Haematologica, Volume 104, Issue 1 by Haematologica

haematologica Journal of 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 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 2019 are as following: Print edition

Institutional Euro 700

Personal Euro 170

Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org). Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature. Direttore responsabile: Prof. 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 Journal of the Ferrata Storti Foundation

Table of Contents Volume 104, Issue 1: January 2019

Cover Figure Bone marrow smear showing hematogones in a 5-year old boy 3 months after allogeneic cord blood transplantation performed for Tlymphoblastic leukemia relapse. Courtesy of Prof. Rosangela Invernizzi.

Editorials 1

Novel insights into the role of aberrantly expressed MNX1 (HLXB9) in infant acute myeloid leukemia Juerg Schwaller

New insights into the causes of thrombotic events in patients with myeloproliferative neoplasms raise the possibility of novel therapeutic approaches Michal Bar-Natan and Ronald Hoffman

Gemtuzumab ozogamicin in acute myeloid leukemia: act 2, with perhaps more to come Johann Hitzler and Elihu Estey

Lenalidomide can be safely combined with chlorambucil and rituximab in older patients with chronic lymphocytic leukemia Candida Vitale and Alessandra Ferrajoli

Review Articles 13

Peering through zebrafish to understand inherited bone marrow failure syndromes Usua Oyarbide et al.

Advances in risk assessment and prophylaxis for central nervous system relapse in diffuse large B-cell lymphoma David Qualls and Jeremy S. Abramson

Articles Hematopoiesis

The homeobox transcription factor HB9 induces senescence and blocks differentiation in hematopoietic stem and progenitor cells Deborah Ingenhag et al.

Iron Metabolism & its Disorders

Macrophage ferroportin is essential for stromal cell proliferation in wound healing Stefania Recalcati et al.

Myelodysplastic Syndromes

Myeloproliferative Neoplasms

Vascular endothelial cell expression of JAK2V617F is sufficient to promote a pro-thrombotic state due to increased P-selectin expression Alexandre Guy et al.

Chronic Myeloid Leukemia

Transcriptional activation of the miR-17-92 cluster is involved in the growth-promoting effects of MYB in human Ph-positive leukemia cells Manuela Spagnuolo et al.

Incidence, outcomes, and risk factors of pleural effusion in patients receiving dasatinib therapy for Philadelphia chromosome-positive leukemia Timothy P. Hughes et al.

Haematologica 2019; vol. 104 no. 1 - January 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation Acute Myeloid Leukemia

102

2-Bromopalmitate targets retinoic acid receptor alpha and overcomes all-trans retinoic acid resistance of acute promyelocytic leukemia Ying Lu et al.

113

Gemtuzumab ozogamicin for de novo acute myeloid leukemia: final efficacy and safety updates from the open-label, phase III ALFA-0701 trial Juliette Lambert et al.

120

Gemtuzumab ozogamicin in children with relapsed or refractory acute myeloid leukemia: a report by Berlin-Frankfurt-Münster study group Naghmeh Niktoreh et al.

Acute Lymphoblastic Leukemia

128

Clinical and molecular characteristics of MEF2D fusion-positive B-cell precursor acute lymphoblastic leukemia in childhood, including a novel translocation resulting in MEF2D-HNRNPH1 gene fusion Kentaro Ohki et al.

Non-Hodgkin Lymphoma

138

A phase 2 study of rituximab, bendamustine, bortezomib and dexamethasone for first-line treatment of older patients with mantle cell lymphoma Rémy Gressin et al.

Chronic Lymphocytic Leukemia

147

Plasma Cell Disorders

155

DOT1L inhibition blocks multiple myeloma cell proliferation by suppressing IRF4-MYC signaling Kazuya Ishiguro et al.

Platelet Biology & its Disorders

166

Clinical factors and biomarkers predict outcome in patients with immune-mediated thrombotic thrombocytopenic purpura Elizabeth M. Staley et al.

Coagulation & its Disorders

176

Computed tomography pulmonary angiography versus ventilation-perfusion lung scanning for diagnosing pulmonary embolism during pregnancy: a systematic review and meta-analysis Cécile Tromeur et al.

Stem Cell Transplantation

189

Machine learning reveals chronic graft-versus-host disease phenotypes and stratifies survival after stem cell transplant for hematologic malignancies Jocelyn S. Gandelman et al.

Cell Therapy & Immunotherapy

197

The allogeneic HLA-DP-restricted T-cell repertoire provoked by allogeneic dendritic cells contains T cells that show restricted recognition of hematopoietic cells including primary malignant cells Aicha Laghmouchi et al.

Blood Transfusion

207

Cold storage of platelets in additive solution: the impact of residual plasma in apheresis platelet concentrates Irene Marini et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/104/1.toc

Intracardiac or intrapulmonary shunts were present in at least 35% of adults with homozygous sickle cell disease followed in an outpatient clinic Bryan C. Hambley et al. http://www.haematologica.org/content/104/1/e1

Haematologica 2019; vol. 104 no. 1 - January 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Effects of erythropoiesis-stimulating agents on overall survival of International Prognostic Scoring System Low/Intermediate-1 risk, transfusion-independent myelodysplastic syndrome patients: a cohort study Emanuela Messa et al. http://www.haematologica.org/content/104/1/e4

Impact of FLT3-ITD length on prognosis of acute myeloid leukemia Song-Bai Liu et al. http://www.haematologica.org/content/104/1/e9

e13

Long-term follow up of pediatric Philadelphia positive acute lymphoblastic leukemia treated with the EsPhALL2004 study: high white blood cell count at diagnosis is the strongest prognostic factor Andrea Biondi et al. http://www.haematologica.org/content/104/1/e13

e17

Pre-clinical evaluation of second generation PIM inhibitors for the treatment of T-cell acute lymphoblastic leukemia and lymphoma Renate De Smedt et al. http://www.haematologica.org/content/104/1/e17

e21

Programmed cell death protein-1 (PD-1)-expression in the microenvironment of classical Hodgkin lymphoma at relapse during anti-PD1-treatment Stephanie Sasse et al. http://www.haematologica.org/content/104/1/e21

e25

MYC protein expression scoring and its impact on the prognosis of aggressive B-cell lymphoma patients Maria R. Ambrosio et al. http://www.haematologica.org/content/104/1/e25

e29

Incidence of upper-extremity deep vein thrombosis in western France: a community-based study AurĂŠlien Delluc et al. http://www.haematologica.org/content/104/1/e29

Case Reports Case Reports are available online only at www.haematologica.org/content/104/1.toc

e32

Lymphoproliferation, immunodeficiency and early-onset inflammatory bowel disease associated with a novel mutation in Caspase 8 Veronika Kanderova et al. http://www.haematologica.org/content/104/1/e32

e35

B-lymphoblastic lymphoma with TCF3-PBX1 fusion gene Mari Kubota-Tanaka et al. http://www.haematologica.org/content/104/1/e35

e38

Spatial clonal evolution leading to ibrutinib resistance and disease progression in chronic lymphocytic leukemia RichĂĄrd Kiss et al. http://www.haematologica.org/content/104/1/e38

Comments Comments are available online only at www.haematologica.org/content/104/1.toc

e42

Programmed cell death protein-1 (PD1) expression in the microenvironment of classical Hodgkin lymphoma is similar between favorable and adverse outcome and does not enrich over serial relapses with conventional chemotherapy Joseph G. Taylor et al. http://www.haematologica.org/content/104/1/e42

e45

Programmed cell death protein1 (PD1)-expression in the microenvironment of classical Hodgkin lymphoma at relapse after conventional chemotherapy and at relapse on anti-PD1 treatment Stephanie Sasse et al. http://www.haematologica.org/content/104/1/e45

Haematologica 2019; vol. 104 no. 1 - January 2019 http://www.haematologica.org/

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

Ancient Greek

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

Scientific Latin

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

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2017 JCR impact factor = 9.090

EDITORIALS Novel insights into the role of aberrantly expressed MNX1 (HLXB9) in infant acute myeloid leukemia Juerg Schwaller University Children’s Hospital beider Basel (UKBB), Department of Biomedicine, University of Basel Childhood Leukemia Group ZLF, Switzerland. E-mail: J.Schwaller@unibas.ch doi:10.3324/haematol.2018.205971

lmost two decades ago, the molecular characterization of a t(7;12)(q36;p13) chromosomal translocation in very young children with acute myeloid leukemia (AML) and poor outcome identified a fusion mRNA potentially encoding for a chimeric protein that contains the pointed (PNT) and ETS domains of the ETS variant 6 (ETV6) gene, also known as TEL1 (Translocating E26 transforming-specific leukemia 1) on 12p13, joined to the regulatory sequences and first exons of the HLXB9 homeobox gene.1 Previous work reported a series of infant AML patients with t(7;12)(q36;p13) with blasts carrying a potential ETV6 translocation (revealed by a split FISH signal).2,3 In fact, the entire HLBX9 gene seems to be transferred onto the der(12) without disruption of the gene itself. A fusion transcript may result from a yet to be identified long-range splicing mechanism and can be found in about 50% of cases. Expression of any reciprocal ETV6HLXB9 fusion transcript has not been reported. These findings suggested that a position effect triggers stable overexpression of HLXB9 in t(7;12)(q36;p13)+ AML.4,5 The Nordic Society for Pediatric Hematology and Oncology (NOPHO) recently reported additional 7 patients to the previously 35 published t(7;12)(q36;p13)+ AML patients. Overall leukemic blasts from 20-30% of infant AML patients (<2 years) carry this translocation associated with trisomy 19 in 86% of the cases.6 In addition to t(7;12)(q36;p13), another translocation, t(6;7)(q23;q35), was identified in an AML cell line (GDM-1) which also resulted in aberrantly high HLXB9 expression via juxtaposition with regions of the MYB gene.7 The homeobox HLXB9 (or HB9) protein is today referenced as MNX1 (motor neuron and pancreas homeobox 1, OMIM: 142994). Putative loss-of-function mutations of MNX1 were identified as the molecular correlates of hereditary sacral agenesis.8 Gene targeting experiments in mice revealed MNX1 to be a critical regulator for normal pancreas9,10 as well as motor neuron development.11 Loss-of-function MNX1 mutations in neonatal diabetes confirmed its role in pancreas development in humans.12 There is increasing evidence to suggest an important role of this critical developmental homeobox transcription factor not only in hematologic malignancies but also in solid cancers;13,14 however, the functional consequences and molecular mechanisms of aberrantly high MNX1 expression for malignant transformation remain poorly understood. In a study published in this edition of Haematologica, Ingenhag et al.15 addressed the oncogenic potential of increased MNX1 expression in non-hematopoietic and hematopoietic cells. They found that lentiviral MNX1 overexpression in human HT1080 fibrosarcoma or mouse NIH3T3 fibroblasts resulted in growth arrest with stalling at the G1/G2 phase of cell cycle, morphological signs of senescence, and increased senescence-associated β-galactosidase activity (SA-β-gal) associated with activation of the tumor suppressor haematologica | 2019; 104(1)

TP53 and its target the cyclin-dependent kinase inhibitor 1A (CDKN1A, aka p21WAF1/CIP1). As oncogene-induced senescence is a hallmark of early malignant transformation of solid tumors, this finding suggests that MNX1 overexpression may result in a pre-cancerous state.16 However, one has to keep in mind that both of the models used are immortalized solid cancer cell lines that may carry potent oncogenes such as mutated NRASQ61K present in HT-1080 (https://portals.broadinstitute.org/ccle/page?cell_line= HT1080_SOFT_TISSUE). Nevertheless, previous work has shown that overexpression of well-characterized AML-associated fusion oncogenes (e.g. PML-RARA, RUNX1-ETO, CBFB-MYH11) induces DNA damage, and activates a CDKN1A-dependent cell cycle arrest and DNA repair in mouse hematopoietic stem and progenitor cells (HSPC).17 It will, therefore, be interesting to verify whether overexpression of MNX1 may also lead to DNA damage activating the TP53-CDKN1A pathway in HSPC. To address the oncogenic potential in hematopoietic cells, Ingenhag et al.15 also lentivirally expressed MNX1 in lineage marker-depleted mouse bone marrow (BM)-derived HSPC. Reconstitution of lethally irradiated syngenic recipients resulted in a reduced overall peripheral blood cellularity with very low contribution to any mature (CD19+ B cells; CD3+ T cells; CD11b+ neutrophils) cell lineage. Although the total cell number was also reduced in the BM, MNX1 expressing cells mostly contributed to the megakaryocytic-erythrocyte progenitor (MEP) cell compartment, whereas no distinct MNX1 expressing population was detected within the granulocytemonocyte progenitor (GMP) compartment. Although some MNX1 expressing immature B cells (B220+CD19+CD93+) were found, no signal for MNX1 expression was found in mature B cells and immature or mature T cells. None of the transplanted mice developed any signs of a hematologic disease during a relatively short observation period of 6.5 months. These observations show for the first time that MNX1 overexpression impacts hematopoietic differentiation of HSPC in vivo but may not be sufficient to induce a hematologic disease. To study the potential oncogenic activity in human hematopoietic cells, Ingenhag et al.15 were able to lentivirally express MNX1 in CD34+ HSPC at comparable levels to what was observed in primary t(7;12)(q36;p13)+ AML cells. Gene expression profiling revealed that, in addition to MNX1, 116 significantly aberrantly regulated genes including de novo expression of HBZ (Hemoglobin Subunit Zeta), HBE1 (Hemoglobin Subunit Epsilon 1) SLC4A1 (Solute Carrier Family 4 Member 1), all tightly associated with erythroid differentiation. Notably, gene set enrichment analysis revealed upregulation of genes related to cell cycle progression, and downregulation of cytoskeleton organization and various intracellular signaling pathways. Previous work by the same group has already suggested that elevated MNX1 levels affect expression of genes 1

Editorials

Figure 1. Schematic representation of the findings by Ingenhag et al.15

implicated in cell-cell interaction and adhesion.18 Interestingly, MNX1 expression also reduced the clonogenic activity of human CD34+ HSPC. Likewise previous experiments also showed no aberrant in vitro clonogenic growth of MNX1 over-expressing mouse HSPC.18 Together, the experiments in mouse and human hematopoietic cells suggest increased MNX1 expression interferes with normal hematopoietic differentiation. Notably, most of the reported infants aberrantly expressing MNX1 were diagnosed with poorly differentiated FAB subtypes. Four out of 42 published patients developed acute megakaryoblastic leukemia (AMKL) suggesting MNX1 impacts on MEP cell maturation.6 Collectively, Ingenhag et al.15 show that aberrantly expressed MNX1 interferes with normal cycle progression, inducing premature senescence in solid cancer cell lines, and interferes with hematopoietic differentiation in vitro and in vivo (Figure 1). This work provides important clues for our understanding of t(7;12)(q36;p13)+ infant AML and sets the stage for additional experiments addressing several open key questions. First, it remains to be demonstrated whether aber2

rant MNX1 expression is sufficient to induce a leukemic phenotype. Although recent large deep sequencing studies did not report the presence of any common co-operating mutations (FLT3-ITD, NPM1, CEBP/A, WT1) in 4 infant AML carrying t(7;12)(q36;p13), the strong association with trisomy 19 in the blasts of the majority of these patients suggests some co-operative events.19 Notably, trisomy 19 has been described as a recurrent chromosomal abnormality in hematologic malignancies, including AMKL.20,21 Second, it will be important to identify the natural target cell of MNX1-mediated transformation. The exclusive association of t(7;12)(q36;p13) strongly suggests a fetal origin. It is somehow reminiscent of inv(16)(p13q24) leading to expression of a CBFA2T3-GLIS2 fusion in AMKL which mostly affects infants. Modeling the transforming activity of this fusion in mice revealed that only expression in fetal liver hematopoietic cells but not in adult BM-derived cells was able to induce a leukemic disease.22,23 Therefore, future studies will ideally follow different fetal-liver hematopoietic stem and/or progenitor cells conditionally over-expressing MNX1 in clonogenic assays and BM reconstitution assays. haematologica | 2019; 104(1)

Editorials

Third, what are the MNX1 target genes that mediate its leukemogenic potential? In a previous study, the same group identified binding to and regulation of the prostaglandin E receptor 2 (PTGER2) by MNX1 overexpressed in the human HL60 AML cell line.24 However, critical targets might significantly differ in a context of fetal liver HSPC. Finally, and most importantly, it needs to be shown whether high expression of MNX1 is critical to maintain a transformed phenotype. Knockdown or genome editing experiments in primary human AML cells (e.g. expanded in immune deficient mice) or conditional expression in transgenic mouse models may show the way. Further exploration of the MNX1 interacting proteome could provide some clues as how to develop strategies for targeted therapeutic intervention. Funding Our lab is supported by the Swiss Cancer League, the Swiss National Science Foundation, the Wilhelm Sander Foundation (Munich, Germany), the San Salvatore Foundation (Lugano, Switzerland), and the Novartis Research Foundation (Basel, Switzerland).

References 1. Beverloo HB, Panagopoulos I, Isaksson M, et al. Fusion of the homeobox gene HLXB9 and the ETV6 gene in infant acute myeloid leukemias with the t(7;12)(q36;p13). Cancer Res. 2001;61(14):5374-5377. 2. Tosi S, Harbott J, Teigler-Schlegel A, et al. t(7;12)(q36;p13), a new recurrent translocation involving ETV6 in infant leukemia. Genes Chromosomes Cancer. 2000;29(4):325-332. 3. Slater RM, von Drunen E, Kroes WG, et al. t(7;12)(q36;p13) and t(7;12)(q32;p13)--translocations involving ETV6 in children 18 months of age or younger with myeloid disorders. Leukemia. 2001;15(6):915920. 4. von Bergh AR, van Drunen E, van Wering ER, et al. High incidence of t(7;12)(q36;p13) in infant AML but not in infant ALL, with a dismal outcome and ectopic expression of HLXB9. Genes Chromosomes Cancer. 2006;45(8):731-739. 5. Ballabio E, Cantarella CD, Federico C, et al. Ectopic expression of the HLXB9 gene is associated with an altered nuclear position in t(7;12) leukaemias. Leukemia. 2009;23(6):1179-1182. 6. Espersen ADL, Noren-Nystrom U, Abrahamsson J, et al. Acute myeloid leukemia (AML) with t(7;12)(q36;p13) is associated with infancy and trisomy 19: Data from Nordic Society for Pediatric Hematology and Oncology (NOPHO-AML) and review of the literature. Genes Chromosomes Cancer. 2018;57(7):359-365. 7. Nagel S, Kaufmann M, Scherr M, Drexler HG, MacLeod RA. Activation

8. 9. 10. 11.

12.

13.

14. 15. 16. 17. 18. 19. 20. 21.

22. 23. 24.

of HLXB9 by juxtaposition with MYB via formation of t(6;7)(q23;q36) in an AML-M4 cell line (GDM-1). Genes Chromosomes Cancer. 2005;42(2):170-178. Ross AJ, Ruiz-Perez V, Wang Y, et al. A homeobox gene, HLXB9, is the major locus for dominantly inherited sacral agenesis. Nat Genet. 1998;20(4):358-361. Harrison KA, Thaler J, Pfaff SL, Gu H, Kehrl JH. Pancreas dorsal lobe agenesis and abnormal islets of Langerhans in Hlxb9-deficient mice. Nat Genet. 1999;23(1):71-75. Li H, Arber S, Jessell TM, Edlund H. Selective agenesis of the dorsal pancreas in mice lacking homeobox gene Hlxb9. Nat Genet. 1999;23(1):6770. Thaler J, Harrison K, Sharma K, Lettieri K, Kehrl J, Pfaff SL. Active suppression of interneuron programs within developing motor neurons revealed by analysis of homeodomain factor HB9. Neuron. 1999;23(4):675-687. Flanagan SE, De Franco E, Lango Allen H, et al. Analysis of transcription factors key for mouse pancreatic development establishes NKX2-2 and MNX1 mutations as causes of neonatal diabetes in man. Cell Metab. 2014;19(1):146-154. Desai SS, Modali SD, Parekh VI, Kebebew E, Agarwal SK. GSK-3beta protein phosphorylates and stabilizes HLXB9 protein in insulinoma cells to form a targetable mechanism of controlling insulinoma cell proliferation. J Biol Chem. 2014;289(9):5386-5398. Zhang L, Wang J, Wang Y, et al. MNX1 Is Oncogenically Upregulated in African-American Prostate Cancer. Cancer Res. 2016;76(21):62906298. Ingenhag D, Reister S, Auer F, et al. The homeobox transcription factor HB9 induces senescence and blocks differentiation in hematopoietic stem and progenitor cells. Haematologica 2019;104(1):35-46. Prieur A, Peeper DS. Cellular senescence in vivo: a barrier to tumorigenesis. Curr Opin Cell Biol. 2008;20(2):150-155. Viale A, De Franco F, Orleth A, et al. Cell-cycle restriction limits DNA damage and maintains self-renewal of leukaemia stem cells. Nature. 2009;457(7225):51-56. Wildenhain S, Ruckert C, Rottgers S, et al. Expression of cell-cell interacting genes distinguishes HLXB9/TEL from MLL-positive childhood acute myeloid leukemia. Leukemia. 2010;24(9):1657-1660. Bolouri H, Farrar JE, Triche T Jr, et al. The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions. Nat Med. 2018;24(1):103-112. Johansson B, Billstrom R, Mauritzson N, Mitelman F. Trisomy 19 as the sole chromosomal anomaly in hematologic neoplasms. Cancer Genet Cytogenet. 1994;74(1):62-65. Dastugue N, Lafage-Pochitaloff M, Pages MP, et al. Cytogenetic profile of childhood and adult megakaryoblastic leukemia (M7): a study of the Groupe Francais de Cytogenetique Hematologique (GFCH). Blood. 2002;100(2):618-626. Dang J, Nance S, Ma J, et al. AMKL chimeric transcription factors are potent inducers of leukemia. Leukemia. 2017;31(10):2228-2234. Lebert-Ghali CE, Neault M, Fournier M, et al. Generation of a novel mouse model recapitulating features of human acute megakaryoblastic leukemia. Exp Hematol. 2018;64:S79. Wildenhain S, Ingenhag D, Ruckert C, et al. Homeobox protein HB9 binds to the prostaglandin E receptor 2 promoter and inhibits intracellular cAMP mobilization in leukemic cells. J Biol Chem. 2012;287 (48):40703-40712.

New insights into the causes of thrombotic events in patients with myeloproliferative neoplasms raise the possibility of novel therapeutic approaches Michal Bar-Natan and Ronald Hoffman Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA E-mail: ronald.hoffman@mssm.edu doi:10.3324/haematol.2018.205989

he Philadelphia chromosome-negative myeloproliferative neoplasms (MPN) include polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (MF). This group of clonal hematological malignancies is associated with a protracted clinical course haematologica | 2019; 104(1)

frequently punctuated by thrombotic events. Such thrombotic events have been previously attributed to excessive numbers of functionally abnormal red cells, platelets and leukocytes. MPN patients are not only at a high risk of developing arterial and venous thromboses, but also throm3

Editorials

boses at unusual sites including the hepatic, portal and splenic veins, the cerebral sinuses and the mesenteric arteries. The mechanism(s) underlying this pro-thrombotic tendency in MPN are incompletely understood and have been the subject of speculation for almost seven decades. Over the last 24 months, several reports have appeared which have shed new light on the mechanisms underlying this thrombotic tendency. They implicate a pro-inflammatory MPN milieu as well as interactions between excessive numbers of qualitatively abnormal blood cells and the vessel endothelium in the generation of these thrombotic events (Figure 1). In this issue of Haematologica, Guy et al.1 demonstrate a role for integrins in the development of thromboses using endothelial cells (EC) engineered to overexpress the MPN driver mutation JAK2V617F in vivo and in vitro. Our initial understanding of the MPN pro-thrombotic state was largely influenced by the seminal observations of Pearson and Wetherley-Mein.2 They demonstrated that the incidence of thrombotic events in PV patients was directly related to the degree of hematocrit elevation. Red cells are the primary determinant of blood viscosity, which increases non-linearly with increasing hematocrit levels at both arterial and venous shear rates. Numerous studies have also suggested that increased red cell numbers increase the margination of platelets along the vessel walls. Recently, Walton et al.,3 using a transfusion-based polycythemia model in healthy mice, showed that polycythemic mice had accelerated rates of arterial thrombus formation and shortened clotting times due to a platelet-dependent increase in thrombus formation. Their data collectively reflect the manner in which red cells independently promote the development of arterial but not venous thrombosis. Klatt et al.,4 however, provided further data indicating that red cells trigger additional events beyond biophysical interactions that accelerate venous thrombosis. They showed that platelet/red cell interactions lead to increased platelet FAS ligand (FASL) exposure which then activates the death receptor (FASR) present on red cells. This ligand/receptor interaction ultimately results in further externalization of red cell phosphatidylserine which promotes the assembly of coagulation factor complexes leading to thrombin generation and the formation of occlusive thrombi. Klatt et al. reported that these events could occur on a collagen surface with low shear rates which resembles a venous system. The consequences of excessive numbers of red cells in MPN patients was validated by Marchioli et al.5 who showed that sustained normalization of hematocrit levels (<45%) in high-risk PV patients was associated with reduced numbers of thrombotic events. Furthermore, Alvarez-Larran et al.6 demonstrated that PV patients with higher phlebotomy requirements were at the highest risk of developing thrombotic events. However, several lines of evidence strongly suggest that additional mechanisms beyond hematocrit elevation are required to explain a number of observed clinical manifestations including: (i) the occurrence of thrombotic events in over a third of patients prior to the diagnosis of PV; (ii) the occurrence of splanchnic vein thromboses, frequently in patients with a JAK2V617F mutation with normal blood counts; (iii) the increased incidence of thrombotic events in normal individuals found to have clonal hematopoiesis of indeterminate potential with a JAK2 mutation; (iv) the persistent rate of thrombosis fol4

lowing normalization of the hematocrit in PV patients; and (v) the increased rate of thrombosis in ET and MF patients without polycythemia. Intuitively, physicians have linked MPN-associated thrombocytosis to the high incidence of thrombotic events, however, the thrombotic risk in ET patients does not seem to be related to the degree of thrombocytosis7 and those patients with extreme degrees of thrombocytosis (>1.5 million) are ironically at a higher risk of bleeding rather than clotting due to the development of a secondary form of von Willebrand disease. The conclusion that additional factors beyond excessive numbers of blood cells contribute to the MPN pro-thrombotic tendency was bolstered by the more recent observation that patients with a JAK2V617F mutation, particularly those individuals with a high variant allele burden, were at a greater risk of developing thrombotic events than those with calreticulin mutations.8 Several groups have provided evidence that mutated JAK2 might affect not only hematopoietic cells but also EC, which raises the possibility that MPN might actually arise in some patients in a primitive cell that resembles the hemogenic endothelium.9 In this issue of Haematologica Guy et al.1 report the construction of several murine models which can be used to evaluate the contribution of EC to the MPN pro-thrombotic state. They demonstrate that mice that were genetically engineered to express JAK2V617F in EC but not hematopoietic cells had a predilection to develop thrombotic events in spite of having normal blood counts and normal rates of thrombin generation. Importantly, this thrombotic tendency was accentuated by the creation of a pro-inflammatory milieu through the administration of low doses of tumor necrosis factor alpha. Using both in vitro and in vivo approaches they next showed that JAK2V617F+ human and murine EC were capable of promoting both leukocyte rolling and adhesion. Although the most common integrins associated with leukocyte adhesion to EC were not upregulated in these mutated EC, Guy et al. did demonstrate increased surface expression of P-selectin (CD62P) and von Willebrand factor (VWF), both of which are contained within Weibel-Palade bodies in EC. Importantly the pro-adhesive properties of the JAK2V617F+ EC were reversed by treatment with either a P-selectin blocking antibody or hydroxyurea, a drug that remains the standard of care for treating high-risk PV and ET patients. The authors concluded that hydroxyurea did not block the effects of P-selectin but rather decreased the release of P-selectin and VWF from Weibel-Palade bodies. The upregulation of P-selectin by mutated EC was attributed to increased STAT3 phosphorylation which is a downstream event of JAK/STAT signaling. Importantly, earlier this year, Guadall et al.10 generated data that supported the findings of Guy et al. using a totally different experimental system. They developed wild-type and JAK2V617F+ EC from immortalized human pluripotent stem cells and showed that JAK2V617F+ EC promoted the adherence of leukocytes and were characterized by increase phosphorylation of STAT3 and overexpression of both VWF and P-selectin. The availability of large numbers of JAK2V617F+ human EC from immortalized human pluripotent stem cells allowed these investigators to document gene expression analyses, demonstrating increased expression of genes associated with inflammation and cell adhesion in JAK2V617F+ human EC. P-selectin has been previously implicated by the haematologica | 2019; 104(1)

Editorials

Figure 1. The mechanism of thrombus formation in myeloproliferative neoplasm. The Jak2V617F mutation causes an increase in endothelial cell (EC) Weibel-Palade body (WPB) degranulation of P-selectin and von Willebrand factor (VWF); translocation of Rap1 towards the cell membrane with activation of the integrins LFA1 and VLA4; and increased neutrophil extracellular trap (NET) formation. In addition, a red blood cell-platelet interaction through FasL/FasR causes externalization of phosphatidylserine (PS). All of these events play a role in thrombus formation. Rap1: Ras-related protein 1; LFA1: lymphocyte function-associated antigen 1; STAT: signal transducer and activator of transcription; PAD4: peptidyl arginine deimidase 4; ICAM-1: intracellular adhesion molecule 1; VCAM-1: vascular cell adhesion molecule 1; VLA4: very late antigen-4: FasL: Fas ligand.

Migliaccio Laboratory to play a critical role in the development not only of thrombosis but also progression to myelofibrosis in a GATA1low mouse model.11 In this animal model, abnormal localization of P-selectin in both megakaryocytes and platelets led to increased platelet/leukocyte interactions, an increased incidence of thrombotic events, and increased release of neutrophil proteases and transforming growth factor-beta which plays a critical role in the development of bone marrow fibrosis, osteosclerosis and disease progression in MPN. Importantly, the progression to MF as well as the increased frequency of thromboses in the GATA1low mice was not observed in mice in which P-selectin was deleted by genetic approaches.12,13 The role of integrins in MPN thrombosis has been further supported by the provocative work of Edelmann et al.14 who showed that JAK2V617F+ granulocytes haematologica | 2019; 104(1)

and monocytes were characterized by increased activation of VLA-4 and/or LFA1. These integrins are cell adhesion molecules which play an essential role in the attachment of leukocytes to EC by interacting with intracellular matrix proteins. The translocation of these two integrins to the granulocyte surface was due to the effects of mutated JAK2 on the insideâ&#x20AC;&#x201C;outside signaling molecule, Rap1. Most importantly these investigators demonstrated that the administration of integrin-blocking antibodies to JAK2V617F+ mice diminished the rate of thrombosis. Additional evidence for the role of neutrophils in thrombosis in MPN was recently offered by Wolach et al.15 with their demonstration that neutrophils from patients with JAK2V617F MPN are primed to form neutrophil extracellular trap, implicated in the pathogenesis and promotion of thrombosis. Moreover, mice with conditional knock-in of 5

Editorials

JAK2V617F have an increased propensity to neutrophil extracellular trap formation and thrombosis. Inhibition of JAKSTAT signaling by ruxolitinib abrogated neutrophil extracellular trap formation and reduced thrombosis in this murine model. The current report of Guy et al. as well as the other reports referred to in this Commentary each delineate the increasingly plausible role of various cell adhesion molecules (selectins and integrins) in MPN-associated thrombosis and in some cases evolution to MF. The question remains, how relevant are these observations in disease models to the pathophysiology of MPN in patients? This is especially relevant to the work dealing with JAK2V617F+ EC. Sozer et al.16,17 previously documented that angiogenic monocytes as well as true EC were JAK2V617F+ in PV patients with splanchnic vein thromboses. Using laser capture microdissection they demonstrated that the EC within the hepatic veins of some PV patients with hepatic vein thrombosis were JAK2V617F+. Furthermore, Rosti et al.18 reported that splenic vein EC were JAK2V617F+ in 67% of patients with MF. To better understand the significance of these intriguing experimental findings, the frequency of JAK2V617F+ MPN patients with mutated EC, the extent of the distribution of these JAK2V617F+ EC within the vasculature of various tissues, and the relationship of these findings to the incidence of thrombosis in MPN require evaluation in larger numbers of patients. It will also be interesting to determine whether other driver mutations in MPN, such as calreticulin, share the same properties which might explain, in part, the different propensity to develop thrombosis relative to that in JAK2V617F-mutated patients. The increased propensity to develop thrombosis in MPN patients is likely multifactorial in origin. An elevated hematocrit and a pro-inflammatory state, as well as a series of cellular interactions mediated by cell adhesion molecules that are expressed by red cells, platelets, leukocytes, monocytes and EC, may all play a role (Figure 1), and combinations of these events at any one time may further increase the risk of developing a thrombotic event. Most importantly, this recent round of studies provides a rationale for the evaluation of blocking antibodies to P-selectin, VLA-4 and LFA-1, which in part are already in clinical use for other conditions,19-21 to further reduce the incidence of not only thrombotic events but also disease progression beyond that achieved with the presently available therapeutic options. The outcomes of such proposed clinical trials, which are at best presently in the planning stages, will be closely watched. Such studies will allow us to assess the importance of each of these membrane proteins in the development of life-threatening clinical events in MPN patients and are likely to increase the therapeutic options for such patients.

References 1. Guy A, Gourdou-Latyszenok V, Le Lay N, et al. Vascular endothelial cell expression of JAK2V617F is sufficient to promote a pro-thrombotic state due to increased P-selectin expression. Haematologica. 2019;104(1):70-81. 2. Pearson TC, Wetherley-Mein G. Vascular occlusive episodes and venous haematocrit in primary proliferative polycythaemia. Lancet. 1978;2(8102):1219-1222. 3. Walton BL, Lehmann M, Skorczewski T, et al. Elevated hematocrit enhances platelet accumulation following vascular injury. Blood. 2017;129(18):2537-2546. 4. Klatt C, Kruger I, Zey S, et al. Platelet-RBC interaction mediated by FasL/FasR induces procoagulant activity important for thrombosis. J Clin Invest. 2018;128(9):3906-3925. 5. Marchioli R, Finazzi G, Specchia G, et al. Cardiovascular events and intensity of treatment in polycythemia vera. N Engl J Med. 2013;368(1):22-33. 6. Alvarez-Larran A, Perez-Encinas M, Ferrer-Marin F, et al. Risk of thrombosis according to need of phlebotomies in patients with polycythemia vera treated with hydroxyurea. Haematologica. 2017;102(1):103-109. 7. Vannucchi AM, Barbui T. Thrombocytosis and thrombosis. Hematology Am Soc Hematol Educ Program. 2007:363-370. 8. Falchi L, Kantarjian HM, Verstovsek S. Assessing the thrombotic risk of patients with essential thrombocythemia in the genomic era. Leukemia. 2017;31(9):1845-1854. 9. Pereira CF, Chang B, Gomes A, et al. Hematopoietic reprogramming in vitro informs in vivo identification of hemogenic precursors to definitive hematopoietic stem cells. Dev Cell. 2016;36(5):525-539. 10. Guadall A, Lesteven E, Letort G, et al. Endothelial cells harbouring the JAK2V617F mutation display pro-adherent and pro-thrombotic features. Thromb Haemost. 2018;118(9):1586-1599. 11. Zetterberg E, Verrucci M, Martelli F, et al. Abnormal P-selectin localization during megakaryocyte development determines thrombosis in the gata1low model of myelofibrosis. Platelets. 2014;25(7):539-547. 12. Centurione L, Di Baldassarre A, Zingariello M, et al. Increased and pathologic emperipolesis of neutrophils within megakaryocytes associated with marrow fibrosis in GATA-1(low) mice. Blood. 2004;104(12):3573-3580. 13. Spangrude GJ, Lewandowski D, Martelli F, et al. P-selectin sustains extramedullary hematopoiesis in the Gata1 low model of myelofibrosis. Stem Cells. 2016;34(1):67-82. 14. Edelmann B, Gupta N, Schnoeder TM, et al. JAK2-V617F promotes venous thrombosis through beta1/beta2 integrin activation. J Clin Invest. 2018;128(10):4359-4371. 15. Wolach O, Sellar RS, Martinod K, et al. Increased neutrophil extracellular trap formation promotes thrombosis in myeloproliferative neoplasms. Sci Transl Med. 2018;10(436). 16. Sozer S, Fiel MI, Schiano T, Xu M, Mascarenhas J, Hoffman R. The presence of JAK2V617F mutation in the liver endothelial cells of patients with Budd-Chiari syndrome. Blood. 2009;113(21):5246-5249. 17. Sozer S, Ishii T, Fiel MI, et al. Human CD34+ cells are capable of generating normal and JAK2V617F positive endothelial like cells in vivo. Blood Cells Mol Dis. 2009;43(3):304-312. 18. Rosti V, Villani L, Riboni R, et al. Spleen endothelial cells from patients with myelofibrosis harbor the JAK2V617F mutation. Blood. 2013;121(2):360-368. 19. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. 20. Schwab N, Schneider-Hohendorf T, Wiendl H. Therapeutic uses of anti-alpha4-integrin (anti-VLA-4) antibodies in multiple sclerosis. Int Immunol. 2015;27(1):47-53. 21. Vincenti F, Mendez R, Pescovitz M, et al. A phase I/II randomized open-label multicenter trial of efalizumab, a humanized anti-CD11a, anti-LFA-1 in renal transplantation. Am J Transplant. 2007;7(7):17701777.

haematologica | 2019; 104(1)

Editorials

Gemtuzumab ozogamicin in acute myeloid leukemia: act 2, with perhaps more to come Johann Hitzler1 and Elihu Estey2 1

Division of Hematology/Oncology, The Hospital for Sick Children and Developmental and Stem Cell Biology, The Hospital for Sick Children Research Institute, Toronto, ON, Canada and 2Division of Hematology University of Washington Medical Center and Clinical Research Division Fred Hutchinson Cancer Research Center Seattle, WA, USA E-mail: eestey@u.washington.edu doi:10.3324/haematol.2018.205948

his editorial discusses two papers, both published in this issue of Haematologica, that expand our knowledge regarding gemtuzumab ozogamicin (GO), both in adults with newly-diagnosed acute myeloid leukemia (AML) and in children with relapsed AML. GO combines the toxin calicheamicin with an antibody to CD33, an antigen found on the surface of hematopoietic cells, including in some instances “stem cells” (Figure 1).1 The relative lack of CD33 expression on the surface of non-hematopoietic cells, except for macrophages lining hepatic sinusoids (Kupffer cells), limits the drug’s principal toxicity to blood forming cells and to a lesser extent Kupffer cells; the latter toxicity results in a sinusoidal obstructive syndrome, most often seen after allogeneic hematopoietic cell transplant (HCT) but largely preventable.2 On September 1, 2017, having originally approved but subsequently withdrawn approval for GO, the US Food and Drug Administration (FDA) re-approved use of GO combined with daunorubicin and cytosine arabinoside (ara-C) given as standard “7+3” for treatment of adults with newly-diagnosed CD33-positive AML, who constitute the great majority of patients with the disease, thus officially ushering in GO’s second act. (https://www.fda.gov/drugs/informationondrugs/approveddrugs/ucm574518.htm). Several months later, the European Medicines Agency (EMA) approved GO for the same indication (https://www.ema.europa.eu/medicines/human/ EPAR/mylotarg-0#authorisation-details-section). The FDA also gave notice of its approval of GO for use as single agent in relapsed or refractory AML and in children. Both the FDA and EMA approvals primarily rested on a French trial (ALFA-0701) randomizing newly diagnosed patients aged 50-70 years to receive 7+3 ± GO 3 mg/m2 on days 1, 4, and 7.3 In a break with precedent, which we discuss below, approval was based on prolongation of eventfree survival (EFS) rather than “overall” survival (hereafter “survival”); the benefit was limited to patients with “favorable” or “intermediate” cytogenetics. Median survival in patients remaining alive was 20 months. It is known, however, that the risk of relapse or death declines precipitously only once 2-3 years have elapsed since achievement of CR.4 Hence the update of the ALFA-0701 trial reported by Lambert et al. in the current issue of Haematologica is particularly noteworthy given the median follow up of 47 months in the 7+3+GO group and 41 months in the controls.5 The essential findings of the original study remain unchanged, validating the FDA and EMA decisions to grant approval for GO. Another report in the current issue of Haematologica by Niktoreh et al. from the Berlin-Frankfurt-Münster study group notes that when GO was given on a “compassionhaematologica | 2019; 104(1)

ate” basis, either alone or with cytarabine, treatment resulted in sufficient reduction in blast count to permit 64% of 76 children aged under 18 years to receive HCT.6 At four years, probability of survival was 18±5%: 27±7% in the HCT and 0% in the non-HCT group. Although there is less experience with GO in children than in adults, this report follows a randomized trial of chemotherapy with or without addition of GO (one dose of 3 mg/m2 given on the first and fourth course of chemotherapy) conducted in over 500 children by the Children’s Oncology Group (COG).7 Results were analogous to those reported in ALFA-0701: a greater effect on EFS than on survival and largely limited to individuals with intermediate or favorable cytogenetics. Both the COG study and the study reported here by Niktoreh et al.6 justify the FDA’s approval of GO in children. The value of GO in relapsed pediatric AML will be clarified by the authors’ ongoing prospective randomized controlled trial (EudraCT Number: 2010018980-41). For many years, an increase in survival had been the sole basis for approval of new drugs in AML. This was sensible, since once failure to enter CR or relapse from CR had occurred, survival was typically limited to 2-3 months. Hence EFS and survival were largely synonymous. However today, the ability to keep people alive once these events have occurred has improved. Probably this is mainly due to improved supportive care, particularly better anti-fungal agents, but also reflects better “salvage” therapies. For example, the NCRI/MRC group in the UK reported that AML patients with intermediate-risk cytogenetics who did not receive HCT in first complete remission (CR1), nonetheless had similar survival as patients receiving HCT in CR1.8 This was a result of the ability to achieve, and then perform HCT in CR2 in the patients in whom relapse occurred after failure to perform HCT in CR1. The severance of EFS time from survival time has important implications. Once an event (no CR or relapse) occurs in a patient randomized to one arm of a phase III trial, there is no assurance that the salvage therapy or supportive care received will be identical to that received by a patient who has an event in the other arm of the trial. Hence what began as a randomized trial degenerates into an observational study, with all the attendant biases. Thus EFS, the primary end point in the ALFA-0701 trial, may be more reliable than survival as an indicator of the efficacy of an intervention such as GO.9 While there was an improvement in survival in the GO arm of ALFA-0701 and in the individual patient metaanalysis performed by Hills et al.,10 that included ALFA0701, these improvements were not “statistically significant”, unlike those in EFS, using P<0.05 as the criterion for 7

Editorials

Figure 1. Mechanism of action of gemtuzumab ozogamicin.

statistical significance. However, comparison of hazard rates using P-values focuses on the relative value of an intervention rather than the absolute value. A good way to examine the latter is as the â&#x20AC;&#x153;number needed to treatâ&#x20AC;? (NNT) to prevent a single event (or a single death); this information is often omitted from reports of clinical trials. Here, however, Lambert et al. note that at one year, 6 people would need to be treated with GO to prevent one event, while at year 3, NNT was 4. Such NNT values imply that most people do not benefit from GO when it is added to 7+3. Can we identify who will benefit? Since GO binds to CD33 it seems plausible that higher levels of CD33 expression might be associated with a better outcome following use of GO. Likewise, since it is known that GO, like agents such anthracyclines, is extruded from AML blasts by ATP-binding cassette transporter proteins mediating multidrug resistance (MDR), there may also be a relationship between extrusion and response to GO. Pollard et al. noted that children in the COG study who were in the highest 75% of CD33 expression had superior EFS regardless of cytogenetic group when given GO but not in the control group (no GO),11 and similar findings 8

have been reported in adults for patients with favorable and intermediate cytogenetics.12,13 A higher dose (e.g. 6 mg/m2 rather than 3 mg/m2) may benefit patients with low, but not high, CD33 expression.13 It should be kept in mind that quantification of neither CD33 expression nor MDR is currently a routine task. Attempts to enhance GO activity by use of cyclosporine A to block MDR do not appear to have been successful.14 More recently, attention has focused on the role of CD33 single nucleotide polymorphisms (SNPs) in affecting response to GO. Such SNPs regulate the expression of CD33 isoforms. SNPs denoted as TT result in a lack of exon 2, resulting in absence of the antibody-binding site for GO on CD33, as well as in diagnostic immunophenotypic panels. Using data from the aforementioned COG study,7 Lamba et al. found only children with the CC SNP (51% of 816 children), which encodes a full exon 2 resulting in the presence of a GO binding site on CD33, benefitted from addition of GO to chemotherapy regardless of cytogenetics or CD33 expression.15 However, this finding has not been confirmed by the UK NCRI/MRC group,16 although a similar proportion of patients (47% of 563 patients) had the CC SNP as in Lamba et al.,15 and the haematologica | 2019; 104(1)

Editorials

favorable cytogenetic group had the expected longer survival when given chemotherapy + GO rather than chemotherapy alone. The Southwest Oncology Group has also failed to notice an effect of CD33 SNPs on outcome in adults (M Othus, 2018, personal communication). Even the minority of patients who benefit from GO might benefit more from development of improved antiCD33 therapeutics.17 One possibility here is use of bispecific antibodies (BiAbs) that engage CD33 but also direct T cells toward AML cells. An obvious model for this approach is blinatumomab,18 a CD3/CD19 molecule built in the Bispecific T-cell Engager (BiTE) format. A first series of CD33/CD3 BiAbs, including the BiTE AMG 330 and the tandem diabody AMV-564, have recently entered clinical tests. Like GO, all CD33 BiAbs (and other CD33-directed therapeutics) currently under investigation recognize the V set domain, which is located distally on CD33. However, preliminary studies with artificial CD33 molecules show that membrane proximal binding enhances the immune effector cell functions of CD33 antibodies (R Walter, 2018, personal communication). Development of antibodies recognizing such proximal sites is likely to be an area of examination in GO’s second and, hopefully, subsequent acts.

References 1. Walter R, Appelbaum F, Estey E, and Bernstein I Acute myeloid leukemia stem cells and CD33-targeted immunotherapy. Blood. 2012;119(26):6198-6208. 2. McKoy J, Angelotta C, Bennett C, et al. Gemtuzumab ozogamicin-associated sinusoidal obstructive syndrome (SOS): an overview from the research on adverse drug events and reports (RADAR) project. Leuk Res. 2007;31(5):599-604. 3. Castaigne S, Pautas C, Terré C, et al. Effect of gemtuzumab ozogamicin on survival of adult patients with de-novo acute myeloid leukaemia (ALFA-0701): a randomised, open-label, phase 3 study. Lancet. 2012;379(9825):1508-1516. 4. de Lima M, Strom S, Keating M, et al. Implications of potential cure in acute myelogenous leukemia: development of subsequent cancer and return to work. Blood. 1997;90(12):4719-4724.

