour the development of factor VIII inhibitors in severe haemophilia A mice. Haemophilia. 2015;21(1):e94-98. Lövgren KM, Søndergaard H, Skov S, Wiinberg B. Joint bleeds increase the inhibitor response to human factor VIII in a rat model of severe haemophilia A. Haemophilia. 2016;22(5):772-779.

1413