5. Lambert J, Pautas C, Terré C et al. Gemtuzumab ozogamicin for de novo acute myeloid leukemia: final efficacy and safety updates from the open-label, phase III ALFA-0701 trial. Haematologica. 2019;104(1):113-119. 6. Niktoreh N, Lerius B, Zimmermann M et al. Gemtuzumab ozogamicin in children with relapsed or refractory acute myeloid leukemia: a report by Berlin-Frankfurt-Münster study group. Haematologica. 2019;104(1):120-127. 7. Gamis AS, Alonzo TA, Meshinchi S, et al. Gemtuzumab ozogamicin in children and adolescents with de novo acute myeloid leukemia improves event-free survival by reducing relapse risk: results from the randomized phase III Children’s Oncology Group trial AAML0531.J Clin Oncol. 2014;32(27):3021-3032. 8. Burnett AK, Goldstone A, Hills RK, et al. Curability of patients with acute myeloid leukemia who did not undergo transplantation in first remission. J Clin Oncol. 2013;31(10):1293-1301. 9. Estey E, Othus M, Lee SJ, Appelbaum FR, Gale RP. New drug approvals in acute myeloid leukemia: what's the best endpoint? Leukemia. 2016;30(3):521-525. 10. Hills RK, Castaigne S, Appelbaum FR, et al. Addition of gemtuzumab ozogamicin to induction chemotherapy in adult patients with acute myeloid leukaemia: a meta-analysis of individual patient data from randomised controlled trials. Lancet Oncol. 2014;15(9)986-996. 11. Pollard JA, Loken M, Gerbing RB, et al. CD33 Expression and Its Association With Gemtuzumab Ozogamicin Response: Results From the Randomized Phase III Children's Oncology Group Trial AAML0531. J Clin Oncol. 2016;34(7):747-755. 12. Olombel G, Guerin E, Guy J, et al. The level of blast CD33 expression positively impacts the effect of gemtuzumab ozogamicin in patients with acute myeloid leukemia Blood. 2016;127(17) 2157-2160. 13. Khan N, Hills RK, Virgo P, et al. Expression of CD33 is a predictive factor for effect of gemtuzumab ozogamicin at different doses in adult acute myeloid leukaemia. Leukemia. 2017;31(5):1059-1068. 14. Tsimberidou A, Estey E, Cortes J, et al. Gemtuzumab, fludarabine, cytarabine, and cyclosporine in patients with newly diagnosed acute myelogenous leukemia or high-risk myelodysplastic syndromes. Cancer. 2003;97(6):1481-1487. 15. Lamba JK, Chauhan L, Shin M, et al. CD33 Splicing Polymorphism Determines Gemtuzumab Ozogamicin Response in De Novo Acute Myeloid Leukemia: Report From Randomized Phase III Children's Oncology Group Trial AAML0531. J Clin Oncol. 2017;35(23):26742682. 16. Gale R, Pope T, Wright M, et al. No evidence that CD33 splicing SNP impacts the response to GO in younger adults with AML treated on UK MRC/NCRI trials. Blood. 2018;131(4):468-471. 17. Walter R Investigational CD33-targeted therapeutics for acute myeloid leukemia. Expert Opin Investig Drugs 2018; 27(4): 339-348 18. Kantarjian H, Stein A, Gökbuget N, et al. Blinatumomab versus Chemotherapy for Advanced Acute Lymphoblastic Leukemia. N Engl J Med. 2017;376(9):836-847.

Lenalidomide can be safely combined with chlorambucil and rituximab in older patients with chronic lymphocytic leukemia Candida Vitale1 and Alessandra Ferrajoli2 1 Division of Hematology, University of Torino, A.O.U. Città della Salute e della Scienza di Torino, Italy and 2Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

E-mail: aferrajo@mdanderson.org doi:10.3324/haematol.2018.206359

he clinical activity of lenalidomide in chronic lymphocytic leukemia (CLL) was first reported more than 10 years ago.1,2 Since then, this agent has been studied in various combinations with anti-CD20 monoclonal antibodies, chemotherapy, chemo-immunotherapy and B-cell receptor (BCR)-targeting agents. These studies have shown clinical responses; however, most importantly, haematologica | 2019; 104(1)

they have also highlighted unique and unexpected toxicities, in particular when lenalidomide was combined with chemo-immunotherapy and targeted agents. In this issue of Haematologica, Kater and colleagues report the experience of the HOVON CLL study group on the feasibility and efficacy of the combination of lenalidomide, chlorambucil, and rituximab in treatment-naïve patients 9

Editorials

with CLL.3 The patients enrolled in this trial were considered ineligible to receive the combination of fludarabine, cyclophosphamide, and rituximab (FCR) because of their older age or the presence of comorbidities. For the first six cycles (induction-I), lenalidomide was given in combination with chlorambucil and rituximab at a starting dose of 2.5 mg, with escalation to 10 mg. The authors report that they were able to administer a median lenalidomide dose of 86.7% of the full dose, with the full dose given to more than 50% of patients. For the next six cycles (induction-II), lenalidomide was given as monotherapy at a dose of 10 mg daily. The median administered dose was 99.7% of the full dose, and the full dose was given to 69% of patients during

cycle 6. The results of this phase 1-2 study showed that the combination of lenolidamide, chlorambucil, and rituximab can be safely administered to patients with CLL: grade 3-4 toxicities were mainly hematologic (grade 3-4 neutropenia occurred in 73% and 64% of patients during induction-I and induction-II, respectively), tumor lysis syndrome did not occur, tumor flare reaction occurred in five (9%) patients (mainly grade 2), and two (4%) patients had a thromboembolic event despite thromboembolic prophylaxis. Of 53 patients in induction-I, eight discontinued treatment because of excessive toxicity, whereas five of 42 patients discontinued treatment during induction-II. The authors also report on the activity of this combination:

Figure 1. Mechanisms of action of lenalidomide. The mechanisms of action of lenalidomide include: (A) direct effects on CLL cells and (B, C) modification of tumor-microenvironment interactions. This figure is reproduced with permission from Mattei R et al., Lenalidomide in chronic lymphocytic leukemia: the present and future in the era of tyrosine kinase inhibitors. Crit Rev Oncol Hematol. 206;97:291-302. NLCs: nurse-like cells; EC: endothelial cells.

haematologica | 2019; 104(1)

Editorials

responses were seen in 83% of the patients treated, the median progression-free survival was 49 months, and the 3year overall survival rate was 95%. Early monotherapy trials showed that lenalidomide is associated with a unique toxicity profile in patients with CLL, causing tumor lysis syndrome, tumor flare reaction, myelosuppression, and, in particular, neutropenia, skin rash, and diarrhea.1,2 Particularly severe events, including deaths, were reported in trials with no dose escalation4 or rapid dose escalation,5 and tumor lysis syndrome occurred in patients with bulky lymphadenopathy despite proper prophylaxis.1 After these early experiences, trials with lenalidomide employed stepwise dose escalation strategies with low starting doses (usually 2.5 mg/day), as in the study presented by Kater et al., and managed tumor flare reactions with non-steroidal anti-inflammatory drugs and corticosteroids. Moreover, careful patient selection is recommended. The population included in the study by Kater et al. mainly consisted of older but fit patients: 98% were 65 years or older, 87% had a Cumulative Illness Rating Scale score of 6 or lower, and all had a glomerular filtration rate of 60 mL/min or higher at the time of entry into the study. In the last decade, several lenalidomide-containing combination regimens have been evaluated in patients with treatment-na誰ve CLL. When lenalidomide was combined with rituximab in the frontline setting, the treatment was generally well tolerated, with the most common grade 3-4 toxicities being neutropenia, anemia, infections, increased transaminase levels, and skin rash.6,7 However, when different partners, such as chemotherapy agents or targeted drugs, were tested in combination with lenalidomide, some trials documented excessive toxicities that led to early termination of the studies. For instance, a phase 1 study investigating the combination of lenalidomide with fludarabine and rituximab was closed early because of unpredictable reactions and unexpectedly persistent myelosuppression, even when very low doses of fludarabine and lenalidomide were given, which made treatment delivery difficult.8 Instead, induction treatment with low-dose lenalidomide together with reduced-dose FCR was demonstrated to be safe.9 In the relapse setting, lenalidomide was evaluated in association with rituximab and ibrutinib; the study investigating this approach showed a high incidence of persistent severe neutropenia that occurred despite growth factor support.10 This unfavorable toxicity profile, together with poor preliminary efficacy data, discouraged further evaluation of this combination. The combination of lenalidomide with rituximab and idelalisib also showed unacceptable liver toxicity in patients with relapsed or refractory indolent lymphoma.11 Regarding efficacy, the single-arm design of the study by Kater et al. does not allow a direct comparison of the triple combination with chlorambucil and rituximab. Acknowledging the limitations of cross-trial comparisons, however, the efficacy of the proposed regimen compares positively with that of chlorambucil plus anti-CD20 monoclonal antibodies. In a study conducted by Strati et al.,7 the combination of lenalidomide plus rituximab produced an overall response rate of 73% in treatment-na誰ve patients, with a complete remission (CR)/CR with incomplete hematologic recovery (CRi) rate of 35%, a median time to treatment failure of 22 months, and a 4-year overall survival rate haematologica | 2019; 104(1)

of 90%. The same treatment combination was explored in a multicenter study, which showed an overall response rate of 88%, of which 15% were CR/CRi, and a median progression-free survival of 19 months.6 It is essential to put the data presented by Kater and colleagues into perspective by considering recent changes in the treatment landscape of CLL brought about by the availability of new targeted drugs, such as BTK inhibitors, PI3K inhibitors, and Bcl-2 antagonists, which have also been studied in older patients. In a recent update of the phase III RESONATE-2 trial of ibrutinib, which enrolled patients aged 65 years and older with previously untreated CLL and without del(17p), researchers reported that at a median follow-up time of 29 months, the overall response rate was 92%, the median duration of progression-free survival had not been reached, and the 24-month progression-free survival rate was 89%.12 Patients carrying abnormalities on chromosome 17 represent a subset of CLL patients with a particularly poor prognosis. In the cohort presented by Kater et al., eight (17%) patients had del(17p), and their progression-free survival rate was lower than that of patients without del(17p) (38% versus 59% at 3 years). Notably, in a phase II study that evaluated ibrutinib in a cohort of treatment-na誰ve CLL patients with TP53 aberrations, the estimated 5-year progression-free survival rate was 74.4%.13 That being said, a credit that pertains exclusively to lenalidomide is the role this drug has had in elucidating tumor-microenvironment interactions in CLL. Phenotypic and functional immune defects are known to be associated with CLL; these defects confer an increased risk of infections and autoimmune phenomena and foster leukemia cell proliferation and survival. Several studies have shown that treatment with lenalidomide modulates the cross-talk between tumor cells and various components of the tumor microenvironment. Examples of these effects include the ability to normalize CD3+ T-cell and Treg numbers in vivo14,15 and to restore immunological synapse formation.16 The antitumoral activities of lenalidomide also appear to be attributable to a direct effect on neoplastic cells; lenalidomide not only enhances immune recognition, but also induces CRBN-mediated upregulation of p21 in vitro17 (Figure 1). The recent progresses in immunotherapy approaches that exploit the ability to engineer the T-cell receptor, such as chimeric antigen receptor (CAR) T-cell therapy, may revitalize interest in the use of immunomodulatory agents, including lenalidomide, in CLL. Immune dysfunctions are thought to be responsible for the lower efficacy of these approaches in CLL than in other lymphoproliferative diseases. Apheresis products from CLL patients and the derived CAR T-cell products exhibit an exhausted phenotype and tend to have reduced potency. It has been demonstrated that certain features of the apheresis product, such as the predominance of early memory/na誰ve T cells and low expression of exhaustion markers, correlate with efficacy.18 It has also been reported that ibrutinib may correct some of the T-cell defects that hinder CAR T-cell production and enhance in vivo function.19 The ability of lenalidomide to enhance CAR T-cell activity has been explored in a mouse model of B-cell lymphoma20 and provides a rationale for future investigations of the immunomodulatory properties of lenalidomide and its derivatives in CLL. 11

Editorials

References 1. Chanan-Khan A, Miller KC, Musial L, et al. Clinical efficacy of lenalidomide in patients with relapsed or refractory chronic lymphocytic leukemia: results of a phase II study. J Clin Oncol. 2006;24(34):53435349. 2. Badoux XC, Keating MJ, Wen S, et al. Lenalidomide as initial therapy of elderly patients with chronic lymphocytic leukemia. Blood. 2011;118(13):3489-3498. 3. Kater AP, van Oers MHJ, van Norden Y, et al. Feasibility and efficacy of addition of individualized-dose lenalidomide to chlorambucil and rituximab as first-line treatment in elderly and FCR-unfit patients with advanced chronic lymphocytic leukemia. Haematologica. 2019;104 (1):147-154. 4. Andritsos LA, Johnson AJ, Lozanski G, et al. Higher doses of lenalidomide are associated with unacceptable toxicity including life-threatening tumor flare in patients with chronic lymphocytic leukemia. J Clin Oncol. 2008;26(15):2519-2525. 5. Chen CI, Bergsagel PL, Paul H, et al. Single-agent lenalidomide in the treatment of previously untreated chronic lymphocytic leukemia. J Clin Oncol. 2011;29(9):1175-1181. 6. James DF, Werner L, Brown JR, et al. Lenalidomide and rituximab for the initial treatment of patients with chronic lymphocytic leukemia: a multicenter clinical-translational study from the Chronic Lymphocytic Leukemia Research Consortium. J Clin Oncol. 2014;32(19):2067-2073. 7. Strati P, Thompson PA, Keating M, et al. A Phase II Study of the combination of lenalidomide and rituximab in patients with treatmentnaĂŻVe and relapsed chronic lymphocytic leukemia. Blood. 2016;128(22):4389-4389. 8. Brown JR, Abramson J, Hochberg E, et al. A phase I study of lenalidomide in combination with fludarabine and rituximab in previously untreated CLL/SLL. Leukemia. 2010;24(11):1972-1975. 9. Mato AR, Foon KA, Feldman T, et al. Reduced-dose fludarabine, cyclophosphamide, and rituximab (FCR-Lite) plus lenalidomide, followed by lenalidomide consolidation/maintenance, in previously untreated chronic lymphocytic leukemia. Am J Hematol. 2015;90 (6):487-492.

10. Ujjani C, Wang H, Skarbnik A, et al. A phase 1 study of lenalidomide and ibrutinib in combination with rituximab in relapsed and refractory CLL. Blood Adv. 2018;2(7):762-768. 11. Cheah CY, Nastoupil LJ, Neelapu SS, Forbes SG, Oki Y, Fowler NH. Lenalidomide, idelalisib, and rituximab are unacceptably toxic in patients with relapsed/refractory indolent lymphoma. Blood. 2015;125(21):3357-3359. 12. Barr PM, Robak T, Owen C, et al. Sustained efficacy and detailed clinical follow-up of first-line ibrutinib treatment in older patients with chronic lymphocytic leukemia: extended phase 3 results from RESONATE-2. Haematologica. 2018;103(9):1502-1510. 13. Ahn IE, Farooqui MZH, Tian X, et al. Depth and durability of response to ibrutinib in CLL: 5-year follow-up of a phase 2 study. Blood. 2018;131(21):2357-2366. 14. Lee B-N, Gao H, Cohen EN, et al. Treatment with lenalidomide modulates T-cell immunophenotype and cytokine production in patients with chronic lymphocytic leukemia. Cancer. 2011;117(17):3999-4008. 15. Strati P, Keating MJ, Wierda WG, et al. Lenalidomide induces long-lasting responses in elderly patients with chronic lymphocytic leukemia. Blood. 2013;122(5):734-737. 16. Ramsay AG, Johnson AJ, Lee AM, et al. Chronic lymphocytic leukemia T cells show impaired immunological synapse formation that can be reversed with an immunomodulating drug. J Clin Invest. 2008;118(7):2427-2437. 17. Fecteau JF, Corral LG, Ghia EM, et al. Lenalidomide inhibits the proliferation of CLL cells via a cereblon/p21(WAF1/Cip1)-dependent mechanism independent of functional p53. Blood. 2014;124(10):1637-1644. 18. Fraietta JA, Lacey SF, Orlando EJ, et al. Determinants of response and resistance to CD19 chimeric antigen receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat Med. 2018;24(5):563-571. 19. Fraietta JA, Beckwith KA, Patel PR, et al. Ibrutinib enhances chimeric antigen receptor T-cell engraftment and efficacy in leukemia. Blood. 2016;127(9):1117-1127. 20. Otahal P, Prukova D, Kral V, et al. Lenalidomide enhances antitumor functions of chimeric antigen receptor modified T cells. Oncoimmunology. 2016;5(4):e1115940.

haematologica | 2019; 104(1)

REVIEW ARTICLE

Peering through zebrafish to understand inherited bone marrow failure syndromes

Ferrata Storti Foundation

Usua Oyarbide,1* Jacek Topczewski2,3 and Seth J. Corey1,4,5*

Department of Pediatrics, Childrenâ&#x20AC;&#x2122;s Hospital of Richmond and Massey Cancer Center at Virginia Commonwealth University, Richmond, VA, USA; 2Department of Pediatrics, Stanley Manne Children's Research Institute, Northwestern University Feinberg School of Medicine, Chicago, IL, USA; 3Department of Biochemistry and Molecular Biology, Medical University of Lublin, Poland; 4Department of Microbiology/Immunology, Virginia Commonwealth University, USA and 5Department of Human and Molecular Genetics, Virginia Commonwealth University, Richmond, USA 1

*Current address: Departments of Pediatrics, Translational Hematology and Oncology Research, and Cancer Biology, Cleveland Clinic, Cleveland, OH, USA

Haematologica 2019 Volume 104(1):13-24

ABSTRACT

nherited bone marrow failure syndromes are experiments of nature characterized by impaired hematopoiesis with cancer and leukemia predisposition. The mutations associated with inherited bone marrow failure syndromes affect fundamental cellular pathways, such as DNA repair, telomere maintenance, or proteostasis. How these disturbed pathways fail to produce sufficient blood cells and lead to leukemogenesis are not understood. The rarity of inherited cytopenias, the paucity of affected primary human hematopoietic cells, and the sometime inadequacy of murine or induced pluripotential stem cell models mean it is difficult to acquire a greater understanding of them. Zebrafish offer a model organism to study gene functions. As vertebrates, zebrafish share with humans many orthologous genes involved in blood disorders. As a model organism, zebrafish provide advantages that include rapid development of transparent embryos, high fecundity (providing large numbers of mutant and normal siblings), and a large collection of mutant and transgenic lines useful for investigating the blood system and other tissues during development. Importantly, recent advances in genomic editing in zebrafish can speedily validate the new genes or novel variants discovered in clinical investigation as causes for marrow failure. Here we review zebrafish as a model organism that phenocopies Fanconi anemia, Diamond-Blackfan anemia, dyskeratosis congenita, ShwachmanDiamond syndrome, congenital amegakaryocytic thrombocytopenia, and severe congenital neutropenia. Two important insights, provided by modeling inherited cytopenias in zebrafish, widen understanding of ribosome biogenesis and TP53 in mediating marrow failure and non-hematologic defects. They suggest that TP53-independent pathways contribute to marrow failure. In addition, zebrafish provide an attractive model organism for drug development.

Correspondence: coreys2@ccf.org

Received: August 20, 2018. Accepted: November 14, 2018. Pre-published: December 20, 2018.

doi:10.3324/haematol.2018.196105 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/13 ÂŠ2019 Ferrata Storti Foundation

Introduction The inherited bone marrow failure syndromes (IBMFs) comprise a diverse group of rare monogenic disorders that are phenotypically heterogeneous. They may involve a single or multiple lineage(s). The classic disorders are: Fanconi anemia (FA), Diamond-Blackfan anemia (DBA), Shwachman-Diamond syndrome (SDS), dyskeratosis congenita (DC), severe congenital neutropenia (SCN), and congenital amegakaryocytic thrombocytopenia (CAMT). Besides their phenotypic characterizations, these syndromes correlate strongly with mutations involving a specific pathway. FA results from mutations in genes encoding components of the DNA damage response,1 DC in telomere maintenance,2 and DBA in ribosome function.3 haematologica | 2019; 104(1)

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

U. Oyarbide et al.

SDS is emerging as a disorder in proteostasis and ribosome maturation (Table 1).4 The molecular basis for how these phenotypically and genotypically heterogeneous conditions result in single or multiple cytopenias remains poorly understood. No common pathway has yet been established, but zebrafish studies have suggested TP53 responses. Activation of the TP53 pathway in mediating marrow failure has been reported for DC,5 FA,6 and a novel bone marrow failure syndrome.7 The TP53 pathway has been suggested to mediate marrow failure for other inherited neutropenias such as SCN and SDS.8 Environmental exposures can accelerate marrow failure, for example, aldehydes producing DNA crosslinks in FA.9 How epigenetics and genetic co-modifiers contribute to these diseases is even less understood. Investigating the molecular basis of the IBMFS will lead to a greater understanding of hematopoiesis, and development and maintenance of non-hematologic tissues. Since the IBMFS constitute leukemia or cancer predisposition syndromes, insights into their pathophysiology will also benefit our understanding, prevention, and perhaps treatment of cancer and age-related genetic changes.

Zebrafish model to study inherited bone marrow failure syndromes Zebrafish (Danio rerio) have gained popularity as a model organism for a number of reasons. Approximately 70% of all human genes have a zebrafish ortholog.10 Genes are orthologs if they evolved from a common gene, and orthologs typically share similar function. (The

Human Genome Organization has adopted a nomenclature for gene and protein expression among different species, which we show using SDS as an example in Table 2.) In addition to lower maintenance and breeding costs, zebrafish provide major advantages to mice: their large clutch size of externally fertilized eggs, transparent embryos, quicker development (all major organs develop and begin functioning during the first 5 days), and short generational time to gamete formation.11 A high degree of genetic and morphological similarity in hematopoiesis between zebrafish and humans suggests that zebrafish can provide valuable insights into the pathogenesis of IBMFS. Developmental hematopoiesis in the zebrafish is comparable to that observed in mice or humans (Figure 1).12-15 One notable difference is that the site of definitive hematopoiesis lies in the zebrafish kidney perivascular space, not the bone marrow. Since the hematopoietic stem cell (HSC) niche provides protection and regulation of self-renewal and differentiation of HSC into blood cells, this difference may be important in non-cell autonomous processes. Studies using zebrafish have facilitated our understanding of vertebrate hematopoiesis and aberrant hematopoiesis in diseases. Hematopoietic and nonhematopoietic lineage-specific transgenic reporter strains are available. They have been useful for the identification and characterization of genes for embryonic hematopoiesis, erythropoiesis, and modeling of human blood diseases (Table 3).16-19 In addition to a collection of zebrafish mutants induced by N-ethyl-N-nitrosourea or

Table 1. Inherited bone marrow failure syndromes.

Disease

Prevalence Male-to-female per 1,000,000 ratio

Diamond-Blackfan anemia (DBA)

Symptoms

Genes involved and their estimated frequency

Cancer predisposition

Erythroid failure, congenital malformations, growth retardation, short stature. Thumbs, upper limbs, hands, and craniofacial, urogenital, and cardiovascular anomalies are also common Abnormal skin pigmentation, nail dystrophy, mucosal leukoplakia, pulmonary fibrosis, and bone marrow failure

RPS19 (25%), RPL5 (7%), RPS26 (6.6), RPL11 (5%), RPL35a (3%), RPS10 (3%), RPS24 (2.4%), RPS17 (1%), RPL15, RPS28, RPS29, RPS7, RPS15, RPS27a, RPS27, RPL9, RPL18, RPL26, RPL27, RPL31, TSR2, GATA1, EPO DKC1 (17-36%), TERC (6-10%), TERT(1-7%), NHP2 (<1%), NOP10 (<1%), CTC1 (1-3%), WRAP53 (3%) and TINF2 (11-24%), ACD, PARN, RTEL1, USB1, TCAB1, POT1, TPP1, WRD79, TR, NOLA2, NOLA3 FANCA (65%), FANCB (<1%), FANCC (14%), FANCG (10%), FANCD1/BRCA2 (<1%), FANCD2 (<1%), FANCE (4%), FANCF (4%), RAD51, FANCC1, FANL, FANCL, FANC, PALPB2, RADC51C, SLX4, FANCQ. BRCA1, FANCT SBDS (90%) DNAJC21 EFL1, SRP54 MPL

AML, MDS, ALL, Hodgkin and non-Hodgkin lymphomas, osteogenic sarcoma, breast cancer, hepatocellular carcinoma, melanoma, fibrohistiocytoma, gastric cancer, colon cancer AML, solid tumors

ELANE, GFI1, HAX1, G6PC3, VPS45, JAG1, CSF3R, WAS, SRP54

AML, MDS

5-7

1:1

Dyskeratosis congenita (DC)

3:1

Fanconi anemia (FA)

1.2:1

Developmental abnormalities in a number of organ systems and bone marrow failure

Shwachman-Diamond syndrome (SDS)

1.7:1

Exocrine pancreatic insufficiency, bone marrow dysfunction and skeletal abnormalities Thrombocytopenia and megakaryocytopenia

Congenital amegakaryocytic thrombocytopenia (CAMT) Severe congenital neutropenia (SCN)

Unknown (less than 100 cases reported) 5

Neutropenia

AML, solid tumors

AML, MDS

AML: acute myeloid leukemia; ALL: acute lymphocytic leukemia; MDS: myelodysplastic syndromes.

haematologica | 2019; 104(1)

IBMFS in zebrafish

viral insertion,20,21 gene function can be studied by transgenic expression or genome editing by transcription activator-like effector nucleases (TALEN) or Cas nucleases acting on clustered, regularly interspaced, short palindromic repeats (CRISPR). Gene expression can be silenced temporarily and early during development by injection of morpholino antisense nucleotides (MO). Zebrafish have provided a useful model organism for a quick validation and study of human disease candidate genes, including those involved in the pathophysiology of IBMFS (Table 4). MO-mediated knockdown was widely used to probe gene function, though this method has limitations. Phenotype of morphants (MO-injected animals) can differ and is often more severe than those of the corresponding mutants. There could be different reasons for this: 1) phenotypic rescue of zygotic mutants by maternal wild-type mRNA; 2) off-target effects of the MO; 3) hypomorphic nature of the mutant allele analyzed; or 4) genetic compensation in mutants but not in morphants (see Stainier et al.22). Moreover, injection of MO can cause Tp53 activation and cell death.23 In some instances, cell death can be prevented by simultaneous blocking of p53 by a second MO. This may lead to a misinterpretation of

results, particularly in processes that depend on the Tp53 DNA damage response pathway (reviewed below). In some cases, results of MO knockdowns were not recapitulated with the genome editing techniques.24 Close examination of the differences in gene expression revealed a novel compensation mechanism that operates only after mutation but not after MO knockdown (Table 5).25

Diamond-Blackfan anemia Diamond-Blackfan anemia is characterized by red cell hypoplasia, erythroid macrocytosis, and markedly reduced erythroid precursors in the bone marrow. Other hematopoietic lineages are usually normal at birth,26 but they may be affected later in childhood/adolescence.27 In

Table 2. Gene and protein nomenclature among species Human Mouse Zebrafish

Gene symbol

Protein symbol

sbds sbds sbds

SBDS SBDS Sbds

Figure 1. Comparison of developmental hematopoiesis in humans and zebrafish. Primitive and definitive hematopoiesis occurs in both species. In human, hematopoietic stem cells (HSC) originate in aorta-gonad-mesonephros (AGM) and placenta, from where they colonize fetal liver and finally the bone marrow. In zebrafish, primitive hematopoiesis starts after hemangioblast formation around 12 hpf in the anterior lateral mesoderm (ALM) and posterior lateral mesoderm (PLM). Later, HSCs originate in the AGM and then mobilize to caudal hematopoietic tissue (CHT) prior to their final destination of the kidney (modified from Teittinen et al.12 and de Jong and Zon14).

haematologica | 2019; 104(1)

U. Oyarbide et al.

addition to severe anemia, individuals with DBA may display physical anomalies that include thumb, upper limb, craniofacial, cardiovascular and kidney malformations, and short stature. DBA patients have a 25% higher risk of developing myelodysplastic syndromes (MDS), acute myeloid leukemia (AML), and osteosarcoma. Diamond-Blackfan anemia is an autosomal dominant disorder with a disease incidence of 5-7 per million live births, equally distributed between genders.28 DBA patients have mutations in approximately 20 genes encoding ribosomal proteins; the most common (25%) is RPS19.29 Frameshift, splice defects, intragenic deletions and insertions, nonsense, as well as missense mutations have all been identified. Mutations involve other ribosomal genes: RPL5 (7%), RPL26 (6.6%), RPL11 (5%), RPS10 (3%), RPS26 (3%), RPL35A (3%), RPS24 (2.4%), RPS17 (1%), RPL15, RPS28, RPS29, RPS7, RPS15, RPS27a, RPS, RPL9, RPL18, RPL26, RPL27, RPL31.3 These findings support DBA as a disorder of ribosomal biogenesis and/or function. Mutations in three non-ribosomal proteins, GATA1, TSR2, and EPO, are also associated to DBA.3,29 It is hypothesized that DBA results from apoptosis due to aberrant activation of TP53 that induces cell cycle arrest or apoptosis in response to ribosomal stress.30 Some of the reports implicating TP53, as reviewed below, are based on MO-mediated effects. In two different studies, rps19-deficient zebrafish were

created using MO. Knockdown of rps19 in zebrafish recapitulates the hematopoietic and developmental phenotypes of DBA, including erythropoietic failure with severe anemia, with cell cycle arrest and increased apoptosis, with p53 upregulation. The rps19-deficient phenotype was rescued by injection of zebrafish rps19 mRNA.31-33 Moreover, these phenotypes were not rescued by expressing rps19 mRNAs with a missense or nonsense mutation found in DBA patients.32 Co-injection of MOs against rps19 and p53, showed a complete rescue of the morphological abnormalities, but did not rescue the hematologic defects. These results suggest that there is an erythroid specificity in Rps19 deficiency in zebrafish, independently of Tp53 activity. (See below for further discussion on Tp53 in DBA pathogenesis).34 Chakraborty et al. analyzed the effect of MO-mediated loss of rpl11 in zebrafish. Knockdown of rpl11 led to morphological defects in the developing brain, head, and eyes, and pericardial edema. These phenotypes appear specific as the investigators were able to suppress the morphant by co-injection of MO-resistant rpl11 mRNA. Similar to the loss of Rsp19 function, knockdown of rpl11 resulted in an upregulation of tp53 and mdm2. Moreover, co-injection of rpl11 and tp53 MO rescued the developmental defects and reduced apoptosis, suggesting that ribosomal dysfunction due to the loss of Rpl11 activates a Tp53-dependent response to prevent faulty embryonic development.

Table 3. Comparison of human, mouse and zebrafish blood systems. Adult HSC Blood cell types Erythrocytes (life span) Platelets (life span) Neutrophils (life span)

Primitive myelopoiesis

Human

Mouse

Zebrafish

Bone marrow Erythrocytes, granulocytes, lymphocytes and platelets Without nucleus (115 days) Platelets (8-9 days) Segmented nuclei with up to four lobes mpo-expressing cells (5.4 days) Yolk sac, AGM, fetal liver

Bone marrow Erythrocytes, granulocytes, lymphocytes and platelets Without nucleus (60 days) Platelets (4 days) Twisted toroid with a central hole mpo expressing cell (12.5 hours)

Kidney marrow Erythrocytes, granulocytes, lymphocytes and thrombocytes With nucleus (at least 10 days) Thrombocytes (4 days) Segmented nuclei with two or three lobes mpo expressing cells (3.5 days)

Yolk sac (E7.25-E10), AGM, fetal liver (after E9.5) Fetal liver (E9.5) Bone marrow Yolk sac (E7.0) Yolk sac (E9.5), fetal liver (E12.5) and then bone marrow Begins at E8.5 N/A Bone marrow AGM next fetal liver and finally bone marrow Bone marrow Thymus (E10-12)

ALM (~11 hpf) and CHT (~24 hpf)

Definitive myelopoiesis Fetal liver and bone marrow Primitive erythropoiesis

Yolk sac (3-4 weeks) Definitive erythopoiesis Yolk sac (4 weeks) Fetal liver (5-6 weeks) and then bone marrow Circulation Begins at 8 weeks Primitive thrombopoiesis N/A Definitive thrombopoiesis Bone marrow Developmental AGM next fetal liver and HSC finally bone marrow B cells Bone marrow T-cell Thymus maturation (8-9 weeks)

Kidney (~HSC starts seeding at 4 dpf) ICM (~12 hpf) CHT (2-6 dpf) and then kidney marrow (4 dpf) Begins at 24 hpf CHT (~48 hpf) Kidney marrow (~5 dpf) AGM next CHT and finally kidney marrow Kidney marrow Thymus (7 dpf)

AGM: aorta-gonadal-mesonephros; ALM: anterior lateral mesoderm; CHT: caudal hematopoietic tissue; dpf: days post fertilization; ICM: intermediate cell mass.

haematologica | 2019; 104(1)

IBMFS in zebrafish Table 4. Comparison of mouse and zebrafish models for inherited bone marrow failure syndromes.

Disease DBA

Phenotype of mouse model Mouse protein similarity with human protein (%)

Zebrafish protein with similarity with human protein (%)

RPS19 (99%)

Phenotype zebrafish morphant Rps19 (86%)

Rps19 KO: embryonic lethal, heterozygous fully compensated

rps19 mutants. Erythroid defects, compensated for the loss of one Rps19 allele developmental defect and tp53 activation, fully compensated in heterozygous. Decreased HSCs.30,90

rps19 morphant. Severe anemia and developmentala bnormalities. Dysregulation of delta Np63 and tp53.31

Rps19 with ENU-induced missense mutation: embryonic lethality in homozygous. Heterozygous, mild anemia and growth retardation. L-leucine improved the anemia. Rps19 deficiency (transgenic line): anemia, leukopenia and bone marrow failure. Loss of p53 rescued the phenotype.91

RPL11 (100%)

Rpl11 (96%)

Rpl11 KO embryonic lethal. Heterozygous, haploinsufficiency: anemia, decreased erythroid progenitors.92

rps11 mutants. Erythroid defects, developmental defects and tp53 activation. Decrease HSCs.30,36,38

RPS29 (100%)

rpl11 morphant. Morphological defects in the developing brain, small head and eyes and pericardial edema. Upregulation of tp53 and mdm2.35

Rps29 (96%)

N/A

rps29 mutant. Severe anemia and increased rps29 morphant. Defects in red blood cell apoptosis. P53 mutations near completely rescued development and an increase in apoptotic cells.45 rps29 morphological and hematopoietic phenotype.93

RPL5 (98%)

Rpl5 (88%)

Rpl5 KO embryonic lethal. Heterozygous fully compensated.96

N/A

RPS24 (90%)

rpl5 morphant. Primitive and definitive hematopoiesis affected and morphological abnormalities.

Rps24 (87%)

Rps24 KO embryonic lethal. Heterozygous fully compensated.91

N/A

RPL35 (98%)

rps24 morphant. Morphological defects: aplasia in the brain, a bent tail and reduced size. Severe anemia, in a tp53-independent manner.42

Rpl35 (92%)

N/A

rpl35 mutant very high tumor incidence (100%).37,43

RPL14 (94%)

Rpl14 (72%)

Conditional deletion of Rps14 (and 8 other genes): anemia, bone marrow apoptosis.91

rpl14 mutant: high number of tumors (74%)43

RPS7 (100%) V156G

rps7 mutant. Hematopoietic and developmental defect. High tumor incidence (47%).36,37,43

RPL35A (99%)

RPS11 (92%) N/A

rps7 morphant. Impaired hematopoiesis and tp53 activation.40

Rpl35a (90%) N/A

RPS27 (100%) N/A

rpl14 morphant. Severe anemia45 and morphological abnormalities.98

RRps7 (96%) Y177S

Rps7 mutations (RPS7 and RPS7 ): small size, abnormal skeleton and eye malformation. No anemia.94 N/A

rpl35a morphants. Morphological defects: aplasia in the brain, a bent tail and reduced size. Severe anemia, in a tp53-independent manner.42

rpl35a morphants. Morphological defects: aplasia in the brain, a bent tail and reduced size. Severe anemia, in a tp53-independent manner.42,95

Rps27 (98%) N/A

rps27 morphant. Defective erythropoiesis and morphological abnormalities.99

Rps11 (91%) rps11 mutants. Erythroid defects and tp53 activation.

N/A

continued on the next page

haematologica | 2019; 104(1)

U. Oyarbide et al. continued from the previous page

DKC1 (91%)

Dkc1 (80%)

Hypomorphic Dkc1 mutant recapitulate in the first and second generations (G1 and G2) the clinical features of DC.96 Î&#x201D;15 Dkc1 mice: growth retardation, increased DNA damage response via ATM/p53 pathway. 97

N/A

dkc1 morphant. Reduced hematopoiesis, increased tp53 expression, and defective ribosomal biogenesis, no detectable changes in telomerase function.50

NOLA1 (96%)

Nola1 (91%)

N/A

nola1 mutant. Reduced hematopoiesis, increased tp53 expression, and defective ribosomal biogenesis, no detectable changes in telomerase function.50

TERT (62%)

Tert (33%)

Transgenic line over-expressing TERT: short telomeres and increased DNA damage.98

tert mutant. Tissue atrophy, premature death, sarcopenia, impaired cell proliferation and accumulation of senescence cells.55,57

FANCD2 (65%) N/A

fancd2 morphant. Shortened body length, microcephaly and abnormally small eyes, which are due to extensive cellular apoptosis. Upregulation of tp53.63,64

BRCA2 (57%)

Brca2 (41%)

BRCA2 mutant: embryonic lethality

100

brca2 mutants. Genomic instability.

RAD51 (98%)

N/A

Rad51 (93%)

Rad51 mutants. Decreased cell proliferation, embryonic lethal.101

rad51 mutants. Only infertile males, size reduction, hypocellular kidney marrow. Double mutants for Rad51 and P53 rescued HSPC defect but showed higher tumor incidence.66

SBDS (97%) Sbds KO: embryonic lethal.

sbds mutant. Size reduction, liver, pancreas and digestive tract atrophy and reduction of neutrophils.78

MPL (80%) CSF3R (73%) Csf3R KO. Low number of neutrophils in peripheral blood. Expression of truncated Csf3r confers a strong clonal advantage to HSCs.104

sbds morphant. Loss of neutrophils, abnormal skeletal architecture and pancreatic hypoplasia. Sbds knockdown phenotype not rescued by loss of tp53.76,77

Mpl (23%)

c-Mpl KO. Decrease platelets and megakaryocytes mpl mutant. Low number of thrombocytes.83

SCN

N/A

Sbds (87%) 102

CAMT

N/A

Fancd2 (53%)

FancD2 KO: reduced fertility, growth retardation and increased incidence of tumors.99

SDS

N/A

Mpl morphant. Low number of thrombocytes.103

Csf3r (44%) csf3r mutant. Reduction in neutrophils and myeloid cells in the kidney marrow.87

SRP54 (99%) N/A

N/A

Srp54 (95%) N/A

srp54 morphant. Loss of neutrophils and chemotaxis, diminished exocrine pancreas.88

*Morphant: an organism that has been treated with a morpholino antisense to temporarily knockdown the expression of a gene.

An increase in tp53 expression and its target genes, cdkn1a and mdm2, was observed in rpl11 morphants. Genes involved in apoptosis (bik, bax, puma, and noxa) were also up-regulated.35 Danilova et al. demonstrated that developmental and hematopoietic defects, and lower expression of Îą-E1 globin and hbae1.1 in Rps19-deficient fish were mediated by Tp53 upregulation. Upregulation of tp53 also occurred in zebrafish mutants for rps8, rps11 and rps18.31 Danilova et al. used a zebrafish rpl11 mutant to characterize the molecular pathways associated with ribosomal deficiency.36 This mutant showed anemia, decreased 18

HSCs, and activation of the Tp53 pathway with altered expression in genes involved in cell cycle arrest (cdkn1a and ccng1) and apoptosis (bax and puma). Moreover, abnormal regulation of metabolic pathways with a shift from glycolysis to aerobic respiration, upregulation of genes involved in gluconeogenesis and insulin levels, decreased biosynthesis, and increased catabolism were observed. Nucleotide metabolism was affected by upregulation of adenosine deaminase (ada) and xanthine dehydrogenase/oxidase (xdh).33,37 They showed that treatment of mutant embryos with an exogenous supply of haematologica | 2019; 104(1)

IBMFS in zebrafish

nucleosides resulted in downregulation of tp53 and its targets with normalization of ada and xdh levels. Interestingly, DBA patients show increased erythrocyte adenosine deaminase activity.38 Zhang et al. generated two zebrafish mutants using TALENs for rps19 and rps11. The knockout of both rps19 and rps11 resulted in the erythroid defects similar to DBA, such as lack of mature red blood cells (RBCs) and Tp53 activation. The mutants had significantly reduced production of globin proteins accompanied by either increased or unaffected level of mRNA transcripts. Furthermore, they observed decreased HSCs at 3 dpf in rps19 mutants and hemoglobin levels by 4 dpf. The authors concluded that this reduction in RBCs may be caused by a decreased cell survival and/or production of definitive HSCs.30 Similarly, Rowel et al. created a 5 bp deletion in rps19 zebrafish mutant using TALENs. Homozygous rps19 mutants showed developmental anomalies and anemia, and were dead by 5 days post fertilization (dpf). However, rps19 heterozygotes showed no difference to their wild-type siblings. Interestingly, exposure to cold stress during the first dpf resulted in a reduced number of RBCs. To further investigate the biological functions of RPS7, Duan et al. used MO to knockdown rps7 in zebrafish.39 In rps7-deficient embryos, mdm2 and tp53 were activated, inducing the expression of downstream target genes involved in p53 pathway (bik, bax and puma, cdkn1a, and ccng1). rps7 morphants showed severe anemia with reduced expression of gata1 and the mature erythroid marker Îąe3 at 24 hours post fertilization (hpf). A marked suppression of hemoglobin at 48 hpf was observed, indicating that the deficiency of Rps7 might cause abnormal proliferation and/or differentiation of erythroid progenitors. There were also severe defects (short body length, tissue necrosis, and curved tail). Furthermore, simultaneous knockdown of the tp53 by co-injecting a tp53 MO

resulted in partial rescue of morphological abnormalities. The lower levels of gata1 and Îą-E1 globin were partially rescued in the co-injected embryos, even though tp53, cdk1a, and mdm2 were still up-regulated.39 The contribution of tp53 to the pathological development of bone marrow failure syndromes may be tissueand mutation-specific. Antunes et al. studied the effect of different rps7 and rpl11 mutations in zebrafish. rps7 mutant showed a stronger phenotype due to less maternal contribution of rps7 comparing to rpl11 mutant. Both mutants had severe anemia, morphological abnormalities, and increased apoptosis. Injection of p53 MOs rescued the apoptosis and the morphological phenotypes; however, it was unable to rescue anemia.40 Taylor et al. showed that rps29 mutants had defects in RBC development and increased apoptosis. Mutant embryos showed upregulation of tp53 and cdk1a expression. Mutation of tp53 in homozygous rps29 mutant embryos reversed the apoptotic and hematologic phenotypes. However, mutated tp53 did not fully rescue the embryonic lethality of rps29 mutants, suggesting that tp53-independent mechanisms were affected by rps29 knockdown.41 Yadav et al. knocked down five ribosomal protein genes (two DBAassociated, rpl35a and rps24, and three non-DBA-associated, rps3, rpl35 and rplp1), and analyzed these deficiencies on morphology and erythrocyte number in the presence and absence of p53 using MOs. They showed that any ribosomal protein deficiency led to anemia in zebrafish. Elimination of Tp53 function did not significantly affect the anemia, despite improving non-hematopoetic phenotypes.42 DBA zebrafish models have helped identify MDM2-ribosomal protein interactions, which may interfere with MDM2 inhibition to p53 function. p53 rescue of severe anemia in ribosomal protein deficiency zebrafish models varies (Table 6).31,35,43,44 Altogether, these findings suggest that there are p53-independent mechanisms

Table 5. Comparison between morphants and mutants.22-24

Morphants versus mutants Morphant Effect Affects Phenotype Time to create Side effects Genetic compensation p53 pathway

Knock down RNA transcripts More severe maternal mRNA block by MO 1-3 days More off-target effects No Affected

Mutant Permanent changes in DNA Genomic DNA Less severe maternal mRNA 6-8 months Less off target effects Yes Non-affected

Table 6. RP deficiency and p53 rescue in zebrafish models.

Severe anemia

Developmental malformations

Type of p53 rescue

p53 rescue of anemia

p53 rescue of other phenotypes

Ref

Yes N/A Yes Yes Yes Yes

Yes Yes Yes Yes Yes Yes

p53 MO p53 MO p53 MO p53 MO p53M214K p53M214K

No N/A Partial No Yes No

Yes Yes Partial Yes Yes Yes

rps19 morphant rpl11 morphant rps7 morphant Rps7 mutant Rps29 mutant rps24 & rpl35a morphants haematologica | 2019; 104(1)

35 40 36 93 42

U. Oyarbide et al.

involved in bone marrow failure. One p53-independent effect may be translational dysfunction. Zebrafish can provide a model organism to identify Tp53-independent pathways that contribute to marrow failure mice or humans. Zebrafish may also be a valuable model organism for drug development for DBA treatment. Several groups have tested the hypothesis that L-leucine and L-arginine can stimulate translation via the mTOR pathway and rescue affected DBA fish. Treatment of rpl19 and rpl14 zebrafish morphants with L-leucine improved developmental defects and hemoglobin levels.45 Yadav et al. rescued the morphological defects of Rpl35a-deficient embryos and were able to improve erythroid cell number.42 They concluded that translation deficit, not Tp53 activation, is the primary defect perturbing erythropoiesis.42 While there have been anecdotal reports of leucine stimulation of erythropoiesis in DBA patients,46 definitive clinical trial results are still pending. Another study found that RAP-011, an activin receptor ligand trap, partially restored erythropoiesis in rpl11 morphants as well as rpl11 and rpl19 mutants.47 Zebrafish also provided an in vivo model for further drug development of SMER28, a small molecule inducer of ATG5-dependent autophagy.48 Given these results, we await clinical translation of SMER28 as a potential treatment for DBA.

Dyskeratosis congenita Dyskeratosis congenita is associated with abnormal skin pigmentation, nail dystrophy, and oral leukoplakia. DC patients may have other organ involvement, including the pulmonary, gastrointestinal, skeletal, neurological, immunological, and ophthalmological systems. Eightyfive percent of DC patients experience bone marrow failure, which accounts for much of the DC-related mortality. Other causes of mortality include infections, pulmonary complications, and hematologic and non-hematologic malignancies.49-51 Dyskeratosis congenita is a genetically heterogeneous disorder, showing autosomal recessive, autosomal dominant, and X-linked inheritance. So far, at least 21 mutated genes have been identified that can cause DC: DKC1, TERC, TERT, NHP2, NOP10, CTC1, WRD79, TR, NOLA2, NOLA3, PARN, TPP1, POT1, CTC1, USB1, TCAB1, RTEL1, ACD, PARN, WRAP53 and TINF2 (http://telomerase.asu.edu/diseases.html).51-53 The X-linked DKC1 has a more severe phenotype compared with the autosomal dominant forms. Although there is a broad consensus that DC results from stem cell renewal failure due to defective telomere maintenance, some mutated genes (e.g. TERT, TERC, and DKC1) are required for prerRNA processing.2,49,50,54 How telomerase activity and impaired ribosomal biogenesis contribute to the pathophysiology of DC is still not known. Telomeres are complex DNA-protein structures at the end of chromosomes, and they shorten with each cell division. When telomeres become critically short, a DNA damage response is activated, causing cell cycle arrest or death. In humans, telomerase-based telomere elongation is the major mechanism that counteracts this process of continuous telomere shortening. In peripheral white blood cells, rapid telomere shortening occurs within the first year of life, followed by a more gradual decline over time.49 Genetic diseases that cause telomerase deficiency are associated with premature aging and cancer susceptibili20

ty. As in humans, zebrafish chromosomes possess telomeres that progressively decline with age, reaching lengths in old age comparable to those observed when telomerase is mutated.55 Several studies have helped to characterize its well-conserved molecular and cellular physiology. Different zebrafish mutants and morphants for telomere and telomerase research showed shorter lifespan, shorter telomeres, and different affected tissues (mainly brain, blood, gut and testes). These results make zebrafish an excellent model to unravel the connection between telomere shortening, tissue regeneration, aging and disease.55,56 Amsterdam et al. isolated the nop10hi2578 mutant allele where a viral insertion within the first intron resulted in nop10 decreased expression. This mutation is homozygously lethal by 5 dpf.21 nop10 encodes for a protein involved in 18S rRNA processing and is also part of the telomerase complex. Pereboom et al. observed that nop10 loss in this mutant line resulted in a failure of the 18S rRNA to be properly processed, which led to the instability of the 40S ribosomal subunit. Due to the loss of 18S RNA, ribosomal proteins cannot be incorporated into a ribosome subunit and interact with other proteins, including the E3 ubiquitin ligase Mdm2. Mdm2 regulates Tp53 by promoting its ubiquitination and degradation. By binding to Mdm2, Rps7 enhances the E3 ubiquitin ligase activity of Mdm2 that promotes the degradation of Rps7. Furthermore, they observed that an increase in Tp53-specific apoptosis is coupled to the increased binding of Mdm2 to the Rps7. They observed that nop10 mutants failed to form HSCs, a phenotype that is rescued by introducing a loss-of-function tp53 mutation. However, they detected no changes in telomere length in nop10 mutants.53 They concluded that the cytopenia(s) of DC could be the result of ribosome biogenesis defects. This would lead to Tp53-mediated apoptosis of HSCs during early development, caused partially by the association of Rps7 with Mdm2.53 Two different approaches were used by Zhang et al. to study DC in zebrafish. First, MO-mediated knockdown was used to study the mechanisms whereby dkc1 morphants result in HSC failure. Second, they performed retroviral-insertional mutagenesis of nola1. NOLA1 encodes for GAR1, involved in rRNA maturation, and is also a key component telomerase complex. No mutations in NOLA1 have been described in DC patients so far, but suspicion should be aroused in individuals with unexplained marrow failure or fibrosis. Both zebrafish models resulted in reduced hematopoiesis with reduction in runx1 and c-myb, increased tp53 expression, and defective ribosomal biogenesis without detectable changes in telomerase function. Their findings suggest that a telomeraseindependent, Tp53-dependent mechanism contribute to hematopoietic failure in DC.50 Henriques et al. and Anchelin et al. studied the zebrafish telomerase reverse transcriptase tert mutant. These mutants develop normally for the first six months, but progressively develop tissue degeneration (gastrointestinal atrophy, loss of body mass, inflammation, a decrease in total blood cells and cell proliferation), and die prematurely. They also observed a Tp53-dependent response with increased transcripts of puma, cdkn1a, and ccng1a. Upregulation of cell cycle arrest inhibitors led to a G1 arrest and senescence. To study the effect of Tp53 in tert mutants, they created a double mutant tert-/-, tp53-/- and haematologica | 2019; 104(1)

IBMFS in zebrafish

observed rescue of cell proliferation, which partially suppressed the degenerative phenotypes.55,57 In another study, Kishi et al. studied the effect of ablation of terfa; they found multiple malformations mainly in brain, spinal cord, and eye.58 Recently, 2 patients with a phenotype overlapping with DBA and DC (pure red cell aplasia, hypogammaglobulinemia, growth retardation, and microcephaly) harbored a de novo TP53 germline mutation that caused a C-terminal truncation in the last exon. This resulted in enhanced p53mediated transcriptional activity. Using an MO that targets the 3’ splice site of intron 10, Toki et al. developed a zebrafish that displayed reduced number of erythrocytes, severe developmental defects, and died at 96 hpf.7

Fanconi anemia Fanconi anemia is mostly an autosomal recessive condition characterized by congenital abnormalities, progressive bone marrow failure, chromosome fragility, and an early onset of cancers such as myelodysplastic syndromes (MDS) /acute myeloid leukemia (AML) and epithelial malignancies. FA is characterized by non-hematologic phenotype, including short stature, microcephaly, microphthalmia, hypogonadism, and infertility. The mechanisms by which FA leads to developmental anomalies in blood, skeleton, eyes, and gonads are poorly understood; however, genotoxic stress by chemicals, mutagens, and viruses may contribute.59,60 Mutations in at least 20 genes can cause FA. However, since some cases of FA cannot be assigned to any of these genes, additional genes still have to be identified.1,59 Proteins encoded by these genes constitute the FA pathway required for the efficient repair of damaged DNA. The FA core complex consists of at least 8 proteins: FANCA, FANCB, FANCC, FANCE, FANCF, FANCG FANCL, and FANCM. These proteins function as an E3 ligase and mediate the activation of the FANCD2 and FANCI (ID) complex. Once monoubiquitinated, the ID complex interacts with a third group of FANC proteins, including BRCA2 (FANCD1), FANCJ (BRIP1), FANCN (PALB2), FANCO (RAD51C), FANCP (SLX4), BRCA1, FAN1, histone H2AX, and RAD51, thereby contributing to DNA repair via homologous recombination.1,59,61,62 Until now, 20 genes have been associated with causing FA: FANCA, FANCB, FANCC, BRCA2, FANCD2, FANCE, FANCF, FANCG, FANCI, FANCJ, FANCL, FANCN, FANCP, FANCQ, RAD51, BRCA1, FANCT, FANCU, FANCV and FANCW. Information about all these genes is available on the public Fanconi Anemia Mutation Database (http://www.rockefeller.edu/fanconi/). Although zebrafish contain the full complement of FA family members found in humans,63 loss-of-function models have been described for only a few. Liu et al. analyzed the zebrafish ortholog of the human FANCD2 gene using MO.64 They demonstrated developmental defects that arose during embryogenesis after fancd2 knockdown, phenocopying the reduction in body length, and smaller head and eyes, which are frequently observed among FA patients. This suggests that the FA pathway plays a similar role in zebrafish and humans. They showed that the defects in fancd2-deficient embryos were the result of inappropriate and selective activation of Tp53-mediated apoptotic pathways in highly proliferative cells.64 Titus et al. characterized the developmental and tissuespecific expression of FA pathway genes in zebrafish.63 haematologica | 2019; 104(1)

They found maternal deposition of mRNA fanc genes can provide Fanc proteins to repair DNA damage encountered in rapid cleavage divisions. Zebrafish fancl mutants develop only as sterile males but without hematopoietic defects. The sex reversal was due to abnormal increase of germ cell apoptosis that compromises survival of developing oocytes and masculinizes the gonads. Interestingly, when the tp53 mutation was introduced, the sex reversal phenotype could be rescued.65 Botthoff et al. created a rad51 knockout zebrafish mutant. In this model, zebrafish lacking rad51 survived to adulthood, but they were all infertile males with fewer HSPCs in the kidney. In earlier stages (2 and 4 dpf), they found that rad51-/- embryos also had a lower number, increased apoptosis, and reduced proliferation of HSPCs compared with their wild-type siblings. To study the role of p53 in the rad51 mutants, they generated a zebrafish with mutations in both genes. After four months post fertilization, HSPCs were the same in wild-type and double mutants. The sex reversal was also corrected, but neither females nor male double mutants were fertile.66

Shwachman-Diamond syndrome Shwachman-Diamond syndrome is an autosomal recessive disorder characterized by exocrine pancreatic insufficiency, bone marrow dysfunction, and skeletal abnormalities. Hematologic abnormalities are a major cause of morbidity and mortality, and include cytopenia(s), MDS, and AML. Neutropenia occurs in approximately 90% of patients and occurs as early as the neonatal period. Skeletal abnormalities, such as metaphyseal chondrodysplasia, thoracic dystrophy, and short stature are common in SDS. In 2003, mutations in the Shwachman–Bodian-Diamond syndrome (SBDS) gene were identified.70 In approximately 90% of cases, SDS is caused by two common mutations in exon 2 of SBDS: 183-184TA→CT introduces an in-frame stop codon (K62X) and 258+2T>C (C84Cfs) disrupts the donor splice site of intron 2, allowing a hypomorph to be produced.67 Fifty percent of cases are compound heterozygotes with respect to these two mutations. Boocock et al. found that both changes correspond to sequences that occur normally in the pseudogene. Both mutations can also occur in the same allele.67 Studies have identified additional changes in the coding sequence of SBDS that led to frameshift and missense mutations. In 2007, Menne et al. characterized the function of the yeast SBDS ortholog Sdo1 in 60S maturation and translational activation of ribosomes.68 SBDS is a protein with a well-documented role in the later steps of ribosome biogenesis. SBDS interacts with the GTPase EFL1 to trigger release of eIF6 from the 60S ribosomal subunit. EIF6 is critical for biogenesis and nuclear export of pre-60S subunits and prevents ribosomal subunit association. Removal of eIF6 is a prerequisite for the association of the 60S with the 40S subunit, and thus for the formation of an actively functioning ribosome.4 Recently, mutations in DNAJC2169,70 and EFL171 have been identified in individuals with SDS-like conditions. All of the SDS-associated mutant genes affect ribosome maturation. These important discoveries advance the concept of SDS as a ribosomopathy, and beg the question as to how ribosomopathies like DBA, SDS, or del (5q) can result in different defects in hematopoietic and non-hematopoietic tissues. There have been no reports of homozygosity for SBDS null alleles, suggesting that human SBDS is essential and 21

U. Oyarbide et al.

that SDS patients carry at least one hypomorphic SBDS allele.67,72-75 This is consistent with the finding that mice homozygous for null alleles of sbds exhibit early embryonic lethality, indicating that SBDS function is an essential for life.30 While conditional knock-outs for sbds have been made, this approach is limited, costly, and time-consuming to generate. Thus, we and others have turned to the zebrafish also to study SDS. Venkatasubramani and Mayer used MO to knockdown sbds in zebrafish embryos, and study the effect in pancreas and myeloid development (Table 4). They observed an alteration in the spatial relationship between endocrine and exocrine pancreas. They also documented abnormal neutrophil distribution in the knockdown zebrafish model.76 In a subsequent study, also using MO, Provost et al. observed that their model fully recapitulated the spectrum of developmental abnormalities observed in SDS patients: loss of neutrophils, skeletal defects, and pancreatic hypoplasia, as well as changes in the ribosomal subunit ratio. In this case, loss of Tp53 did not rescue the developmental defects associated with loss of sbds in zebrafish morphants.77 Our recent work showed that sbds mutants obtained by CRISPR/Cas9 editing phenocopied SDS and displayed neutropenia, growth retardation, and atrophy of the pancreas.78

Congenital amegakaryocytic thrombocytopenia Congenital amegakaryocytic thrombocytopenia is a rare autosomal recessive condition characterized by thrombocytopenia, absence of megakaryocytes, and occasional evolution to aplastic anemia or leukemia.79,80 Mutations in MPL have been described as the cause of CAMT.81 MPL gene encodes for myeloproliferative leukemia protein (CD110), the receptor for thrombopoietin. Mice with genetic ablation of Mpl showed normal development but a deficiency in megakaryocytes and severe thrombocytopenia.82 In zebrafish, disruption of mpl caused a severe reduction in thrombocytes (platelet equivalents), bleeding, and a decrease in HSCs. By phenocopying the human disease, affected zebrafish provide an accurate model to study this disease and for drug screening.83 Reduction in HSCs and repopulation defects in affected zebrafish demonstrate that c-Mpl function in hematopoiesis is highly conserved. Moreover, the partial rescue of thrombocyte number by IL-11 provides a model to finely dissect JAK/STAT signaling in thrombopoiesis.

Severe congenital neutropenia Severe congenital neutropenia is a group of heterogeneous genetic disorders characterized by a maturation arrest at the promyelocyte stage of granulopoiesis and a high propensity to develop MDS/AML.84 Over the past eighteen years, the following mutations have been identi-

References 1. Rosenberg PS, Alter BP, Ebell W. Cancer risks in Fanconi anemia: findings from the German Fanconi Anemia Registry. Haematologica. 2008;93(4):511-517. 2. Alter BP, Giri N, Savage SA, Rosenberg PS. Cancer in dyskeratosis congenita. Blood. 2009;113(26):6549-6557. 3. Da Costa L, O'Donohue MF, van

fied as causing SCN: ELANE, GFI1, HAX1, VPS45, JAGN, CSF3R, and WAS. ELANE is the most commonly mutated gene in SCN, but there is no zebrafish ortholog. However, zebrafish has proven to be a powerful model to validate and characterize the function of newly described gene candidates for SCN. Vacuolar Protein Sorting 45 Homolog (VPS45) encodes a protein associated with protein trafficking into distinct organelles. Biallelic mutations in this gene are the cause of SCN5. A zebrafish model of vps45 knockdown also showed a large decrease in neutrophils.85 Mutations in CSF3R cause SCN7.86 Pazhakh et al. mutated csf3r in zebrafish to study the effect on neutrophil production. They found that csf3r zebrafish mutants survive until adulthood with a 50% reduction in neutrophils and a substantial reduction in myeloid cells in the kidney marrow.87 Recently, SRP54 mutations have been identified as the second most common cause of SCN (with some features of SDS).88,89 Knockdown of SRP54 in zebrafish recapitulated the human phenotype of neutropenia, chemotaxis defect, and pancreatic exocrine insufficiency.88

Conclusions Despite the identification of specific gene mutations and pathway involvement for the great majority of patients with IBMFS, little is known about how they result in single or multiple lineage cytopenias. Furthermore, very little is known about co-operating mutations that effect transformation to MDS, AML, or solid tumors. Patient-based studies are problematic owing to the rarity of these disorders and to the long latency before bone marrow failure or malignancy. Zebrafish provide a relatively inexpensive, rapidly developing, vertebrate model organism. Despite some differences in their respective hematopoietic organs, mutations or silencing of relevant zebrafish genes phenocopies human IBMFS. Studies on gene mutations or suppression in zebrafish have validated the role of ribosome biogenesis, and advanced the hypothesis that the TP53 pathway plays a major role in the pathophysiology of some of the IBMFS. Zebrafish modeling may also contribute to drug development, as suggested by studies on Lleucine and SMER28 for DBA. Acknowledgments SJC is supported by funding from NIH R01 HL128173, NIH R21 CA159203, Department of Defense Bone Marrow Failure Idea Development Award BM140102, Shwachman-Diamond Syndrome Foundation, Connorâ&#x20AC;&#x2122;s Heroes, and the CURE Childhood Cancer Foundation. Due to space restrictions, the authors deeply regret not being able to cite all of our colleaguesâ&#x20AC;&#x2122; publications.

Dooijeweert B, et al. Molecular approaches to diagnose Diamond-Blackfan anemia: The EuroDBA experience. Eur J Med Genet. 2017 Oct 26. [Epub ahead of print] 4. Finch AJ, Hilcenko C, Basse N, et al. Uncoupling of GTP hydrolysis from eIF6 release on the ribosome causes Shwachman-Diamond syndrome. Genes Dev. 2011;25(9):917-929. 5. Carrillo J, Gonzalez A, Manguan-Garcia C, Pintado-Berninches L, Perona R. p53 path-

way activation by telomere attrition in XDC primary fibroblasts occurs in the absence of ribosome biogenesis failure and as a consequence of DNA damage. Clin Transl Oncol. 2014;16(6):529-538. 6. Ceccaldi R, Parmar K, Mouly E, et al. Bone marrow failure in Fanconi anemia is triggered by an exacerbated p53/p21 DNA damage response that impairs hematopoietic stem and progenitor cells. Cell Stem Cell. 2012;11(1):36-49.

haematologica | 2019; 104(1)

IBMFS in zebrafish 7. Toki T, Yoshida K, Wang R, et al. De Novo Mutations Activating Germline TP53 in an Inherited Bone-Marrow-Failure Syndrome. Am J Hum Genet 2018;103(3):440-447. 8. Glaubach T, Minella AC, Corey SJ. Cellular stress pathways in pediatric bone marrow failure syndromes: many roads lead to neutropenia. Pediatr Res. 2014;75(1-2):189-195. 9. Garaycoechea JI, Crossan GP, Langevin F, Daly M, Arends MJ, Patel KJ. Genotoxic consequences of endogenous aldehydes on mouse haematopoietic stem cell function. Nature. 2012;489(7417):571-575. 10. Howe K, Clark MD, Torroja CF, et al. The zebrafish reference genome sequence and its relationship to the human genome. Nature. 2013;496(7446):498-503. 11. Veldman MB, Lin S. Zebrafish as a Developmental Model Organism for Pediatric Research. Pediatr Res. 2008;64(5):470-476. 12. Teittinen KJ, Grönroos T, Parikka M, Rämet M, Lohi O. The zebrafish as a tool in leukemia research. Leuk Res. 2012;36(9): 1082-1088. 13. Davidson AJ, Zon LI. The ‘definitive’ (and ‘primitive’) guide to zebrafish hematopoiesis. Oncogene. 2004;23(43): 7233-7246. 14. de Jong JLO, Zon LI. Use of the Zebrafish System to Study Primitive and Definitive Hematopoiesis. Annu Rev Genet. 2005;39(1):481-501. 15. Palis J. Primitive and definitive erythropoiesis in mammals. Front Physiol. 2014;5:3. 16. Carradice D, Lieschke GJ. Zebrafish in hematology: sushi or science? Blood. 2008;111(7):3331-3342. 17. Chen AT, Zon LI. Zebrafish blood stem cells. J Cell Biochem. 2009;108(1):35-42. 18. Ellett F, Lieschke GJ. Zebrafish as a model for vertebrate hematopoiesis. Curr Opin Pharmacol. 2010;10(5):563-570. 19. Bowman TV, Zon LI. Lessons from the Niche for Generation and Expansion of Hematopoietic Stem Cells. Drug Discov Today Ther Strateg 2009;6(4):135-140. 20. Solnica-Krezel L, Schier AF, Driever W. Efficient Recovery of Enu-Induced Mutations from the Zebrafish Germline. Genetics. 1994;136(4):1401-1420. 21. Amsterdam A, Burgess S, Golling G, et al. A large-scale insertional mutagenesis screen in zebrafish. Genes Dev. 1999;13(20):27132724. 22. Stainier DYR, Raz E, Lawson ND, et al. Guidelines for morpholino use in zebrafish. PLoS Genet. 2017;13(10):e1007000. 23. Robu ME, Larson JD, Nasevicius A, et al. p53 Activation by Knockdown Technologies. PLoS Genet. 2007;3(5):e78. 24. Kok FO, Shin M, Ni CW, et al. Reverse genetic screening reveals poor correlation between morpholino-induced and mutant phenotypes in zebrafish. Dev Cell. 2015;32(1):97-108. 25. Rossi A, Kontarakis Z, Gerri C, et al. Genetic compensation induced by deleterious mutations but not gene knockdowns. Nature. 2015;524(7564):230-233. 26. Chiabrando D, Tolosano E. Diamond Blackfan Anemia at the crossroad between ribosome biogenesis and heme metabolism. Adv Hematol. 2010;2010:790632. 27. Giri N, Kang E, Tisdale JF, et al. Clinical and laboratory evidence for a trilineage haematopoietic defect in patients with refractory Diamond-Blackfan anaemia. Br J Haematol. 2000;108(1):167-175. 28. Landowski M, O'Donohue MF, Buros C, et al. Novel deletion of RPL15 identified by

haematologica | 2019; 104(1)

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43. 44.

array-comparative genomic hybridization in Diamond-Blackfan anemia. Hum Genet. 2013;132(11):1265-1274. Da Costa L, Narla A, Mohandas N. An update on the pathogenesis and diagnosis of Diamond–Blackfan anemia. F1000Res. 2018;7. Zhang Y, Ear J, Yang Z, Morimoto K, Zhang B, Lin S. Defects of protein production in erythroid cells revealed in a zebrafish Diamond–Blackfan anemia model for mutation in RPS19. Cell Death Dis. 2014;5(7):e1352. Danilova N, Sakamoto KM, Lin S. Ribosomal protein S19 deficiency in zebrafish leads to developmental abnormalities and defective erythropoiesis through activation of p53 protein family. Blood. 2008;112(13):5228-5237. Uechi T, Nakajima Y, Chakraborty A, Torihara H, Higa S, Kenmochi N. Deficiency of ribosomal protein S19 during early embryogenesis leads to reduction of erythrocytes in a zebrafish model of DiamondBlackfan anemia. Hum Mol Genet. 2008;17(20):3204-3211. Danilova N, Bibikova E, Covey TM, et al. The role of the DNA damage response in zebrafish and cellular models of Diamond Blackfan anemia. Dis Model Mech. 2014;7(7):895-905. Torihara H, Uechi T, Chakraborty A, Shinya M, Sakai N, Kenmochi N. Erythropoiesis failure due to RPS19 deficiency is independent of an activated Tp53 response in a zebrafish model of Diamond-Blackfan anaemia. Br J Haematol. 2011;152(5):648654. Chakraborty A, Uechi T, Higa S, Torihara H, Kenmochi N. Loss of Ribosomal Protein L11 Affects Zebrafish Embryonic Development through a p53-Dependent Apoptotic Response. PLoS One. 2009;4(1):e4152. Amsterdam A, Nissen RM, Sun Z, Swindell EC, Farrington S, Hopkins N. Identification of 315 genes essential for early zebrafish development. Proc Natl Acad Sci USA. 2004;101(35):12792-12797. Danilova N, Sakamoto KM, Lin S. Ribosomal protein L11 mutation in zebrafish leads to haematopoietic and metabolic defects. Br J Haematol. 2011;152(2): 217-228. Danilova N, Gazda HT. Ribosomopathies: how a common root can cause a tree of pathologies. Dis Model Mech. 2015;8 (9):1013-1026. Duan J, Ba Q, Wang Z, et al. Knockdown of ribosomal protein S7 causes developmental abnormalities via p53 dependent and independent pathways in zebrafish. Int J Biochem Cell Biol. 2011;43(8):1218-1227. Antunes AT, Goos YJ, Pereboom TC, et al. Ribosomal Protein Mutations Result in Constitutive p53 Protein Degradation through Impairment of the AKT Pathway. PLoS Genet. 2015;11(7):e1005326. Taylor AM, Humphries JM, White RM, Murphey RD, Burns CE, Zon LI. Hematopoietic defects in rps29 mutant zebrafish depend upon p53 activation. Exp Hematol. 2012;40(3):228-237.e5. Yadav GV, Chakraborty A, Uechi T, Kenmochi N. Ribosomal protein deficiency causes Tp53-independent erythropoiesis failure in zebrafish. Int J Biochem Cell Biol. 2014;49:1-7. Amsterdam A, Sadler KC, Lai K, et al. Many Ribosomal Protein Genes Are Cancer Genes in Zebrafish. PLoS Biol. 2004;2(5):e139. Zhang Y, Lu H. Signaling to p53: ribosomal

45.

46.

47.

48.

49.

50.

51. 52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

proteins find their way. Cancer Cell. 2009;16(5):369-377. Payne EM, Virgilio M, Narla A, et al. LLeucine improves the anemia and developmental defects associated with DiamondBlackfan anemia and del(5q) MDS by activating the mTOR pathway. Blood. 2012;120(11):2214-2224. Pospisilova D, Cmejlova J, Hak J, Adam T, Cmejla R. Successful treatment of a Diamond-Blackfan anemia patient with amino acid leucine. Haematologica. 2007;92(5):e66-e67. Ear J, Huang H, Wilson T, et al. RAP-011 improves erythropoiesis in zebrafish model of Diamond-Blackfan anemia through antagonizing lefty1. Blood. 2015;126(7):880890. Doulatov S, Vo LT, Macari ER, et al. Drug discovery for Diamond-Blackfan anemia using reprogrammed hematopoietic progenitors. Sci Transl Med. 2017;9(376). Du HY, Pumbo E, Ivanovich J, et al. TERC and TERT gene mutations in patients with bone marrow failure and the significance of telomere length measurements. Blood. 2008;113(2):309-316. Zhang Y, Morimoto K, Danilova N, Zhang B, Lin S. Zebrafish Models for Dyskeratosis Congenita Reveal Critical Roles of p53 Activation Contributing to Hematopoietic Defects through RNA Processing. PLoS One. 2012;7(1):e30188. Ruggero D, Shimamura A. Marrow failure: a window into ribosome biology. Blood. 2014;124(18):2784-2792. Ballew BJ, Yeager M, Jacobs K, et al. Germline mutations of regulator of telomere elongation helicase 1, RTEL1, in Dyskeratosis congenita. Hum Genet. 2013;132(4):473-480. Pereboom TC, van Weele LJ, Bondt A, MacInnes AW. A zebrafish model of dyskeratosis congenita reveals hematopoietic stem cell formation failure resulting from ribosomal protein-mediated p53 stabilization. Blood. 2011;118(20):5458-5465. Freed EF, Bleichert F, Dutca LM, Baserga SJ. When ribosomes go bad: diseases of ribosome biogenesis. Mol Biosyst. 2010;6(3): 481-493. Henriques CM, Carneiro MC, Tenente IM, Jacinto A, Ferreira MG. Telomerase Is Required for Zebrafish Lifespan. PLoS Genet. 2013;9(1):e1003214. Carneiro MC, de Castro IP, Ferreira MG. Telomeres in aging and disease: lessons from zebrafish. Dis Model Mech. 2016;9(7):737748. Anchelin M, Alcaraz-Perez F, Martinez CM, Bernabe-Garcia M, Mulero V, Cayuela ML. Premature aging in telomerase-deficient zebrafish. Dis Model Mech. 2013;6(5):11011112. Kishi S, Bayliss PE, Uchiyama J, et al. The Identification of Zebrafish Mutants Showing Alterations in SenescenceAssociated Biomarkers. PLoS Genet. 2008;4(8):e1000152. Rodríguez-Marí A, Postlethwait JH. The Role of Fanconi Anemia/BRCA Genes in Zebrafish Sex Determination. The Zebrafish: Disease Models and Chemical Screens: Elsevier BV; 2011:461-490. Soulier J. Detection of somatic mosaicism and classification of Fanconi anemia patients by analysis of the FA/BRCA pathway. Blood. 2004;105(3):1329-1336. Geiselhart A, Lier A, Walter D, Milsom MD. Disrupted Signaling through the Fanconi Anemia Pathway Leads to Dysfunctional

U. Oyarbide et al.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

Hematopoietic Stem Cell Biology: Underlying Mechanisms and Potential Therapeutic Strategies. Anemia. 2012; 2012:1-18. Dong H, Nebert DW, Bruford EA, Thompson DC, Joenje H, Vasiliou V. Update of the human and mouse Fanconi anemia genes. Hum Genomics. 2015;9:32. Titus TA, Selvig DR, Qin B, et al. The Fanconi anemia gene network is conserved from zebrafish to human. Gene. 2006;371 (2):211-223. Liu TX, Howlett NG, Deng M, et al. Knockdown of zebrafish Fancd2 causes developmental abnormalities via p53dependent apoptosis. Dev Cell. 2003;5(6):903-914. Rodríguez-Marí A, Cañestro C, BreMiller RA, et al. Sex Reversal in Zebrafish fancl Mutants Is Caused by Tp53-Mediated Germ Cell Apoptosis. PLoS Genet. 2010;6(7): e1001034. Botthof JG, Bielczyk-Maczynska E, Ferreira L, Cvejic A. Loss of the homologous recombination gene rad51 leads to Fanconi anemia-like symptoms in zebrafish. Proc Natl Acad Sci USA. 2017;114(22):E4452-e4461. Boocock GR, Morrison JA, Popovic M, et al. Mutations in SBDS are associated with Shwachman-Diamond syndrome. Nat Genet. 2003;33(1):97-101. Menne TF, Goyenechea B, Sánchez-Puig N, et al. The Shwachman-Bodian-Diamond syndrome protein mediates translational activation of ribosomes in yeast. Nature Genet. 2007;39(4):486-495. Dhanraj S, Matveev A, Li H, et al. Biallelic mutations in DNAJC21 cause ShwachmanDiamond syndrome. Blood. 2017;129(11): 1557-1562. Tummala H, Walne AJ, Williams M, et al. DNAJC21 Mutations Link a Cancer-Prone Bone Marrow Failure Syndrome to Corruption in 60S Ribosome Subunit Maturation. Am J Hum Genet. 2016;99 (1):115-124. Stepensky P, Chacon-Flores M, Kim KH, et al. Mutations in EFL1, an SBDS partner, are associated with infantile pancytopenia, exocrine pancreatic insufficiency and skeletal anomalies in aShwachman-Diamond like syndrome. J Med Genet. 2017;54(8):558566. Erdos M, Alapi K, Balogh I, et al. Severe Shwachman-Diamond syndrome phenotype caused by compound heterozygous missense mutations in the SBDS gene. Exp Hematol. 2006;34(11):1517-1521. Kuijpers TW, Alders M, Tool AT, Mellink C, Roos D, Hennekam RC. Hematologic abnormalities in Shwachman Diamond syndrome: lack of genotype-phenotype relationship. Blood. 2005;106(1):356-361. Makitie O, Ellis L, Durie PR, et al. Skeletal phenotype in patients with ShwachmanDiamond syndrome and mutations in SBDS. Clin Genet. 2004;65(2):101-112. Dror Y, Donadieu J, Koglmeier J, et al. Draft consensus guidelines for diagnosis and treat-

76.

77.

78.

79. 80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

ment of Shwachman-Diamond syndrome. Ann N Y Acad Sci. 2011;1242:40-55. Venkatasubramani N, Mayer AN. A Zebrafish Model for the ShwachmanDiamond Syndrome (SDS). Pediatr Res. 2008;63(4):348-352. Provost E, Wehner KA, Zhong X, et al. Ribosomal biogenesis genes play an essential and p53-independent role in zebrafish pancreas development. Development. 2012;139(17):3232-3241. Oyarbide U, Kell MJ, Farinas J, Topczewski J, Corey S. Gene disruption of zebrafish Sbds phenocopies human ShwachmanDiamond Syndrome but suggests more global and lineage defects. Blood. 2016;128(22):336. Ballmaier M. c-mpl mutations are the cause of congenital amegakaryocytic thrombocytopenia. Blood. 2001;97(1):139-146. Rose MJ, Nicol KK, Skeens MA, Gross TG, Kerlin BA. Congenital amegakaryocytic thrombocytopenia: the diagnostic importance of combining pathology with molecular genetics. Pediatr Blood Cancer. 2008;50(6):1263-1265. Ihara K, Ishii E, Eguchi M, et al. Identification of mutations in the c-mpl gene in congenital amegakaryocytic thrombocytopenia. Proc Natl Acad Sci USA. 1999;96(6):3132-3136. Alexander WS, Roberts AW, Nicola NA, Li R, Metcalf D. Deficiencies in progenitor cells of multiple hematopoietic lineages and defective megakaryocytopoiesis in mice lacking the thrombopoietic receptor c-Mpl. Blood. 1996;87(6):2162-2170. Lin Q, Zhang Y, Zhou R, et al. Establishment of a congenital amegakaryocytic thrombocytopenia model and a thrombocyte-specific reporter line in zebrafish. Leukemia. 2017;31(5):1206-1216. Skokowa J, Germeshausen M, Zeidler C, Welte K. Severe congenital neutropenia: inheritance and pathophysiology. Curr Opin Hematol. 2007;14(1):22-28. Vilboux T, Lev A, Malicdan MCV, et al. A Congenital Neutrophil Defect Syndrome Associated with Mutations in VPS45. N Engl J Med. 2013;369(1):54-65. Ward AC, van Aesch YM, Gits J, et al. Novel point mutation in the extracellular domain of the granulocyte colony-stimulating factor (G-CSF) receptor in a case of severe congenital neutropenia hyporesponsive to G-CSF treatment. J Exp Med. 1999;190(4):497-507. Pazhakh V, Clark S, Keightley MC, Lieschke GJ. A GCSFR/CSF3R zebrafish mutant models the persistent basal neutrophil deficiency of severe congenital neutropenia. Sci Rep. 2017;7:44455. Carapito R, Konantz M, Paillard C, et al. Mutations in signal recognition particle SRP54 cause syndromic neutropenia with Shwachman-Diamond-like features. J Clin Invest. 2017;127(11):4090-4103. Bellanné-Chantelot C, Schmaltz-Panneau B, Marty C, et al. Mutations in SRP54 gene cause severe congenital neutropenia as well

as Shwachman-Diamond-like syndrome. Blood. 201;132(12):1318-1331. 90. Rowell J, Pietka G, Virgilio M, Pena O, Hockings C, Payne E. A Zebrafish Model of Diamond-Blackfan Anemia Results in Bone Marrow Failure and Demonstrates Defective Translation in Erythroid Cells By Ribosome Footprinting. Blood. 2017;130 (Suppl 1):871. 91. McGowan KA, Mason PJ. Animal models of Diamond Blackfan Anemia. Semin Hematol. 2011;48(2):106-116. 92. Morgado-Palacin L, Varetti G, Llanos S, Gomez-Lopez G, Martinez D, Serrano M. Partial Loss of Rpl11 in Adult Mice Recapitulates Diamond-Blackfan Anemia and Promotes Lymphomagenesis. Cell Rep. 2015;13(4):712-722. 93. Taylor AM, Humphries JM, White RM, Murphey RD, Burns CE, Zon LI. Hematopoietic defects in rps29 mutant zebrafish depend upon p53 activation. Exp Hematol. 2012;40(3):228-237.e5. 94. Watkins-Chow DE, Cooke J, Pidsley R, et al. Mutation of the diamond-blackfan anemia gene Rps7 in mouse results in morphological and neuroanatomical phenotypes. PLoS Genet. 2013;9(1):e1003094. 95. Uechi T, Nakajima Y, Nakao A, et al. Ribosomal protein gene knockdown causes developmental defects in zebrafish. PLoS One. 2006;1:e37. 96. Ruggero D, Grisendi S, Piazza F, et al. Dyskeratosis congenita and cancer in mice deficient in ribosomal RNA modification. Science. 2003;299(5604):259-262. 97. Gu BW, Bessler M, Mason PJ. A pathogenic dyskerin mutation impairs proliferation and activates a DNA damage response independent of telomere length in mice. Proc Natl Acad Sci USA. 2008;105(29):10173-10178. 98. Jaskelioff M, Muller FL, Paik J-H, et al. Telomerase reactivation reverses tissue degeneration in aged telomerase deficient mice. Nature. 2011;469(7328):102-106. 99. Parmar K, D'Andrea A, Niedernhofer LJ. Mouse models of Fanconi anemia. Mutat Res. 2009;668(1-2):133-140. 100. Rodriguez-Mari A, Postlethwait JH. The role of Fanconi anemia/BRCA genes in zebrafish sex determination. Methods Cell Biol. 2011;105:461-490. 101. Lim DS, Hasty P. A mutation in mouse rad51 results in an early embryonic lethal that is suppressed by a mutation in p53. Mol Cell Biol. 1996;16(12):7133-7143. 102. Zhang S, Shi M, Hui CC, Rommens JM. Loss of the mouse ortholog of the Shwachman-Diamond syndrome gene (Sbds) results in early embryonic lethality. Mol Cell Biol. 2006;26(17):6656-6663. 103. Lim KH, Chang YC, Chiang YH, et al. Expression of CALR mutants causes mpldependent thrombocytosis in zebrafish. Blood Cancer J. 2016;6(10):e481. 104. Liu F, Kunter G, Krem MM, et al. Csf3r mutations in mice confer a strong clonal HSC advantage via activation of Stat5. J Clin Invest. 2008;118(3):946-955.

haematologica | 2019; 104(1)

REVIEW ARTICLE

Advances in risk assessment and prophylaxis for central nervous system relapse in diffuse large B-cell lymphoma

Ferrata Storti Foundation

David Qualls and Jeremy S. Abramson

Center for Lymphoma, Massachusetts General Hospital Cancer Center, Boston, MA, USA

ABSTRACT

Haematologica 2019 Volume 104(1):25-34

entral nervous sytem recurrence of diffuse large B-cell lymphoma is an uncommon but devastating event, making identification of patients at high risk for relapse within the central nervous system essential for clinicians. Modern risk stratification includes both clinical and biological features. A validated clinical risk model employing the five traditional International Prognostic Index risk factors plus renal or adrenal involvement can identify a high-risk patient population with a central nervous system recurrence risk of greater than 10%. Lymphoma involvement of certain discrete extranodal sites such as the testis also confers increased risk, even in stage I disease. Adverse biological risk factors for central nervous system relapse include presence of translocations of MYC, BCL2 and/or BCL6, in so-called double- or triple-hit lymphoma. Immunohistochemically detectable co-expression of MYC and BCL2 in the absence of translocations also portends an increased risk of relapse within the central nervous system, particularly in the setting of the activated B-cell-like subtype of diffuse large B-cell lymphoma. The role, method, and timing of prophylactic therapy remain controversial based on the available data. We review both intrathecal and systemic strategies for prophylaxis in high-risk patients. Our preference is for systemic methotrexate in concert with standard chemoimmunotherapy in the majority of cases. Several novel agents have also demonstrated clinical activity in primary and secondary central nervous system lymphoma and warrant future investigation in the prophylactic setting.

Correspondence: jabramson@mgh.harvard.edu

Received: July 25, 2018. Accepted: November 15, 2018. Pre-published: December 20, 2018.

Introduction

doi:10.3324/haematol.2018.195834

Diffuse large B-cell lymphoma (DLBCL) is the most common adult non-Hodgkin lymphoma, accounting for approximately one-third of all newly diagnosed cases in the United States.1,2 The prognosis has improved substantially since the introduction of rituximab nearly 20 years ago, with 5-year overall survival rates of approximately 70% depending on the baseline characteristics of the patients and their disease.3-5 Despite improvements in outcomes, a minority of patients with DLBCL will still suffer relapse within the central nervous system (CNS), which carries nearly universally poor outcomes with a median survival following diagnosis of CNS involvement of only 2-5 months.6-10 Indeed, secondary CNS lymphoma has long represented a great unmet medical need within the field of oncology, a need that has been particularly difficult to address since CNS involvement is usually an exclusion criterion for participation in clinical trials of novel agents. Given the significant morbidity and mortality associated with this event, much attention has been devoted to the identification of high-risk patients, and evaluation of therapies to mitigate the risk of CNS recurrence. Another important area of investigation involves the diagnosis and treatment of occult leptomeningeal disease: lymphoma detected by cerebrospinal fluid (CSF) cytology or flow cytometry, without overt clinical signs or symptoms of CNS involvement by lymphoma. Occult disease has been shown to be significantly asso-

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/25

haematologica | 2019; 104(1)

ÂŠ2019 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.

D. Qualls et al.

ciated with CNS relapse risk, as well as mortality.11-15 Few data are available to inform the optimnal intensity of therapy required to eradicate occult CNS disease, but our practice is to treat occult CNS involvement by lymphoma in the same way as we treat active secondary CNS lymphoma. The rate and patterns of CNS involvement with DLBCL have evolved with the introduction of rituximab-containing therapy. Most studies have found a slight decrease in the incidence of CNS relapse with rituximab use, with modern rates of 2-4%.7,16-19 This decrease is likely due to superior control of systemic disease, as well as a benefit from minimal penetration of the rituximab antibody in the CSF.7,9,20 Accordingly, localization of CNS relapse in the modern era has shifted to the brain parenchyma in the majority of cases, whereas relapse in the leptomeningeal compartment predominated prior to the introduction of rituximab.7,20-21 Relapses within the CNS generally occur early in the treatment course, often presenting prior to completion of initial therapy or shortly thereafter.16,19 A significant proportion (14-48%) of patients with CNS relapse also have systemic relapse at the time of diagnosis.11,12,22,23 While isolated CNS relapse may be prevented with effective CNS prophylaxis, concomitant systemic and CNS relapse likely represents a failure of systemic treatment,7,16 and patients with concurrent CNS and systemic relapse appear to have a worse prognosis than those with CNS relapse alone.24 Current studies do not reliably differentiate between isolated and concomitant systemic/CNS relapse when exploring the efficacy of CNS prophylaxis, so this remains an important area for future investigation. CNS-directed prophylactic therapy has been widely utilized in DLBCL. This practice is based largely on the established benefit in other high-grade lymphomas with a high risk of CNS involvement, particularly Burkitt lymphoma and acute lymphoblastic leukemia.25,26 Data demonstrating a benefit of prophylactic strategies in DLBCL, however, have been limited and occasionally conflicting. This is largely due to the low overall risk of CNS events, as well as reliance on retrospective analyses and underpowered subset analyses of prospective studies, which suffer from small sample sizes and significant heterogeneity in indications and methods for prophylaxis. As a result, selection of appropriate patients for CNS prophylaxis, as well as the type and timing of prophylactic strategies, remain highly controversial. In this review, we attempt to consolidate current information and examine the most recent advances regarding CNS risk assessment and approaches to CNS prophylaxis in patients with DLBCL.

Evaluating risk of central nervous system recurrence Optimal CNS prophylaxis in DLBCL relies first upon accurately identifying the small proportion of high-risk patients who should be targeted for CNS evaluation and intervention. Modern risk stratification includes both traditional clinical and laboratory assessments, as well as incorporation of pathological and molecular characteristics of the patientâ&#x20AC;&#x2122;s disease.

Clinical risk models The risk of CNS involvement in patients with DLBCL is concentrated in high-risk populations with certain patientand disease-specific characteristics. Studies performed in 26

the early 2000s demonstrated that patients with high International Prognostic Index (IPI) scores were at greater risk of CNS involvement.10,27-29 The original IPI consisted of five risk factors: age >60 years, elevated lactate dehydrogenase, Eastern Cooperative Oncology Group (ECOG) Performance Status >1, advanced stage disease, and involvement of more than one extranodal site. More recently, the German High Grade NHL Study Group (DSHNHL) and British Columbia Cancer Agency (BCCA) developed and validated the CNS International Prognostic Index (CNS-IPI) as a risk stratification tool to predict risk of CNS recurrence.18 They examined an initial cohort consisting of 2164 patients enrolled in DSHNHL clinical trials and identified five risk factors for CNS disease based on multivariate analysis: these include four of the five original IPI risk factors (with the exception of >1 extranodal site), plus either kidney or adrenal involvement. The presence of multiple extranodal sites was likely not a significant risk factor in multivariable analysis because of the overlap with advanced stage, but it was nonetheless retained in the final six-factor model for ease of application. Patients are stratified as having low (0-1 points), intermediate (2-3 points), or high (4-6 points) risk disease, which predicted 2-year rates of CNS relapse of 0.8%, 3.9%, and 12%, respectively (Figure 1). The model was validated on 1597 patients in the BCCA database. Notably, additional risk factors for CNS recurrence were identified in the validation cohort based on multivariable analysis, which likely reflect differences in the populations of patients between the two cohorts; additional risk factors included >1 extranodal sites, and involvement of the testis, pericardium, orbit, or bone marrow. Subsequent studies have confirmed the utility of the CNS-IPI, particularly in validating risk associated with the highest risk cohort with four to six risk factors.17,19,30

High-risk extranodal sites Certain extranodal sites have been implicated as discrete risk factors for subsequent CNS relapse, though few have been found to be independently predictive of CNS involvement on multivariable analyses. Testicular involvement in DLBCL is the most well-established, with studies demonstrating CNS relapse rates of 12-25%, even in stage I completely resected disease.31-34 Unlike other systemic DLBCL which typically recur early, primary testicular DLBCL can relapse in the CNS as late as 10 years after initial diagnosis, and occurs most commonly within the brain parenchyma.31,32 There appears to be a pathophysiological relationship between primary testicular DLBCL and primary CNS DLBCL. Both diseases tend to be activated B-cell (ABC)-like by transcriptional profiling, and share genetic features including oncogenic toll-like receptor signaling based on MYD88 mutations or NFKBIZ amplification, B-cell receptor pathway activation, and BCL-6 deregulation.35 Altered adhesion molecule expression also likely contributes to the predilection of primary testicular DLBCL for immune privileged sites such as the testis and CNS, while the loss of HLA-DR and surface immunoglobulin and increased expression of PD-L1 and PD-L2 via 9p24.1 amplification contribute to immune escape.35,36 Additional sites of extranodal involvement have historically been associated with greater risk of CNS relapse, although on multivariate analysis these have not been consistently predictive in the modern era. These sites haematologica | 2019; 104(1)

CNS prophylaxis in DLBCL

include the bone marrow, paranasal sinus, orbit, pericardium, ovary, uterus, and breast.37-41 Involvement of paranasal sinuses was identified as a significant risk in the pre-rituximab era, but recent data have demonstrated that the risk of CNS events is not increased in the era of rituximabbased therapy, and there is not a clear role for either intrathecal prophylaxis or consolidative radiation therapy in these patients.42 Kidney and adrenal involvement had previously been identified as high-risk locations, and are now included in the CNS-IPI, as previously discussed.43,44

oped. The use of pretreatment positron emission tomography to predict CNS relapse has been proposed, with elevated total lesion glycolysis found to be predictive of an increased risk of CNS relapse on multivariable analysis.53 Additional biomarkers emerging from pathological analyses have been utilized to stratify risk, including ITGA10, CXCR5 and nuclear PTEN.54,55 Further research in these areas in concert with existing biomarkers and clinical risk stratification tools should be performed to determine the clinical utility of these studies before they are incorporated into routine clinical care.

Biological risk factors As our understanding of the molecular pathogenesis of DLBCL has advanced, specific biological features have become increasingly important predictors of CNS risk. DLBCL harboring a MYC rearrangement has been associated with an increased risk of CNS recurrence and an inferior overall survival relative to other forms of DLBCL.45,46 A MYC rearrangement rarely occurs as a sole genetic abnormality, however, and the dominant prognostic impact appears to be conferred by the rearrangement in concert with additional genetic aberrations.45,46 High-grade B-cell lymphoma with translocations of MYC and BCL2 and/or BCL6, also known as double- or triple-hit lymphoma, represents about 5% of all newly diagnosed large B-cell lymphomas and carries a poor prognosis with a median overall survival of less than 2 years.47,48 CNS involvement is common at either diagnosis or relapse, and has been reported in as many as 50% of affected patients.48-50 Compared to routine DLBCL therapy, more aggressive initial chemoimmunotherapy regimens such as dose-adjusted EPOCH-R (etoposide, prednisolone, vincristine, cyclophosphamide, doxorubicin plus rituximab) are typically employed, along with intrathecal CNS prophylaxis. This is based primarily on retrospective analyses which have demonstrated improved progression-free survival with more aggressive induction regimens and have suggested improved survival for patients receiving intrathecal therapy in this disease in which CNS recurrences commonly involve the leptomeningeal compartment.47 Another area of interest has been DLBCL with immunohistochemically detectable expression of MYC and BCL2 without associated translocations, otherwise called double-expressing lymphoma. Dual protein expression is significantly more common than double-hit lymphoma, occurring in approximately 30% of cases of DLBCL.51,52 A retrospective analysis of double-expressing lymphoma by the BCCA found a CNS recurrence risk of approximately 9%, compared to only 2% in non-double-expressing cases of DLBCL.51 This risk was modified, however, based on cell of origin, and by risk stratification according to the CNS-IPI score. The risk of CNS events in double-expressing lymphoma appears limited to the ABC-like subset of DLBCL in which the CNS relapse risk is approximately 15%, while there is no apparent increased risk in doubleexpressing germinal center B-cell (GCB)-like disease. Risk also appears confined to the intermediate- and high-risk CNS-IPI patients in whom the CNS relapse rate approximates 12% and 22%, respectively, without any increased risk among low-risk CNS-IPI patients with doubleexpressing lymphoma. These analyses reinforce the complexity of assigning risk of CNS relapse in DLBCL, which warrants attention to clinical, histopathological, and molecular factors for optimal risk estimation. Novel methods for risk stratification are being develhaematologica | 2019; 104(1)

Baseline central nervous system evaluation CNS recurrence of DLBCL typically occurs early in the disease course, either during systemic treatment or within several months of completing treatment.8,28 This suggests that subclinical involvement of the CNS by DLBCL is likely present at the time of diagnosis in such cases. Early identification of patients with CNS involvement is crucial, as the treatment and prognosis may be significantly altered based on this knowledge. As such, evaluation of the CNS via CSF analysis and/or neuroimaging should be considered in patients with high-risk features or neurological symptoms. CSF analysis consists of conventional cytology and flow cytometry. Cytology alone has low sensitivity for CNS disease at less than 60% for leptomeningeal disease and virtually no ability to detect parenchymal lymphoma.56 The addition of flow cytometry significantly increases sensitivity for detection of occult CNS involvement, which can be found in approximately 10% of high-risk patients and is associated with a high rate of subsequent CNS progression and poor overall survival.1115 An important caveat is that these data include patients predominantly treated in the pre-rituximab era, at a time when most CNS relapses involved the leptomeningeal compartment.28,57,58 Since the introduction of rituximabbased chemoimmunotherapy, the incidence of CNS

Figure 1. Kaplan-Meier curve depicting risk of central nervous system relapse based on the Central Nervous System International Prognostic Index score. DSHNHL: German High Grade Lymphoma Study Group cohort, BCCA: British Columbia Cancer Agency cohort. Schmitz, et al. J Clin Oncol 2016.18

D. Qualls et al.

relapses has decreased and these relapses more commonly involve the brain parenchyma where CSF evaluation provides a lower diagnostic yield.7,20,21 As such, routine evaluation of the CSF in patients without neurological symptoms in the modern era remains controversial. Our own practice is to evaluate the CSF at baseline in all patients with neurological symptoms, in patients with disease infiltrating neural foramina, and in most patients with double- or triple-hit lymphoma, as these patients are at particularly high risk and often relapse within the leptomeninges. For high-risk asymptomatic DLBCL patients without double- or triple-hit cytogenetics who we are treating with intrathecal methotrexate for CNS prophylaxis, we evaluate their baseline CSF at the time of their initial intrathecal injection. For high-risk patients receiving CNS prophylaxis with high-dose systemic methotrexate, we consider whether a finding of an occult positive CSF would alter the patient’s treatment plan as CNS-directed therapy is already planned. Patients with relative contraindications to CNS prophylaxis, or less clear indications for prophylaxis in whom the risk/benefit ratio is not as well defined, may benefit from CSF analysis to assist in clinical decision-making. An alternative approach is to perform baseline CNS evaluation with CSF cytology and flow cytometry in all patients considered at high-risk of CNS relapse. Our practice has shifted away from this strategy in all highrisk patients as the rate of leptomeningeal relapse has declined in the rituximab era, decreasing the yield of broadly applied CSF testing. Furthermore, it remains unclear whether additional CNS-directed therapy for occult disease is needed if these patients are already receiving empiric systemic high-dose methotrexate as CNS prophylaxis. For patients receiving intrathecal rather than systemic CNS-directed chemotherapy, however, the finding of occult lymphoma within the CSF should prompt intensification of the intrathecal treatment regimen beyond what would be administered for prophylaxis alone. Ultimately the decision of whether to perform baseline CSF evaluation should be personalized to the patient based on that person’s discrete CNS risk factors, symptomatology, medical comorbidities, and treatment plan. One potential area for future research is the use of polymerase chain reaction to evaluate the presence of occult CNS disease. Recent studies in primary CNS lymphoma have shown that assessments for micro-RNA (miRNA) and U2 small nuclear RNA fragments (RNU21f) via polymerase chain reaction were able to detect primary CNS DLBCL with high sensitivity and specificity.59,60 Studies performed in other cancers with CNS involvement have also demonstrated the utility of nextgeneration sequencing of CSF cell-free DNA in detecting and characterizing CNS disease.61,62 While similar studies still need to be performed in secondary CNS lymphoma, it is possible that utilizing these techniques could further improve our ability to assess occult CNS involvement. At initial diagnosis, a careful history and neurological examination should be performed in all patients. The presence of any neurological signs or symptoms warrants magnetic resonance imaging evaluation of the brain and/or spine based on the relevant clinical finding. In the absence of neurological signs or symptoms of concern, there are insufficient data to recommend routine baseline neuroimaging. 28

Central nervous system prophylaxis Despite the widespread use of CNS prophylaxis in patients determined to be at high risk of CNS recurrence, its efficacy remains controversial. There are no randomized controlled trials designed specifically to determine whether prophylactic strategies reduce CNS events. Most relevant data, therefore, come from subset analyses of clinical trials which are not powered to determine the impact of CNS prophylaxis, and from retrospective analyses that are susceptible to selection and reporting biases.63 Interpretation of these data is further confounded by significantly varying protocols with different indications, timing, dosing, and chemotherapeutic agents employed for prophylaxis.

Indications for central nervous system prophylaxis Ultimately the goal of CNS prophylaxis is to minimize the incidence of CNS relapse, while allocating such therapy to those at highest risk and sparing those with low-risk disease unnecessary toxicity. As discussed above, there is a spectrum of risk associated with specific disease features and patients’ characteristics, and the threshold for use of prophylaxis varies from clinician to clinician. Based on the aforementioned discussion of biological and clinical risk stratification, prophylactic CNS-directed therapy should be considered for nearly all patients with double- or triple-hit lymphoma, DLBCL patients with a high-risk CNS-IPI score (4-6 risk factors), and intermediate-risk patients (2-3 risk factors) who are ABC-subtype with dual expression of MYC and BCL2. We also recommend CNS prophylaxis in patients with disease in selected high-risk anatomic locations, including primary testicular DLBCL, orbital disease involving the globe or posterior compartment, and disease directly infiltrating spinal neuroforamina. CNS prophylaxis in patients with multiple other discrete extranodal locations remains more controversial based on available data and should be personalized in the context of the overall clinical and biological risk factors for the patient.

Intrathecal therapy Intrathecal chemotherapy, particularly methotrexate, has been the most widely employed method of prophylaxis. Despite extensive data available, however, a protective benefit favoring this approach for prevention of CNS relapse has never been established, either before or after the introduction of rituximab. Three prospective studies in the rituximab era found no benefit from intrathecal methotrexate among patients defined as high risk. The RICOVER-60 trial conducted by the DSHNHL compared CHOP-14 (cyclophosphamide, doxorubicin, vincristine and prednisolone at 14-day intervals) to R-CHOP-14 (CHOP-14 plus rituximab) in patients over 60 years of age and recommended CNS prophylaxis with intrathecal methotrexate in patients with involvement of the testes, head or upper neck. Among 1222 subjects, 273 received at least one cycle of intrathecal methotrexate. Notably, only 57% of patients on this trial who met criteria for CNS prophylaxis actually received it, perhaps reflecting a lack of enthusiasm for this approach by treating investigators. When comparing CNS recurrence rates within this targeted population based on administration of prophylaxis, no significant preventive benefit could be identified.16 These findings were replicathaematologica | 2019; 104(1)

CNS prophylaxis in DLBCL

ed in a broader analysis of 2210 patients treated on additional prospective clinical trials by the DSHNHL, 620 of whom received rituximab-based treatment: intrathecal methotrexate again yielded no reduction in risk of CNS events.64 Similarly, a randomized controlled trial comparing RCHOP-14 versus R-CHOP-21 (cyclophosphamide, doxorubicin, vincristine and prednisolone plus rituximab at 21day intervals) included 984 patients, of whom 177 received CNS prophylaxis with the vast majority (92%) receiving intrathecal methotrexate. When stratified based on CNS-IPI, patients treated with intrathecal methotrexate had similar rates of CNS relapse, progression-free and overall survival compared with those not given prophylaxis.19 Several additional retrospective analyses involving large numbers of patients treated with rituximab-based chemoimmunotherapy in the modern era have likewise failed to demonstrate an association between intrathecal methotrexate use and reduction in CNS relapse rates.22,65-67 The lack of clinical benefit observed with intrathecal methotrexate may be explained by the pharmacokinetics of this drug. Historic experiments found that methotrexate concentrations within the neuroaxis varied widely between different patients when the drug was administered via lumbar puncture. In one study, two of nine patients given intrathecal methotrexate did not meet the target therapeutic concentration at any time, and five of the nine did not sustain therapeutic concentrations for 24 hours.68 Another study monitoring the distribution of radionuclide Indium showed that it could take up to 24 hours for intrathecal injections of Indium to appear in the ventricles, suggesting that intrathecal therapy injected at the lumbar sac may fail to protect the cerebral leptomeninges due to uneven distribution.69 Furthermore, intrathecal methotrexate fails to achieve therapeutic concentrations within the brain parenchyma,70 which could lead to reduced efficacy in this site where the majority of CNS relapses occur in the rituximab era.6,7,20,21 It is important to note the significant limitations of studies evaluating the efficacy of intrathecal methotrexate. While available data including non-randomized prospective and retrospective studies have not reliably demonstrated lower CNS relapse rates in patients receiving intrathecal therapy, these data are heterogeneous and cannot be considered definitive in their conclusions. Only appropriately powered randomized trials can truly exclude the possibility that intrathecal methotrexate reduces the risk of CNS relapse, although such trials would be extremely difficult to conduct. In cases in which alternative therapies, such as high-dose intravenous methotrexate, cannot be administered, intrathecal methotrexate remains a reasonable option. One scenario in which intrathecal methotrexate remains an appropriate standard therapy is when administered with the dose-adjusted EPOCH-R regimen in patients with Burkitt lymphoma or high grade B-cell lymphoma (including double-hit and triple-hit lymphoma), in which it has been the exclusively studied method of CNS protection. Intrathecal methotrexate has been demonstrated to improve the clinical outcome in these histological types of lymphoma which frequently relapse in the CSF.47,71

Systemic chemotherapy Effective systemic therapy which crosses the bloodbrain barrier may overcome the liabilities of intrathecal haematologica | 2019; 104(1)

therapy and achieve even and predictable concentrations throughout the entire neuroaxis, including both the leptomeningeal and parenchymal compartments.72 Consideration of systemic CNS prophylaxis is derived largely from experience in primary CNS DLBCL, in which high-dose systemic methotrexate improves progressionfree and overall survival and remains the standard backbone of first-line treatment.73,74 The efficacy of systemic methotrexate as prophylactic therapy for the CNS has also been validated in acute lymphoblastic leukemia75 and Burkitt lymphoma76,77 in which it remains an accepted standard of care. These results have been corroborated in DLBCL with the phase III GELA trial comparing CHOP-21 (cyclophosphamide, doxorubicin, vincristine and prednisolone at 21day intervals) against the intensive ACVBP regimen (doxorubicin, cyclophosphamide, vindesine, bleomycin, prednisone induction followed by sequential consolidation therapy) in patients with aggressive non-Hodgkin lymphoma and IPI >1.58 The ACVBP arm included four doses of intrathecal methotrexate, plus two infusions of highdose systemic methotrexate at 3000 mg/m2. Results of the trial were notable for significantly fewer CNS recurrences in the ACVBP arm compared to the CHOP arm (2.7% versus 8%; P=0.004), as well as an overall survival benefit. Greater systemic disease control with the more intensive ACVBP regimen may well account for some of the observed benefit over CHOP, but the lower rate of isolated CNS relapse suggests that CNS prophylaxis may also have played an important role. Given the lack of appreciable benefit in numerous prior studies evaluating intrathecal methotrexate alone, it is reasonable to consider that the intravenous methotrexate contributed to the reduction in the rate of CNS recurrence. We described our retrospective experience adding systemic high-dose methotrexate to R-CHOP as CNS prophylaxis in selected high-risk patients with DLBCL.80 Patients received at least one dose of intravenous methotrexate at a dose of 3500 mg/m2 administered on day 15 of alternating cycles (i.e. cycles 2, 4, and 6) of R-CHOP. The population had a significant proportion of high-risk patients with IPI scores of 3-5 in 68%, elevated lactate dehydrogenase concentration in 73%, more than one extranodal site of involvement in 62% of subjects, and frequent involvement of high-risk locations including the kidneys, adrenal glands, testes, bone marrow, or the epidural space. Among 65 high-risk patients, two CNS recurrences (3%) occurred (one at 4 months and the other at 9 months). The median follow-up for the entire population was 33 months, and the 3-year progression-free survival was 76%. Toxicities noted within this population included 26 patients (39%) with creatinine elevation above the upper limit of normal, although only one patient required temporary hemodialysis and subsequently recovered renal function. Renal toxicity led to discontinuation of methotrexate in nine patients (14%), all of whom recovered baseline renal function. In eight patients (12%) the subsequent R-CHOP cycle had to be delayed by 1-3 weeks because of toxicity (nephrotoxicity in 4, mucositis in 2, and cytopenias in 2). Despite these adverse events, the study demonstrated that patients with normal baseline renal function could tolerate high-dose methotrexate treatment intercalated with R-CHOP therapy, and that this was associated with a lower rate of CNS relapse than may be expected based on their high-risk features at baseline.18 29

D. Qualls et al.

Two other retrospective studies evaluated high-dose methotrexate for CNS prophylaxis in DLBCL. In one Italian center, high-dose methotrexate (with or without intrathecal liposomal cytarabine at the discretion of the treating physician) was administered after completion of all cycles of R-CHOP for three or four cycles in patients deemed at high risk of CNS recurrence.66 These patients were then retrospectively compared to patients with highrisk features treated with no CNS prophylaxis. At a median of 60 months, 12% of patients who did not receive prophylaxis had had a CNS relapse versus 2.5% of those who received prophylaxis (P=0.03), suggesting that CNS prophylaxis was beneficial. Of note, there were differences in risk factors between the two populations with more patients being defined as having high-risk disease due to advanced stage and elevated lactate dehydrogenase concentration in the group that received no prophylaxis, while high-risk anatomic locations including testis, kidney and orbit were enriched in the prophylaxed population. Such differences in patient selection complicate the interpretation of all retrospective analyses, so that conclusions can be considered suggestive but not definitive. That said, a third retrospective analysis reported concordant results with lower rates of CNS relapse in patients treated with a

combination of high-dose intravenous methotrexate and intrathecal methotrexate compared to intrathecal methotrexate alone with a hazard ratio for CNS relapse at 3 years of 0.26 (95% confidence interval: 0.08 â&#x20AC;&#x201C; 0.81) based on multivariate analysis.79 Two prospective trials have incorporated high-dose methotrexate and cytarabine, in addition to other CNSactive agents, for high-risk patients with DLBCL. A phase II trial of R-CODOX-M/IVAC (cyclophosphamide, vincristine, doxorubicin, methotrexate, ifosfamide, etoposide, and cytarabine) was performed in patients with newly diagnosed DLBCL and an IPI score of â&#x2030;Ľ3.80 Among 96 patients with no CNS involvement at diagnosis, 41 had CNS-IPI scores of 2-3 (intermediate risk) and 55 had CNSIPI scores of 4-6 (high risk); the rates of CNS relapse in these groups at 2 years were 0% and 6%, respectively, which are lower than might have been predicted without CNS-directed therapy, although the concomitant toxicity of these intensive regimens must be taken into account. The Nordic Lymphoma Study Group performed a phase II study in patients with high-risk DLBCL or grade 3 follicular lymphoma, with age-adjusted IPI scores of 2-3.23 Treatment consisted of six cycles of R-CHOEP-14 (RCHOP-14 plus etoposide) followed by cytarabine at 3000

Figure 2. Suggested approach to central nervous system risk stratification and prophylaxis in newly diagnosed diffuse large B-cell lymphoma. DLBCL: diffuse large B-cell lymphoma; CNS-IPI: Central Nervous System International Prognostic Index; DHL: double-hit lymphoma; COO: cell of origin; ABC: activated B-cell; DEL: doubleexpressing lymphoma; THL: triple-hit lymphoma; LP: lumbar puncture; MRI: magnetic resonance imaging; CNS: central nervous system; CSF: cerebrospinal fluid; MTX: methotrexate; CrCl: creatine clearance; HD-MTX: high-dose methotrexate; IT-MTX: intrathecal methotrexate.

haematologica | 2019; 104(1)

CNS prophylaxis in DLBCL

mg/m2 twice daily for 2 days, followed 3 weeks later by methotrexate at a dose of 3000 mg/m2. Among 156 patients, there were three toxicity-related deaths. The rate of CNS relapse was 4.5%, and all relapses occurred within 6 months of diagnosis; this CNS recurrence rate is comparable to that in historical controls. It is interesting to note, however, that all relapses occurred very early in the disease course and the CNS-active agents were administered only following induction therapy. This supports the hypothesis that occult CNS involvement may already have been present early in the disease course, and that earlier administration of systemic methotrexate may have conferred greater benefit. The optimal timing of CNS prophylaxis, particularly with high-dose intravenous methotrexate, remains incompletely elucidated. CNS relapse typically occurs during or within 5-6 months of induction treatment.8,28 The incidence of early CNS events forms the rationale for introduction of CNS-directed therapy concurrently with systemic induction treatment, but this must be balanced with risks of toxicity due to concomitant therapy. For high-risk DLBCL patients who are appropriate candidates for highdose systemic methotrexate, we favor inclusion at a dose of 3500 mg/m2 on day 15 of the 21-day R-CHOP cycle for up to a total of three doses, usually administered in alternating R-CHOP cycles. An alternative schedule is to administer three or four doses of intravenous methotrexate immediately following completion of R-CHOP, although this risks earlier CNS progression. Methotrexate can be administered at 10- to 14-day intervals when given as monotherapy following completion of R-CHOP. A recommended algorithm for risk stratification and prophylaxis is shown in Figure 2. Safe administration of high-dose systemic methotrexate for CNS prophylaxis relies upon careful attention to the selection and supportive care of patients. This treatment should be avoided in patients with a poor performance status, and those with impaired renal function or significant effusions or ascites which may serve as reservoirs for methotrexate and prolong toxicity. In order to minimize risk of toxicity, patients are pre-treated with hydration and alkalinization, which continues after methotrexate infusion to accelerate clearance. Methotrexate at a dose of 3000-3500 mg/m2 is typically administered over 2-4 h, with leucovorin rescue commencing 24 h after the beginning of the methotrexate infusion, and continuing every 6 h for 12-16 doses as the methotrexate clears. It is essential to monitor methotrexate levels along with electrolytes and renal function in order to ensure that the drug is cleared rapidly, which helps to avoid toxicity. There are no data to support systemic prophylaxis with chemotherapeutic agents other than methotrexate at this time. Cytarabine has activity as a single agent and in combination with high-dose methotrexate in primary CNS lymphoma, but has not been validated to add benefit as prophylaxis in systemic DLBCL.74,81 Etoposide is a widely used lymphoma therapy which may attain cytotoxic concentrations in the CSF and had historically been associated with a reduced risk of CNS recurrence, raising the prospect of benefit in the prophylactic setting.8,82 Subsequent studies of etoposide with rituximab-containing therapy in the modern era, however, found no significant benefit in reducing CNS risk, so etoposide also cannot be recommended as a component of prophylactic therapy.18,83 haematologica | 2019; 104(1)

Two novel agents, ibrutinib and lenalidomide, have demonstrated activity in relapsed DLBCL, particularly in ABC-like DLBCL which characterizes most cases of primary and secondary CNS lymphoma. Additionally, ibrutinib appears especially promising in ABC-like DLBCL harboring both MYD88 and CD79B mutations, a mutational pattern commonly observed in primary CNS DLBCL.84-86 Given this biological rationale, ibrutinib and lenalidomide have both been preliminarily investigated in primary CNS DLBCL in which they have demonstrated the ability to cross the blood-brain barrier and induce remissions.87,88 Based on these findings, BTK inhibitors and lenalidomide warrant evaluation in the therapy of secondary CNS lymphoma as well. Whether incorporation of one or both of these novel agents into upfront therapy in high-risk patients will reduce the risk of CNS relapse remains unknown, but will likely be elucidated by randomized trials adding these agents to R-CHOP in ABC-like DLBCL, which have been completed and await reporting. Two phase II trials in which lenalidomide was added to RCHOP included 136 patients with CNS-IPI intermediateand high-risk scores present in 71.3% and 18.4%, respectively.89 Prophylactic intrathecal methotrexate was employed in only 14% of patients. At a median follow-up of 48 months, only one of the 136 patients had experienced a CNS relapse, which is a promising early result. In addition to evaluating these agents in patients with active CNS DLBCL, studies will be helpful in determining the potential benefit of these agents in preventing CNS relapse, and in determining which patients could derive the most benefit from these novel therapies. The advent of immune checkpoint inhibitors in the treatment of solid tumors and Hodgkin lymphoma90 has garnered interest in their use for DLBCL. While PD-L1 expression is uncommon in systemic DLBCL,91 higher rates of PD-L1 expression have been noted in primary mediastinal large B-cell lymphoma,92 primary CNS DLBCL, and primary testicular DLBCL.35 A small case series including four patients with relapsed/refractory primary CNS DLBCL and one patient with CNS relapse of primary testicular DLBCL showed clinical and radiographic responses in all patients treated with the PD-1 inhibitor nivolumab.93 These early data warrant further investigation to determine whether select subsets of high-risk DLBCL patients may benefit from immune checkpoint inhibition to reduce the risk of CNS relapse.

Conclusions In patients with DLBCL, relapse within the CNS remains a rare but devastating complication. There are significant limitations to determining the optimal methods of risk stratification and prophylaxis against CNS relapse, including the infrequency of the event, heterogeneity of existing literature, and inability to enroll sufficient numbers of patients in appropriately powered clinical trials with the primary outcome of CNS relapse. As a result, recommendations and guidelines remain largely empiric in nature. Based on the available data and clinical experience, the optimal approach is to first consider patient- and diseasespecific risk factors and identify patients at highest risk for CNS relapse. Proper risk stratification should include calculation of patientsâ&#x20AC;&#x2122; CNS-IPI score, consideration of extranodal sites of disease, and identification of disease-specific 31

D. Qualls et al.

biological factors including double- or triple-hit translocations and double-expresser status in concert with determination of the cell of origin. Baseline CNS evaluation may include CSF studies with cytology and flow cytometry based on patient-specific risk factors, and imaging of the neuroaxis if neurological signs or symptoms are present. For those patients at high risk of CNS relapse, prophylactic therapy should be considered. Our preference is to employ systemic methotrexate as first-line CNS prophylaxis if the patient is an appropriate candidate, with intrathecal methotrexate reserved for high-risk patients

References 12. 1. Morton LM, Wang SS, Devesa SS, et al. Lymphoma incidence patterns by WHO subtype in the United States, 1992-2001. Blood. 2006;107(1):265-276. 2. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. 3. Ziepert M, Hasenclever D, Kuhnt E, et al. Standard International prognostic index remains a valid predictor of outcome for patients with aggressive CD20+ B-cell lymphoma in the rituximab era. J Clin Oncol. 2010;28(14):2373-2380. 4. Smith A, Crouch S, Howell D, et al. Impact of age and socioeconomic status on treatment and survival from aggressive lymphoma: a UK population-based study of diffuse large B-cell lymphoma. Cancer Epidemiol. 2015;39(6):1103-1112. 5. Sehn LH, Berry B, Chhanabhai M, et al. The revised International Prognostic Index (RIPI) is a better predictor of outcome than the standard IPI for patients with diffuse large Bcell lymphoma treated with R-CHOP. Blood. 2007;109(5):1857-1861. 6. Ghose A, Elias HK, Guha G, et al. Influence of rituximab on central nervous system relapse in diffuse large B-cell lymphoma and role of prophylaxis--a systematic review of prospective studies. Clin Lymphoma Myeloma Leuk. 2015;15(8):451-457. 7. Zhang J, Chen B, Xu X. Impact of rituximab on incidence of and risk factors for central nervous system relapse in patients with diffuse large B-cell lymphoma: a systematic review and meta-analysis. Leuk Lymphoma. 2014;55(3):509-514. 8. Boehme V, Zeynalova S, Kloess M, et al. Incidence and risk factors of central nervous system recurrence in aggressive lymphoma-a survey of 1693 patients treated in protocols of the German High-Grade NonHodgkin's Lymphoma Study Group (DSHNHL). Ann Oncol. 2007;18(1):149-157. 9. Bjorkholm M, Hagberg H, Holte H, et al. Central nervous system occurrence in elderly patients with aggressive lymphoma and a long-term follow-up. Ann Oncol. 2007;18(6):1085-1089. 10. Bernstein SH, Unger JM, Leblanc M, et al. Natural history of CNS relapse in patients with aggressive non-Hodgkin's lymphoma: a 20-year follow-up analysis of SWOG 8516 -- the Southwest Oncology Group. J Clin Oncol. 2009;27(1):114-119. 11. Wilson WH, Bromberg JE, Stetler-Stevenson M, et al. Detection and outcome of occult leptomeningeal disease in diffuse large B-cell

13.

14.

15.

16.

17.

18.

19.

20.

who are ineligible for systemic methotrexate because of renal dysfunction or other comorbidities. Early administration of CNS prophylaxis, during the course of initial therapy, may treat occult CNS disease and prevent early treatment failure. Finally, new targeted therapies and immune-modulating agents may provide novel opportunities for the treatment and prevention of CNS relapse in the future. Inclusion of patients with secondary CNS lymphoma in clinical trials of promising agents would significantly accelerate the rate of progress in this currently unmet medical need.

lymphoma and Burkitt lymphoma. Haematologica, 2014;99(7):1228-1235. Hegde U, Filie A, Little RF, et al. High incidence of occult leptomeningeal disease detected by flow cytometry in newly diagnosed aggressive B-cell lymphomas at risk for central nervous system involvement: the role of flow cytometry versus cytology. Blood. 2005;105(2):496-502. Benevolo G, Stacchini A, Spina M, et al. Final results of a multicenter trial addressing role of CSF flow cytometric analysis in NHL patients at high risk for CNS dissemination. Blood. 2012;120(16):3222-3228. Alvarez R, Dupuis J, Plonquet A, et al. Clinical relevance of flow cytometric immunophenotyping of the cerebrospinal fluid in patients with diffuse large B-cell lymphoma. Ann Oncol. 2012;23(5):12741279. Sancho JM, Orfao A, Quijano S, et al. Clinical significance of occult cerebrospinal fluid involvement assessed by flow cytometry in non-Hodgkin's lymphoma patients at high risk of central nervous system disease in the rituximab era. Eur J Haematol. 2010;85(4):321-328. Boehme V, Schmitz N, Zeynalova S, et al. CNS events in elderly patients with aggressive lymphoma treated with modern chemotherapy (CHOP-14) with or without rituximab: an analysis of patients treated in the RICOVER-60 trial of the German HighGrade Non-Hodgkin Lymphoma Study Group (DSHNHL). Blood. 2009;113(17): 3896-3902. El-Galaly TC, Villa D, Michaelsen TY, et al. The number of extranodal sites assessed by PET/CT scan is a powerful predictor of CNS relapse for patients with diffuse large B-cell lymphoma: An international multicenter study of 1532 patients treated with chemoimmunotherapy. Eur J Cancer. 2017;75:195-203. Schmitz N, Zeynalova S, Nickelsen M, et al. CNS International Prognostic Index: a risk model for CNS relapse in patients with diffuse large B-cell lymphoma treated with RCHOP. J Clin Oncol. 2016;34(26):3150-3156. Gleeson M, Counsell N, Cunningham D, et al. Central nervous system relapse of diffuse large B-cell lymphoma in the rituximab era: results of the UK NCRI R-CHOP-14 versus 21 trial. Ann Oncol. 2017;28(10):2511-2516. Guirguis HR, Cheung MC, Mahrous M, et al. Impact of central nervous system (CNS) prophylaxis on the incidence and risk factors for CNS relapse in patients with diffuse large B-cell lymphoma treated in the rituximab era: a single centre experience and review of the literature. Br J Haematol. 2012;159(1):39-49.

21. Mitrovic Z, Bast M, Bierman PJ, et al. The addition of rituximab reduces the incidence of secondary central nervous system involvement in patients with diffuse large Bcell lymphoma. Br J Haematol. 2012;157 (3):401-403. 22. Kumar A, Vanderplas A, LaCasce AS, et al. Lack of benefit of central nervous system prophylaxis for diffuse large B-cell lymphoma in the rituximab era: findings from a large national database. Cancer. 2012;118 (11):2944-2951. 23. Holte H, Leppa S, Bjorkholm M, et al. Dosedensified chemoimmunotherapy followed by systemic central nervous system prophylaxis for younger high-risk diffuse large Bcell/follicular grade 3 lymphoma patients: results of a phase II Nordic Lymphoma Group study. Ann Oncol. 2013;24(5):13851392. 24. Uni M, Kagoya Y, Nannya Y, et al. Central nervous system relapse in patients with diffuse large B-cell lymphoma: analysis of incidence and prognostic factors. Leuk Lymphoma. 2015;56(6):1869-1871. 25. Blum KA, Lozanski G, Byrd JC. Adult Burkitt leukemia and lymphoma. Blood. 2004;104(10):3009-3020. 26. Cortes J, O'Brien SM, Pierce S, et al. The value of high-dose systemic chemotherapy and intrathecal therapy for central nervous system prophylaxis in different risk groups of adult acute lymphoblastic leukemia. Blood. 1995;86(6):2091-2097. 27. Haioun C, Besson C, Lepage E, et al. Incidence and risk factors of central nervous system relapse in histologically aggressive non-Hodgkin's lymphoma uniformly treated and receiving intrathecal central nervous system prophylaxis: a GELA study on 974 patients. Ann Oncol. 2017;11(6):685-690. 28. Hollender A, Kvaloy S, Nome O, et al. Central nervous system involvement following diagnosis of non-Hodgkin's lymphoma: a risk model. Ann Oncol. 2002;13 (7):1099-1107. 29. A predictive model for aggressive nonHodgkin's lymphoma. N Engl J Med. 1993; 329(14):987-994. 30. Tomita N, Yokoyama M, Yamamoto W, et al. The standard international prognostic index for predicting the risk of CNS involvement in DLBCL without specific prophylaxis. Leuk Lymphoma. 2018;59(1):97-104. 31. Kridel R, Telio D, Villa D, et al. Diffuse large B-cell lymphoma with testicular involvement: outcome and risk of CNS relapse in the rituximab era. Br J Haematol. 2017;176 (2):210-221. 32. Zucca E, Conconi A, Mughal TI, et al. Patterns of outcome and prognostic factors in primary large-cell lymphoma of the testis

haematologica | 2019; 104(1)

CNS prophylaxis in DLBCL

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

in a survey by the International Extranodal Lymphoma Study Group. J Clin Oncol. 2003;21(1):20-27. Mazloom A, Fowler N, Medeiros LJ, et al. Outcome of patients with diffuse large Bcell lymphoma of the testis by era of treatment: the M. D. Anderson Cancer Center experience. Leuk Lymphoma. 2010;51(7): 1217-1224. Fonseca R, Habermann TM, Colgan JP, et al. Testicular lymphoma is associated with a high incidence of extranodal recurrence. Cancer. 2000;88(1):154-161. 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. Deng L, Xu-Monette ZY, Loghavi S, et al. Primary testicular diffuse large B-cell lymphoma displays distinct clinical and biological features for treatment failure in rituximab era: a report from the International PTL Consortium. Leukemia. 2016;30(2):361-372. Yao Z, Deng L, Xu-Monette ZY, et al. Concordant bone marrow involvement of diffuse large B-cell lymphoma represents a distinct clinical and biological entity in the era of immunotherapy. Leukemia. 2018;32(2):353-363. Laskin JJ, Savage KJ, Voss N, et al. Primary paranasal sinus lymphoma: natural history and improved outcome with central nervous system chemoprophylaxis. Leuk Lymphoma. 2005;46(12):1721-1727. Lee GW, Go SI, Kim SH, et al. Clinical outcome and prognosis of patients with primary sinonasal tract diffuse large B-cell lymphoma treated with rituximab-cyclophosphamide, doxorubicin, vincristine and prednisone chemotherapy: a study by the Consortium for Improving Survival of Lymphoma. Leuk Lymphoma. 2015;56(4): 1020-1026. El-Galaly TC, Cheah CY, Hutchings M, et al. Uterine, but not ovarian, female reproductive organ involvement at presentation by diffuse large B-cell lymphoma is associated with poor outcomes and a high frequency of secondary CNS involvement. Br J Haematol. 2016;175(5):876-883. Aviv A, Tadmor T, Polliack A. Primary diffuse large B-cell lymphoma of the breast: looking at pathogenesis, clinical issues and therapeutic options. Ann Oncol. 2013;24(9): 2236-2244. Murawski N, Held G, Ziepert M, et al. The role of radiotherapy and intrathecal CNS prophylaxis in extralymphatic craniofacial aggressive B-cell lymphomas. Blood. 2014;124(5):720-728. Lehners N, Kramer I, Schwarzbich MA, et al. Analysis of clinical characteristics and outcome of patients with previously untreated diffuse large B-cell lymphoma and renal involvement in the rituximab era. Leuk Lymphoma. 2016;57(11):2619-2625. Tomita N, Yokoyama M, Yamamoto W, et al. Central nervous system event in patients with diffuse large B-cell lymphoma in the rituximab era. Cancer Sci. 2012;103(2):245251. Savage KJ, Johnson NA, Ben-Neriah S, et al. MYC gene rearrangements are associated with a poor prognosis in diffuse large B-cell lymphoma patients treated with R-CHOP chemotherapy. Blood. 2009;114(17):35333537. Barrans S, Crouch S, Smith A, et al. Rearrangement of MYC is associated with poor prognosis in patients with diffuse large

haematologica | 2019; 104(1)

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

B-cell lymphoma treated in the era of rituximab. J Clin Oncol. 2010;28(20):3360-3365. Petrich AM, Gandhi M, Jovanovic B, et al. Impact of induction regimen and stem cell transplantation on outcomes in double-hit lymphoma: a multicenter retrospective analysis. Blood. 2014;124(15):2354-2361. Oki Y, Noorani M, Lin P, et al. Double hit lymphoma: the MD Anderson Cancer Center clinical experience. Br J Haematol. 2014;166(6):891-901. Kanungo A, Medeiros LJ, Abruzzo LV, et al. Lymphoid neoplasms associated with concurrent t(14;18) and 8q24/c-MYC translocation generally have a poor prognosis. Mod Pathol. 2006;19(1):25-33. Le Gouill S, Talmant P, Touzeau C, et al. The clinical presentation and prognosis of diffuse large B-cell lymphoma with t(14;18) and 8q24/c-MYC rearrangement. Haematologica. 2007;92(10):1335-1342. Savage KJ, Slack GW, Mottok A, et al. Impact of dual expression of MYC and BCL2 by immunohistochemistry on the risk of CNS relapse in DLBCL. Blood. 2016;127(18):2182-2188. Hu S, Xu-Monette ZY, Tzankov A, et al. MYC/BCL2 protein coexpression contributes to the inferior survival of activated B-cell subtype of diffuse large B-cell lymphoma and demonstrates high-risk gene expression signatures: a report from The International DLBCL Rituximab-CHOP Consortium Program. Blood. 2013;121(20): 4021-4031. Song YS, Lee WW, Lee JS, et al. Prediction of Central nervous system relapse of diffuse large B-cell lymphoma using pretherapeutic [18F]2-fluoro-2-deoxyglucose (FDG) positron emission tomography/computed tomography. Medicine. 2015;94(44):e1978. Lemma SA, Pasanen AK, Haapasaari KM, et al. Similar chemokine receptor profiles in lymphomas with central nervous system involvement - possible biomarkers for patient selection for central nervous system prophylaxis, a retrospective study. Eur J Haematol. 2016;96(5):492-501. Lemma SA, Kuusisto M, Haapasaari KM, et al. Integrin alpha 10, CD44, PTEN, cadherin11 and lactoferrin expressions are potential biomarkers for selecting patients in need of central nervous system prophylaxis in diffuse large B-cell lymphoma. Carcinogenesis. 2017;38(8):812-820. Glass JP, Melamed M, Chernik NL, et al. Malignant cells in cerebrospinal fluid (CSF): the meaning of a positive CSF cytology. Neurology. 1979;29(10):1369-1375. van Besien K, Ha CS, Murphy S, et al. Risk factors, treatment, and outcome of central nervous system recurrence in adults with intermediate-grade and immunoblastic lymphoma. Blood. 1998;91(4):1178-1184. Tilly H, Lepage E, Coiffier B, et al. Intensive conventional chemotherapy (ACVBP regimen) compared with standard CHOP for poor-prognosis aggressive non-Hodgkin lymphoma. Blood. 2003;102(13):4284-4289. Baraniskin A, Kuhnhenn J, Schlegel U, et al. Identification of microRNAs in the cerebrospinal fluid as marker for primary diffuse large B-cell lymphoma of the central nervous system. Blood. 2011;117(11):3140-3146. Baraniskin A, Zaslavska E, NopelDunnebacke S, et al. Circulating U2 small nuclear RNA fragments as a novel diagnostic biomarker for primary central nervous system lymphoma. Neuro Oncol. 2016;18(3): 361-367.

61. Martinez-Ricarte F, Mayor R, Martinez-Saez E, et al. Molecular diagnosis of diffuse gliomas through sequencing of cell-free circulating tumor DNA from cerebrospinal fluid. Clin Cancer Res. 2018;24(12):28122819. 62. Pentsova EI, Shah RH, Tang J, et al. Evaluating cancer of the central nervous system through next-generation sequencing of cerebrospinal fluid. J Clin Oncol. 2016;34 (20):2404-2415. 63. Hutchings M, Ladetto M, Buske C, et al. ESMO Consensus Conference on malignant lymphoma: management of 'ultra-high-risk' patients. Ann Oncol. 2018;29(8):1687-1700. 64. Schmitz N, Zeynalova S, Glass B, et al. CNS disease in younger patients with aggressive B-cell lymphoma: an analysis of patients treated on the Mabthera International Trial and trials of the German High-Grade NonHodgkin Lymphoma Study Group. Ann Oncol. 2012;23(5):1267-1273. 65. Tomita N, Takasaki H, Ishiyama Y, et al. Intrathecal methotrexate prophylaxis and central nervous system relapse in patients with diffuse large B-cell lymphoma following rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone. Leuk Lymphoma. 2015;56(3):725-729. 66. Ferreri AJ, Bruno-Ventre M, Donadoni G, et al. Risk-tailored CNS prophylaxis in a mono-institutional series of 200 patients with diffuse large B-cell lymphoma treated in the rituximab era. Br J Haematol. 2015;168 (5):654-662. 67. Tai WM, Chung J, Tang PL, et al. Central nervous system (CNS) relapse in diffuse large B cell lymphoma (DLBCL): pre- and post-rituximab. Ann Hematol. 2011;90(7): 809-818. 68. Shapiro WR, Young DF, Mehta BM. Methotrexate: distribution in cerebrospinal fluid after intravenous, ventricular and lumbar injections. N Engl J Med. 1975;293(4): 161-166. 69. Chamberlain MC. Radioisotope CSF flow studies in leptomeningeal metastases. J Neurooncol. 1998;38(2-3):135-140. 70. Blasberg RG, Patlak C, Fenstermacher JD. Intrathecal chemotherapy: brain tissue profiles after ventriculocisternal perfusion. J Pharmacol Exp Ther. 1975;195(1):73-83. 71. Dunleavy K, Pittaluga S, Shovlin M, et al. Low-intensity therapy in adults with Burkitt's lymphoma. N Engl J Med. 2013;369(20):1915-1925. 72. Niemann A, Muhlisch J, Fruhwald MC, et al. Therapeutic drug monitoring of methotrexate in cerebrospinal fluid after systemic high-dose infusion in children: can the burden of intrathecal methotrexate be reduced? Ther Drug Monit. 2010;32(4):467-475. 73. Ferreri AJ, Reni M, Pasini F, et al. A multicenter study of treatment of primary CNS lymphoma. Neurology. 2002;58(10):15131520. 74. Ferreri AJ, Reni M, Foppoli M, et al. Highdose cytarabine plus high-dose methotrexate versus high-dose methotrexate alone in patients with primary CNS lymphoma: a randomised phase 2 trial. Lancet. 2009;374 (9700):1512-1520. 75. Pui CH, Campana D, Pei D, et al. Treating childhood acute lymphoblastic leukemia without cranial irradiation. N Engl J Med. 2009;360(26):2730-2741. 76. Magrath I, Adde M, Shad A, et al. Adults and children with small non-cleaved-cell lymphoma have a similar excellent outcome when treated with the same chemotherapy

D. Qualls et al. regimen. J Clin Oncol. 1996;14(3):925-934. 77. Rizzieri DA, Johnson JL, Byrd JC, et al. Improved efficacy using rituximab and brief duration, high intensity chemotherapy with filgrastim support for Burkitt or aggressive lymphomas: cancer and Leukemia Group B study 10 002. Br J Haematol. 2014;165(1): 102-111. 78. Abramson JS, Hellmann M, Barnes JA, et al. Intravenous methotrexate as central nervous system (CNS) prophylaxis is associated with a low risk of CNS recurrence in high-risk patients with diffuse large B-cell lymphoma. Cancer. 2010;116(18):4283-4290. 79. Cheah CY, Herbert KE, O'Rourke K, et al. A multicentre retrospective comparison of central nervous system prophylaxis strategies among patients with high-risk diffuse large B-cell lymphoma. Br J Cancer. 2014;111(6):1072-1079. 80. Phillips E, Kirkwood A, Lawrie A, et al. Low rates of CNS relapse in high risk DLBCL patients treated with R-CODOX-M and RIVAC: results from a phase 2 UK NCRI/bloodwise trial. Blood. 2016;128 (22):1855. 81. Herzig RH, Hines JD, Herzig GP, et al. Cerebellar toxicity with high-dose cytosine

arabinoside. J Clin Oncol. 1987;5(6):927-932. 82. Relling MV, Mahmoud HH, Pui CH, et al. Etoposide achieves potentially cytotoxic concentrations in CSF of children with acute lymphoblastic leukemia. J Clin Oncol. 1996;14(2):399-404. 83. Malecek MK, Petrich AM, Rozell S, et al. Frequency, risk factors, and outcomes of central nervous system relapse in lymphoma patients treated with dose-adjusted EPOCH plus rituximab. Am J Hematol. 2017;92 (11):1156-1162. 84. Ngo VN, Young RM, Schmitz R, et al. Oncogenically active MYD88 mutations in human lymphoma. Nature. 2011;470 (7332):115-119. 85. Davis RE, Ngo VN, Lenz G, et al. Chronic active B cell receptor signaling in diffuse large B cell lymphoma. Nature. 2010;463 (7277):88-92. 86. Lenz G, Davis RE, Ngo VN, et al. Oncogenic CARD11 mutations in human diffuse large B cell lymphoma. Science. 2008;31 (5870):1676-1679. 87. Lionakis MS, Dunleavy K, Roschewski M, et al. Inhibition of B cell receptor signaling by ibrutinib in primary CNS lymphoma. Cancer Cell. 2017;31(6):833-843.e5.

88. Houillier C, Choquet S, Touitou V, et al. Lenalidomide monotherapy as salvage treatment for recurrent primary CNS lymphoma. Neurology. 2015;84(3):325-326. 89. Ayed AO, Chiappella A, Pederson L, et al. CNS relapse in patients with DLBCL treated with lenalidomide plus R-CHOP (R2CHOP): analysis from two phase 2 studies. Blood Cancer J. 2018;8(7):63. 90. Ansell SM, Lesokhin AM, Borrello I, et al. PD-1 blockade with nivolumab in relapsed or refractory Hodgkin's lymphoma. N Engl J Med. 2015;372(4):311-319. 91. 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. 92. Twa DD, 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. 93. Nayak L, Iwamoto FM, LaCasce A, et al. PD-1 blockade with nivolumab in relapsed/refractory primary central nervous system and testicular lymphoma. Blood. 2017;129(23):3071-3073.

haematologica | 2019; 104(1)

ARTICLE

Hematopoiesis

The homeobox transcription factor HB9 induces senescence and blocks differentiation in hematopoietic stem and progenitor cells

Ferrata Storti Foundation

Deborah Ingenhag,1 Sven Reister,1 Franziska Auer,1 Sanil Bhatia,1 Sarah Wildenhain,1 Daniel Picard,1,2 Marc Remke,1,2 Jessica I. Hoell,1 Andreas Kloetgen,1,3 Dennis Sohn,4 Reiner U. Jänicke,4 Gesine Koegler,5 Arndt Borkhardt1 and Julia Hauer1

Department of Pediatric Oncology, Hematology and Clinical Immunology, Medical Faculty of Heinrich-Heine-University, Düsseldorf; 2Department of Pediatric NeuroOncogenomics, German Cancer Consortium (DKTK) and German Cancer Research Center (DKFZ), Heidelberg; 3Computational Biology of Infection Research, Helmholtz Center for Infection Research, Braunschweig; 4Laboratory of Molecular Radiooncology, Clinic and Policlinic for Radiation Therapy and Radiooncology, Medical Faculty of Heinrich-Heine-University, Düsseldorf and 5Institute for Transplantation Diagnostics and Cell Therapeutics, Medical Faculty of Heinrich-Heine-University, Düsseldorf, Germany 1

Haematologica 2019 Volume 104(1):35-46

ABSTRACT

he homeobox gene HLXB9 encodes for the transcription factor HB9, which is essential for pancreatic as well as motor neuronal development. Beside its physiological expression pattern, aberrant HB9 expression has been observed in several neoplasias. Especially in infant translocation t(7;12) acute myeloid leukemia, aberrant HB9 expression is the only known molecular hallmark and is assumed to be a key factor in leukemic transformation. However, so far, only poor functional data exist addressing the oncogenic potential of HB9 or its influence on hematopoiesis. We investigated the influence of HB9 on cell proliferation and cell cycle in vitro, as well as on hematopoietic stem cell differentiation in vivo using murine and human model systems. In vitro, HB9 expression led to premature senescence in human HT1080 and murine NIH3T3 cells, providing for the first time evidence for an oncogenic potential of HB9. Onset of senescence was characterized by induction of the p53-p21 tumor suppressor network, resulting in growth arrest, accompanied by morphological transformation and expression of senescence-associated β-galactosidase. In vivo, HB9-transduced primary murine hematopoietic stem and progenitor cells underwent a profound differentiation arrest and accumulated at the megakaryocyte/erythrocyte progenitor stage. In line, gene expression analyses revealed de novo expression of erythropoiesis-related genes in human CD34+ hematopoietic stem and progenitor cells upon HB9 expression. In summary, the novel findings of HB9-dependent premature senescence and myeloidbiased perturbed hematopoietic differentiation, for the first time shed light on the oncogenic properties of HB9 in translocation t(7;12) acute myeloid leukemia.

Introduction Senescence serves as a tumor-suppressive mechanism and prevents proliferation of cells which have acquired an irreversible DNA-damage.1 Physiologically this results from continued telomere shortening during each round of replication and is therefore called replicative senescence. Onset of senescence is characterized by induction of tumor-suppressor networks such as p53-p21, followed by cell cycle arrest, morphological transformation, and increased β-galactosidase activity.1 Induction of senescence prior to the replication limit is termed premature senescence. In this case, DNA-damage is caused by genotoxic or replicative stress, for example due to mutagenic agents or oncogene expression.2 This was shown for strong oncogenes like RAS and MYC, which induce senescence in fibroblasts in the absence of other transforming mutations, so called oncogeneinduced senescence.3,4 haematologica | 2019; 104(1)

Correspondence: Julia.Hauer@med.uni-duesseldorf.de

Received: January 25, 2018. Accepted: July 30, 2018. Pre-published: August 9, 2018. doi:10.3324/haematol.2018.189407 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/35 ©2019 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.

D. Ingenhag et al.

HLXB9, also known as MNX1 (motor neuron and pancreas homeobox 1), belongs to the ANTP class of homeobox genes.5 It is located on chromosome 7q36, spanning 5.8 kb and comprising 3 exons. The corresponding 401 aa protein is named HB9; this is highly conserved and functions as a transcription factor.6 Physiologically, HB9 is expressed during embryogenesis and is essential for the formation of the dorsal pancreatic bud and B-cell maturation.7-9 In addition, HB9 plays an important role in neuronal development by promoting motor neuron differentiation.10,11 A deregulated HB9 expression has been found in several tumor types. In poorly differentiated hepatocellular carcinomas, microarray analyses identified HB9 as the strongest differentially expressed gene compared to nonneoplastic hepatic controls.12 Also in transcriptome analysis of prostate cancer biopsies from African-Americans, HB9 was the most highly up-regulated protein coding gene compared to matched benign tissues.13 In hematopoietic neoplasias, HB9 is aberrantly highly expressed in translocation t(7;12) acute myeloid leukemia (AML), which accounts for up to 30% of infant AML.14,15 Translocation t(7;12) AML patients have a very dismal prognosis, with a 3-year event-free survival of 0%, regardless of the treatment approach.15,16 Since its first description in 2000, aberrant HB9 expression remains the only known molecular hallmark of translocation t(7;12) AML,17,18 but only poor functional data exist regarding its oncogenic properties and how, if at all, aberrant HB9 expression influences hematopoiesis, thereby contributing to leukemogenesis. Early expression studies reported HB9 expression in healthy CD34+ hematopoietic stem and progenitor cells (HSPCs),19 but could not be validated by studies of our and other groups.15,20,21 Hence, a physiological function of HB9 in HSPCs remains a subject of debate. Morphologically, translocation t(7;12) AML blast cells are less differentiated (FAB subtype M0 or M2), accompanied by expression of stem cell markers like CD34 and CD117,15,22 indicating a very early differentiation block. Gene expression profiling of HB9+ blast cells revealed a modulation of cell-cell interaction and cell adhesion.22 In previous studies, we had used the AML cell line HL-60 for stable HB9 overexpression to identify potential HB9 target genes by combined ChIP-on-chip and expression analyses.21 As HL-60 cells represent an already transformed AML cell line model, harboring several genetic aberrations like loss of p53 and MYC replication,23 it is difficult to come to any conclusions about the oncogenic potential of HB9 and its influence on primary hematopoietic cells in vivo with respect to translocation t(7;12) leukemogenesis. Thus, in our current study, we evaluated the oncogenic potential of HB9 by its effect on proliferation and cell cycle regulation. Furthermore, we performed for the first time in vivo hematopoietic reconstitution experiments to investigate the influence of HB9 expression on hematopoietic cell differentiation with regard to translocation t(7;12) AML.

perature and immediately analyzed by flow cytometry (FACSCalibur, BD Biosciences, Heidelberg, Germany).

β-galactosidase staining

Six days after transduction, cells were stained for β-galactosidase activity using the Senescence Cells Histochemical Staining Kit (Sigma-Aldrich, Taufkirchen, Germany) according to the manufacturer’s instructions. A total of 300 cells were counted for each replicate and the frequency of positive cells was determined. Images were taken using an Axiovert 200 microscope (Zeiss, Jena, Germany).

Bone marrow transplantation Bone marrow cells were harvested from femurs and tibiae of 8-10-week old male C57BL/6 mice and Lineage+ cells were depleted using the Lineage Cell Depletion Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Lin– cells were cultured in Stemspan SFEM (Stemcell Technologies, Cologne, Germany), supplemented with 1% penicillin-streptomycin, 100 ng/mL SCF, 100 ng/mL FLT3L, 100 ng/mL TPO, 25 ng/mL IL6 and 25 ng/mL IL11 (PeproTech, Hamburg, Germany ). Twenty-four hours (h) after isolation, Lin– cells were lentivirally transduced (MOI 50), together with hexadimethrine bromide (5 mg/mL; Sigma-Aldrich) on RetroNectin (Takara, Frankfurt, Germany) coated plates, and spinoculated at 1000 g for 2 h at 32°C. One day after transduction, Lin– cells were transplanted via tail-vein injection into myeloablative-irradiated 810-week old female B6.SJL-Ptprc Pepc/BoyJ mice.

Isolation of CD34+ cord blood cells The use of primary human CD34+ cord blood cells for this study was approved by the ethics committee of the medical faculty of the Heinrich-Heine-University, Duesseldorf, and was carried out in accordance with the Declaration of Helsinki. Cord blood samples were obtained from healthy donors after informed consent (José Carreras Stem Cell Bank, University Hospital Duesseldorf, Germany) and mononuclear cells were enriched by density gradient centrifugation with Ficoll Paque Plus (GE Healthcare, Frankfurt, Germany). CD34+ cells were enriched using CD34 MicroBead Kit and MACS technology (Miltenyi Biotec) according to the manufacturer’s instructions and cultivated in X-Vivo 20 (Lonza, Cologne, Germany) supplemented with 100 ng/mL SCF, 100 ng/mL FLT3L, 100 ng/mL TPO and 20 ng/mL IL3 (Peprotech).

Statistical analysis Statistical significance was determined from adequately powered sample sizes of similar variation using two-tailed unpaired Student t-tests and was defined as P≤0.05 (*), P≤0.01 (**) and P≤0.001 (***). Sample sizes are given in the figure legends.

Other methods The details of other methods, including virus production, immunofluorescence, siRNA transfection, cell preparation and flow cytometric analysis, as well as western blot, PCR, quantitative Reverse Transcriptase-PCR and RNA-Seq-analysis are given in the Online Supplementary Appendix. The RNA-Seq data generated in this study can be found with accession number GSE117060 in the NCBI GEO database.

Methods Results Cell cycle analysis 3x105 cells were washed twice with PBS and resuspended in hypotonic buffer solution, containing 0.1% Triton-X 100, 0.1% sodium-citrate and 50 µg/mL propidium iodide. After resuspension, cells were incubated for 10 minutes in the dark at room tem36

Aberrant HB9 expression induces premature senescence In order to investigate the influence of HB9 on cellular proliferation and cell cycle, human HT1080 and murine haematologica | 2019; 104(1)

HB9 induces senescence

NIH3T3 cells were lentivirally transduced with HB9-GFP or GFP (Online Supplementary Figure S1) and subjected to downstream analysis. Cell models were selected due to high transduction efficiency, as well as a short doubling time, and have already been used in several studies regarding proliferation and cell cycle analysis.24-26 The transduction efficiency reached almost 100% in both models with a high transgene expression, whereas no endogenous HB9 was detectable (Figure 1A). While control vector-transduced cells ([GFP]) showed a normal exponential growth, HB9-transduced cells ([HB9]) arrested within 72 h after transduction (HT1080: P=0.01; NIH3T3: P=0.05) (Figure 1B). Cell-cycle analysis revealed a significant decrease of cells in the S-phase in HT1080[HB9] (9.6% vs. 4.9%; P=0.005), stalling the cells in G1- and G2-phase (Figure 1C). NIH3T3[HB9] corroborated a significant decrease of cells in the S-phase (12.7% vs. 8.5%; P=0.038). Furthermore, a significant decrease of cells in the G1-phase (61.6% vs.

51.6%; P=0.037) and an increase of aneuploid cells (>4n; 3.4% vs. 13.3%; P=0.002) was observed (Figure 1C). Contiguous with the growth arrest, HB9-transduced cells underwent morphological changes, becoming flattened, enlarged and multinuclear (Figure 2A). As these morphological properties were reminiscent of senescent cells, we assessed senescence-associated β-galactosidase activity (SA-β-gal).27 The frequency of SA-β-gal+ cells was significantly increased in HT1080[HB9] (10-fold; P≤0.01) and NIH3T3[HB9] (133-fold; P≤0.01), compared to control (Figure 2B). Immunofluorescence staining confirmed HB9 localization within the nucleus, which is essential for its function as a transcription factor (Figure 2C). Further, costaining with phalloidin depicted an increase of cytoskeleton and presence of stress fibers as additional characteristics of senescent cells in HT1080[HB9] and NIH3T3[HB9].28 In addition, HT1080[HB9] displayed a senescence-associated nuclear actin accumulation.29

Figure 1. Proliferation and cell cycle analysis of HB9-transduced HT1080 and NIH3T3 cells. (A) Flow cytometric analysis of GFP-expression was used to determine the transduction efficiency. HB9-expression was analyzed by western blot of non- (nt), control- (GFP) and HB9-vector-transduced NIH3T3 and HT1080 cells. (B) Proliferation study of HB9- or GFP-transduced HT1080 and NIH3T3 cells. Cells were seeded at 24 hours (h) after transduction and counted over a 4-day period (n=3). (C) Cell cycle analysis of HB9- or GFP-expressing cells 72 h after lentiviral transduction (n=3).

haematologica | 2019; 104(1)

D. Ingenhag et al.

HB9 activates the p53-p21 tumor suppressor network The tumor suppressor p53 plays a prominent role in the G1/S checkpoint control and in the mediation of premature senescence via activation of the cyclin-dependent kinase inhibitor p21.4,30 Therefore, we investigated the p53/p21 status in the HB9 expressing cell line models. Immunoblotting revealed phosphorylation of p53 at serine 15 upon HB9 expression, indicating activation of the p53-signaling pathway in response to DNA-damage.31 As a result of phosphorylation, p53 accumulates, leading to induction of its downstream mediator p21 (Figure 3A).30,32 Gamma-irradiated HT1080 and Etoposide-treated NIH3T3 cells served as positive control for a DNA-dam-

age dependent p53-pathway activation.31,33 We used a siRNA-mediated p53 knockdown model to determine whether p53 is essential for the HB9-dependent growth arrest. Therefore, HT1080 cells were transfected with p53 or non-targeting siRNA-pools prior to transduction and subjected to proliferation analysis. The siRNA mediated knockdown resulted in a distinct p53 protein reduction for 72 h (Online Supplementary Figure S2), correlating with a reduced pathway activation (Online Supplementary Figure S3). As expected, knockdown of p53 prevented the growth inhibitory effect caused by HB9, as p53-knockdown cells showed a significantly increased proliferation rate compared to non-targeting control (Figure 3B). In line,

Figure 2. Analysis of the morphology of HB9-transduced HT1080 and NIH3T3 cells. (A) Microscopic analysis regarding the morphology of HB9- and GFP-transduced cells (scale bar=50mm). 150 cells were counted from each cell line to determine the percentage of multinuclear cells (n=3). (B) SAÎ˛-gal staining was performed seven days after transduction. 150 cells were counted each to determine the percentage of Î˛-gal+ cells (n=3). (C) Immunofluorescence staining. F-actin was stained by Phalloidin (shown in green) to detect senescent cell characteristics, such as presence of stress fibers and nuclear actin accumulation. DNA was stained by DAPI (shown in blue). HB9 is shown in red (scale bar=20mm; shown is one representative experiment out of three).

haematologica | 2019; 104(1)

HB9 induces senescence

no significant reduction (P=0.1) of the S-phase was observed in HT1080[HB9] transfected with p53-siRNA compared to GFP-control (Figure 3C). These data confirm that the HB9-dependent growth arrest is mediated via activation of p53-signaling. Thus, with regard to onset of a tumor suppressor network, resulting in cell cycle arrest, morphological transformation and expression of SA-β-gal, HB9 induces premature senescence in both cell line models. In general, gene expression, which triggers senescence, has the potential to initiate or promote carcinogenesis.34 The induction of premature senescence following oncogene expression is known as oncogene-induced senescence.2 While expression of these oncogenes in vitro results in growth inhibition, pre-malignant neoplasias arise in vivo.35 Thus, our next step was to investigate the oncogenic potential of HB9 together with its influence on hematopoietic differentiation in vivo.

HB9-expressing HSPCs undergo differentiation arrest and accumulate at the megakaryocyte/erythrocyte progenitor stage To this end, we developed a murine HB9+ transplantation model. Lineage– (Lin–) cells, obtained from CD45.2+ donor mice, were lentivirally transduced with HB9 or control vector and transplanted into myeloablative-irradiated

CD45.1+ recipient mice. Transduction of Lin– HSPCs resulted in 15-20% GFP+ or HB9-GFP+ cells, respectively (Figure 4A), and comparable or even higher GFP expression levels were detected in Lin[HB9] compared to Lin[GFP] by qRT-PCR (Online Supplementary Figure S4A). We also confirmed co-expression of HB9 in Lin[HB9], and, as expected, no endogenous HB9 expression was detected in Lin[GFP] (Figure 4B and Online Supplementary Figure S4B). Twelve weeks after transplantation, peripheral blood cells of recipient mice were analyzed for GFPexpression. Flow cytometric analysis confirmed complete hematopoietic reconstitution by transplanted HSPCs, as more than 90% of the cells were positive for CD45.2 (Figure 4C). In Lin[GFP]-transplanted mice (B6[GFP]), we detected GFP+ T, B and myeloid cells (GFP+: 30.1-85.8% T cells, 20.7-69.8% B cells, 8.8-70.2% myeloid cells), whereas no distinct HB9-GFP+ cell population was present in any lineage of Lin[HB9]-transplanted mice (B6[HB9]) (Figure 4C). Using PCR-based analysis, we were able to confirm genomic integration and transcription of the expression cassettes in peripheral blood cells of B6[HB9] and B6[GFP] (Online Supplementary Figure S5). In line with flow cytometry, qRT-PCR showed a significantly decreased GFP-expression (13-fold; P=0.015) in B6[HB9] compared to B6[GFP] (Online Supplementary Figure S5B). Regarding cell frequencies, B6[HB9] mice showed an over-

Figure 3. Analysis of p53-pathway activation in HB9-transduced HT1080 and NIH3T3. (A) Western blot analysis of p53, phospho-p53 Ser15 (P-p53) and of its downstream effector p21 in HB9- or GFP-transduced HT1080 and NIH3T3 cells. γ-irradiated (HT1080) or Etoposide-treated (NIH3T3) cells serve as control for p53 pathway activation. Shown is one representative experiment out of three. (B) Proliferation and (C) cell cycle analysis of HB9-transduced HT1080, treated with non-targeting (nt) and p53-targeting siRNA, compared to GFP-transduced cells treated with non-targeting siRNA (n=3). ns: not significant.

haematologica | 2019; 104(1)

D. Ingenhag et al.

all decrease in T cells (Figure 5A), valid for both CD4+ and CD8+ T cells (Figure 5B), while the frequency of B cells varied. Addressing the myeloid lineage, an increased frequency of monocytes and granulocytes was observed (Figure 5A). Overall blood cellularity was decreased in B6[HB9] regardless of lineage contribution (Figure 5C). Serial blood draws confirmed these observations during the entire monitoring period (6.5 months). To investigate whether a differentiation blockage is causative for reduced blood cellularity as well as absence of HB9-GFP+ mature blood cells, we performed comprehensive FACS analysis from HSCs to mature B cells, T cells, and myeloid cells. In contrast to the mature cell pool, HB9-GFP-expressing cells were detected within the Lin–Sca-1+c-Kit+ (LSK)/HSC compartment (Figure 6A). To evaluate the differentiation potential of HB9-expressing HSCs, Lin– cells were further analyzed regarding IL7Rα, which is expressed by lymphoid but not myeloid committed progenitors. HB9-GFP expression was decreased in lymphoid committed progenitors (Lin–IL7Rα+) compared to that in myeloid committed progenitors (Lin–IL7Rα–), while B6[GFP] showed comparable amounts of GFP+ cells in both progenitor populations (Figure 6B). Due to the enrichment of HB9-expressing cells in the myeloid progenitor subset, this compartment was further analyzed. Sub-gating of the myeloid progenitor population into common myeloid progenitors (CMP), megakaryocyte/erythrocyte progenitors (MEP) and granulocyte/macrophage progenitors (GMP) revealed that HB9GFP-expressing cells contributed almost exclusively to the MEP subset as the amount of HB9-GFP+ MEP was signifi-

cantly enriched compared to CMP and GMP (Figure 6C). In line with peripheral blood, total white blood cell number of bone marrow and thus CMP, GMP and MEP was decreased in B6[HB9] compared to B6[GFP], but without affecting cell frequencies (Online Supplementary Figure S6). Analysis of terminal MEP maturation was not applicable, as erythrocytes as well as thrombocytes are anuclear cell types and therefore do not carry the transgene. We further assessed the differentiation potential of HB9-GFP+ GMP cells, together with the frequency of the myeloid cell types originating from this. According to peripheral blood analysis, no distinct HB9-GFP+ cell population was detectable within the granulocyte and the monocyte/macrophage population in the bone marrow or spleen of B6[HB9] compared to B6[GFP] (Online Supplementary Figures S7-S9 and S11A-C). The frequency of macrophages/monocytes was slightly increased in B6[HB9] compared to B6[GFP], which is congruent to peripheral blood analysis, while, in contrast, the frequency of granulocytes was comparable (Online Supplementary Figure S12A and B). With respect to lymphoid differentiation we assessed HB9-GFP expression as well as frequency of immature and mature B cells and T cells. While at the immature B220+CD19+CD93+ B-cell stage a low frequency of HB9GFP+ cells was still detectable, no distinct HB9-GFP+ cell population was identified within the more mature B220+CD19+CD93– B-cell population in bone marrow, or within mature B cells of the spleen and peripheral blood

Figure 4. Transplantation of HB9-transduced hematopoietic stem and progenitor cells and monitoring of hematopoietic reconstitution. (A) Flow cytometric analysis of GFP expression in Lineage– cells, non-transduced (Lin[nt]), or transduced with HB9 (Lin[HB9]) or control vector (Lin[GFP]). Shown is one representative experiment. (B) Detection of HB9 expression by qRT-PCR in Lin[GFP] and Lin[HB9] (n=3). (C) Flow cytometric analysis of CD45.2 and GFP expression in peripheral blood cells of Lin[GFP]- (B6[GFP]) or Lin[HB9]-transplanted mice (B6[HB9]).

haematologica | 2019; 104(1)

HB9 induces senescence

(Online Supplementary Figure S7-S9 and S11D and E). Furthermore, B6[HB9] mice showed an overall increased frequency of immature CD93+ and a decreased frequency of more mature CD93â&#x20AC;&#x201C; B cells in bone marrow compared to B6[GFP], but without affecting B-cell frequency in the periphery (Online Supplementary Figure S12C). In contrast to immature B cells, no HB9-GFP+ cells were detected within immature CD4+CD8+ thymocytes (Online Supplementary Figures S10 and S11F). Congruent to the other lineages, no distinct HB9-GFP+ cell population was found within the mature T-cell compartment of bone marrow, spleen or peripheral blood (Online Supplementary Figure S7-S9 and S11G). In line with our results obtained from peripheral blood, the frequency of CD8+ T cells was slightly decreased in bone marrow and spleen of B6[HB9] compared to B6[GFP], while the frequency of CD4+ T cells was comparable (Online Supplementary Figure S12D). In

summary, HB9+ HSCs showed an impaired differentiation capacity, resulting in an overall differentiation blockage, leading to reduced bone marrow as well as peripheral blood cellularity and accumulation of HB9+ cells at the MEP stage.

HB9 triggers expression of erythropoiesis-related genes and leads to decreased clonogenic potential in HSPCs To further analyze the influence of HB9 on hematopoietic cells, regarding gene expression as well as clonogenicity, a highly purified population of transduced cells is mandatory. Thus, these experiments were performed with primary human CD34+ HSPCs, as these cells display a better GFP signal-to-noise ratio compared to their murine counterpart, allowing GFP+ cell FACS-sort upon transduction.

Figure 5. Frequency of lymphoid and myeloid cell types in peripheral blood of B6[GFP] and B6[HB9]. (A) Flow cytometric analysis was used to determine the frequency of T cells (CD3+), B cells (CD19+), monocytes (CD11b+Gr-1â&#x20AC;&#x201D;) and granulocytes (CD11b+Gr-1+) in peripheral blood cells of B6[GFP] and B6[HB9] 12, 19 and 26 weeks after transplantation (n=6). (B) Frequency of CD3+CD4+ and CD3+CD8+ T cells in peripheral blood cells of B6[GFP] and B6[HB9] 26 weeks after transplantation (n=6). (C) Leukocyte count/mL peripheral blood of B6[GFP] and B6[HB9] 12, 19 and 26 weeks after transplantation (n=6).

haematologica | 2019; 104(1)

D. Ingenhag et al.

CD34[HB9] cells displayed transgene expression levels (Figure 7A) comparable to those in translocation t(7;12) AML blast cells,22 and no endogenous HB9 expression was detectable (Figure 7B). Genome-wide expression profiling revealed 117 genes, which were significantly differentially expressed (P≤0.05, FC≥1.8) in CD34[HB9] compared to CD34[GFP] (Online

Supplementary Table S1 and Online Supplementary Figure S13). The gene with the highest fold change was HB9/MNX1, confirming transgene expression. Six genes were de novo expressed in CD34[HB9] compared to CD34[GFP] (Figure 7C), of which hemoglobin subunit zeta (HBZ), as well as solute carrier family 4 member 1 (SLC4A1), are restricted to the erythroid lineage, thereby

Figure 6. Flow cytometric analysis of GFP expression in the hematopoietic stem and progenitor cell compartment. (A) Lineage–c-Kit+Sca-1+ (LSK) hematopoietic stem cells were analyzed for GFP expression. Shown is one representative experiment. (B) Lymphoid (Lin–IL7Rα+) and myeloid (Lin–IL7Rα–) progenitors were analyzed for GFP expression and the percentage of GFP+ cells was determined for each compartment (n=5). (C) Flow cytometric analysis of GFP expression in the myeloid progenitor compartment, regarding MEP (Lin- IL7Rα– c-Kit+ Sca-1- FcγRlow CD34-), CMP (Lin– IL7Rα- c-Kit+ Sca-1- FcγRlow CD34+) and GMP (Lin– IL7Rα– c-Kit+ Sca-1– FcγRhigh CD34+) (n=5).

haematologica | 2019; 104(1)

HB9 induces senescence

confirming the HB9-dependent erythroid-biased differentiation observed in vivo. HBZ is an embryonic α-like globin gene, which is expressed in primitive erythroid cells. As hemoglobin is a tetramer of 2 α-like globin polypeptide chains and 2 β-like globin chains, also the embryonic β-like globin gene HBE1 was significantly up-regulated (4.6-fold; P=0.019). SLC4A1 is part of the anion exchanger family and is expressed in the erythrocyte plasma membrane, where it functions as a chloride/bicarbonate exchanger. Thus, HB9 expression in CD34+ HSPCs initiates de novo expression of genes related to erythropoiesis, which is in line with the differentiation bias towards the megakaryocytic/erythroid cell lineage observed in vivo. The computational gene set enrichment analysis software was used to statistically determine altered pathways related to HB9 expression associated with biological process-related gene ontology terms. Among the positive enriched biological processes, mitosis-related processes showed the strongest enrichment, while, in contrast, cytoskeleton organization-related processes showed the strongest negative enrichment (Online Supplementary Table S2 and Online Supplementary Figure S14). This phenotype

resembles our observations made in vitro, where HB9 led to cell cycle arrest and multinuclearity as a consequence of senescence-response. With regard to clonogenic potential, HB9-transduced CD34+ HSPCs showed a decreased clonogenic capacity compared to GFP-transduced cells (Online Supplementary Figure S15), corresponding to reduced cellularity in bone marrow (Online Supplementary Figure S6A) and peripheral blood (Figure 5C) of B6[HB9] mice.

Discussion The homeobox gene HB9 is expressed during early embryonic development, being essential for pancreatic as well as motor neuronal differentiation.7-11 Aberrant HB9 expression has been detected in distinct tumor entities; especially in translocation t(7;12) AML it is assumed to be a key factor in leukemic transformation.15,18 However, up to now only poor functional studies exist addressing its oncogenic potential or its influence on hematopoiesis. We evaluated the oncogenic potential of HB9 regarding its influence on proliferation and cell cycle using a model of

Figure 7. De novo gene expression in human CD34+ HSPCs upon HB9 transduction. (A) qRT-PCR analysis of HB9 expression, normalized to β-Actin, in cord blood derived CD34+ cells, transduced with GFP (CD34[GFP]) or HB9-GFP (CD34[HB9]). (B) Gel electrophoretic analysis of the qRT-PCR products. (C) De novo expressed genes in CD34[HB9] compared to CD34[GFP]; normalized read counts of three independent RNA-Seq experiments are depicted.

haematologica | 2019; 104(1)

D. Ingenhag et al.

human HT1080 and murine NIH3T3 cells. In both cell-line models HB9-expression led to growth arrest within 72 h after transduction, mediated via the p53-p21 signaling axis. In line with this, siRNA-mediated p53-knockdown abolished the HB9-dependent growth inhibitory effect. Regarding cell cycle analysis, HT1080[HB9] showed a significant decrease in cells at the S-phase, stalling the cells at G1/G2-phase of the cell cycle. NIH3T3[HB9] corroborated that effect, and in addition exhibited a decrease in cells in the G1-phase and an increase in aneuploid cells, which can be explained as a dose-dependent effect, as NIH3T3 showed lower HB9 expression levels upon transduction compared to HT1080. This leads to less activation of p53signaling, resulting in an incomplete G1/G2 arrest, so that NIH3T3[HB9] cells have the chance to re-enter the cell cycle, irrespective of an inappropriate replication, leading to an increase of aneuploid cells (> 4n), while there is a decrease in diploid cells at G1- and S-phase. Besides onset of a tumor-suppressor network, resulting in growth arrest, HB9-expressing cells exhibited typical morphological changes, becoming flattened, enlarged and multinuclear, as well as expression of SA-β-gal, thus fulfilling all main criteria of senescence.1 Cellular senescence is initiated in response to replicative stress, resulting from irreversible DNA damage, and can be differentiated into replicative senescence and premature senescence. Induction of premature senescence has been shown for many potent oncogenes, such as RAS, RAF, MEK and BRAF.4,36-38 In contrast to replicative senescence, oncogene-induced premature senescence occurs independently of telomer attrition, as a result of a strong mitogenic signal, which leads to an inappropriate replication of the DNA during the S-phase of the cell cycle and thus activation of DNA damage response.2 In line with this, HB9 triggered a DNA damage response in HT1080 and NIH3T3, indicated by phosphorylation of p53 at Ser15, as this specific phosphorylation site is a target for the DNA-damage kinases ATM and ATR.31,39 Thus HB9 expression induces DNA damage, which in turn activates p53-signaling as part of the DNA damage response. Furthermore, HB9-dependent replicative stress led to an increase in aneuploid cells in NIH3T3[HB9], which has already been shown for MYC oncogene.3 This was further validated in HB9-transduced CD34+ HSPCs: RNA-Seq analyses revealed a positive enrichment of biological processes related to DNA/RNA processing, as well as cell cycle and mitosis, and a negative enrichment of processes related to post-translational phosphorylation and intracellular signaling (Online Supplementary Table S2 and Online Supplementary Figure S16), which correlates to the replicative stress-dependent expression profile of MYC-driven lymphoma.40 Oncogenes, which induce premature senescence in vitro, are known to cause pre-malignant cell populations in vivo.35 This was shown for murine lung adenomas (RASV12), T-cell lymphomas (RASV12), prostate tumors (PTEN), as well as human benign melanocytic naevi (BRAFE600).38,41-43 With respect to translocation t(7;12) leukemogenesis, we set up a murine bone marrow transplantation model to investigate whether ectopic HB9-expression affects hematopoietic cell differentiation in vivo. Therefore, murine HSPCs were transduced with either HB9-GFP or GFP vector and transplanted into myeloablative irradiated recipient mice. HB9-transduction of Lin– HSPCs yielded 44

expression levels comparable to that of translocation t(7;12) AML blast cells.22 Similar to human HSPCs, no endogenous HB9-expression was detectable in murine Lin– HSPCs.15,20,21 In vivo, HB9-expressing HSPCs underwent an overall differentiation arrest, as no HB9-GFP+ cells were detected within the mature B-, T- and myeloid-cell pool of recipient mice. Analysis of the HSPC compartment revealed an early blockage of the lymphoid lineage, while HB9-expressing HSPCs showed a strong myeloid lineage commitment. Within the myeloid progenitor population HB9-expressing cells were significantly enriched in the MEP, compared to the CMP and GMP compartment, thus revealing proliferation of HB9+ cells arrested at the MEP stage. Instead a differentiation blockage without proliferation would have resulted in comparable frequencies of GMP and MEP, whereas a megakaryocytic- and/or erythroid-biased differentiation would have resulted in comparable frequencies of CMP and MEP. A myeloid-biased differentiation in combination with a diminished potential of lymphoid differentiation is associated with an aged hematopoietic system. This is assumed to result from aged HSCs, which show an impaired lymphoid differentiation capacity, while the myeloid differentiation capacity is maintained or even increased.44,45 Consistently, the incidence of myeloid malignancies increases with age. Recent studies highlight not only a myeloid, but in particular a megakaryocytic/erythroid bias in aged human and murine HSCs46 directly correlating to our results of a megakaryocytic/erythroid-biased differentiation and de novo expression of erythropoiesis-related genes in HB9+ HSPCs. Furthermore, the in vivo experiments have shown that the impaired differentiation capacity of HB9+ HSCs results in a decreased bone marrow and peripheral blood cellularity throughout the entire monitoring period. This is in line with our findings of HB9-dependent reduced clonogenicity in human CD34+ HSPCs, and further strengthens the assumption that HB9 induces senescence in hematopoietic cells. Onset of HB9dependent senescence in hematopoietic cells is further supported by RNA-Seq data of CD34+ HSPCs. A strong positive enrichment of mitosis-related processes, together with a contradictory negative enrichment of cytoskeleton organization-related processes, reflects the senescenceassociated multinuclear phenotype observed in the cellline models, at molecular basis in CD34+ HSPCs. With regard to translocation t(7;12) AML, our data may suggest that HB9-expression dictates the development of exclusively myeloid leukemia in translocation t(7;12)-positive AML patients.15 Although most translocation t(7;12) AML blast cells show a very undifferentiated state, some are defined as erythroblastic17,47 as well as megakaryoblastic leukemia,48,49 correlating to the megakaryocytic/erythroid-biased differentiation due to HB9 expression. With regard to leukemogenesis, secondary genetic alterations are necessary, as sole HB9-expression did not result in complete transformation and progression to AML. Based on our in vitro results, HB9 expression led to an increase in aneuploid cells, which is correlated to genetic instability. Thus, via induction of genetic instability, HB9 may increase the chance for secondary genetic alterations, which are necessary for complete cellular transformation. Initial screening studies using a panel of frequently mutated genes in AML (NPM1, CEPBA, MLL, WT1, FLT3, NRAS, K-RAS, PTPN11 and KIT) did not succeed in identifying secondary recurrent genetic alterations in translocahaematologica | 2019; 104(1)

HB9 induces senescence

tion t(7;12) AML.50 Thus, whole exome or whole genome analysis is necessary to identify additional relevant mutations in translocation t(7;12) AML. In summary, the induction of premature senescence, together with perturbed hematopoietic differentiation, for the first time, sheds light on the oncogenic properties of HB9 in translocation t(7;12) AML. This will help us to find novel approaches for the treatment of fatal translocation t(7;12) AML, especially because resistance to therapy is a hallmark of this AML subtype.

References 1. Kuilman T, Michaloglou C, Mooi WJ, Peeper DS. The essence of senescence. Genes Dev. 2010;24(22):2463-2479. 2. Courtois-Cox S, Jones SL, Cichowski K. Many roads lead to oncogene-induced senescence. Oncogene. 2008;27(20):28012809. 3. Felsher DW, Zetterberg A, Zhu J, Tlsty T, Bishop JM. Overexpression of MYC causes p53-dependent G2 arrest of normal fibroblasts. Proc Natl Acad Sci USA. 2000; 97(19):10544-10548. 4. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell. 1997;88(5):593-602. 5. Holland PW, Booth HA, Bruford EA. Classification and nomenclature of all human homeobox genes. BMC Biol. 2007; 5:47. 6. Harrison KA, Druey KM, Deguchi Y, Tuscano JM, Kehrl JH. A novel human homeobox gene distantly related to proboscipedia is expressed in lymphoid and pancreatic tissues. J Biol Chem. 1994; 269(31):19968-19975. 7. Dalgin G, Ward AB, Hao le T, Beattie CE, Nechiporuk A, Prince VE. Zebrafish mnx1 controls cell fate choice in the developing endocrine pancreas. Development. 2011; 138(21):4597-4608. 8. Harrison KA, Thaler J, Pfaff SL, Gu H, Kehrl JH. Pancreas dorsal lobe agenesis and abnormal islets of Langerhans in Hlxb9deficient mice. Nat Genet. 1999;23(1):7175. 9. Li H, Arber S, Jessell TM, Edlund H. Selective agenesis of the dorsal pancreas in mice lacking homeobox gene Hlxb9. Nat Genet. 1999;23(1):67-70. 10. Arber S, Han B, Mendelsohn M, Smith M, Jessell TM, Sockanathan S. Requirement for the homeobox gene Hb9 in the consolidation of motor neuron identity. Neuron. 1999;23(4):659-674. 11. Thaler J, Harrison K, Sharma K, Lettieri K, Kehrl J, Pfaff SL. Active suppression of interneuron programs within developing motor neurons revealed by analysis of homeodomain factor HB9. Neuron. 1999; 23(4):675-687. 12. Wilkens L, Jaggi R, Hammer C, Inderbitzin D, Giger O, von Neuhoff N. The homeobox gene HLXB9 is upregulated in a morphological subset of poorly differentiated hepatocellular carcinoma. Virchows Arch. 2011;458(6):697-708. 13. Zhang L, Wang J, Wang Y, et al. MNX1 Is Oncogenically Upregulated in African-

haematologica | 2019; 104(1)

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

Funding JH has been supported by the German Cancer Aid (110997), the German José Carreras Leukemia Foundation (DJCLS 02R/2016), the Deutsche Kinderkrebsstiftung (DKS 2016.17) and the “Forschungskommission” of the medical faculty of Heinrich-Heine-University. Acknowledgments We are indebted to all members of our group for useful discussions and critical reading of the manuscript. We thank Silke Furlan for excellent technical support.

American Prostate Cancer. Cancer Res. 2016;76(21):6290-6298. Masetti R, Vendemini F, Zama D, Biagi C, Pession A, Locatelli F. Acute myeloid leukemia in infants: biology and treatment. Front Pediatr. 2015;3:37. von Bergh AR, van Drunen E, van Wering ER, et al. High incidence of t(7;12)(q36;p13) in infant AML but not in infant ALL, with a dismal outcome and ectopic expression of HLXB9. Genes Chromosomes Cancer. 2006;45(8):731-739. Beverloo HB, Panagopoulos I, Isaksson M, et al. Fusion of the homeobox gene HLXB9 and the ETV6 gene in infant acute myeloid leukemias with the t(7;12)(q36;p13). Cancer Res. 2001;61(14):5374-5377. Tosi S, Harbott J, Teigler-Schlegel A, et al. t(7;12)(q36;p13), a new recurrent translocation involving ETV6 in infant leukemia. Genes Chromosomes Cancer. 2000; 29(4):325-332. Tosi S, Mostafa Kamel Y, Owoka T, Federico C, Truong TH, Saccone S. Paediatric acute myeloid leukaemia with the t(7;12)(q36;p13) rearrangement: a review of the biological and clinical management aspects. Biomark Res. 2015;3:21. Deguchi Y, Kehrl JH. Selective expression of two homeobox genes in CD34-positive cells from human bone marrow. Blood. 1991;78(2):323-328. Nagel S, Kaufmann M, Scherr M, Drexler HG, MacLeod RA. Activation of HLXB9 by juxtaposition with MYB via formation of t(6;7)(q23;q36) in an AML-M4 cell line (GDM-1). Genes Chromosomes Cancer. 2005;42(2):170-178. Wildenhain S, Ingenhag D, Ruckert C, et al. Homeobox protein HB9 binds to the prostaglandin E receptor 2 promoter and inhibits intracellular cAMP mobilization in leukemic cells. J Biol Chem. 2012; 287(48):40703-40712. Wildenhain S, Ruckert C, Rottgers S, et al. Expression of cell-cell interacting genes distinguishes HLXB9/TEL from MLL-positive childhood acute myeloid leukemia. Leukemia. 2010;24(9):1657-1660. Dalton WT Jr, Ahearn MJ, McCredie KB, Freireich EJ, Stass SA, Trujillo JM. HL-60 cell line was derived from a patient with FAB-M2 and not FAB-M3. Blood. 1988;71(1):242-247. Byun HS, Cho EW, Kim JS, et al. Thioredoxin overexpression in HT-1080 cells induced cellular senescence and sensitization to gamma radiation. FEBS letters. 2005;579(19):4055-4062. Hackenbeck T, Knaup KX, Schietke R, et al. HIF-1 or HIF-2 induction is sufficient to achieve cell cycle arrest in NIH3T3 mouse

26.

27.

28.

29.

30.

31.

32.

33. 34. 35. 36.

37.

38.

39.

fibroblasts independent from hypoxia. Cell Cycle. 2009;8(9):1386-1395. Rao W, Xie G, Zhang Y, et al. OVA66, a tumor associated protein, induces oncogenic transformation of NIH3T3 cells. PloS One. 2014;9(3):e85705. Dimri GP, Lee X, Basile G, et al. A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc Natl Acad Sci USA. 1995;92(20):9363-9367. Nishio K, Inoue A. Senescence-associated alterations of cytoskeleton: extraordinary production of vimentin that anchors cytoplasmic p53 in senescent human fibroblasts. Histochem Cell Biol. 2005;123(3):263273. Kwak IH, Kim HS, Choi OR, Ryu MS, Lim IK. Nuclear accumulation of globular actin as a cellular senescence marker. Cancer Res. 2004;64(2):572-580. Jackson JG, Pereira-Smith OM. p53 is preferentially recruited to the promoters of growth arrest genes p21 and GADD45 during replicative senescence of normal human fibroblasts. Cancer Res. 2006;66(17):83568360. Shieh SY, Ikeda M, Taya Y, Prives C. DNA damage-induced phosphorylation of p53 alleviates inhibition by MDM2. Cell. 1997;91(3):325-334. Harper JW, Adami GR, Wei N, Keyomarsi K, Elledge SJ. The p21 Cdk-interacting protein Cip1 is a potent inhibitor of G1 cyclindependent kinases. Cell. 1993;75(4):805816. Baldwin EL, Osheroff N. Etoposide, topoisomerase II and cancer. Curr Med Chem Anticancer Agents. 2005;5(4):363-372. Rodier F, Campisi J. Four faces of cellular senescence. J Cell Biol. 2011;192(4):547556. Prieur A, Peeper DS. Cellular senescence in vivo: a barrier to tumorigenesis. Curr Opin Cell Biol. 2008;20(2):150-155. Zhu J, Woods D, McMahon M, Bishop JM. Senescence of human fibroblasts induced by oncogenic Raf. Genes Dev. 1998; 12(19):2997-3007. Lin AW, Barradas M, Stone JC, van Aelst L, Serrano M, Lowe SW. Premature senescence involving p53 and p16 is activated in response to constitutive MEK/MAPK mitogenic signaling. Genes Dev. 1998; 12(19):3008-3019. Michaloglou C, Vredeveld LC, Soengas MS, et al. BRAFE600-associated senescence-like cell cycle arrest of human naevi. Nature. 2005;436(7051):720-724. Calabrese V, Mallette FA, DeschenesSimard X, et al. SOCS1 links cytokine signaling to p53 and senescence. Mol Cell. 2009;36(5):754-767.

D. Ingenhag et al. 40. Sabo A, Kress TR, Pelizzola M, et al. Selective transcriptional regulation by Myc in cellular growth control and lymphomagenesis. Nature. 2014;511(7510):488-492. 41. Collado M, Gil J, Efeyan A, et al. Tumour biology: senescence in premalignant tumours. Nature. 2005;436(7051):642. 42. Braig M, Lee S, Loddenkemper C, et al. Oncogene-induced senescence as an initial barrier in lymphoma development. Nature. 2005;436(7051):660-665. 43. Chen Z, Trotman LC, Shaffer D, et al. Crucial role of p53-dependent cellular senescence in suppression of Pten-deficient tumorigenesis. Nature. 2005; 436(7051): 725-730. 44. Akunuru S, Geiger H. Aging, Clonality, and

45. 46.

47.

48.

Rejuvenation of Hematopoietic Stem Cells. Trends Mol Med. 2016;22(8):701-712. Rossi DJ, Jamieson CH, Weissman IL. Stems cells and the pathways to aging and cancer. Cell. 2008;132(4):681-696. Rundberg Nilsson A, Soneji S, Adolfsson S, Bryder D, Pronk CJ. Human and Murine Hematopoietic Stem Cell Aging Is Associated with Functional Impairments and Intrinsic Megakaryocytic/Erythroid Bias. PloS One. 2016;11(7):e0158369. Satake N, Maseki N, Nishiyama M, et al. Chromosome abnormalities and MLL rearrangements in acute myeloid leukemia of infants. Leukemia. 1999;13(7):10131017. Slater RM, von Drunen E, Kroes WG, et al.

t(7;12)(q36;p13) and t(7;12)(q32;p13)-translocations involving ETV6 in children 18 months of age or younger with myeloid disorders. Leukemia. 2001;15(6):915-920. 49. Taketani T, Taki T, Sako M, Ishii T, Yamaguchi S, Hayashi Y. MNX1-ETV6 fusion gene in an acute megakaryoblastic leukemia and expression of the MNX1 gene in leukemia and normal B cell lines. Cancer Genet Cytogenet. 2008;186(2):115119. 50. Balgobind BV, Hollink IH, Arentsen-Peters ST, et al. Integrative analysis of type-I and type-II aberrations underscores the genetic heterogeneity of pediatric acute myeloid leukemia. Haematologica. 2011; 96(10): 1478-1487.

haematologica | 2019; 104(1)

ARTICLE

Iron metabolism & its Disorders

Macrophage ferroportin is essential for stromal cell proliferation in wound healing

Ferrata Storti Foundation

Stefania Recalcati,1# Elena Gammella,1# Paolo Buratti,1 Andrea Doni,2 Achille Anselmo,2 Massimo Locati2,3 and Gaetano Cairo1

1 Department of Biomedical Sciences for Health, University of Milan; 2Humanitas Clinical and Research Center, Rozzano and 3Department of Medical Biotechnologies and Translational Medicine, University of Milan, Italy

# SR and EG contributed equally to this work

Haematologica 2019 Volume 104(1):47-58

ABSTRACT

ron recycling by macrophages is essential for erythropoiesis, but may also be relevant for iron redistribution to neighboring cells at the local tissue level. Using mice with iron retention in macrophages due to targeted inactivation of the iron exporter ferroportin, we investigated the role of macrophage iron release in hair follicle cycling and wound healing, a complex process leading to major clinical problems, if impaired. Genetic deletion of ferroportin in macrophages resulted in iron deficiency and decreased proliferation in epithelial cells, which consequently impaired hair follicle growth and caused transient alopecia. Hair loss was not related to systemic iron deficiency or anemia, thus indicating the necessity of local iron release from macrophages. Inactivation of macrophage ferroportin also led to delayed skin wound healing with defective granulation tissue formation and diminished fibroplasia. Iron retention in macrophages had no impact on the inflammatory processes accompanying wound healing, but affected stromal cell proliferation, blood and lymphatic vessel formation, and fibrogenesis. Our findings reveal that iron/ferroportin plays a largely underestimated role in macrophage trophic function in skin homeostasis and repair.

Introduction Tissue resident macrophages play an important role both in tissue homeostasis, by supporting neighboring parenchymal cells with trophic signals and nutrients, and in tissue repair following injury.1-4 In the skin context, macrophages are critical regulators of hair follicle growth5 and cutaneous wound healing, two events with many similarities.6 Indeed, perifollicular macrophages prompt the entry of hair follicle stem cells into the anagen phase of growth,7 while selective ablation of macrophages impairs the wound healing response.8 Although wound macrophages display a mixed phenotypic and functional profile, the initial phase of an injury is characterized by the prevalence of pro-inflammatory, classically activated M1 macrophages, which are associated with the production of oxygen radicals and pro-inflammatory cytokines. Conversely, at later stages during resolution of inflammation and tissue repair, alternatively polarized M2 macrophages oriented to tissue repair and remodeling, predominate.1,9 This M1 to M2 switch is required for normal healing.2 Macrophages are also at the cross-road of iron traffic.10,11 Iron-recycling macrophages provide iron for erythropoiesis by clearing senescent erythrocytes.12 Conversely, iron sequestration by pro-inflammatory macrophages is a wellknown mechanism of efficient bacteriostasis in host defense.13 In line with their different functions in homeostatic and inflammatory conditions, polarized macrophages show considerable differences in their transcriptional profiles,14 including a distinct regulation of genes related to iron metabolism.11,15 Iron retention by M1 macrophages correlates with high expression of the iron storage protein ferritin. Conversely, M2 macrophages display increased heme uptake and production of anti-inflammatory mediators via heme oxygenase-dependent heme haematologica | 2019; 104(1)

Correspondence: gaetano.cairo@unimi.it or massimo.locati@humanitasresearch.it or massimo.locati@unimi.it Received: May 14, 2018. Accepted: August 14, 2018. Pre-published: August 16, 2018. doi:10.3324/haematol.2018.197517 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/47 ÂŠ2019 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.

S. Recalcati et al.

catabolism, as well as high expression of the iron exporter ferroportin (FPN).16 A tight control of iron metabolism is needed for appropriate tissue homeostasis and healing. Excess iron, both in macrophages and in the extracellular milieu, has a deleterious effect on tissue repair,17 and heme iron has pro-oxidant and pro-inflammatory properties, so that its clearance and degradation by M2 macrophages contributes to resolution.18 However, it is also conceivable that increased iron retention in macrophages leads to lower iron availability for neighboring cells, thus compromising the trophic role of macrophages. In fact, given the necessity of iron for many essential biological functions, including cell replication,19,20 defective iron release can jeopardize iron-dependent functions essential for cutaneous homeostasis and efficient tissue restoration. Decreased iron availability could impair the growth of fibroblasts, as well as epithelial and endothelial cells during new tissue formation. Moreover, the hydroxylases necessary for efficient collagen assembly during the repair phase are irondependent enzymes.21 In this study, we investigated the role of macrophage iron metabolism in tissue homeostasis and repair exploiting a mouse line with iron retention in macrophages caused by targeted FPN inactivation in cells of the myeloid lineage, thus avoiding artefactual systemic iron overload and other confounding elements, such as increased local iron accumulation in other cell types. Using the skin as a model tissue, we show that macrophage-dependent FPN-mediated iron release is required for hair growth in homeostatic conditions and for efficient wound healing, a process which is essential for survival and also clinically relevant, as nonhealing wounds are a major clinical problem associated with various human diseases.22,23

Methods Animals The crossing of mice carrying a floxed Fpn allele (Fpnfl/fl),24 provided by Dr Nancy Andrews, with mice expressing Cre under the control of the LysM promoter in the C57BL/6J background25 in order to generate mice with specific FPN-macrophage inactivation (Fpn1fl/flLysCre+/-) is described in detail in the Online Supplementary Material. Procedures involving animals handling and care conformed with protocols approved by the Humanitas Clinical and Research Center in compliance with national (DL 116, GU suppl. 40, 18-2-1992; DL 26, GU 4-3-2014) and international law and policies (EEC Council Directive 2010/63/EU, OJ L 276/33, 22â&#x20AC;&#x201C;09-2010; National Institutes of Health Guide for the Care and Use of Laboratory Animals, US National Research Council, 2011). The study was approved by the Italian Ministry of Health. All efforts were made to minimize the number of animals used and their suffering.

Results Ferroportin deletion in macrophages causes hair follicle alterations and alopecia To generate mice that lack FPN in macrophages, we crossed Fpnfl/fl mice24 with LysMCre mice25 to create myeloid cell-specific FPN knockout mice. The phenotypic characterization of these Fpn1fl/flLysCre+/- mice is described in the Online Supplementary Material and illustrated in Online Supplementary Figure S1. FPN deletion in macrophages resulted in a significant decrease of hematologic parameters, such as hemoglobin level and hematocrit (Figure 1A,B), and red blood cell count, mean corpuscular volume and mean cell hemoglobin (Online Supplementary Table S1), at weaning (3 weeks after birth). Thereafter, both parameters rapidly returned to normal levels and remained almost unaltered until 18 weeks. In line with the mild anemia observed in 2-month old 129/SvEvTac mice lacking macrophage FPN,26 in weaned Fpn1fl/flLysCre+/- mice hemoglobin levels and hematocrit were lower than in Fpn1fl/flLysCre-/- mice at all time-points, although the difference never reached statistical significance (Figure 1A,B). Serum iron levels and transferrin saturation showed a tendency to decrease with age, but were never statistically different between Fpn1fl/flLysCre+/- mice and their control Fpn1fl/flLysCre-/- littermates. The hepatic expression of hepcidin (HAMP), which regulates systemic iron homeostasis by inhibiting FPN,20 showed age-related variations but was not different between Fpn1fl/flLysCre+/- mice and their control littermates. Similarly, we did not detect significant differences in skin HAMP mRNA levels, which were much lower than in the liver (Figure 1B), while hepcidin levels in skin lysates were undetectable. Accordingly, the expression of Fam132b mRNA encoding for erythroferrone, the erythroid regulator of hepcidin,27 was unchanged in both spleen and bone marrow (Figure 1B and Online Supplementary Figure S2). Fpn1fl/flLysCre+/- mice showed diffuse alopecia with sparing of the head in 100% of both male and female mice until the fourth week of age (Figure 1A,C). Histological analysis in 3-week old mice showed no differences between the two genotypes in any organ evaluated, with the exception of the increased iron accumulation in spleen and liver macrophages (Online Supplementary Figure S1) and a moderate/severe and diffuse/multifocal to coalescing dilatation of hair follicles, which contained remnants of hair shafts and keratin, and slight acanthosis of the superficial epidermis (Figure 1C). Alopecia gradually disappeared and hair re-growth was evident starting 2 weeks after weaning (Figure 1A), but minor skin alterations were still detectable in adult Fpn1fl/flLysCre+/- mice, which had a reduced number of hair follicles, multifocal areas of hair shaft rarefaction and a thin hypodermis with an apparent increase of adipose tissue (Figure 1D). Taken together, these results indicate that targeted FPN deletion in macrophages results in severe alterations of the hair follicle and transient alopecia.

Statistical analysis Results are expressed as the mean Âą standard error of mean. Statistical significance between two groups was assessed by an unpaired two-tailed Mann-Whitney test or Student t test with Prism software (GraphPad). For comparison of more than two groups, data were analyzed using one-way analysis of variance (ANOVA). Full details of the Methods are available in the Online Supplementary Material. 48

Alopecia is not related to systemic iron deficiency In Fpn1fl/flLysCre+/- mice hair regrowth was not complete until 3 weeks after weaning, while hemoglobin levels and hematocrit had already returned to normal after 1 week (Figure 1A) and at all time-points there was no difference in serum iron availability between Fpn1fl/flLysCre+/- and control littermates. This suggested that alopecia in Fpn1fl/flLysCre+/mice was not a local reflection of systemic iron deficiency haematologica | 2019; 104(1)

Stromal cells need macrophage iron

Figure 1. Transient alopecia and anemia are present in Fpn1fl/flLysCre+/- mice. (A) Hemoglobin (Hb) levels and hematocrit (Hct) in 3- to 6-week (w) old mice (mean ± SEM of 50 mice for each group; ***P<0.0001, **P<0.001). The histogram at the bottom shows the percentage alopecia at different time-points (mean ± SEM of 50 mice for each group; ***P<0.0001, **P<0.001 versus Fpn1fl/flLysCre-/-). (B) Top: Hb levels, Hct, serum iron and transferrin saturation (TS) in 3-, 6-, 12-, and 18week old mice (mean ± SEM of 50 mice for each group; ***P<0.0001, **P<0.001, *P<0.01). Bottom: hepcidin (HAMP) expression in the liver (solid bars) and skin (striped bars) and spleen erythroferrone (Fam132b) mRNA levels of 3-, 6-, 12-, and 18-week old mice. mRNA levels were measured by quantitative real time polymerase chain reaction and normalized to the housekeeping gene 18S RNA. Data are presented as mean ± SEM of 10 mice for each group. (C) Representative appearance of 3-week old Fpn1fl/flLysCre-/- (left) and Fpn1fl/flLysCre+/- (right) mice and representative histology (dorsal area) of the same mice. Magnification 10X, in the inset 20X. (D) Representative histology of the skin (dorsal area) of adult (12-week old) mice. Tissue sections were stained with hematoxylin and eosin. Magnification: 10X.

haematologica | 2019; 104(1)

S. Recalcati et al.

but a consequence of iron sequestration in skin macrophages, ultimately resulting in impaired hair follicle growth. To further assess this issue, we analyzed hematologic parameters and alopecia after exposure to low iron diet according to the protocol outlined in Figure 2A. Until week 5, both Fpn1fl/flLysCre+/- and control littermates were anemic, and hemoglobin levels and other hematologic parameters were always slightly and not significantly lower in animals lacking macrophage FPN (Figure 2B,C, Online Supplementary Table S1). Serum iron levels and transferrin saturation were lower in 3-week old Fpn1fl/flLysCre+/mice than in control littermates, although not statistically significantly so, but returned to normal levels at 6 weeks without differences between the two strains (Figure 2C). Liver hepcidin expression was repressed by the iron-deficient diet, but was not different between Fpn1fl/flLysCre+/mice and their control littermates (Figure 2C). HAMP mRNA levels in the skin were below the threshold of detection. A significant increase in erythroferrone expression was found in Fpn1fl/flLysCre+/- pups at 3 weeks (Figure 2C and Online Supplementary Figure S2), which is indicative of higher erythropoietic activity. After the introduction of the normal diet, both hepcidin and erythroferrone expression returned to normal levels (Figure 2C and Online Supplementary Figure S2). Mice with loss of macrophage FPN were grossly affected by diffuse alopecia of the trunk throughout the period of exposure to the low iron diet, whereas Fpn1fl/flLysCre-/- mice, despite low serum iron availability, did not develop alopecia (Figure 2B). Remarkably, alopecia did not appear in Fpn1fl/flLysCre-/- mice even after exposure to the iron-deficient diet for 11 weeks. In Fpn1fl/flLysCre+/- mice, after reintroduction of a normal diet, hemoglobin and serum iron returned to normal levels 2 weeks before the restoration of normally haired skin (Figure 2B,C), thus indicating that alopecia and hypoferremia/anemia are not associated. Histological analysis showed that in Fpn1fl/flLysCre+/- mice challenged with the low iron diet alopecia was associated with severe follicular keratosis with intraluminal accumulation of keratin and distorted hair shafts and subsequent dilation of the hair follicles. Conversely, no relevant histopathological changes were found in the haired skin of the Fpn1fl/flLysCre-/- mice maintained in the same dietary conditions (Figure 2D). Both in Fpn1fl/flLysCre+/- and Fpn1fl/flLysCre-/- mice maintained under iron deprivation conditions for 5 weeks and subsequently fed a normal diet for 2 weeks, skin histology showed that hair follicles were in the anagen stage, but in Fpn1fl/flLysCre+/- mice hair shafts did not exit the follicular ostia and follicular keratosis/dilation, increased epidermal hyperplasia and dermal inflammation were observed (Figure 2E).

Ferroportin deletion in macrophages leads to epithelial iron deficiency and decreased proliferation in cutaneous hair follicles Since we showed that iron released by macrophages via FPN supports in vitro cell proliferation,15 an important role for FPN in skin macrophages could be to mediate the release of sufficient iron in the microenvironment for cell multiplication. Indeed, confocal microscopy revealed a significantly lower expression of the proliferation marker Ki67 in the epithelial cells of the hair bulbs of 3-week old Fpn1fl/flLysCre+/- mice (Figure 3). Conversely, in the same cells we found a strong increase of transferrin receptor (TfR1) expression, which is indicative of cellular iron dep50

rivation (Figure 3). Notably, F4/80+ macrophages, which are abundant in the skin stroma but with no differences in number between Fpn1fl/flLysCre+/- and Fpn1fl/flLysCre-/- mice (Figure 3), expressed lower levels of both Ki67 and TfR1 but had an increased content of both the L and H subunits of the iron storage protein ferritin (Figure 3 and Online Supplementary Figure S3) as compared to epithelial cells. Qualitative analysis also showed that in Fpn1fl/flLysCre-/mice ferritin is detectable only in epithelial cells (Online Supplementary Figure S3), whereas in Fpn1fl/flLysCre+/- mice ferritin expression is particularly strong in F4/80 + macrophages. These results suggest that iron retention in resident macrophages, by starving neighboring hair follicle cells of iron and hence inhibiting their proliferation, has detrimental effects on tissue homeostasis.

Ferroportin deletion in macrophages compromises wound healing Resident macrophages support parenchymal cells with trophic signals, particularly under conditions characterized by increased cell proliferation, such as during tissue repair following injury.2 To test the role of macrophagederived iron in this context, we investigated the wound healing process after incisional skin damage during the entire time course of repair, i.e. the early-inflammatory [2 days post injury (dpi)], middle-proliferative (7 dpi), and late-remodeling phases (12 dpi). We first investigated FPN expression in FACS-sorted macrophages from wounds; in Fpn1fl/flLysCre-/- mice FPN mRNA levels progressively increased during repair (Figure 4A), suggesting a predominant role of FPN in the late phases, whereas, as expected, FPN mRNA was always barely detectable in Fpn1fl/flLysCre+/- mice. The analysis of other iron-related genes showed a rise in the expression of TfR1, which mediates iron uptake, and a decrease of ferritin H subunit during the middle-late phase of repair in macrophages of Fpn1fl/flLysCre-/- mice, but not Fpn1fl/flLysCre+/- mice, which is evidence of iron deposition in these cells (Figure 4A). Accordingly, histological analysis showed iron accumulation in wound macrophages of Fpn1fl/flLysCre+/- mice (Figure 4B). Hepcidin-dependent FPN modulation should not play a role in wound healing, as no difference was seen in liver HAMP expression between Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice during wound repair (Online Supplementary Figure S4A) and HAMP expression in FACSsorted macrophages was undetectable. Hepcidin levels in the wound lysate, which were much lower than in serum, were not different between Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice (Online Supplementary Figure S4B). Macroscopic analysis of wound size showed that the process of closure was considerably delayed in Fpn1fl/flLysCre+/- mice than in control littermates, with significantly wider lesions at all time points and a lag of 3-5 days at 3 dpi through 12 dpi (Figure 4C). Histological analysis performed according to the criteria described in Online Supplementary Table S2 supported this observation, as Fpn1fl/flLysCre+/- mice displayed a more prolonged inflammatory response and delayed granulation tissue formation, associated with diminished fibroplasia, whereas mononuclear cells and granulocytes were unchanged (Figure 4D).

Ferroportin deletion in macrophages has no impact on leukocyte recruitment and activation in the wound Given the role of leukocytes in tissue repair,23 we evaluated leukocyte recruitment in our experimental setting. haematologica | 2019; 104(1)

Stromal cells need macrophage iron

Figure 2. Alopecia in Fpn1fl/flLysCre+/- mice is not related to iron deficiency/anemia. (A) Schematic overview of the feeding protocol: pups were fed by dams kept on an iron-deficient diet for 3 weeks until weaning and then maintained on a low iron diet for another 2 weeks followed by a normal diet. (B) Hemoglobin (Hb) levels and hematocrit (Hct) in 3- to 8-week old mice (mean ± SEM of 10 mice for each group). The histogram at the bottom shows the degree of alopecia at different time-points. ***P<0.0001 versus Fpn1fl/flLysCre-/-. (C) Top: Hb levels, Hct, serum iron and transferrin saturation (TS) in 3- to 18-week old mice (mean ± SEM of 10 mice for each group; ***P<0.0001). Bottom: hepcidin (HAMP) and erythroferrone (Fam132b) mRNA levels in the liver and spleen, respectively, of 3- to 18-week old mice. Expression in 3week old mice fed the normal diet is shown in comparison. mRNA levels were measured by quantitative real-time polymerase chain reaction and normalized to the housekeeping gene 18S RNA. Data are presented as mean ± SEM of 10 mice for each group; ***P<0.0001, **P<0.001. (D) Representative histology of the skin (dorsal area) of 3-week old Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice maintained on an iron-deficient diet. Magnification 20X. (E) Representative histology of the skin (dorsal region) after 5 weeks of an iron deficient diet plus 2 weeks of a normal diet. Tissue sections were stained with hematoxylin and eosin. Magnification: 10X; 20X in the insets.

haematologica | 2019; 104(1)

S. Recalcati et al.

Neutrophils (Ly6G+ cells) and eosinophils (CCR3+ cells) were abundant at 2 and 7 dpi and decreased thereafter, whereas an inverse trend was evident for T cells (CD3+ cells) and macrophages (F4/80+ cells), which increased at 12 dpi (Figure 5A). The accumulation kinetics of these cells, which are typical of skin wound healing,22 were not affect-

ed by the presence or absence of FPN in macrophages. Macrophages with different functional orientations have specific roles in the overlapping phases of wound repair.1,9 As iron accumulation in macrophages might favor the expression of inflammatory mediators.17,26,28,29 we evaluated the levels of inflammatory cytokines in

Figure 3. Epithelial iron deficiency and decreased proliferation in cutaneous hair follicles of Fpn1fl/flLysCre+/- mice. Expression and localization of Ki67, F4/80, transferrin receptor (TfR1) and L ferritin subunit (FtL) in cutaneous tissue of Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice was assessed by confocal microscopy. Representative confocal microscopy images for merged signals, Ki67, TfR1, F4/80 and FtL are shown. Quantification of confocal images (5-9 fields of vision/mouse, 3 mice/group) is also reported. ***P<0.0001 versus Fpn1fl/flLysCre-/-. Arrowheads indicate hair bulbs. Bars: 100 mm. Magnification: 40X.

haematologica | 2019; 104(1)

Stromal cells need macrophage iron

Figure 4. Skin repair is delayed in Fpn1fl/flLysCre+/- mice. (A) Macrophages (CD45+/CD11b+/Ly6C+/F4/80+) were sorted by FACS from wounded skin tissue of eight animals/group from Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice at 2, 7 and 12 days post-injury (dpi). Ferroportin (FPN), transferrin receptor 1 (TfR1) and H-ferritin (FtH) mRNA expression was assessed by quantitative real time polymerase chain reaction and normalized to the housekeeping gene 18S RNA. Data are presented as mean Âą SEM; *P<0.01, **P<0.001, ***P<0.0001. (B) Representative histology of Perls' Prussian blue iron staining of dorsal skin samples at 12 dpi in Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/-mice. Magnification 40X. A semi-quantitative evaluation of Perls' iron staining is shown on the right; n=6 for each group; ***P<0.0001. (C) Kinetic analysis of skin excisional wound areas. Values represent mean Âą SEM of 24 values for each group; *P<0.01, **P<0.001 versus Fpn1fl/flLysCre-/-. One representative experiment (6 mice/group) out of four is shown. The inset shows representative macroscopic images of Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice skin wounds at 7 dpi. (D) Histological grading of wounds, based on separate evaluation of distinct features of the wound healing process, at 2, 7 and 12 dpi. IGT: immature granulation tissue, MGT: mature granulation tissue. The semiquantitative score was defined as described in Online Supplementary Table S2; n= 12 for each group; ***P<0.0001, **P<0.001 versus Fpn1fl/flLysCre-/-.

haematologica | 2019; 104(1)

S. Recalcati et al.

Figure 5. Wound-infiltrating leukocytes and macrophage polarization are not altered in Fpn1fl/flLysCre+/- mice. (A) Frequencies of neutrophils (CD11b+/Ly6G+), eosinophils (CD11b+/CCR3+), T lymphocytes (CD3+), macrophages (CD11b+/F4/80+) on CD45+ cells. Frequencies of M1 (MHCII+/CD206-) and M2 (MHCII-/CD206+) polarized macrophages on total live macrophages. Dots and black lines represent single animals and the mean ± SEM, respectively; n=10 mice for each group; ***P<0.0001, **P<0.001, *P<0.01). (B) Bone marrow-derived macrophages from Fpn1fl/flLysCre-/- and Fpn1fl/flLysCre+/- mice were polarized to M1 and M2 macrophages and relative iNOS, TNFα, Arg1, YM1, FPN, TfR1 and CD163 mRNA levels were measured by quantitative real-time polymerase chain reaction at 24 h; results were normalized to the housekeeping gene 18S RNA (mean ± SEM of 12 mice for each group; ***P<0.0001, **P<0.001, *P<0.01).

haematologica | 2019; 104(1)

Stromal cells need macrophage iron

wound lysates, but did not detect significant differences between the two mouse lines at any time-point (Online Supplementary Figure S5). Since iron accumulation in macrophages of Fpn1fl/flLysCre+/- mice could affect the polarization of these cells during the healing process,15 we investigated the distribution of the different polarized macrophages. As expected, an increase in

MHCII+/CD206- M1 macrophages was detected already in the middle-proliferative phase, while a significant increase in MHCII-/CD206+ M2 macrophages was evident only in the late-remodeling phase, but no difference was found between Fpn1fl/flLysCre+/- mice and their control littermates (Figure 5A). Moreover, we evaluated the expression of polarization markers in bone marrow-

Figure 6. Vessel and stromal cell reduction accompanied by iron deficiency and decreased proliferation in wounds of Fpn1fl/flLysCre+/- mice. Expression of CD31, Lyve-1, collagen-1, PDFGR, ÎąSMA, Ki67 and TfR1 after skin wounding at 7 dpi was assessed by confocal microscopy and the positive area expressed as %. Each circle represents an analysis from a single confocal image (5-9 fields of vision/mouse, 6 mice/group), ***P<0.0001, **P<0.001. Representative confocal microscopy images are shown. Bars: 100 mm. Magnification: 40X.

haematologica | 2019; 104(1)

S. Recalcati et al.

derived macrophages exposed in vitro to polarization stimuli, but again no difference was observed in markers for both M1 (iNOS and TNFα) and M2 (Arg1 and YM1) macrophages (Figure 5B). The expression of iron-related genes in bone marrow-derived macrophages of Fpn1fl/flLysCre-/- mice mirrored the pattern previously observed in human polarized macrophages,15 with elevated expression of FPN, TfR1 and the hemoglobin/haptoglobin complex receptor CD163 in M2 macrophages. This finding is in line with the prominent expression of FPN in macrophages during the late phase of repair, when the M2 cell infiltrate is increased (Figures 4A and 5A). Deletion of macrophage FPN resulted in lower expression of TfR1 and CD163 transcript levels in M2 macrophages (Figure 5B), possibly as a consequence of iron accumulation.

Ferroportin deletion in macrophages affects stromal cells during wound healing Since FPN deletion in macrophages significantly affected the wound healing process but iron accumulation in FPN-deficient macrophages did not alter the inflammatory processes associated with wound healing, we evaluated whether the defective iron release by FPN-deficient macrophages affected the biology of surrounding stromal cells in the wound tissue. Confocal analysis at 7 dpi showed reduced expression in Fpn1fl/flLysCre+/- mice, as compared to control littermates, of blood (CD31) and lymph vessel (Lyve-1) endothelium markers (Figure 6). This was accompanied by decreased expression of platelet-derived growth factor receptor-α, a marker of mesenchymal cells, and lower levels of collagen I and alpha smooth muscle actin (αSMA), which are markers of activated fibroblasts and myofibroblasts, respectively (Figure 6). Moreover, in the absence of macrophage FPN, surrounding stromal cells were iron-deficient, as indicated by upregulation of TfR1 expression, and proliferated less than control counterparts, as shown by decreased Ki67 expression (Figure 6). Taken together, these results indicate that the iron retention in macrophages caused by FPN deletion impairs blood vessel formation and stromal cell proliferation, leading to delayed skin repair.

Discussion The role of erythrophagocytic macrophages as a source of iron for erythropoiesis is well established.16 However, macrophages may also be involved in iron redistribution at a local tissue level, thus affecting neighboring cells. We previously showed that FPN-mediated iron release from human macrophages supports in vitro cell proliferation.15 In the present study, we showed that a steady supply of iron released by macrophage FPN is essential for tissue homeostasis in two conditions, follicular development and wound healing, which share many similarities, including fast cell growth rate.6 Our study, therefore, underlines a new iron-related function of macrophages in tissue homeostasis and regeneration, in line with the increasing recognition of these cells’ considerable functional polyvalence and trophic role, in addition to established immunological functions.30 We did not address the effect of FPN gene deletion in other myeloid cells affected in the LysM conditional model here adopted as their contribution to iron storage and release is negligible com56

pared to that of macrophages.11,13 Our findings showing impaired hair follicle growth in mice with FPN deficiency in macrophages are in line with the report of similar hair and skin lesions in mice with altered expression of other proteins of iron metabolism,3133 although in these other settings the presence of systemic iron deficiency/anemia did not allow the relative contributions of circulating iron versus local availability of macrophage-derived iron to be distinguished. We showed that the alopecia in mice with loss of macrophage FPN was not related to limited systemic iron availability, as evidenced by the lack of differences in serum iron availability and the similar hepatic and local hepcidin expression (Figure 1). Evidence that local iron release from macrophages, which are abundant in skin tissue (Figure 3), is more important than systemic iron levels was also provided by the persistence of alopecia after the return of normal hemoglobin and body iron levels (Figures 1 and 2), and by the absence of hair loss in hypoferremic and anemic Fpn1fl/flLysCre-/- mice (Figure 2), even when fed an iron-deficient diet for a long time. In line with the alopecia and delayed entry of the hair follicle into anagen exhibited by mice overexpressing H ferritin,34 we provide evidence that iron release from macrophages is required to sustain the rapid multiplication of hair follicle cells (Figure 3). In the absence of macrophage FPN, follicle epithelial cells are iron-deficient, as demonstrated by the increased expression of TfR1, and have a lower replication rate, as indicated by reduced Ki67 levels. The discrepancy with the lack of alopecia in a similar model of macrophage-specific FPN inactivation reported by Zhang and colleagues26 could be explained by the different iron content of the standard diet used (157 ppm versus 232, respectively) and by the different genetic backgrounds of the mice. Indeed, the role of dietary iron absorption, which is more important in mice than in humans,35 in correcting the alopecia was also indicated by the effect of switching to chow diet at weaning. Alopecia may result from insufficient iron availability caused by decreased local iron release (this study), Matriptase-dependent severe systemic iron deficiency32,33 (although the role of local FPN was not addressed in those studies) and iron sequestration in ferritin.34 The absence of alopecia in Fpn1fl/flLysCre-/- mice kept on an iron-deficient diet for almost 3 months suggests that local iron release may provide iron more directly in a paracrine fashion when circulating iron levels fall. We conclude that the essential role of macrophages in hair follicle cycling5 is not only related to their production of growth stimulators, such as Wnt,7 but also to the ability to supply the growing tissue with iron. Macrophages are part of the nurturing niche of stem cells in various tissues,36 including tumors. The results reported here in skin hair follicles raise the possibility that macrophage-dependent iron provision has a more general role in different stem cell niches. We also report here a similar role for macrophagederived iron during skin wound healing, a complex tissue repair process consisting of overlapping phases of inflammation and tissue remodeling in which macrophages play a key role.2 The use of mice lacking FPN selectively in cells of the myeloid lineage allowed us to define the role of macrophage iron in wound healing in the absence of the systemic iron overload and large local iron accumulation present in other models.17,28 In the present setting, the disruption of iron export from local macrophages delayed wound healing, apparently by preventing neighboring haematologica | 2019; 104(1)

Stromal cells need macrophage iron

mesenchymal and stromal cells from receiving the iron supply necessary for growth/differentiation. In line with the higher FPN expression in M2 than in M1 macrophages,15,37 the lack of macrophage FPN exerts its major effects in the middle-late phase of repair when the M1 to M2 switch occurs. Indeed, in the late stages, normal M2 skin macrophages export iron through enhanced FPN expression, whereas FPN-deficient macrophages accumulate iron with concomitant induction of ferritin and repression of TfR1 (Figure 4A). The lower fibrosis score (Figure 4D) and the decreased expression of collagen-1 and αSMA (Figure 6) show that the stromal component is compromised, as the absence of macrophage FPN resulted in iron deprivation and impaired proliferation of stromal cells (Figure 6). In this context, fibroblasts may not receive enough iron, which can be among the paracrine factors secreted by M2 macrophages to favor cell multiplication.38 A detrimental effect on collagen synthesis and assembly, which require iron-dependent prolyl hydroxylases,21 or other iron-dependent functions such as dihydroxy-docosahexaenoic acid production,39 may contribute to defective repair. Our results also show that macrophage iron is essential for the development of the vascular network during tissue healing, as both lymphatic and blood vessels were reduced (Figure 6). Although the decrease of vascular structure caused by macrophage depletion was previously ascribed to the reduced production of vascular endothelial growth factor and transforming growth factor-β,9 the latter being also involved in extracellular matrix deposition and αSMA expression,40 in our experimental model the levels of these growth factors were unchanged (Online Supplementary Figure S5). Similarly, given that hemoglobin levels of adult Fpn1fl/flLysCre+/- mice were normal, defective oxygenation as a possible factor involved in impaired vascularization can be ruled out. Therefore, our results showing reduced neovessel density, reduced granulation tissue formation and decreased fibrosis in the absence of macrophage iron release, in the face of unchanged levels of prominent angiogenic and fibrogenic factors, such as vascular endothelial growth factor and transforming growth factor-β, support the relevance of the trophic role of macrophage-derived iron in the wound milieu. Understanding the role of iron in macrophage production of inflammatory molecules has been hampered by contradictory findings. An increased inflammatory response was found in iron-depleted macrophages,41 but not in equally iron-deficient macrophages from HFE-/mice,42 and other studies showed that iron levels correlate positively with the synthesis of pro-inflammatory cytokines.26,43 In addition, a pro-inflammatory state has been shown in macrophages and macrophage/microglia cells exposed to heme or iron28,29 and in hemorrhagic areas within tumors.44 Similarly, decreased iron release from

References 1. Brancato SK, Albina JE. Wound macrophages as key regulators of repair: origin, phenotype, and function. Am J Pathol. 2011;178(1):19-25. 2. Wynn TA, Vannella KM. Macrophages in tissue repair, regeneration, and fibrosis.

haematologica | 2019; 104(1)

macrophages, associated with pro-inflammatory activation and defective M2 polarization, impaired wound healing in chronic venous leg ulcers.17 Conversely, we found that iron retention in macrophages has no impact on leukocyte recruitment and activation as well as macrophage polarization (Figure 5 and Online Supplementary Figure S5). Moreover, in vitro polarized bone marrow-derived macrophages from the two mouse lines did not show differential expression of M1 and M2 markers (Figure 5B). Therefore, in our model, iron accumulation does not exacerbate the pro-inflammatory phenotype of wound healing-associated macrophages, in keeping with a recent study showing that iron did not increase M1 polarization of RAW264.7 macrophages.45 The conflicting results may be related to the different experimental models, the heterogeneity of macrophages and the exposure to different iron sources, such as heme iron which is highly toxic.46 In the absence of FPN, macrophages from Fpn1fl/flLysCre+/- mice accumulate iron in ferritin, which increases less than 2-fold (Figure 4A), but iron deposition seems less massive than in conditions such as chronic ulcers,17 in which iron content may increase 20-fold,47 or hemolysis.29,44 In our experimental setting iron accumulation may, therefore, be insufficient to interfere with the M1/M2 switch and favor a proinflammatory state. A recent study demonstrated that FPN downregulation in macrophages impaired skeletal muscle regeneration after injury,48 but the effect of increased iron accumulation on the inflammatory profile of macrophages was not addressed. In conclusion, the results of our study indicate that local macrophage FPN, by supplying iron to cells in the microenvironment, affects both the physiological context of follicular anagen and the pathophysiological context of wound healing. In its absence, stromal cells are iron-deficient and their proliferation is impaired (Figures 3 and 6). The importance of local iron recycling is underlined by the lack of changes in hepatic and skin hepcidin. A similar requirement for iron provided locally by macrophages has been described for the repair of skeletal muscle cells, in which iron retention in macrophages, by impairing myoblast proliferation, results in smaller myofibers.48 Iron should, therefore, be added to the list of trophic mediators produced locally by macrophages that stimulate the growth, differentiation and activity of neighboring parenchymal and stromal cells in order to maintain tissue homeostasis or repair. Acknowledgments This work was supported by grants from the Italian Association for Cancer Research (AIRC-IG 2016 #19213 to ML) and MIUR (COFIN to GC). The authors would like to thank Nancy Andrews for providing Fpnfl/fl mice, Alberto Mantovani for support and helpful comments, and Eugenio Scanziani and Camilla Recordati for their help with the histological analysis.

Immunity. 2016;44(3):450-462. 3. Mantovani A, Biswas SK, Galdiero MR, Sica A, Locati M. Macrophage plasticity and polarization in tissue repair and remodelling. J Pathol. 2013;229(2):176-185. 4. Biswas SK, Mantovani A. Orchestration of metabolism by macrophages. Cell Metab. 2012;15(4):432-437. 5. Stenn KS, Paus R. Controls of hair follicle

cycling. Physiol Rev. 2001;81(1):449-494. 6. Ansell DM, Kloepper JE, Thomason HA, Paus R, Hardman MJ. Exploring the "hair growth-wound healing connection": anagen phase promotes wound re-epithelialization. J Invest Dermatol. 2011;131(2):518-528. 7. Castellana D, Paus R, Perez-Moreno M. Macrophages contribute to the cyclic activation of adult hair follicle stem cells. PLoS

S. Recalcati et al. Biol. 2014;12(12):e1002002. 8. Mirza R, DiPietro LA, Koh TJ. Selective and specific macrophage ablation is detrimental to wound healing in mice. Am J Pathol. 2009;175(6):2454-2462. 9. Lucas T, Waisman A, Ranjan R, et al. Differential roles of macrophages in diverse phases of skin repair. J Immunol. 2010;184(7):3964-3977. 10. Recalcati S, Locati M, Cairo G. Systemic and cellular consequences of macrophage control of iron metabolism. Semin Immunol. 2012;24(6):393-398. 11. Cairo G, Recalcati S, Mantovani A, Locati M. Iron trafficking and metabolism in macrophages: contribution to the polarized phenotype. Trends Immunol. 2011;32(6): 241-247. 12. Soares MP, Hamza I. Macrophages and iron metabolism. Immunity. 2016;44(3):492-504. 13. Ganz T, Nemeth E. Iron homeostasis in host defence and inflammation. Nat Rev Immunol. 2015;15(8):500-510. 14. Martinez FO, Gordon S, Locati M, Mantovani A. Transcriptional profiling of the human monocyte-to-macrophage differentiation and polarization: new molecules and patterns of gene expression. J Immunol. 2006;177(10):7303-7311. 15. Recalcati S, Locati M, Marini A, et al. Differential regulation of iron homeostasis during human macrophage polarized activation. Eur J Immunol. 2010;40(3):824-835. 16. Drakesmith H, Nemeth E, Ganz T. Ironing out ferroportin. Cell Metab. 2015;22(5):777787. 17. Sindrilaru A, Peters T, Wieschalka S, et al. An unrestrained proinflammatory M1 macrophage population induced by iron impairs wound healing in humans and mice. J Clin Invest. 2011;121(3):985-997. 18. Lundvig DM, Immenschuh S, Wagener FA. Heme oxygenase, inflammation, and fibrosis: the good, the bad, and the ugly? Front Pharmacol. 2012;3:81. 19. Cairo G, Bernuzzi F, Recalcati S. A precious metal: Iron, an essential nutrient for all cells. Genes Nutr. 2006;1(1):25-39. 20. Muckenthaler MU, Rivella S, Hentze MW, Galy B. A red carpet for iron metabolism. Cell. 2017;168(3):344-361. 21. Markolovic S, Wilkins SE, Schofield CJ. Protein hydroxylation catalyzed by 2-oxoglutarate-dependent oxygenases. J Biol Chem. 2015;290(34):20712-20722. 22. Singer AJ, Clark RA. Cutaneous wound healing. N Engl J Med. 1999;341(10):738-746. 23. Gurtner GC, Werner S, Barrandon Y,

24.

25.

26.

27.

28.

29.

30.

31.

32. 33.

34.

35. 36.

Longaker MT. Wound repair and regeneration. Nature. 2008;453(7193):314-321. Donovan A, Lima CA, Pinkus JL, et al. The iron exporter ferroportin/Slc40a1 is essential for iron homeostasis. Cell Metab. 2005;1(3):191-200. Clausen BE, Burkhardt C, Reith W, Renkawitz R, Fรถrster I. Conditional gene targeting in macrophages and granulocytes using LysMcre mice. Transgenic Res. 1999;8(4):265-277. Zhang Z, Zhang F, An P, et al. Ferroportin1 deficiency in mouse macrophages impairs iron homeostasis and inflammatory responses. Blood. 2011;118(7):1912-1922. Kautz L, Jung G, Valore EV, Rivella S, Nemeth E, Ganz T. Identification of erythroferrone as an erythroid regulator of iron metabolism. Nat Genet. 2014;46(7):678-684. Kroner A, Greenhalgh AD, Zarruk JG, Passos Dos Santos R, Gaestel M, David S. TNF and increased intracellular iron alter macrophage polarization to a detrimental M1 phenotype in the injured spinal cord. Neuron. 2014;83(5):1098-1116. Vinchi F, Costa da Silva M, Ingoglia G, et al. Hemopexin therapy reverts heme-induced proinflammatory phenotypic switching of macrophages in a mouse model of sickle cell disease. Blood. 2016;127(4):473-486. Wynn TA, Chawla A, Pollard JW. Macrophage biology in development, homeostasis and disease. Nature. 2013;496(7446):445-455. Nicolas G, Bennoun M, Porteu A, et al. Severe iron deficiency anemia in transgenic mice expressing liver hepcidin. Proc Natl Acad Sci USA. 2002;99(7):4596-4601. Du X, She E, Gelbart T, et al. The serine protease TMPRSS6 is required to sense iron deficiency. Science. 2008;320(5879):1088-1092. Folgueras AR, de Lara FM, Pendas AM, et al. Membrane-bound serine protease matriptase-2 (Tmprss6) is an essential regulator of iron homeostasis. Blood. 2008;112(6):25392545. Hasegawa S, Harada K, Morokoshi Y, Tsukamoto S, Furukawa T, Saga T. Growth retardation and hair loss in transgenic mice overexpressing human H-ferritin gene. Transgenic Res. 2013;22(3):651-658. Ganz T. Systemic iron homeostasis. Physiol Rev. 2013;93(4):1721-1741. Kaur S, Raggatt LJ, Batoon L, Hume DA, Levesque JP, Pettit AR. Role of bone marrow macrophages in controlling homeostasis and repair in bone and bone marrow niches. Semin Cell Dev Biol. 2017;61:12-21.

37. Corna G, Campana L, Pignatti E, et al. Polarization dictates iron handling by inflammatory and alternatively activated macrophages. Haematologica. 2010;95(11): 1814-1822. 38. Ploeger DT, Hosper NA, Schipper M, Koerts JA, de Rond S, Bank RA. Cell plasticity in wound healing: paracrine factors of M1/ M2 polarized macrophages influence the phenotypical state of dermal fibroblasts. Cell Commun Signal. 2013;11(1):29. 39. Lu Y, Tian H, Hong S. Novel 14,21-dihydroxy-docosahexaenoic acids: structures, formation pathways, and enhancement of wound healing. J Lipid Res. 2010;51(5):923392. 40. Hinz B. Formation and function of the myofibroblast during tissue repair. J Invest Dermatol. 2007;127(3):526-537. 41. Pagani A, Nai A, Corna G, et al. Low hepcidin accounts for the proinflammatory status associated with iron deficiency. Blood. 2011;118(3):736-746. 42. Roy CN, Custodio AO, de Graaf J, et al. An Hfe-dependent pathway mediates hyposideremia in response to lipopolysaccharide-induced inflammation in mice. Nat Genet. 2004;36(5):481-485. 43. Wang L, Johnson EE, Shi HN, Walker WA, Wessling-Resnick M, Cherayil BJ. Attenuated inflammatory responses in hemochromatosis reveal a role for iron in the regulation of macrophage cytokine translation. J Immunol. 2008;181(4):27232731. 44. Costa da Silva M, Breckwoldt MO, Vinchi F, et al. Iron induces anti-tumor activity in tumor-associated macrophages. Front Immunol. 2017;8:1479. 45. Gan ZS, Wang QQ, Li JH, Wang XL, Wang YZ, Du HH. Iron reduces M1 macrophage polarization in RAW264.7 macrophages associated with inhibition of STAT1. Mediators Inflamm. 2017;2017:8570818. 46. Gozzelino R, Jeney V, Soares MP. Mechanisms of cell protection by heme oxygenase-1. Annu Rev Pharmacol Toxicol. 2010;50:323-354. 47. Ackerman Z, Seidenbaum M, Loewenthal E, Rubinow A. Overload of iron in the skin of patients with varicose ulcers. Possible contributing role of iron accumulation in progression of the disease. Arch Dermatol. 1988;124(9):1376-1378. 48. Corna G, Caserta I, Monno A, et al. The repair of skeletal muscle requires iron recycling through macrophage ferroportin. J Immunol. 2016;197(5):1914-1925.

haematologica | 2019; 104(1)

ARTICLE

Myelodysplastic Syndromes

Fetal hemoglobin induction during decitabine treatment of elderly patients with high-risk myelodysplastic syndrome or acute myeloid leukemia: a potential dynamic biomarker of outcome Julia Stomper,1 Gabriele Ihorst,2 Stefan Suciu,3 Philipp N. Sander,1,4 Heiko Becker,1 Pierre W. Wijermans,5 Christoph Plass,6,7 Dieter Weichenhan,6 Emmanuel Bissé,8 Rainer Claus1,9 and Michael Lübbert1,10

Department of Hematology, Oncology, and Stem Cell Transplantation, Faculty of Medicine and Medical Center, University of Freiburg, Germany; 2Clinical Trials Unit, Faculty of Medicine and Medical Center, University of Freiburg, Germany; 3EORTC Headquarters, Brussels, Belgium; 4Department of Chemistry, The Scripps Research Institute, La Jolla, CA, USA; 5Department of Hematology, Haga Hospital, The Hague, the Netherlands; 6DKFZ Heidelberg, Division of Epigenomics and Cancer Risk Factors, Heidelberg, Germany; 7 German Cancer Research Consortium (DKTK), Heidelberg, Germany; 8Institute for Clinical Chemistry and Laboratory Medicine, Faculty of Medicine and Medical Center, University of Freiburg, Germany; 9Department of Internal Medicine II, Hematology/Oncology, Augsburg Medical Center, Germany and 10German Cancer Research Consortium (DKTK), Freiburg, Germany 1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):59-69

ABSTRACT

ematologic responses to hypomethylating agents are often delayed in patients with myelodysplastic syndrome or acute myeloid leukemia. Fetal hemoglobin is a potential novel biomarker of response: recently, we demonstrated that a high fetal hemoglobin level prior to decitabine treatment was associated with superior outcome. Here we investigated whether early fetal hemoglobin induction during decitabine treatment also had prognostic value, and studied the potential of decitabine to induce erythroid differentiation and fetal hemoglobin expression in vitro. Fetal hemoglobin levels were measured by high-performance liquid chromatography in patients with higher-risk myelodysplastic syndrome (n=16) and acute myeloid leukemia (n=37) before treatment and after each course of decitabine. Levels above 1.0% were considered induced. Patients achieving complete or partial remission as best response had attained a median fetal hemoglobin of 1.9% after two courses of treatment, whereas the median value in patients who did not reach complete or partial remission was 0.8% (P=0.015). Fetal hemoglobin induction after two courses of decitabine treatment was associated with early platelet doubling (P=0.006), and its subsequent decrease with hematologic relapse. In patients with myelodysplastic syndrome, induction of fetal hemoglobin after course 2 of treatment was associated with longer overall survival: median of 22.9 versus 7.3 months in patients with or without induction of fetal hemoglobin, respectively [hazard ratio=0.2 (95% confidence interval: 0.1-0.9); P=0.03]. In vitro decitabine treatment of two bi-potential myeloid leukemia cell lines (K562 and HEL) resulted in induction of an erythroid (not megakaryocytic) differentiation program, and of fetal hemoglobin mRNA and protein, associated with GATA1 gene demethylation and upregulation. In conclusion, fetal hemoglobin may provide a useful dynamic biomarker during hypomethylating agent therapy in patients with myelodysplastic syndrome or acute myeloid leukemia. haematologica | 2019; 104(1)

Correspondence: michael.luebbert@uniklinik-freiburg.de

Received: December 30, 2017. Accepted: August 28, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2017.187278 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/59 ©2019 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.

J. Stomper et al.

Introduction DNA-hypomethylating agents (HMAs) alter gene expression of malignant cells by gene reactivation, e.g. of epigenetically silenced tumor suppressor genes resulting in induction of apoptosis or senescence,1-4 or of cancer/testis antigens and retroviral sequences, thus eliciting an immune response against cancer cells,5,6 and by reduction of oncogene overexpression.7 In cell line models of acute myeloid leukemia (AML), they also induce granulocytic maturation although this effect has yet to be validated in vivo.8 Probably the earliest proof of therapeutic gene reactivation by an HMA was the demonstration of induction of fetal hemoglobin (HbF) in a patient with severe beta-thalassemia through demethylation and transcriptional activation of the gamma-globin gene locus,9 the developmental regulation of which is governed by DNA methylation. This ability of HMAs to de-repress epigenetically silenced gamma-globin expression in (polyclonal) erythropoietic precursors has also been observed in patients with solid tumors10,11 and has prompted clinical trials on the use of HMAs in hemoglobinopathies.12,13 Several groups have shown that in patients with myelodysplastic syndrome (MDS) or AML receiving azacitidine, HbF is induced and may be a marker of treatment effect.14,15 Very recently, we observed elevated HbF levels also in MDS/AML patients treated with decitabine.16 We now dissected the kinetics of the in vivo induction of HbF in these patients with high-risk MDS or AML, in order to evaluate the prognostic value of HbF induction. We observed that HbF levels after two courses of treatment were of prognostic value regarding hematologic responses of the different lineages, and, in MDS, for overall survival from this time point on [hazard ratio (HR)=0.2; 95% confidence interval (95% CI): 0.1-0.9; P=0.03]. The cellular source of HbF in responders appeared to be predominantly non-malignant erythroid precursors. To model HbF reactivation in vitro, we investigated two transformed leukemia cell lines for the induction of an erythroid differentiation program that also encompassed HbF protein synthesis induced by decitabine.

Statistical methods We used non-parametric Spearman correlation coefficients (rs) to assess the association between HbF levels and other variables. To address the hypothesis that HbF after course 2 of decitabine treatment may predict hematologic responses after course 4, we employed linear regression models with HbF levels as the independent variable, and platelet count, hemoglobin concentration, neutrophil counts, and bone marrow blasts as dependent variables. Results are displayed graphically. The evaluation of possible outliers was based on studentized residuals.18 No potential outliers were excluded from the statistical analyses. The HbF value distributions between responders and non-responders were compared using the Wilcoxon two-sample test. Before-after differences in binary variables such as elevated HbF (yes/no) were assessed statically using the McNemar test. Time-to-event endpoints comprised overall survival, progression-free survival, and AML-free survival. We performed landmark analyses starting at the time of HbF measurements after two cycles of decitabine treatment in order to avoid the so-called immortal-time bias.19 Thus, they were relevant only for patients who reached this time point. The Kaplan-Meier method was used to estimate distributions of overall, progression-free and AML-free survival according to HbF value (> versus ≤1%), and to compute median estimations. A log-rank test was used to assess the prognostic importance of HbF value, and a Cox model to estimate the hazard ratio (HR) and its 95% confidence interval (95% CI). Statistical analyses (all conducted at the CTU Freiburg, Germany) were performed with SAS 9.2 (SAS Institute Inc., Cary, NC, USA), and IBM SPSS Statistics 22.

Cell lines and in vitro treatment K562 and HEL cells (DMSZ, Braunschweig, Germany) were cultured in RPMI1640 containing 10% heat-inactivated fetal calf serum. Cells were treated with three 24-h pulses of 100 and 20 nM decitabine (Sigma Aldrich) and harvested 144 h after the first treatment, as described previously.20 Hemin (Sigma Aldrich), a positive control for induction of erythroid differentiation,21 was added to the culture medium at a final concentration of 50 mM.22 Cell viability was tested by 0.4% trypan blue staining. Megakaryocytic differentiation was induced by exposure to 5 nM phorbol ester (PMA, Sigma Aldrich) for 48 h, as described previously.23 . Further information on the methods, including transcriptome profiling, immunoblotting, and methyl-CpG immunoprecipitation sequencing, is provided in the Online Supplement.

Methods Patients, treatments and response evaluations Details on the patients, the treatments given and evaluation of responses are provided in the Online Supplement. Both the therapeutic studies and the translational investigations were approved by the institutional review board (ethics committee) of the University of Freiburg Medical Center. Patients (all treated at the Freiburg study site) provided their written informed consent to be included in the respective clinical studies, and to related translational investigations according to the Declaration of Helsinki.

Hemoglobin quantification by high-performance liquid chromatography HbF levels were measured by high-performance liquid chromatography (HPLC) before treatment and after the end of each treatment course (every 6 weeks) as described previously.17 Patients were grouped by normal (HbF ≤1%) or elevated HbF levels (HbF >1%), in accordance with reference values for the University of Freiburg Medical Center Central Laboratory. 60

Results Kinetics of in vivo induction of HbF in patients with myelodysplastic syndrome/acute myeloid leukemia receiving decitabine In 40 patients (15 with higher-risk MDS, 25 with AML), HbF was measured serially twice or more during treatment (in a single patient, HbF before treatment was not available). The patients’ characteristics and those of their diseases are presented in Table 1. When analyzing HbF kinetics over time (Figure 1A, Online Supplementary Figure S1), the first on-treatment time point with robust HbF induction was at the end of the second course of decitabine treatment. Compared to the median pre-treatment HbF of 0.8%, we observed overall induction to 1.1% after course 2. The median fold-change of HbF concentration at the end of the second course of decitabine treatment as compared to pre-treatment levels was 1.67 (P=0.088). haematologica | 2019; 104(1)

HbF induction in AML/MDS and outcome prediction

By grouping patients according to pre-treatment HbF levels (normal versus elevated) and subsequent HbF levels (normal versus increased), four different patterns could be discerned. In patients with normal pre-treatment HbF, HbF was either induced to levels above the upper limit of normal (group A, n=11) or remained within the normal range (group B, n=12); among patients with elevated pretreatment levels, HbF either remained elevated also in subsequent measurements (group C, n=13, in most cases after an initial drop following the first treatment course) or dropped to normal levels, without subsequent elevation above the upper limit of normal (group D, n=3). Two vignettes representative of each of the four different patterns of HbF kinetics are shown in Figure 1B. The number of treatment courses varied between these four groups: median numbers were 5, 2, 7 and 3 courses in groups A, B, C and D, respectively (range: 2-11, 2-6, 2-23, and 2-6 courses, respectively). Consequently, more patients in groups A and C attained hematologic responses over time, including complete remission (with suppression of the abnormal clone in patients with initial cytogenetic abnormalities), compared to patients in groups B and D. HbF levels were elevated in significantly more patients during treatment (groups A+C: n=24, 61.5%) than prior to treatment (groups C+D: n=16, 41.0%; P=0.033).

observed when restricting the same kind of analysis to MDS and AML patients who had received more than four courses of decitabine (data not shown). In patients achieving platelet doubling already after one course of decitabine, the median HbF after course 2 was 1.9%, versus 0.8% in patients without this platelet response (P=0.006) (Figure 2E). In contrast, no correlation was observed between HbF levels and platelet counts before treatment initiation: rs=-0.13 (P=0.63) in MDS (n=16) and rs=0.02 (P=0.91) in AML patients (n=36).

Early HbF induction by decitabine has predictive value for subsequent hematologic stabilization and response

Table 1. Characteristics of MDS and AML patients and their disease at baseline.

Among patients achieving complete or partial remission as their best overall response, a median HbF of 1.9% (1.9% in MDS, 2.0% in AML) was observed at this time point compared to a median of 0.8% (1.0% and 0.6%, respectively) in patients not attaining complete or partial remission (P=0.015) (Figure 1C, Online Supplementary Figure S2). Given that robust HbF induction during decitabine treatment was first observed after two cycles of treatment, we chose this time point for further analyses regarding the potential value of HbF induction for predicting subsequent responses. Using linear regression models and correlation analyses, we assessed the association between HbF levels after course 2 of decitabine treatment and responses in the different hematopoietic lineages after course 4 in all 40 MDS/AML patients (3 patients did not receive more than 2 courses, but had peripheral blood counts available 3 months after course 2: one underwent hematopoietic stem cell transplantation, one changed to a different study after course 2, and one received thalidomide and best supportive care at that point). Higher HbF levels after two courses were associated with significantly higher platelet counts after four courses (rs=0.49, P=0.01) (Figure 2A). As for neutrophil counts, a trend towards higher counts after course 4 in patients with elevated HbF levels after course 2 was noted (rs=0.35, P=0.08) (Figure 2B). A borderline significant association was also observed between HbF levels after two courses of treatment and hemoglobin levels after four courses (rs=0.36, P=0.08) (Figure 2C). Interrogating the potential prognostic value of HbF induction for bone marrow blast suppression after four courses in the entire cohort of 21 MDS/AML patients with available blast counts, an association was noted (rs=-0.48, P=0.03) (Figure 2D). When looking at MDS and AML patients separately, the overall results were similar in both cohorts, particularly for blast suppression (Online Supplementary Figure S4A-D). Moreover, similar results were haematologica | 2019; 104(1)

Association of HbF induction by decitabine and survival outcomes in patients with myelodysplastic syndrome/acute myeloid leukemia Overall, progression-free and AML-free survival were measured from the time of HbF determination after completion of two courses of decitabine treatment. In the MDS cohort, overall survival was significantly longer in the nine patients in whom HbF was elevated above the upper limit of normal at that time point than in the six patients with HbF in the normal range: median 22.9 versus 7.3 months, respectively (HR=0.21; 95% CI: 0.05-0.87; P=0.03) (Figure 3A). Censoring the six patients who had

Age (years; median and range) Sex Male Female FAB subtype

MDS

AML

74 (66 - 77)

73 (62 - 82)

8 7

15 10

RA: 1 RAEB: 8 RAEB-t: 4 CMMoL: 2

M1: 4 M2: 2 M4: 5 M5: 1 M6: 3 N/A: 10 2.2 (0.5 - 27.1) 0.33 (0 - 14.09) 19 (0 - 96)

WBC (x 109/L; median and range) 2.4 (0.8 - 18.8) ANC (x 109/L; median and range) 0.93 (0.16 - 2.40) Peripheral blood blasts 1 (0 - 13) (%; median and range) Bone marrow blasts 18 (2 - 30) 55 (10 - 95) (%; median and range) Hb (g/dL; median and range) 8.4 (5.0 - 11.3) 8.7 (5.6 - 12.4) PLT (x 109/L; median and range) 35 (10 - 151) 34 (7 - 229) sLDH (U/l; median and range) 235 (102 - 842) 271 (134 - 1276) Cytogenetics Normal: 3 Intermediate risk: 7 Sole 5q-: 1 Poor risk: 6 Sole -7: 1 < 10 normal metaphases: 4 Complex: 6 No metaphases: 1 No metaphases: 4 N/A: 7 MDS: myelodysplastic syndrome; AML: acute myeloid leukemia. RA, refractory anemia; RAEB: refractory anemia with blast excess; RAEB-t: refractory anemia with blast excess in transformation (i.e. AML according to the World Health Organization classification); CMMoL: chronic myelomonocytic leukemia. FAB (French-American-British) subtypes: M1: acute myeloblastic leukemia with minimal maturation; M2: acute myeloblastic leukemia with maturation; M4: acute myelomonocytic leukemia; M5: acute monocytic leukemia; M6: acute erythroid leukemia. N/A: not assessed; WBC: white blood cells; ANC: absolute neutrophil counts; Hb: hemoglobin; PLT: platelets; sLDH: serum lactate dehydrogenase.

J. Stomper et al.

Figure 1. HbF kinetics in patients with myelodysplastic syndrome/acute myeloid leukemia during treatment with decitabine. (A) Box plot of HbF values prior to treatment with decitabine and after each of the first eight treatment courses in patients with myelodysplastic syndrome (n=15) or acute myeloid leukemia (n=25) for whom there was more than one measurement of HbF level. HbF was measured prior to treatment (= cycle 0) in 39 of these patients, after courses 1 to 8 (= cycles 1, 2, 3, etc.) in 38, 32, 24, 18, 14, 10, 7, and 9 of these patients, respectively. Outliers are plotted as individual points, extreme values as stars. For each of cycles 0, 1, and 3, one extreme value is not displayed in the diagram, and one outlier is not shown for cycle 2. (B) Two representative vignettes of each of the four groups of HbF responses to decitabine treatment are shown: initially normal HbF induced to levels above (group A) or remaining within the normal range during treatment (group B); already elevated pre-treatment HbF remaining elevated (group C) or decreasing to normal levels (group D). Shaded areas within the charts indicate values outside the normal range. The end of each decitabine treatment course is indicated with a numbered arrow. Hb: hemoglobin; HbF: fetal hemoglobin; PLT: platelets; WBC: white blood cells. (C) Box plot of HbF values at the end of the second course of decitabine treatment in 32 MDS/AML patients showing higher HbF levels in patients achieving complete or partial remission (CR, PR) as best overall response (right) compared to patients who did not attain CR/PR (left).

haematologica | 2019; 104(1)

HbF induction in AML/MDS and outcome prediction

Figure 2. HbF levels after two courses of decitabine treatment are associated with responses of different hematologic lineages after course 4, and with early platelet doubling. HbF values after treatment course 2 (x-axis) in patients with myelodysplastic syndrome (MDS)/acute myeloid leukemia (AML) who received at least two cycles of decitabine were plotted against (A) platelet counts (n=25) (y-axis), (B) absolute neutrophil counts (n=25), (C) hemoglobin levels (n=25), and (D) the percentage of bone marrow blasts (n=21), as determined after course 4. Linear regression equations are represented as lines and were estimated as (A) platelets=36+41*HbF, (B) absolute neutrophil count=1243+340*HbF, (C) hemoglobin=9.4+0.5*HbF, (D) percentage bone marrow blasts=25-6*HbF. (E) Box plot demonstrating significantly higher HbF values after the second course of decitabine treatment in MDS/AML patients who achieved platelet doubling already after one course of decitabine.

undergone hematopoietic stem cell transplantation at the time of their transplant, this difference lost statistical significance (P=0.098) (Online Supplementary Figure S6). The secondary endpoints, progression-free survival and AMLfree survival were also investigated and showed a trend to a more favorable outcome for patients with elevated HbF: the median progression-free survival was 7.7 months versus 2.4 months (HR=0.32; 95% CI: 0.10-1.10; P=0.07) (Figure 3B) and the median AML-free survival was 13.1 months versus 7.6 months (HR=0.42; 95% CI: 0.13-1.38; P=0.15) (Figure 3C). In the AML cohort, 17 patients had HbF determinations available after two courses of treatment and could be included in a survival analysis. There was no significant difference in overall survival between those with elevated haematologica | 2019; 104(1)

HbF (n=9) and those with normal HbF (n=8): the median overall survival was 17.3 and 11.6 months, respectively (HR=0.68; 95% CI: 0.23-1.96; P=0.47) (Figure 3D).

Elevated HbF levels are observed in decitabine-treated patients with myelodysplastic syndrome/acute myeloid leukemia with bone marrow blast clearance, and decline at relapse To investigate whether malignant cells or normal, emerging erythroid precursors synthesize the elevated HbF in response to HMAs, linear regression analyses were performed (Figure 4A). This showed an association between HbF induction after four treatment courses and a decreasing percentage of blasts at this time point in 17 MDS/AML patients. 63

J. Stomper et al. A

Figure 3. Elevated HbF after course 2 of decitabine treatment is associated with improved survival in patients with myelodysplastic syndrome. Kaplan-Meier survival estimates were determined according to whether HbF after course 2 of decitabine treatment was in the normal range (i.e. 0-1.0% of total hemoglobin) or elevated (i.e. >1.0%). (A) Overall survival of 15 patients with myelodysplastic syndrome (MDS), two in the group with elevated HbF censored at 100 months (still alive in remission following allografting). (B) Progression-free survival of 15 MDS patients, one in the group with elevated HbF censored (still alive in remission following allografting). (C) Acute myeloid leukemia (AML)-free survival of 15 MDS patients, one in the group with elevated HbF censored (still alive in remission following allografting). (D) Overall survival of 17 AML patients, one in the group with elevated HbF censored (still alive in remission following allografting).

Looking in more detail at the subgroup of 18 patients who received more than four courses of treatment, it was noted that six attained not only a multilineage hematologic response but also bone marrow blast clearance. Interestingly, in all of them HbF levels were, or became, elevated at the time of remission. In four of these patients, the complete remission status could be confirmed by cytogenetics: chromosomal abnormalities present at baseline were no longer detectable at that point (cytogenetic remission; in several of these patients, suppression of the abnormal clone was confirmed by fluorescence in situ hybridization). Taken together, these analyses strongly suggested that HbF induction observed after four courses of treatment occurred preferentially in the emerging non-clonal erythroid cells, in which epigenetically silenced beta-globinlike genes were reactivated by treatment with the HMA. For seven MDS/AML patients, HbF measurements were obtained both after the end of decitabine course 2 and at the time of hematologic relapse. Relative to an elevated HbF level after course 2 (median 1.9%; range, 1.5-4.6%), a decrease was observed in all seven patients at the time of relapse (median 1.1%; range, 0.2-1.9%). In two of these 64

patients (both of whom had a complex karyotype), the decline in HbF level preceded the hematologic relapse, implicating it as a potential early predictor of relapse in a subgroup of patients. Notably, the initial increase after course 2 compared to pre-treatment HbF levels (available for 6 patients: median 1.2%; range, 0.3-3.9%) became reversed at relapse in five of the six patients (Figure 4B).

Decitabine triggers an erythroid but not a megakaryocytic maturation program in bi-potential myeloid leukemia cells To model the effects of decitabine in vitro, two myeloblastic cell lines (K562, HEL) with bi-lineage differentiation potential were treated with decitabine at nontoxic concentrations. Striking morphological changes included polyploidy and cytoplasmic maturation indicative of partial erythroid differentiation (Figure 5A), confirmed by benzidine staining in K562 cells: decitabine- and hemin- but not PMA-treated cells became hemoglobinpositive (Figure 5B). K562 cells treated with PMA developed morphological changes indicative of megakaryocytic differentiation and induction of surface CD41/61 whereas lack of CD41 detection by FACS analysis after decitabine haematologica | 2019; 104(1)

HbF induction in AML/MDS and outcome prediction

and hemin indicated absence of megakaryocytic differentiation (Figure 5C). Transcriptome profiling by mRNA expression arrays in K562 cells revealed the greatest similarity between heminand decitabine-treated cells (Figure 6A,B). Notably, transcriptome changes of PMA-treated K562 cells were most extensive and clustered most distantly from those of all other treatments (Figure 6A,B). In HEL cells, variable probes also differed considerably between decitabine and PMA treatment (Online Supplementary Figures S8 and S9). Gene ontology analyses of upregulated transcripts consistently identified terms related to erythropoiesis and iron metabolism among the top regulated transcript groups in decitabine-treated cells whereas terms related to megakaryocyte lineage differentiation did not show significance (data not shown). Among the transcripts upregulated by decitabine, almost all major globin genes could be identified (Figure 6C). Decitabine-treated K562 showed a >7-fold induction of alpha-1/alpha-2-globin transcription whereas PMA treatment did not alter alpha-globin transcription. Gamma-globin transcription was induced by all three treatments, with decitabine exhibiting the most pronounced effect. A strong induction was also noted for zeta-globin transcription upon decitabine and hemin, but not PMA, treatment. Similar patterns were noted for HEL (data not shown). To prove induction of functionally competent HbF molecules in K562 cells by decitabine treatment, we quantified hemoglobin tetramers by cation-exchange chromatography. In two independent experiments, a reproducible timeand dose-dependent (albeit modest) induction of HbF tetramers was noted (Figure 6D).

Decitabine-induced globin transcription in K562 cells is associated with GATA1 gene demethylation and upregulation To determine whether transcriptional upregulation of both the hemoglobin gamma 1 (HBG1) and 2 (HBG2) gene was a consequence of direct DNA methylation changes at the locus control region or at direct upstream regulatory regions of both genes, DNA methylation was assessed globally by methyl-CpG immunoprecipitation, as previously described.24 No enrichment could be demonstrated in untreated or in decitabine-treated K562 cells, indicating absence or only very low levels of DNA methylation at the HBG1 and HBG2 promoter and at the locus control region (Online Supplementary Figure S10). The gamma-globin locus is known to be devoid of a CpG island,25 and was shown to be unmethylated when selected CpG sites were interrogated for their methylation status by using methylation-sensitive restriction enzymes.26 The presence of activating H3K4me3 and H3K9ac histone marks (from K562 data sets of the ENCODE consortium27) and absence of the repressive mark H3K27me3 (data not shown) indicated that transcriptional activity was already present in untreated cells at both loci, which could be demonstrated by the elevated HBG1/HBG2 probe intensity in the mRNA expression array data (Figure 6C). GATA1 has been identified as a key regulatory factor which binds prominently to the locus control region of the beta-globin locus (Online Supplementary Figure S10). In order to understand how HMA treatment might contribute to the transcriptional upregulation of HBG1 and HBG2 without altering pre-existing low levels of local DNA methylation in K562 cells, we assessed DNA methyhaematologica | 2019; 104(1)

Figure 4. Induced HbF seems to be preferentially derived from non-malignant erythroid cells. (A) Linear regression model showing the association between HbF induction after four courses of treatment and a decreasing percentage of blasts at this time in 17 patients with myelodysplastic syndrome (MDS)/acute myeloid leukemia (AML). (B) Comparison of HbF levels prior to treatment, after course 2 of decitabine, and at relapse. Three MDS and three AML patients had HbF measurements available at all three time points. In five of the six patients, the initial increase in HbF after course 2 compared to pre-treatment levels became reversed at relapse. In four of these five patients, the level of HbF at relapse was even lower than that at the start of treatment.

lation at the GATA1 gene locus. Upon decitabine treatment, DNA methylation was substantially reduced in K562 cells at the GATA1 promoter region and at an upstream regulatory region (Figure 7A). Western blot analysis of K562 cells after 3 days of decitabine treatment demonstrated a dose-dependent, 2- to 3-fold upregulation of GATA1 protein at day 3 (Figure 7B). After 6 days of treatment, this effect was less pronounced. In contrast, PMA treatment strongly repressed GATA1 expression, an effect only modestly antagonized by decitabine.

Discussion Clinically meaningful pharmacological induction of developmentally silenced HbF gene expression9,28 is a prime example of what has been termed epigenetic therapy. However, the broad clinical application of this approach in hemoglobinopathies has been limited by concerns of long-term mutagenic effects in these chronic disorders. The recent advent of DNA-hypomethylating treatment using azanucleoside DNA methyltransferase 65

J. Stomper et al.

Figure 5. Decitabine induces erythroid differentiation and hemoglobin synthesis in K562 and HEL early myeloid progenitor cells. (A) K562 and HEL cells were treated with three 24 h pulses of 100 nM decitabine. Morphological signs of erythroid differentiation upon decitabine (DAC) treatment were determined by cytospin preparation and May-Grünwald staining. (B) Hemoglobin synthesis in K562 cells was determined by benzidine staining. Cells were treated with DAC (three 24 h pulses, 100 nM), hemin (50 mM for 48 h), phorbol 12-myristate 13-acetate (PMA, 5 nM) and a combination of DAC and PMA at these concentrations. For each experimental point, 300 cells were counted, with blue cells being considered hemoglobin-positive. (C) Flow cytometry analysis of CD41/61 surface markers (highly specific for megakaryocytic differentiation) showed differentiation with PMA but not DAC treatment.

Figure 6. Decitabine treatment induces an erythroid transcriptome signature, globin chain expression and fetal hemoglobin assembly in K562 cells. (A) Transcriptome analyses were performed in K562 cells upon treatment (as described in the legend to Figure 5) with decitabine (DAC, 100 nM), hemin (25 and 50 mM), phorbol 12-myristate 13-acetate PMA (PMA, 5 nM) or without treatment (untreated #1 – no treatment, untreated #2 – phosphatebuffered saline as a vehicle control) using mRNA microarrays (HG-U133plus 2.0 gene chip, Affymetrix). The most variable probes (relative standard deviation among all samples ≥2) are displayed as a heatmap with unsupervised hierarchical clustering (Euclidian distance). (B) Venn diagram displaying unique and shared ≥2-fold upregulated and ≥2-fold downregulated probes between the three treatments in K562 cells. (C) Relative mean probe intensities representing expression of globin chain transcripts: HBA1/HBA2, alphaglobin 1 and 2; HBB, beta-globin; HBG1/HBG2, gamma-globin 1 (Ggamma) and 2 (A-gamma); HBD, deltaglobin; HBE1, epsilon-globin 1; HBM, mu-globin; HBQ1, theta-globin, HBZ, zeta-globin) in untreated K562 cells and upon DAC, hemin and PMA treatment, respectively. (D) Increase of functional HbF tetramers determined by high-performance liquid chromatography as a fraction of total cellular protein upon DAC treatment (50 and 100 nM) after 72 and 144 h, respectively. Note the time-dependent incremental rise in HbF (dose-dependent at 72 h).

haematologica | 2019; 104(1)

HbF induction in AML/MDS and outcome prediction

Figure 7. Decitabine induces gene demethylation and expression of the erythroid-specific transcription factor GATA1 in K562 cells. (A) Schematic representation of the GATA1 gene locus. The GATA1 gene is depicted as a black line from 5â&#x20AC;&#x2122; (left) to 3â&#x20AC;&#x2122; (right) with exons represented as boxes. CpG density (CpGs) is indicated by vertical bars. Transcription factor (TF) binding sites were determined through combined TF chromatin immunoprecipitation sequencing (ChIP-seq) experiments from different tissues/cell lines published by the ENCODE consortium. Likewise, activating histone marks H3K9ac and H3K4me3 are taken from ChIPseq experiments in untreated K562 cells which were published and made publicly available from the ENCODE consortium. DNA methylation was assessed by enrichment through methylCpG immunoprecipitation sequencing experiments in untreated K562 cells (K562 untr.) and in K562 cells treated with decitabine (K562 DAC). (B) K562 cells were treated with three 24 h pulses of 100 nM or 200 nM decitabine and harvested on days 3 and 6. PMA 5 nM was added to the culture media for the final 48 h. Immunoblotting of whole cell lysates with rat anti-GATA1. Immunoblotting with mouse anti-beta-actin was performed to control for lane loading.

inhibitors in MDS and AML has re-kindled an interest in HbF regulation in vivo. Specifically, we and others have recently described HbF induction in patients with MDS or AML receiving the DNA methyltransferase inhibitors azacitidine or decitabine.14-16,29 Since HMAs have to be administered over repeated treatment cycles in order to induce responses (which then occur only in about half of the patients), early markers predicting subsequent outcome are urgently needed in order to better advise patients whether to continue HMA treatment or switch to alternative therapy. In order to investigate induction of HbF as a potential predictor of outcome, we chose the time point at the end of two courses of decitabine treatment, i.e. approximately 12 weeks after starting treatment, as the most meaningful: it disclosed the first robust HbF induction (Figure 1A), and was still early enough to be useful for predicting later outcome (whereas information at later time points would be a "self-fulfilling prophecy" regarding response and survival). Indeed, higher HbF at this time point was associated with overall objective responses and, particularly in MDS patients, with improvement of the different hematopoietic lineages after four courses of treatment. Notably, MDS patients with elevated HbF after two haematologica | 2019; 104(1)

course of decitabine treatment had longer overall survival and trends to longer progression-free and AML-free survival. For AML patients, the median overall survival was also nominally longer in this group, but the difference was not statistically significant (possibly because at that time point, half of the AML patients had already died). Time point optimization in AML patients therefore appears necessary, particularly in view of the presently used decitabine treatment schedules of drug administration 5 or 10 days every 4 weeks. It was encouraging to note that higher HbF levels after two courses of treatment were associated with platelet doubling after one course, one of the few established early dynamic predictors of outcome with HMAs.30,31 In order to assess whether these results are applicable in clinical practice, and given the fact that a different decitabine dose is commonly used in MDS and AML patients at present, these results warrant validation in a larger cohort of patients treated with current dosing schemes. For this purpose, serial measurements of HbF have recently been performed in decitabine-treated AML patients randomized within the "inDACtion versus induction" Intergroup AML trial 1301 of the EORTC Leukemia Group (NCT02172872). 67

J. Stomper et al.

With the perspective of exploiting HbF as a potential biomarker for response to HMA therapy, it might be used for selecting a cohort of MDS/AML patients who exhibit a significant increase in HbF after two cycles of decitabine treatment as candidates for a future clinical trial. Such a trial could have the objective to increase the effectiveness of HMA therapy by dose-schedule modifications and/or in combination with other epigenetic or other agents to increase the duration of response. In practical terms, HbF measurement during HMA treatment needs to be compared in different laboratories (at this study’s central laboratory, HPLC quantification of different rare Hb species, including HbF, has been optimized for linear quantification also of smaller amounts). However, baseline HbF is also a promising predictor of HMA response and outcome, as recently demonstrated in the same cohorts of patients.16 At first sight it appears counter-intuitive that pre-treatment high expression may be biologically linked to outcome. However, as already discussed,16 elevated pre-treatment HbF levels may reflect incomplete silencing by methylation (and possibly other epigenetic mechanisms) of the gamma-globin locus, providing a potential surrogate marker for higher de novo sensitivity to HMAs. Interestingly, Cross et al.32 made a similar observation studying long interspersed element (LINE)-1 methylation in AML patients treated with azacitidine: lower LINE-1 methylation prior to treatment, but not early hypomethylation under treatment, predicted hematologic response to this in vivo hypomethylation. What is the cellular source of HbF during the different treatment phases in MDS and AML patients receiving HMAs? In untreated MDS patients clonal erythroid progenitors,33 and in untreated AML patients residual normal, non-clonal erythroid progenitors may be the source of HbF-containing erythrocytes.34 In contrast, in patients attaining a complete or cytogenetic remission, it is more likely that the target cell of HMAs (presumably achieving hypomethylation and transcriptional de-repression of the beta-globin-like gene locus) is part of the non-clonal erythropoiesis, as in patients with solid tumors who show HbF induction during HMA treatment.10,11 The increase in platelet count after decitabine treatment indicates that decitabine reduced or eliminated the suppressive action of the malignant cells on normal hematopoiesis, with subsequent expansion of normal hematopoietic stem cells, which undergo differentiation. At the time of hematologic relapse, the decline in HbF levels might be due to the recurrence of the malignant clone. This contrasts with the model of increased HbF levels in juvenile myelomonocytic leukemia resulting from epigenetic dysregulation of beta-like-globin genes in leukemic cells.35 Serial immunohistochemical bone marrow studies for HbF expression in MDS/AML patients receiving HMAs are warranted to determine the cell of origin of HbF production during the different phases of treatment. Modeling the effects of decitabine on the erythroid versus megakaryocytic lineage in two bi-potential myeloid cell lines, activation of an erythroid but not megakaryocytic gene expression program was observed, including induction of gamma-globin expression and induction, albeit modest, of HbF tetramer formation. We could demonstrate that decitabine treatment regulated many of the genes also regulated by hemin (including mRNA for erythroid-specific transcription factors and beta-like-globin genes), and induced GATA1 at the protein level, con68

comitantly with demethylation at several cis-regulatory regions known to be important for the regulation of this gene.36,37 Notably, the overlap between the transcriptome changes induced by decitabine versus hemin was more marked in the downregulated genes compared to the upregulated ones. Despite demonstrating GATA1 gene demethylation following decitabine treatment, we are unable to conclude that GATA1 induction is a direct consequence of demethylation or is occurring during erythroid differentiation triggered via other factors. Here, a similar DNA methylation analysis of K562 cells treated with hemin instead of decitabine would address this "cause or consequence" question. Taken together, the cell line experiments suggested that increased levels of HbF can also occur because of the effects of decitabine on malignant cells of the erythroid lineage. However, since a cell line is not a good model of the normal functional hematopoietic hierarchy, no conclusion can be drawn as to why erythroid rather than megakaryocytic differentiation was observed. Is there a clinical relevance of induction of HbF beyond HMA treatment? Very recently, in a preclinical study a novel, specific inhibitor of histone deacetylase 1/2 also demonstrated a strong propensity to induce HbF.38 Furthermore, it is well established that inhibitors of the first histone lysine-specific demethylase (LSD)1, including novel, highly specific LSD1/KDM1A inhibitors such as RN-1, are able to reactivate a silenced beta-globin-like gene locus.39 Furthermore, UNC0638, a selective inhibitor of the histone methyltransferases EHMT1 and EHM2, has the ability to induce HbF expression, and this potency is enhanced when the drug is combined with decitabine or the histone deacetylase inhibitor entinostat.40 Very recently, pomalidomide was also shown to be able to induce HbF in patients with multiple myeloma.41 Thus, serial HbF measurements in these different clinical settings may be interesting in order to determine whether the kinetics of this parameter is predictive of treatment response. In conclusion, the technically simple test of assaying HbF levels warrants further prospective studies since the time to best response in patients treated with HMAs is often in the range of 4-6 months. Earlier tailoring of treatment is, therefore, highly desirable. It will be of interest to determine the predictive value of HbF levels compared to other, already established predictors of HMA response such as hematologic parameters, genetic and DNA methylation markers. Acknowledgments The authors wish to thank Tobias Berg, Dietmar Pfeifer, Heike Pahl, Claudia Schmoor, Roland Schüle, Christian Flotho and others who kindly provided helpful input and critical discussion during the course of this study. We really appreciate the experimental contributions of Lena Pados (née Kasten), and the efforts of Thomas Epting (hemoglobin quantification) and Ljudmila Bogatyreva (statistics). Funding This work was supported by the DFG, SFB 992 (MEDEP, project C04). Further research funding: DFG-SPP 1463 (ML, CP), DFG-FOR 2674 (A01: CP; A05: ML/HB; A09: CP/ML), Wilhelm-Sander-Stiftung (grant 1999.032.2), German Cancer Aid (DKH 111210: HB; Max Eder stipend DKH 110461: RC). haematologica | 2019; 104(1)

HbF induction in AML/MDS and outcome prediction

References 15. 1. Daskalakis M, Nguyen TT, Nguyen C, et al. Demethylation of a hypermethylated P15/INK4B gene in patients with myelodysplastic syndrome by 5-aza-2′-deoxycytidine (decitabine) treatment. Blood. 2002;100(8): 2957-2964. 2. Tsai HC, Li H, Van Neste L, et al. Transient low doses of DNA-demethylating agents exert durable antitumor effects on hematological and epithelial tumor cells. Cancer Cell. 2012;21(3):430-446. 3. Klco JM, Spencer DH, Lamprecht TL, et al. Genomic impact of transient low-dose decitabine treatment on primary AML cells. Blood. 2013;121(9):1633-1643. 4. Claus R, Pfeifer D, Almstedt M, et al. Decitabine induces very early in vivo DNA methylation changes in blasts from patients with acute myeloid leukemia. Leuk Res. 2013;37(2):190-196. 5. Chiappinelli KB, Strissel PL, Desrichard A, et al. Inhibiting DNA methylation causes an interferon response in cancer via dsRNA including endogenous retroviruses. Cell. 2015;162(5):974-986. 6. Roulois D, Loo Yau H, Singhania R, et al. DNA-demethylating agents target colorectal cancer cells by inducing viral mimicry by endogenous transcripts. Cell. 2015;162(5): 961-973. 7. Yang X, Han H, De Carvalho DD, Lay FD, Jones PA, Liang G. Gene body methylation can alter gene expression and is a therapeutic target in cancer. Cancer Cell. 2014;26(4):577-590. 8. Lübbert M, Daskalakis M, Kunzmann R, Engelhardt M, Guo Y, Wijermans P. Nonclonal neutrophil responses after successful treatment of myelodysplasia with low-dose 5-aza-2'-deoxycytidine (decitabine). Leuk Res. 2004;28(12):12671271. 9. Ley TJ, DeSimone J, Anagnou NP, et al. 5azacytidine selectively increases gammaglobin synthesis in a patient with beta+ thalassemia. N Engl J Med. 1982;307(24):14691475. 10. Chu BF, Karpenko MJ, Liu Z, et al. Phase I study of 5-aza-2'-deoxycytidine in combination with valproic acid in non-small-cell lung cancer. Cancer Chemother Pharmacol. 2013;71(1):115-121. 11. Gollob JA, Sciambi CJ, Peterson BL, et al. Phase I trial of sequential low-dose 5-aza-2'deoxycytidine plus high-dose intravenous bolus interleukin-2 in patients with melanoma or renal cell carcinoma. Clin Cancer Res. 2006;12(15):4619-4627. 12. Koshy M, Dorn L, Bressler L, et al. 2-deoxy 5-azacytidine and fetal hemoglobin induction in sickle cell anemia. Blood. 2000;96 (7):2379-2384. 13. Olivieri NF, Saunthararajah Y, Thayalasuthan V, et al. A pilot study of subcutaneous decitabine in beta-thalassemia intermedia. Blood. 2011;118(10):2708-2711. 14. Tan P, Wei A, Mithraprabhu S, et al. Dual epigenetic targeting with panobinostat and azacitidine in acute myeloid leukemia and

haematologica | 2019; 104(1)

16.

17.

18. 19.

20.

21.

22.

23.

24.

25. 26.

27. 28.

29.

high-risk myelodysplastic syndrome. Blood Cancer J. 2014;4:e170. Gore SD, Baylin S, Sugar E, et al. Combined DNA methyltransferase and histone deacetylase inhibition in the treatment of myeloid neoplasms. Cancer Res. 2006;66(12):6361-6369. Lübbert M, Ihorst G, Sander PN, et al. Elevated fetal haemoglobin is a predictor of better outcome in MDS/AML patients receiving 5-aza-2'-deoxycytidine (decitabine). Br J Haematol. 2017;176(4): 609-617. Bissé E, Wieland H. High-performance liquid chromatographic separation of human haemoglobins. Simultaneous quantitation of foetal and glycated haemoglobins. J Chromatogr. 1988;434(1):95-110. Belsley DA KE, Welsh RE. Regression Diagnostics. New York: John Wiley & Sons, 1980. Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response and other comparisons of time-to-event by outcome variables. J Clin Oncol. 2008;26(24):39133915. Claus R, Fliegauf M, Stock M, Duque JA, Kolanczyk M, Lübbert M. Inhibitors of DNA methylation and histone deacetylation independently relieve AML1/ETO-mediated lysozyme repression. J Leukoc Biol. 2006;80 (6):1462-1472. Miller CW, Young K, Dumenil D, Alter BP, Schofield JM, Bank A. Specific globin mRNAs in human erythroleukemia (K562) cells. Blood. 1984;63(1):195-200. Rutherford T, Clegg JB, Higgs DR, Jones RW, Thompson J, Weatherall DJ. Embryonic erythroid differentiation in the human leukemic cell line K562. Proc Natl Acad Sci USA. 1981;78(1):348-352. Guilbert LJ, Nelson DJ, Hamilton JA, Williams N. The nature of 12-O-tetradecanoylphorbol-13-acetate (TPA)-stimulated hemopoiesis, colony stimulating factor (CSF) requirement for colony formation, and the effect of TPA on [125I]CSF-1 binding to macrophages. J Cell Physiol. 1983;115(3): 276-282. Baer C, Claus R, Frenzel LP, et al. Extensive promoter DNA hypermethylation and hypomethylation is associated with aberrant microRNA expression in chronic lymphocytic leukemia. Cancer Res. 2012;72(15):3775-3785. Bird AP. CpG-rich islands and the function of DNA methylation. Nature. 1986;321 (6067):209-213. van der Ploeg LH, Flavell RA. DNA methylation in the human gamma delta beta-globin locus in erythroid and nonerythroid tissues. Cell. 1980;19(4):947-958. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57-74. Saunthararajah Y. LD, DeSimone J. Epigenetic regulation of globin genes and disturbances in hemoglobinopathies. In: Lübbert M., Jones PA, ed. Epigenetic Therapy of Cancer. Berlin, Heidelberg: Springer, 2014. Press KR, Uy N, Keefer J, et al. Clinical eval-

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

uation of combined azacitidine and entinostat on the induction of fetal hemoglobin in patients with acute myeloid leukemias and myelodysplastic syndromes. Leuk Lymphoma. 2017:2018;59(3):755-757. van den Bosch J, Lübbert M, Verhoef G, Wijermans PW. The effects of 5-aza-2'deoxycytidine (decitabine) on the platelet count in patients with intermediate and high-risk myelodysplastic syndromes. Leuk Res. 2004;28(8):785-790. van der Helm LH, Alhan C, Wijermans PW, et al. Platelet doubling after the first azacitidine cycle is a promising predictor for response in myelodysplastic syndromes (MDS), chronic myelomonocytic leukaemia (CMML) and acute myeloid leukaemia (AML) patients in the Dutch azacitidine compassionate named patient programme. Br J Haematol. 2011;155(5):599-606. Cross M, Bach E, Tran T, et al. Pretreatment long interspersed element (LINE)-1 methylation levels, not early hypomethylation under treatment, predict hematological response to azacitidine in elderly patients with acute myeloid leukemia. Onco Targets Ther. 2013;6:741-748. Choi JW, Kim Y, Fujino M, Ito M. Significance of fetal hemoglobin-containing erythroblasts (F blasts) and the F blast/F cell ratio in myelodysplastic syndromes. Leukemia. 2002;16(8):1478-1483. Rautonen J, Siimes MA. Initial blood fetal hemoglobin concentration is elevated and is associated with prognosis in children with acute lymphoid or myeloid leukemia. Blut. 1990;61(1):17-20. Fluhr S, Krombholz CF, Meier A, et al. Epigenetic dysregulation of the erythropoietic transcription factor KLF1 and the betalike globin locus in juvenile myelomonocytic leukemia. Epigenetics. 2017;12(8):715-723. Takai J, Moriguchi T, Suzuki M, Yu L, Ohneda K, Yamamoto M. The Gata1 5' region harbors distinct cis-regulatory modules that direct gene activation in erythroid cells and gene inactivation in HSCs. Blood. 2013;122(20):3450-3460. Moriguchi T, Suzuki M, Yu L, Takai J, Ohneda K, Yamamoto M. Progenitor stagespecific activity of a cis-acting double GATA motif for Gata1 gene expression. Mol Cell Biol. 2015;35(5):805-815. Shearstone JR, Golonzhka O, Chonkar A, et al. Chemical inhibition of histone deacetylases 1 and 2 induces fetal hemoglobin through activation of GATA2. PLoS One. 2016;11(4):e0153767. Rivers A, Vaitkus K, Ibanez V, et al. The LSD1 inhibitor RN-1 recapitulates the fetal pattern of hemoglobin synthesis in baboons (P. anubis). Haematologica. 2016;101(6):688697. Renneville A, Van Galen P, Canver MC, et al. EHMT1 and EHMT2 inhibition induces fetal hemoglobin expression. Blood. 2015;126(16):1930-1939. Dulmovits BM, Appiah-Kubi AO, Papoin J, et al. Pomalidomide reverses gamma-globin silencing through the transcriptional reprogramming of adult hematopoietic progenitors. Blood. 2016;127(11):1481-1492.

ARTICLE Ferrata Storti Foundation

Myeloproliferative Neoplasms

Vascular endothelial cell expression of JAK2V617F is sufficient to promote a pro-thrombotic state due to increased P-selectin expression

Alexandre Guy,1 Virginie Gourdou-Latyszenok,1 Nicolas Le Lay,2 Claire Peghaire,1 Badr Kilani,1 Juliana Vieira Dias,1 Cécile Duplaa,1 Marie-Ange Renault,1 Cécile Denis,3 Jean Luc Villeval,4 Yacine Boulaftali,2 Martine Jandrot-Perrus,2 Thierry Couffinhal1,5 and Chloe James1,6

Univ. Bordeaux, Inserm, UMR1034, Biology of Cardiovascular Diseases, Pessac; 2Univ. Paris Diderot, Inserm, UMRS1148, LVTS, Paris; 3Inserm U1176, Hemostasis Inflammation Thrombosis, Le Kremlin-Bicêtre; 4Univ. Paris XI, Inserm U1170, Gustave Roussy Institute, Villejuif; 5CHU de Bordeaux, Service des Maladies Cardiaques et Vasculaires, Pessac and 6CHU de Bordeaux, Laboratoire d’Hématologie, Pessac, France 1

Haematologica 2019 Volume 104(1):70-81

ABSTRACT

Correspondence: chloe.james@inserm.fr.

Received: April 10, 2018. Accepted: August 23, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2018.195321 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/70 ©2019 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.

hrombosis is the main cause of morbidity and mortality in patients with JAK2V617F myeloproliferative neoplasms. Recent studies have reported the presence of JAK2V617F in endothelial cells of some patients with myeloproliferative neoplasms. We investigated the role of endothelial cells that express JAK2V617F in thrombus formation using an in vitro model of human endothelial cells overexpressing JAK2V617F and an in vivo model of mice with endothelial-specific JAK2V617F expression. Interestingly, these mice displayed a higher propensity for thrombus. When deciphering the mechanisms by which JAK2V617F-expressing endothelial cells promote thrombosis, we observed that they have a proadhesive phenotype associated with increased endothelial P-selectin exposure, secondary to degranulation of Weibel-Palade bodies. We demonstrated that P-selectin blockade was sufficient to reduce the increased propensity of thrombosis. Moreover, treatment with hydroxyurea also reduced thrombosis and decreased the pathological interaction between leukocytes and JAK2V617F-expressing endothelial cells through direct reduction of endothelial P-selectin expression. Taken together, our data provide evidence that JAK2V617F-expressing endothelial cells promote thrombosis through induction of endothelial P-selectin expression, which can be reversed by hydroxyurea. Our findings increase our understanding of thrombosis in patients with myeloproliferative neoplasms, at least those with JAK2V617F-positive endothelial cells, and highlight a new role for hydroxyurea. This novel finding provides the proof of concept that an acquired genetic mutation can affect the pro-thrombotic nature of endothelial cells, suggesting that other mutations in endothelial cells could be causal in thrombotic disorders of unknown cause, which account for 50% of recurrent venous thromboses.

Introduction Myeloproliferative neoplasms (MPN) are acquired clonal hematopoietic stem cell disorders, characterized by an increase in one or more myeloid lineages. The Philadelphia chromosome-negative MPN include polycythemia vera with an excess of red blood cells, essential thrombocythemia with an increase of platelets, and primary myelofibrosis.1 More than 90% of patients with polycythemia vera and half of those with essential thrombocythemia and primary myelofibrosis carry a mutation in the Janus kinase 2 (JAK2) gene, i.e. JAK2V617F.2-5 JAK2 is a tyrosine kinase that initiates intracellular signaling of various type 1 cytokine receptors, such as erythropoietin and thrombopoietin receptors.6 The JAK2V617F mutation is responsible for constitutive activation of JAK2 kinase, resulting in subsequent activation of its downstream signaling pathways, ultimately leading to overproduction of myeloid cells. haematologica | 2019; 104(1)

JAK2V617F endothelial cells are pro-thrombotic

Arterial and venous thromboses are the main causes of morbidity in Philadelphia chromosome-negative MPN with reported incidences ranging from 12-39% in polycythemia vera and 11-25% in essential thrombocythemia.7 The pathogenesis of thrombosis in patients with MPN is complex and still largely elusive.8 A variety of blood cells have been reported to participate in the pathophysiology of thrombosis in these neoplasms: (i) platelets isolated from MPN patients show signs of enhanced in vivo activation;9 (ii) leukocytes are activated and hyperleukocytosis is an independent risk factor for thrombosis;9-11 and (iii) red blood cells from patients with polycythemia vera display increased adhesion to the endothelium.12 However, there is evidence that JAK2V617F can be present not only in blood cells but also in endothelial cells (EC) from JAK2V617F-positive MPN patients.13-15 Under physiological conditions, the endothelium maintains a hemostatic balance between pro-thrombotic and anti-thrombotic factors. When stimulated by extrinsic factors such as inflammatory cytokines, hypoxia or antiphospholipid antibodies, EC become activated and promote thrombosis.16 Whether EC can become pro-thrombotic due to intrinsic modifications such as genetic mutations, has not yet been demonstrated. In the current study, we tested whether vascular EC expression of JAK2V617F is sufficient to promote a prothrombotic state. Specifically, we investigated the hemostatic properties of JAK2V617F-expressing EC (hereafter, simply JAK2V617F EC) and these cells’ role in thrombus formation using human EC overexpressing human JAK2V617F, and mice with EC-specific JAK2V617F expression.

Methods In vitro static adhesion of normal blood cells on endothelial cells Blood from healthy volunteers was collected into test-tubes containing EDTA after informed consent had been obtained. Mononuclear cells and neutrophils were isolated by Pancoll density gradient or neutrophil separation medium (Polymorphprep, Fresenius Kabi) and marked with CellTracker Orange. Cells (105) were plated over confluent human umbilical vein endothelial cells (HUVEC) for 1 h at 37°C. After removal of non-adherent cells, the adherent cells were visualized using a fluorescence microscope (AxioObserver, Zeiss), and images were analyzed by ZEN imaging software (Zeiss). Experiments were performed using three to six wells per condition with 12 images taken in each well. The method of quantification has been described elsewhere15,17 Where noted, we added 10 mg/mL P-selectin blocking antibody (AK4 clone, BioLegend) 30 min before plating blood cells on HUVEC.

In vitro neutrophil adhesion on human umbilical vein endothelial cells under flow conditions Channels (Vena8 Endothelial + Cellix) were coated with human fibronectin (100 ng/mL) before infusing 3x106/mL HUVEC.18 After 2 h at 37°C, cells were exposed for 48 h to two-period flow (Kima Pump, Cellix) with alternating perfusion for 3 min (at 600 mL/min) and resting for 20 min (no flow). After 48 h, a confluent monolayer had developed. Neutrophils (3x106/mL) were infused with a 10 mL/min venous flux and imaged using a Zeiss microscope, 20X in phase contrast, for 5 min. Channels were washed with medium and images of every channel were recorded. Neutrophil quantification was performed blindly and the neutrophil velocity was studied with FIJI (Image J).

haematologica | 2019; 104(1)

In vivo leukocyte adhesion to mesenteric venules Pdgfb-iCreERT2;JAK2V617F/WT mice and Pdgfb-iCreERT2;JAK2V617F/WT mice were used 15-20 days after tamoxifen injection. Mice were injected, intraperitoneally, with 250 ng tumor necrosis factor-alpha (TNF-α) and, 4h later, anesthetized with ketamine/xylazine. Leukocytes were stained with 3.6 mg/kg Rhodamine 6G (SigmaAldrich). For each mouse, five mesenteric venules of 150 to 250 µm diameter were observed for 90 sec within 30 min after the surgical procedure, using a fluorescent microscope (AXIO Zoom.V16, Zeiss). Where noted, 25 mg of P-selectin blocking antibody (RB40 clone, BD Biosciences) or isotype control (A110-1 clone, BD Biosciences) were injected 5 min before starting the analysis.

Mouse model of thrombosis To induce platelet activation, the mice were given an intraperitoneal injection of 75 µg/kg collagen (Nycomed Pharma) and 30 mg/kg epinephrine (Helena Laboratories) 3 min before euthanasia.19,20 To increase inflammation, 250 ng TNF-α (RD Systems) were injected 4 h before the animals were sacrificed. After euthanasia, intracardiac puncture was performed and phosphatebuffered saline was perfused for 3 min, followed by 10% formalin for another 3 min. Where noted, 25 mg P-selectin blocking antibody were injected 4 h prior to euthanasia.

Statistics Results are expressed as the mean ± standard error of mean (SEM). Statistical significance was calculated using the Student t test or Mann-Whitney statistical test to compare differences between two groups. For multiple groups, we used one-way analysis of variance (ANOVA) followed by the Tukey post-hoc test or two-way ANOVA followed by the Sidak post-hoc test. GraphPad Prism 6 software was used. A P value <0.05 was considered statistically significant.

Animals Animal experiments were performed in accordance with the guidelines provided by the Institutional Animal Care and Use Committee at Inserm (Agreement number C45-234-6).

Results Pdgfb-iCreERT2;JAK2V617F/WT mice are a reliable model to investigate JAK2V617F endothelial cell properties To analyze the specific role of JAK2V617F EC in thrombus formation, we crossed Pdgfb-iCreERT2 mice with conditional flexed JAK2 (JAK2V617F) mice,21 to allow heterozygous expression of JAK2V617F in EC after tamoxifen administration. Since oral tamoxifen administration has been shown to result in recombination activity in megakaryocytes within 2 days,22 all experiments were performed and analyzed 15-20 days after tamoxifen injection, to allow time for recombined megakaryocytes to mature and be cleared. We first analyzed the efficiency of recombination using Pdgfb-iCreERT2;mT/mG mice, and observed that all vessels in the mesentery and retina were positive for green fluorescent protein (GFP) (Online Supplementary Figure S1A-C). To confirm EC JAK2V617F expression in Pdgfb-iCreERT2;JAK2V617F/WT mice, we sorted CD31-positive cells from the kidney and observed a 50% ratio of JAK2V617F:total JAK2 in accordance with a heterozygous expression of JAK2V617F in all EC (data not shown). In sorted CD31-positive lung EC, western blot analysis demonstrated an increased level of phosphorylation of the downstream effector AKT in line with increased activation of 71

A. Guy et al.

the JAK pathway (Online Supplementary Figure S1D,E). Because myeloproliferative syndromes have been described in EC-specific mouse models,23,24 we ensured that our experimental conditions did not lead to abnormal hematopoiesis. We did not observe any difference in blood cell counts in Pdgfb-iCreERT2;JAK2V617F/WT mice, suggesting that the hematopoietic system was not significant-

ly altered (Online Supplementary Table S1). To definitively exclude hematopoietic expression of JAK2V617F in PdgfbiCreERT2;JAK2V617F/WT mice, we crossed PdgfbiCreERT2;JAK2V617F/WT mice with mT/mG mice to generate Pdgfb-iCreERT2;JAK2V617F/WT;mT/mG mice permitting coexpression of JAK2V617F and GFP in cells after Cre-mediated excision (Online Supplementary Figure S1F). Flow cytome-

Spontaneous

Collagen/ Epinephrine

TNF-alpha

Figure 1. The presence of JAK2V617F in endothelial cells promotes thrombus formation. (A) In Pdgfb-iCreERT2;JAK2V617F/WT mice, thrombus formation occurs spontaneously and is increased after weak platelet activation by low-dose collagen plus epinephrine, or injection of tumor necrosis factor (TNF)-alpha. (B) Carstairs staining of pulmonary thrombi in control mice (left) and Pdgfb-iCreERT2;JAK2V617F/WT mice (right) injected with TNF-alpha. Black arrows indicate thrombi. The clear arrow head indicates fibrin deposition. Scale bar: 500 mm. (C) Representative image of a thrombus formed by neutrophils (green) and platelets (yellow) in PdgfbiCreERT2;JAK2V617F/WT mice. Scale bar in the left image: 50 mM. Scale bar in the right image: 10 mM. Results are expressed as mean value Âą SEM. Statistical significance was determined by the Mann-Whitney test. *P<0.05; **P<0.01.

haematologica | 2019; 104(1)

JAK2V617F endothelial cells are pro-thrombotic

try analysis did not reveal any GFP expression in granulocytes, platelets or red blood cells in PdgfbiCreERT2;JAK2V617F/WT;mT/mG mice 20 days after tamoxifen administration (Online Supplementary Figure S1F), confirming that Cre recombination under the Pdgfb promoter is highly restricted to EC 15-20 days after tamoxifen administration.

The expression of JAK2V617F by endothelial cells leads to increased thrombus formation To investigate whether Pdgfb-iCreERT2;JAK2V617F/WT mice displayed a greater propensity to thrombosis, we examined pulmonary thrombus formation using experimental conditions that allowed assessment of EC involvement, without exposure of the sub-endothelium: (i) spontaneous thrombosis under basal conditions, (ii) a model of mild thrombosis with administration of low doses of collagen together with epinephrine to induce

weak activation of platelets and vasoconstriction and better demonstrate an intrinsic pro-thrombotic phenotype of JAK2V617F EC, and (iii) a weak inflammatory trigger of thrombosis with injection of a low dose of TNF-α (250 ng/mouse). A dose of 500 ng TNF-α is commonly used to trigger inflammation25-27 but we chose the lower dose to reveal a potential hypersensitivity to inflammation. To quantify thrombi, we performed Carstairs staining which allows visualization of platelets, red blood cells, and fibrin.28 Small, spontaneously formed thrombi were observed in the lungs of Pdgfb-iCreERT2;JAK2V617F/WT mice, but not in littermate JAK2V617F/WT control mice (Figure 1A). In both models of mild induction of thrombosis, we observed a significant increase in thrombus formation in Pdgfb-iCreERT2;JAK2V617F/WT mice compared with that in controls (Figure 1A-C). Together, these results demonstrate that JAK2V617F EC have a pro-thrombotic phenotype.

Figure 2. JAK2V617F-expressing endothelial cells have normal anticoagulant activity. (A,B) There was no statistical difference in either (A) the rate or (B) the extent of tissue factor (TF) -triggered thrombin generation at the surface of endothelial cells which were or were not activated with tissue necrosis factor (TNF)-alpha (n=3 for all conditions). (C-E) There was no statistical difference in (C) the rate of thrombin-triggered protein C activation at the surface of endothelial cells (n=3 for all conditions), (D) the production of nitrite or (E) the production of 6-keto prostaglandin F1a (n=3 for all conditions). Results are expressed as mean value ± SEM from three experiments. Statistical significance was determined by the Mann-Whitney test.

haematologica | 2019; 104(1)

A. Guy et al.

JAK2V617F endothelial cells have normal anticoagulant activity Next, we sought to decipher the mechanisms by which JAK2V617F expression in EC leads to a pro-thrombotic phenotype. To assess whether the expression of JAK2V617F by EC decreased these cells’ anticoagulant properties or even triggered pro-coagulant properties, we measured thrombin generation at the surface of HUVEC transduced with lentivirus encoding for either human JAK2V617F or wildtype JAK2 (JAK2WT), or with an empty lentivirus as a control. Our data show that HUVEC were efficiently transduced

as: (i) 95% of JAK2V617F or JAK2WT HUVEC were GFP-positive as assessed by flow cytometry (Online Supplementary Figure S2A) and (ii) the mRNA JAK2V617F mutational burden was 99.5% in JAK2V617F HUVEC (Online Supplementary Figure S2B). Western blot analysis revealed induced protein expression of JAK2 in JAK2V617F- and JAK2WT-transduced cells. The introduction of JAK2V617F led to increased phosphorylation of JAK2, STAT3 and AKT, in agreement with hyperactivation of the JAK/STAT pathway (Online Supplementary Figure S2C). Under resting conditions, measurement of thrombin generation at the surface of

Figure 3. JAK2V617F-expressing endothelial cells have a pro-adhesive phenotype in vitro and in vivo. (A,B) Under static conditions (A) normal mononuclear cells and (B) neutrophils adhere more to human umbilical vein endothelial cells (HUVEC) expressing JAK2V617F than to JAK2WT HUVEC. HUVEC transduced with lentiviruses containing green fluorescent protein alone were used as negative controls. Statistical significance determined by one-way analysis of variance and the post-hoc Tukey test. Results are expressed as mean value ± SEM from three experiments. (C,D) Under flow conditions (C) adhesion of neutrophils is increased in the presence of JAK2V617F HUVEC and (D) neutrophil velocity is reduced. Each dot represents one cell. Statistical significance was determined by the Student t-test. Results are expressed as mean value ± SEM from three experiments. (E) Representative images of adhesion of neutrophils on JAK2WT and JAK2V617F HUVEC under flow conditions. (F) Representative images of leukocyte adhesion in mesenteric venules of control and Pdgfb-iCreERT2;JAK2V617F/WT mice treated with tumor necrosis factor (TNF). (G) Leukocyte rolling and (H) adhesion are increased in mesenteric venules of Pdgfb-iCreERT2;JAK2V617F/WT mice treated with TNF. Results are expressed as mean value ± SEM. Statistical significance was determined by the Mann-Whitney test. *P<0.05; ***P<0.001. ****P<0.0001.

haematologica | 2019; 104(1)

JAK2V617F endothelial cells are pro-thrombotic

control, JAK2WT- or JAK2V617F-lentivirus infected HUVEC did not reveal any significant differences in the kinetics or extent of thrombin generation. These results indicate that there is not a significant gain of pro-coagulant activity in response to JAK2WT- or JAK2V617F-induced HUVEC expression (Figure 2A,B). We next hypothesized that JAK2V617F HUVEC might acquire a procoagulant phenotype due to exposure to circulating inflammatory stimuli. We thus repeated the experiments after overnight activation with 10 ng/mL TNF-α, but did not observe any difference (Figure 2A,B). We also measured the rate of thrombin-triggered protein C activation and did not observe any difference between the cells (Figure 2C). Finally, we quantified the production of nitrite and prostaglandin (6-ketoprostaglandin 1-α) and did not find any difference between JAK2WT and JAK2V617F HUVEC which were or were not stimulated by TNF-α (Figure 2D,E). To assess whether JAK2V617F EC increase hemostatic potential in vivo, we used whole blood thromboelastometry in Pdgfb-iCreERT2;JAK2V617F/WT mice. No difference was observed in clotting time, clot formation time, 10-minute amplitude or alpha angle after stimulation of the extrinsic pathway (Online Supplementary Figure S3).

JAK2V617F endothelial cells have a pro-adhesive phenotype in vitro and in vivo Exposing EC to inflammatory stimuli leads to expression of adhesion molecules that allow rolling and adhesion of leukocytes, a phenomenon that is thought to participate in the pathogenesis of thrombosis. Using JAK2V617Ftransduced HUVEC, we observed an increase in static adhesion of normal mononuclear cells (Figure 3A) and normal polymorphonuclear neutrophils isolated from healthy subjects (Figure 3B), as previously reported with JAK2V617F EC from patients. Under flow conditions, we also observed that more normal polymorphonuclear neutrophils rolled and stably adhered to JAK2V617F HUVEC than to JAK2WT HUVEC (Figure 3C-E). To assess whether this pro-adhesive phenotype was also present in vivo, we injected Pdgfb-iCreERT2;JAK2V617F/WT mice with rhodamine to track leukocytes. Using intravital microscopy, we observed mesenteric venules and quantified leukocyte adhesion and rolling. We observed that both leukocyte rolling and adhesion were significantly increased in PdgfbiCreERT2;JAK2V617F/WT mice previously exposed to low-dose TNF-α (Figure 3F-H).

haematologica | 2019; 104(1)

Figure 4. P-selectin expression is increased in JAK2V617F-expressing endothelial cells. (A) There was no modification of cell surface expression of the adhesion molecules, ICAM-1, VCAM-1, and E-selectin on JAK2V617F human umbilical vein endothelial cells (HUVEC). Statistical significance was determined by the Student t-test. Results are expressed as mean value ± SEM from three experiments. (B) Representative images of P-selectin staining (green) in carotid endothelial cells. Nuclei are stained with DAPI (blue) and VE-cadherin (red). Scale bar: 50 mm. (C) Cell surface expression of mouse P-selectin is increased in carotid endothelial cells from PdgfbiCreERT2;JAK2V617F/WT mice whether or not they received tumor necrosis factor (TNF)-alpha. Each dot represents one image (4 images per mouse). Results are expressed as mean value ± SEM. Statistical significance was determined by the Mann-Whitney test. **P<0.01. (D) The ratio between soluble P-selectin concentration and platelet count is significantly increased in Pdgfb-iCreERT2;JAK2V617F/WT mice. Results are expressed as mean value ± SEM. Statistical significance was determined by the Mann-Whitney test. *P<0.05.

A. Guy et al.

P-selectin expression is increased in JAK2V617F endothelial cells The adhesion of leukocytes to EC is mediated by cell adhesion molecules and selectins.29 Flow cytometry analysis demonstrated that JAK2V617F HUVEC expressed intercellular adhesion molecule (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1) and E-selectin at the same level as JAK2WT HUVEC, whether or not they were previously activated with TNF-α (Figure 4A and Online Supplementary Figure S4). Immunostaining of non-permeabilized carotid arteries from Pdgfb-iCreERT2;JAK2V617F/WT mice showed an increased exposure of P-selectin at the EC surface in vivo, independently of prior administration of TNF-α (Figure 4B,C). Conversely we observed increased levels of soluble P-selectin in the plasma of Pdgfb-iCreERT2;JAK2V617F/WT mice (Online Supplementary Figure S5A). As most soluble P-selectin is of platelet origin, we ruled out an increase of soluble P-selectin due to increased platelet count (Figure 4D). Finally, we excluded increased platelet activation in Pdgfb-iCreERT2;JAK2V617F/WT mice by quantifying soluble platelet factor 4 (Online Supplementary Figure S5B). Collectively, our results support the notion of increased membrane-attached and plasma soluble P-selectin of endothelial origin without an increase in EC expression of ICAM-1, VCAM-1 or E-selectin.

blocking antibody (Figure 6A,B). Quantification of leukocytes in Pdgfb-iCreERT2;JAK2V617F/WT mice treated with TNF-α and the P-selectin blocking antibody revealed a complete inhibition of leukocyte rolling in control mice, which was expected given the well-established role of P-selectin in leukocyte rolling (Figure 6C). In mutant mice, administration of P-selectin blocking antibody completely reversed the pathologically increased leukocyte adhesion (Figure 6C,D). To examine whether increased P-selectin was also responsible for thrombus formation, we used the model of low-dose TNF-α-induced lung thrombus formation (as shown in Figure 1). We observed that pre-treatment of the mice with the P-selectin blocking antibody

Endothelial expression and release of Von Willebrand factor is increased in JAK2V617F endothelial cells in vitro and in vivo P-selectin is stored within EC in exocytotic organelles called Weibel-Palade bodies together with von Willebrand factor (vWF). Exocytosis of Weibel-Palade bodies leads to cell surface expression of vWF and P-selectin. Since we had observed an increase in endothelial P-selectin expression, we next investigated whether vWF was also increased. In vitro, using immunostaining on non-permeabilized HUVEC, we observed increased vWF expression at the surface of JAK2V617F HUVEC, spontaneously and after overnight activation with 10 ng/mL TNF-α (Figure 5A). Furthermore, quantification of vWF in the conditioned media revealed higher amounts of vWF released by JAK2V617F HUVEC (Figure 5B), a difference that was even greater when the cells had been treated with TNF-α. These results were confirmed in vivo with higher levels of vWF antigen in Pdgfb-iCreERT2;JAK2V617F/WT mice than in control mice (Figure 5C). There was no difference in the distribution of vWF multimers between PdgfbiCreERT2;JAK2V617F/WT mice and control animals. These data demonstrate that endothelial JAK2V617F increased the levels of vWF protein and the release of soluble vWF, in association with increased P-selectin expression at the cell surface, as a consequence of increased degranulation of Weibel-Palade bodies.

Increased P-selectin exposure is involved in the pro-adhesive and pro-thrombotic phenotype of JAK2V617F endothelial cells To investigate a potential causal link between increased P-selectin expression at the EC surface and the pro-adhesive phenotype of JAK2V617F EC, we reproduced the same experiments as previously described, but in the presence of a P-selectin blocking antibody. Our in vitro approach with JAK2V617F HUVEC showed a complete reversion of the hyper-adhesive properties of JAK2V617F HUVEC when these cells had been exposed for 30 min to the P-selectin 76

Figure 5. von Willebrand factor expression is increased in JAK2V617F-expressing endothelial cells. (A) Cell surface expression of von Willebrand factor (vWF) (red) is higher in JAK2V617F human umbilical vein endothelial cells (HUVEC) (right) than in JAK2WT HUVEC (left). Nuclei are stained by DAPI (blue). Scale bar: 20 mm. (B) JAK2V617F HUVEC secrete more vWF than JAK2WT HUVEC in the absence or presence of tumor necrosis factor (TNF)-alpha. Statistical significance was determined by Mann-Whitney test. Results are mean value ± SEM from three experiments. (C) Plasma level of vWF antigen (Ag) is increased in PdgfbiCreERT2;JAK2V617F/WT mice. Results are expressed as mean value ± SEM. Statistical significance was determined by the Student t-test. *P<0.05; ***P<0.001.

haematologica | 2019; 104(1)

JAK2V617F endothelial cells are pro-thrombotic

completely abrogated thrombus formation in PdgfbiCreERT2;JAK2V617F/WT mice, but had no effect in control mice (Figure 6E). These results demonstrate that the prothrombotic phenotype of JAK2V617F EC is primarily the consequence of increased adhesive properties, due to overexpression of membrane P-selectin.

Treatment with hydroxyurea abrogates tumor necrosis factor-induced thrombosis in Pdgfb-iCreERT2;JAK2V617F/WT mice through decreased P-selectin expression Hydroxyurea is an anti-metabolite frequently used in MPN to reduce the occurrence of thrombosis. Its antithrombotic effect is reported to occur via the reduction of blood cell counts. Hydroxyurea is also used in sickle cell disease to reduce vaso-occlusive crises, and its beneficial effect is in part mediated by a direct effect on EC, with a

reduction in leukocyte adhesion.30 We hypothesized that hydroxyurea might be capable of reducing the pro-thrombotic effect of JAK2V617F EC. To test this, we treated PdgfbiCreERT2;JAK2V617F/WT and control mice with hydroxyurea for 10 days, injected a low dose of TNF-α and quantified thrombus formation in the lungs after 4 h. We observed a significant inhibition of thrombus formation in Pdgfb-iCreERT2;JAK2V617F/WT mice (Figure 7A). We then investigated whether hydroxyurea had modified the adhesiveness of the EC. We observed a reduction in leukocyte rolling and adhesion in mesenteric venules 4 h after TNF-α administration in Pdgfb-iCreERT2;JAK2V617F/WT mice treated with hydroxyurea (Figure 7B,C). To further determine whether this was due to a direct effect on EC, we treated JAK2V617F HUVEC with 100 mM hydroxyurea for 24 h, washed them and analyzed neutrophil adhesion in vitro. Interestingly, there was a significant reduction of neu-

haematologica | 2019; 104(1)

Figure 6. Increased endothelial P-selectin expression is responsible for the pro-adhesive phenotype of JAK2V617F-expressing endothelial cells. (A,B) Under static conditions, increased adhesion of (A) normal mononuclear cells and (B) neutrophils on JAK2V617F human umbilical vein endothelial cells is reversed in the presence of a P-selectin blocking antibody (Ab). Results are mean value ± SEM from three experiments. In Pdgfb-iCreERT2;JAK2V617F/WT mice, increased (C) rolling and (D) adhesion of leukocytes is abolished in the presence of a P-selectin blocking antibody. (E) Increased thrombus formation in Pdgfb-iCreERT2;JAK2V617F/WT mice is abrogated in the presence of a P-selectin blocking antibody. Results are expressed as mean value ± SEM. Statistical significance was determined by two-way analysis of variance and a Sidak post-hoc test. *P<0.05; **P<0.01; ****P<0.0001.

A. Guy et al.

trophil adhesion on JAK2V617F HUVEC pretreated with hydroxyurea (Figure 7D). Having previously demonstrated that the pro-adhesive phenotype of JAK2V617F EC was mediated by P-selectin, we examined whether the protective effect of hydroxyurea acted through reduction of endothelial P-selectin. As we had previously shown that soluble P-selectin was a reflection of endothelial P-selectin in Pdgfb-iCreERT2;JAK2V617F/WT mice, we measured levels of soluble P-selectin after treatment with hydroxyurea. We observed that the P-selectin:platelet ratio was significantly lower in hydroxyurea-treated Pdgfb-iCreERT2;JAK2V617F/WT mice than in untreated mice (Figure 7E). In these same mice, we used immunostaining to quantify P-selectin expression at the surface of carotid EC. We observed a sig-

nificant decrease of endothelial membrane P-selectin expression, confirming a direct effect of hydroxyurea on JAK2V617F EC (Figure 7F). This effect is in part secondary to JAK/STAT pathway inhibition, as hydroxyurea administration reduces the level of STAT3 phosphorylation in JAK2V617F HUVEC (Online Supplementary Figure S6). Finally, we observed decreased vWF concentrations in the supernatant of JAK2V617F HUVEC treated with hydroxyurea, indicating a reduced release of Weibel Palade bodies (Figure 7G). Collectively, in vitro and in vivo results demonstrate that hydroxyurea has a direct effect on JAK2V617F EC, decreasing endothelial P-selectin release and surface expression, thus decreasing the pro-thrombotic phenotype of the EC.

Figure 7. Treatment with hydroxyurea decreases the pro-thrombotic and pro-adhesive phenotype of JAK2V617F-expressing endothelial cells in vitro and in vivo via inhibiting P-selectin and von Willebrand factor expression. (A) Treatment with hydroxyurea for 15 days decreases tumor necrosis factor-induced thrombosis in the lungs of Pdgfb-iCreERT2;JAK2V617F/WT mice. Treatment with hydroxyurea decreases (B) rolling and (C) adhesion of leukocytes on mesenteric venules in Pdgfb-iCreERT2;JAK2V617F/WT mice treated with tumor necrosis factor. Results are expressed as mean value ± SEM. Statistical significance was determined by the Mann-Whitney test. *P<0.05, ***P<0.001. (D) Pre-treatment of JAK2V617F HUVEC with hydroxurea (HU) decreases static adhesion of neutrophils. Results are expressed as mean value ± SEM. Statistical significance determined by two-way ANOVA analysis of variance and the Sidak post-hoc test. (E) Treatment with hydroxyurea for 15 days led to a decrease of the ratio of soluble P-selectin:number of platelets in plasma of Pdgfb-iCreERT2;JAK2V617F/WT mice treated with tumor necrosis factor. (F) Hydroxyurea decreases the expression of P-selectin at the surface of carotid JAK2V617F endothelial cells. Each dot represents one image with four images per mouse (n=4). (G) Treatment of JAK2V617F HUVEC with hydroxyurea decreases secretion of von Willebrand factor (vWF). Results are expressed as mean value ± SEM. Statistical significance was determined by the Mann-Whitney test. *P<0.05.

haematologica | 2019; 104(1)

JAK2V617F endothelial cells are pro-thrombotic

Discussion Despite significant advances in deciphering the molecular mechanisms responsible for the occurrence and transformation of MPN, the mechanisms that lead to thrombosis, the main cause of morbidity and mortality, remain largely elusive. Recent identification of the JAK2V617F mutation in EC of patients with MPN13-15 opened new perspectives in the pathogenesis of thrombosis in MPN. Here we demonstrate that JAK2V617F-positive EC promote spontaneous thrombosis under basal conditions and have an increased thrombotic response to weak inflammatory stimuli. We describe that the mechanism that leads to thrombosis involves endothelial P-selectin release and cell surface exposure and subsequent leukocyte rolling and adhesion. We also describe that treatment with hydroxy-urea decreases P-selectin endothelial expression and thrombus formation in mice expressing JAK2V617F only in EC. We used a mouse model that allows expression of JAK2V617F only in EC to ensure that the phenotype observed in these mice was solely due to the presence of JAK2V617F EC. We are aware that there are differences between our model and patients with MPN, who often have JAK2V617F positive blood cells together with some JAK2V617F EC. Moreover, not all patients have JAK2V617F EC and, for those that do, it is currently not known where they are located. Our mouse model is not, therefore, representative of the human situation and is not meant to be a model of MPN; however, it does allow precise characterization of the properties of JAK2V617F EC. The link between P-selectin expression and thrombosis has been described previously, and in most cases Pselectin originates from platelets.28,31,32 The mechanism of P-selectin-mediated thrombosis involves neutrophil activation, either through tissue factor expression and activation of the extrinsic coagulation pathway,28 or through priming for neutrophil extracellular trap formation,33 a process that is now well-recognized to participate in thrombus formation.34 An increase in endothelial P-selectin and subsequent thrombosis were seen in response to venous flow reduction and local hypoxia.28 Here, we show for the first time that EC can have constitutively increased expression of P-selectin, even without any hypoxic or inflammatory stimuli. Further studies are now required to decipher the specific intracellular mechanism responsible for increased P-selectin expression in JAK2V617F-expressing EC. One possible mechanism could be increased STAT3 phosphorylation, as STAT3 activation has been shown to upregulate expression of P-selectin in EC35 (Online Supplementary Figure S7). Given that MPN are acquired hematologic malignancies, the description of pro-thrombotic JAK2V617F EC raises the question of their origin. JAK2V617F EC have been found in microdissected vessels13,14 and after culture of circulating endothelial progenitors, both in colony-forming unit endothelial cell (CFU-EC)15,36-38 and endothelial colonyforming cells (E-CFC).15,38 The finding that JAK2V617F CFUEC are present in all MPN patients was not surprising given that CFU-EC are of hematopoietic origin. However, E-CFC are not of hematopoietic origin and the presence of JAK2V617F EC thus suggests the existence of a progenitor cell of hematopoietic and endothelial lineages. Such cells certainly exist in the embryo but their existence in adults is a matter of debate. In the case of microdissection experhaematologica | 2019; 104(1)

iments,13,14 it is possible that the JAK2V617F EC that have been microdissected derive from CFU-EC.39 The presence of JAK2V617F EC of real endothelial origin in MPN is probably rare, but the presence of JAK2V617F EC of hematopoietic origin is common. Demonstrating that JAK2V617F EC have a pro-thrombotic phenotype is thus particularly relevant to our understanding of the pathogenesis of thrombosis in MPN. Our study has important therapeutic implications. We demonstrate that treatment with hydroxyurea inhibits the pathological hyper-adhesive phenotype of JAK2V617F EC. The results presented here challenge current thinking, according to which the anti-thrombotic effect of hydroxy-urea in MPN is only mediated by lowering the blood cell count. However, a direct effect of hydroxyurea on EC has been reported40,41 and is thought to occur via stimulation of the nitric oxide-cyclic GMP pathway in EC, and reduction of leukocyte rolling.30 Moreover, in patients with sickle cell disease, treatment with hydroxyurea efficiently decreases the frequency of vaso-occlusive crises.42 These findings suggest that hydroxyurea should be considered for patients with MPN and a history of thrombosis. This is often the case, as hydroxyurea is the first-line therapy in high-risk patients. However, in young patients, hydroxy-urea is usually avoided because of its potential leukemic effect, even if large, retrospective studies have not confirmed this effect.43-45 The antithrombotic and leukemic effects of hydroxyurea should thus be addressed in a prospective cohort of young MPN patients with thrombosis. Another therapeutic implication of our work comes from the demonstration that increased endothelial P-selectin favors thrombosis in MPN, as was shown in sickle cell disease.46 Very recently, a clinical trial demonstrated that treatment with an anti-P-selectin antibody, crizanlizumab, efficiently prevented pain crises in patients with sickle cell disease.47 It is thus tempting to speculate that such treatment could have therapeutic benefits in MPN patients at high risk of thrombosis, who need to receive anticoagulant treatment for a MPN-related thrombosis (such as splanchnic thrombosis). Our description of the pro-thrombotic profile of JAK2V617F EC may appear to contradict with the results of Etheridge et al.48 These authors used Tie2-Cre/FF1 mice, which allow constitutive expression of human JAK2V617F in blood and EC. These mice were irradiated then transplanted with the bone marrow of wild-type mice to generate mice in which only the EC express JAK2V617F. Etheridge et al. reported a hemorrhagic diathesis due to acquired von Willebrand disease. There are several possible reasons for the apparently contradictory results. Studies have shown the importance of the JAK2V617F :JAK2WT ratio for the MPN phenotype49 and Etheridge et al. used transplanted mice with a human JAK2V617F :mouse JAK2WT ratio of 1:5 in EC, whereas we used a tamoxifen-inducible mouse model that allows heterozygous expression of mouse JAK2V617F/JAK2WT specifically in EC. Another major difference lies in the models of thrombosis, as Etheridge et al. used models in which the endothelial layer is disrupted whereas we chose three models in which the endothelial barrier is maintained intact. Of note, Etheridge et al. reported an overall increase in the content of vWF in EC, which is in accordance with our results. Altogether, we believe that our work is clinically relevant for three reasons; (i) it suggests that specific biomark79

A. Guy et al.

ers of EC activation should be intensively investigated to determine which MPN patients are at high thrombotic risk; (ii) it opens the way to new therapeutic options in MPN, such as hydroxyurea in patients at high risk of thrombosis or with raised levels of markers of endothelial activation, as well as anti-P-selectin antibody instead of or in addition to the standard of care to prevent thrombosis in high-risk MPN patients; and (iii) it provides the proof of concept that an acquired genetic mutation can alter the phenotype of EC, as shown for the pro-thrombotic phenotype acquired upon expression of the JAK2 mutation. This suggests that other activating mutations in EC could be causal in thrombotic disorders of unknown cause, which account for 50% of recurrent venous thromboses. Acknowledgments The authors wish to thank Myriam Petit and Annabel Reynaud and Beatrice Jaspard-Vinassa for their assistance in histology, Beatrice Jaspard-Vinassa for her help with western

References 1. 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. 2. Baxter EJ, Scott LM, Campbell PJ, et al. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. Lancet. 2005;365(9464):1054-1061. 3. James C, Ugo V, Le Couedic JP, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera. Nature. 2005;434(7037): 1144-1148. 4. Kralovics R, Passamonti F, Buser AS, et al. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med. 2005;352(17):1779-1790. 5. Levine RL, Wadleigh M, Cools J, et al. Activating mutation in the tyrosine kinase JAK2 in polycythemia vera, essential thrombocythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell. 2005;7(4):387-397. 6. Lu X, Huang LJ, Lodish HF. Dimerization by a cytokine receptor is necessary for constitutive activation of JAK2V617F. J Biol Chem. 2008;283(9):5258-5266. 7. Elliott MA, Tefferi A. Pathogenesis and management of bleeding in essential thrombocythemia and polycythemia vera. Curr Hematol Rep. 2004;3(5):344-351. 8. Barbui T, Finazzi G, Falanga A. Myeloproliferative neoplasms and thrombosis. Blood. 2013;122(13):2176-2184. 9. Falanga A, Marchetti M, Vignoli A, Balducci D, Barbui T. Leukocyte-platelet interaction in patients with essential thrombocythemia and polycythemia vera. Exp Hematol. 2005;33(5):523-530. 10. Barbui T, Carobbio A, Rambaldi A, Finazzi G. Perspectives on thrombosis in essential thrombocythemia and polycythemia vera: is leukocytosis a causative factor? Blood. 2009;114(4):759-763. 11. Landolfi R, Di Gennaro L, Barbui T, et al. Leukocytosis as a major thrombotic risk factor in patients with polycythemia vera. Blood. 2007;109(6):2446-2452.

blots, Jeremie Teillon for his help in image quantification, Veronique Ollivier for her assistance with in vitro flow experiments, Sylvain Grolleau for his help in mice genotyping, Olivier Mansier for his help with the quantitative polymerase chain reaction, Jerome Guignard for his help with mice and the vectorology platform for lentiviruses production (Inserm U1035). We are grateful to William Vainchenker (Institut Gustave Roussy, Inserm U1009), Pierre-Emmanuel Rautou (Inserm U970) and JeanFranĂ§ois Viallard (Inserm U1034) for their helpful discussions. We are also grateful to the French Intergroup Myeloproliferative (FIM). Funding This study was supported by research grants from ANRDFG JAKPOT (NÂ° ANR-14-CE35-0022-02), INSERM, The Fondation Bettencourt Schueller and the Aquitaine Region. AG was supported by a research grant from INSERM (Poste Accueil INSERM) and VGL was supported by ANR-DFG JAKPOT.

12. De Grandis M, Cambot M, Wautier MP, et al. JAK2V617F activates Lu/BCAM-mediated red cell adhesion in polycythemia vera through an EpoR-independent Rap1/Akt pathway. Blood. 2013;121(4):658-665. 13. Rosti V, Villani L, Riboni R, et al. Spleen endothelial cells from patients with myelofibrosis harbor the JAK2V617F mutation. Blood. 2012;121(2):360-368. 14. Sozer S, Fiel MI, Schiano T, Xu M, Mascarenhas J, Hoffman R. The presence of JAK2V617F mutation in the liver endothelial cells of patients with BuddChiari syndrome. Blood. 2009;113(21): 5246-5249. 15. Teofili L, Martini M, Iachininoto MG, et al. Endothelial progenitor cells are clonal and exhibit the JAK2(V617F) mutation in a subset of thrombotic patients with Ph-negative myeloproliferative neoplasms. Blood. 2011; 117(9):2700-2707. 16. Wakefield TW, Myers DD, Henke PK. Mechanisms of venous thrombosis and resolution. Arterioscler Thromb Vasc Biol. 2008;28(3):387-391. 17. Kaneider NC, Forster E, Mosheimer B, Sturn DH, Wiedermann CJ. Syndecan-4dependent signaling in the inhibition of endotoxin-induced endothelial adherence of neutrophils by antithrombin. Thromb Haemost. 2003;90(6):1150-1157. 18. Burns MP, DePaola N. Flow-conditioned HUVECs support clustered leukocyte adhesion by coexpressing ICAM-1 and Eselectin. Am J Physiol Heart Circ Physiol. 2005;288(1):H194-204. 19. Miszti-Blasius K, Debreceni IB, Felszeghy S, Dezso B, Kappelmayer J. Lack of Pselectin glycoprotein ligand-1 protects mice from thrombosis after collagen/epinephrine challenge. Thromb Res. 2011; 127(3):228-234. 20. Severin S, Gratacap MP, Lenain N, et al. Deficiency of Src homology 2 domain-containing inositol 5-phosphatase 1 affects platelet responses and thrombus growth. J Clin Invest. 2007;117(4):944-952. 21. Hasan S, Lacout C, Marty C, et al. JAK2V617F expression in mice amplifies early hematopoietic cells and gives them a competitive advantage that is hampered by

IFNalpha. Blood. 2013;122(8):1464-1477. 22. Claxton S, Kostourou V, Jadeja S, Chambon P, Hodivala-Dilke K, Fruttiger M. Efficient, inducible Cre-recombinase activation in vascular endothelium. Genesis. 2008; 46(2):74-80. 23. Thambyrajah R, Mazan M, Patel R, et al. GFI1 proteins orchestrate the emergence of haematopoietic stem cells through recruitment of LSD1. Nat Cell Biol. 2016;18(1):2132. 24. Wang L, Benedito R, Bixel MG, et al. Identification of a clonally expanding haematopoietic compartment in bone marrow. EMBO J. 2013;32(2):219-230. 25. Chauhan AK, Kisucka J, Brill A, Walsh MT, Scheiflinger F, Wagner DD. ADAMTS13: a new link between thrombosis and inflammation. J Exp Med. 2008;205(9):2065-2074. 26. Wegmann F, Petri B, Khandoga AG, et al. ESAM supports neutrophil extravasation, activation of Rho, and VEGF-induced vascular permeability. J Exp Med. 2006; 203(7):1671-1677. 27. Yago T, Tsukamoto H, Liu Z, Wang Y, Thompson LF, McEver RP. Multi-inhibitory effects of A2A adenosine receptor signaling on neutrophil adhesion under flow. J Immunol. 2015;195(8):3880-3889. 28. von Bruhl ML, Stark K, Steinhart A, et al. Monocytes, neutrophils, and platelets cooperate to initiate and propagate venous thrombosis in mice in vivo. J Exp Med. 2012;209(4):819-835. 29. Ley K, Laudanna C, Cybulsky MI, Nourshargh S. Getting to the site of inflammation: the leukocyte adhesion cascade updated. Nat Rev Immunol. 2007;7(9):678689. 30. Almeida CB, Scheiermann C, Jang JE, et al. Hydroxyurea and a cGMP-amplifying agent have immediate benefits on acute vaso-occlusive events in sickle cell disease mice. Blood. 2012;120(14):2879-2888. 31. Andre P, Hartwell D, Hrachovinova I, Saffaripour S, Wagner DD. Pro-coagulant state resulting from high levels of soluble Pselectin in blood. Proc Natl Acad Sci USA. 2000; 97(25):13835-13840. 32. Zetterberg E, Verrucci M, Martelli F, et al. Abnormal P-selectin localization during

haematologica | 2019; 104(1)

JAK2V617F endothelial cells are pro-thrombotic

33.

34. 35.

36.

37.

38.

megakaryocyte development determines thrombosis in the gata1low model of myelofibrosis. Platelets. 2014;25(7):539547. Etulain J, Martinod K, Wong SL, Cifuni SM, Schattner M, Wagner DD. P-selectin promotes neutrophil extracellular trap formation in mice. Blood. 2015;126(2):242-246. Martinod K, Wagner DD. Thrombosis: tangled up in NETs. Blood. 2014;123(18):27682776. Kim KJ, Kwon SH, Yun JH, et al. STAT3 activation in endothelial cells is important for tumor metastasis via increased cell adhesion molecule expression. Oncogene. 2017;36(39):5445-5459. Helman R, Pereira WO, Marti LC, et al. Granulocyte whole exome sequencing and endothelial JAK2V617F in patients with JAK2V617F positive Budd-Chiari syndrome without myeloproliferative neoplasm. Br J Haematol. 2018;180(3):443-445. Piaggio G, Rosti V, Corselli M, et al. Endothelial colony-forming cells from patients with chronic myeloproliferative disorders lack the disease-specific molecular clonality marker. Blood. 2009;114(14):3127-3130. Yoder MC, Mead LE, Prater D, et al.

haematologica | 2019; 104(1)

39.

40.

41.

42.

43.

Redefining endothelial progenitor cells via clonal analysis and hematopoietic stem/progenitor cell principals. Blood. 2007;109(5):1801-1809. Sozer S, Ishii T, Fiel MI, et al. Human CD34+ cells are capable of generating normal and JAK2V617F positive endothelial like cells in vivo. Blood Cells Mol Dis. 2009;43(3):304-312. Cokic VP, Beleslin-Cokic BB, Tomic M, Stojilkovic SS, Noguchi CT, Schechter AN. Hydroxyurea induces the eNOS-cGMP pathway in endothelial cells. Blood. 2006;108(1):184-191. Cokic VP, Beleslin-Cokic BB, Noguchi CT, Schechter AN. Hydroxyurea increases eNOS protein levels through inhibition of proteasome activity. Nitric Oxide. 2007;16(3):371-378. Charache S, Terrin ML, Moore RD, et al. Effect of hydroxyurea on the frequency of painful crises in sickle cell anemia. Investigators of the multicenter study of hydroxyurea in sickle cell anemia. N Engl J Med. 1995;332(20):1317-1322. Gangat N, Wolanskyj AP, McClure RF, et al. Risk stratification for survival and leukemic transformation in essential thrombocythemia: a single institutional study of 605

patients. Leukemia. 2007;21(2):270-276. 44. Finazzi G, Caruso V, Marchioli R, et al. Acute leukemia in polycythemia vera: an analysis of 1638 patients enrolled in a prospective observational study. Blood. 2005;105(7):2664-2670. 45. Tefferi A, Rumi E, Finazzi G, et al. Survival and prognosis among 1545 patients with contemporary polycythemia vera: an international study. Leukemia. 2013;27(9):18741881. 46. Matsui NM, Borsig L, Rosen SD, Yaghmai M, Varki A, Embury SH. P-selectin mediates the adhesion of sickle erythrocytes to the endothelium. Blood. 2001;98(6):19551962. 47. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. 48. Etheridge SL, Roh ME, Cosgrove ME, et al. JAK2V617F-positive endothelial cells contribute to clotting abnormalities in myeloproliferative neoplasms. Proc Natl Acad Sci USA. 2014;111(6):2295-2300. 49. Tiedt R, Hao-Shen H, Sobas MA, et al. Ratio of mutant JAK2-V617F to wild-type Jak2 determines the MPD phenotypes in transgenic mice. Blood. 2008;111(8):3931-3940.

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):82-92

Chronic Myeloid Leukemia

Transcriptional activation of the miR-17-92 cluster is involved in the growth-promoting effects of MYB in human Ph-positive leukemia cells Manuela Spagnuolo,1‡ Giulia Regazzo,1‡ Marco De Dominici,2‡ Andrea Sacconi,1 Andrea Pelosi,1† Etleva Korita,1 Francesco Marchesi,3 Francesco Pisani,3 Alessandra Magenta,4 Valentina Lulli,5 Iole Cordone,6 Andrea Mengarelli,3 Sabrina Strano,1 Giovanni Blandino,1 Maria G. Rizzo1* and Bruno Calabretta2*#

Department of Research, Advanced Diagnostics and Technological Innovation, Oncogenomic and Epigenetic Unit, Translational Research Area, IRCCS - Regina Elena National Cancer Institute, Rome, Italy; 2Department of Cancer Biology and Sidney Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA, USA; 3 Department of Clinical and Experimental Oncology–Hematology and Stem Cell Transplant Unit, IRCCS - Regina Elena National Cancer Institute, Rome, Italy; 4Istituto Dermopatico dell'Immacolata-IRCCS, FLMM, Laboratorio di Patologia Vascolare, Rome, Italy; 5Department of Oncology and Molecular Medicine, Istituto Superiore di Sanità, Rome, Italy and 6Department of Research, Advanced Diagnostics and Technological Innovation, Clinical Pathology Unit, IRCCS - Regina Elena National Cancer Institute, Rome, Italy 1

† Present address: Department of Immunology, IRCCS, Ospedale Pediatrico Bambino Gesù, Rome, Italy. ‡MS, GR and MDeD contributed equally to this work. #MGR and BC contributed equally as last authors.

ABSTRACT

Correspondence: maria.rizzo@ifo.gov.it or bruno.calabretta@jefferson.edu Received: February 13, 2018. Accepted: July 27, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.191213 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/82 ©2019 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.

icroRNAs, non-coding regulators of gene expression, are likely to function as important downstream effectors of many transcription factors including MYB. Optimal levels of MYB are required for transformation/maintenance of BCR-ABL-expressing cells. We investigated whether MYB silencing modulates microRNA expression in Philadelphia-positive (Ph+) leukemia cells and if MYB-regulated microRNAs are important for the “MYB addiction” of these cells. Thirtyfive microRNAs were modulated by MYB silencing in lymphoid and erythro-myeloid chronic myeloid leukemia-blast crisis BV173 and K562 cells; 15 of these were concordantly modulated in both lines. We focused on the miR-17-92 cluster because of its oncogenic role in tumors and found that: i) it is a direct MYB target; ii) it partially rescued the impaired proliferation and enhanced apoptosis of MYB-silenced BV173 cells. Moreover, we identified FRZB, a Wnt/β-catenin pathway inhibitor, as a novel target of the miR-17-92 cluster. High expression of MYB in blast cells from 2 Ph+ leukemia patients correlated positively with the miR-17-92 cluster and inversely with FRZB. This expression pattern was also observed in a microarray dataset of 122 Ph+ acute lymphoblastic leukemias. In vivo experiments in NOD scid gamma mice injected with BV173 cells confirmed that FRZB functions as a Wnt/βcatenin inhibitor even as they failed to demonstrate that this pathway is important for BV173-dependent leukemogenesis. These studies illustrate the global effects of MYB expression on the microRNAs profile of Ph+ cells and supports the concept that the “MYB addiction” of these cells is, in part, caused by modulation of microRNA-regulated pathways affecting cell proliferation and survival. Introduction The Philadelphia chromosome (Ph) is the typical chromosomal abnormality of chronic myeloid leukemia (CML) patients.1 It is also detected in a subset of B-cell acute lymphoblastic leukemia (ALL), and less frequently in acute myeloid (AML) and mixed-phenotype acute (MPAL) leukemias.1 The hallmark of the Ph chromosome is the translocation of the proto-oncogene ABL1 from chromosome 9 to the haematologica | 2019; 104(1)

MYB effects on miRNA profile of Ph+ leukemia cells

breakpoint cluster region gene (BCR) on chromosome 22, generating the BCR-ABL1 fusion gene. Such a gene encodes the p190, p210 or the p230 BCR-ABL1 isoforms; these chimeric proteins have constitutively active tyrosine kinase activity and promote the aberrant activation of signaling pathways causing enhanced cell proliferation and resistance to cell death.2 We identified several transcription factors (TFs) whose expression/activity is regulated by BCR-ABL1 oncoproteins and is required for BCRABL1-dependent leukemogenesis.3-6 One such TF is MYB, the prototypical TF of the Myb family,7 essential for fetal and adult hematopoiesis8,9 and required for colony formation of myeloid leukemia blasts, a subset of T-cell leukemia, and BCR-ABL1-transformed myeloid and B cells.6,10-12 In vitro and in mice, BCR-ABL1-transformed cells are more dependent on MYB expression than their normal counterparts,6,12 supporting the concept that certain leukemic cells are “addicted” to MYB.10,11,13 This concept was validated in MLL-AF9-associated AML where partial and transient MYB suppression phenocopies MLL-AF9 withdrawal, eradicating aggressive AML in vivo without preventing normal myelopoiesis.14 MicroRNAs (miRNAs) are small molecules of approximately 22 nucleotides that reprogram gene expression, promoting mRNA degradation and blocking mRNA translation.15 MiRNAs may be especially important in regulating the expression of TFs such as MYB that has distinct biological effects in normal hematopoiesis and in leukemic cells based on its expression levels.15,16 Regulation of MYB expression through miRNAs has been reported previously.17-20 Levels of MYB expression may be differentially controlled by multiple miRNAs and, conversely, MYB could control the expression of different miRNAs9,17-21 to execute lineage-specific developmental choices at critical junctions during hematopoiesis. In particular, overexpression of miR-15 reduced MYB levels in vitro, suppressing erythroid and myeloid colony formation.17 MYB is a direct target of miR-150, playing a key role at different stages of B-cell development.18,20 To gain more information on the role of MYB-regulated miRNAs in leukemic cells, we investigated changes in miRNA levels induced by MYB silencing in Philadelphiapositive (Ph+) cells. We found that, upon MYB silencing, 15 miRNAs are modulated in K562 and in BV173 Ph+ cells. Among these, the miR-17-92 cluster was regulated transcriptionally by MYB through binding to its 5’ regulatory region. Restoring miR-17-92 expression in MYB-silenced BV173 cells partly rescued the reduced proliferation and enhanced apoptosis of these cells. The reduced expression of the miR-17-92 cluster in MYB-silenced Ph+ cells was associated with upregulation of FRZB, an inhibitor of the Wnt/β-catenin pathway, critical for the maintenance of BCR-ABL1-transformed stem cells.22

Methods Cell lines Philadelphia-positive BV173, SUP-B15 and K562 cells were used for the experiments performed in this study. Culture condition, infection with viral vectors to obtain derivative cell lines, transfection, microarray and transcriptional profiling, cell proliferation, cell viability, cell cycle analysis, apoptosis assays, western blotting, RNA isolation and analysis by quan-

haematologica | 2019; 104(1)

titative real-time PCR (qRT-PCR), chromatin immunoprecipitation (ChIP) assays and luciferase assay techniques are all described in the Online Supplementary Methods and Online Supplementary Table S1. Details of statistical/bioinformatic analysis are also described in the Online Supplementary Appendix.

Patients Bone marrow cells were obtained, after informed consent, from 2 Ph+ patients, one with CML-blast crisis with the p210 BCR-ABL isoform, and another with a de novo ALL with the p190 BCR-ABL isoform. In both cases, no additional chromosomal abnormalities were detected by cytogenetic analysis. The study was approved by the Ethical Committee of the Regina Elena National Cancer Institute of Rome, in compliance with the Declaration of Helsinki.

In vivo studies assessing the effects of ectopic FRZB expression Mice were injected in the tail vein with 2x106 BV173-ShMYB 7TFP pUltra-Empty Vector (EV) cells or BV173-ShMYB 7TFP pUltra-hot-FRZB cells (FRZB). Five weeks after the injection, the percentage of circulating leukemia cells was assessed by flow cytometry detection of peripheral blood GFP+mCherry+ cells using the LSR-Fortessa. Mice were sacrificed when moribund and the survival time recorded. For in vivo β-catenin activity analysis, 106 GFP+mCherry+ cells (estimated by flow cytometry) were purified from the bone marrow or the spleen of a mouse injected with EVtransduced or FRZB-expressing BV173 cells, lysed and analyzed for luciferase activity by using the Dual Luciferase Reporter Assay System (Cat. # E1910) and the signal was acquired using a Zylux Femtomaster FB 12 luminometer. Details of the in vivo studies are available in the Online Supplementary Appendix.

Results Differential expression of microRNAs in MYB-silenced Philadelphia-positive leukemic cells We showed previously that optimal levels of MYB expression are required for transformation and maintenance of BCR-ABL-expressing cells.6,12 Since miRNAs are exquisite regulators of gene expression, it is likely that MYB-regulated miRNAs are important for the “MYB addiction” of BCR-ABL-transformed cells. To this end, we performed microarray hybridization studies on RNA from the CML-lymphoid blast crisis BV173 and CML-erythromyeloid blast crisis K562 Ph+ cell lines transduced with the doxycycline (Doxy)-inducible lentiviral vector pLVTSHMYB ShRNA (BV173-ShMYB and K562-ShMYB).23 Compared to untreated (not treated; NT) control cells, Doxy treatment essentially abolished MYB expression in BV173- and K562-ShMYB cells (Figure 1A, upper panel). Unsupervised hierarchical clustering analysis shows expression levels of 519 miRNAs in NT and Doxy-treated [24 hours (h)] BV173- and K562-ShMYB cells (Figure 1A, lower panel). Of these, 125 and 66 were differentially expressed (P≤0.05) in MYB-silenced BV173- and K562-ShMYB cells, respectively (Figure 1B). Of the 35 miRNAs whose expression was altered in both Ph+ cell lines, 15 were modulated concordantly (Online Supplementary Table S2) and 20 discordantly in the two lines (Online Supplementary Table S3). 83

M. Spagnuolo et al.

Real-time PCR analysis of differentially expressed miRNAs in doxycycline-treated BV173-, SUP-B15- and K562-ShMYB cells To validate the results of the miRNA microarray analysis, expression levels of 5 miRNAs (miR-17, miR-18a, miR7, miR-324 and miR-4284) down-regulated by MYB silencing in both cell lines were assessed by qRT-PCR.

These miRNAs were selected based on the fold change of their expression in MYB-silenced cells and their role in tumors.24-28 In agreement with the microarray data, expression of all 5 miRNAs was significantly down-regulated after 48 h Doxy treatment (Figure 1C and D). Of note, levels of miR-17 and miR-7 were significantly down-regulated in K562-ShMYB cells after 24 h Doxy treatment (Figure

E F

Figure 1. miRNA expression profile of MYB-silenced Philadelphia-positive (Ph+) leukemia cells and expression levels of miR-17-92 cluster members. (A) (Upper panels) Western blots of a representative experiment showing specific knockdown of MYB in Doxycycline (Doxy)-treated cells; (lower panel) heat map of differentially expressed miRNAs in Doxy-treated [24 hours (h)] K562-ShMYB and BV173-ShMYB cells. MiRNA expression levels are shown as color variations. Higher and lower values are represented by red and green points, respectively. Pairwise distances between rows and between columns were computed by Euclide distance metric. (B) Venn diagram of differentially expressed miRNAs: 35 miRNAs are commonly modulated in the indicated cell lines. (C and D) qRT-PCR of 5 selected miRNAs from the 15 miRNAs modulated in the same direction in untreated (Not Treated; NT) or Doxy-treated (24-48 h) K562- and BV173-ShMYB cells. Samples were normalized for RNU44 expression. Relative expression was calculated using the comparative Ct method. Data are the average of three independent experiments; error bars indicate SEM. P-values (*P≤0.05; **P≤0.01) were determined using Student t-test. (E) Schematic representation of members of miR-17-92 cluster included in the MIR17HG gene on Chr13q31.3. Arrows represent the direction of miRNA modulation based on the microarray experiment in K562-ShMYB (white) and BV173-ShMYB (black). (F and G) qRT-PCR of the indicated members of miR-17-92 cluster in NT or Doxy-treated (24-48 h) K562-ShMYB and BV173-ShMYB cells. Samples were normalized for RNU44 expression. QRT-PCR was performed in triplicate, including no-template controls. Relative expression was calculated using the comparative Ct method. Data are the average of three independent experiments; error bars indicate Standard Error of Mean. P-values were determined using Student t-test. *P≤0.05; **P≤0.01.

haematologica | 2019; 104(1)

MYB effects on miRNA profile of Ph+ leukemia cells

1C). Since miR-17 and miR-18a belong to the miR-17-92 cluster (Figure 1E) which is involved in BCR-ABL-dependent transformation,29 we also assessed levels of cluster members miR-19a-3p,-19b-3p,-20a-5p and miR-92-5p. These miRNAs were among those expressed in both cell lines in our microarray assay (Online Supplementary Tables S3 and S4). In contrast with the microarray data, qRT-PCR analysis showed that levels of all 4 miRNAs were downregulated in Doxy-treated (48 h) K562-ShMYB cells, whereas only the expression of miR-19a-3p, miR-19b-3p and miR-92-5p (24 h) was decreased in Doxy-treated BV173-ShMYB cells (Figure 1F and G). These conflicting results may depend on the greater sensitivity of the qRTPCR compared to the microarray assay. We also assessed the effects of MYB silencing on the expression of the miR17-92 cluster in the Ph+ ALL cell line SUP-B15 which expresses the p190 BCR-ABL isoform. In this line, Doxy treatment (24 and 48 h) to silence MYB expression induced a statistically significant decrease of miR-17, miR18a, miR-19a and miR-19b levels (Online Supplementary Figure S1). Specificity of the effects of MYB silencing on the expression of the miR-17-92 cluster were demonstrated by using a BV173 derivative line expressing a mutant MYB cDNA harboring synonymous point mutations in the sequence targeted by the MYB shRNA (shRNA-resis-

tant MYB BV173 cell line). Upon Doxy treatment to silence endogenous MYB expression, we found that, in contrast to the parental line (BV173-ShMYB), expression of members of miR-17-92 cluster was not modulated in the BV173 line expressing the MYB cDNA not targetable by the MYB ShRNA (Online Supplementary Figure S2). Thus, Doxy-induced changes in the expression of the miR-17-92 cluster are a specific consequence of MYB silencing.

MYB binds the promoter of the miR-17-92 cluster To investigate whether MYB could directly regulate transcription of the miR-17-92 cluster, we analyzed the MIR17HG promoter for the presence of putative MYB binding sites (MBS). Using MatInspector (www.genomatix.de/matinspector.html), we scanned 4000 bp upstream of the MIR17HG gene and identified several putative MBS (Figure 2A). We focused on 5 MBS with the highest matrix similarity score (Online Supplementary Table S5). Genomic positions of these MBSs relative to MIR17HG Transcriptional Start Site (TSS) are indicated in Figure 2A. To assess whether MYB binds these regions in vivo, ChIP assays were performed in NT and Doxytreated BV173- and K562-ShMYB cells and de-cross-linked DNA amplified with primers flanking genomic regions

Figure 2. MYB binding to the promoter of the miR-17-92 cluster. (A) Schematic representation of 4000 bp regulatory regions upstream of the MIR17HG promoter. A transcription start site (TSS) is indicated. Arrows indicate the promoter region amplified by the specific primer pair used for qPCR amplification of immunoprecipitated chromatin. (B) ChIP analysis of the MIR17HG promoter using the indicated MYB antibody in untreated (Not Treated; NT) or Doxycycline (Doxy)-treated BV173ShMYB and K562-ShMYB cells. Results of qPCR are analyzed with the comparative Ct method. Values of each immunoprecipitated sample are expressed as percentage relative to their respective input and by subtracting the values obtained in the negative controls (no antibody). Error bars represent Standard Error of Mean (SEM) (n=3); P-values (*P≤0.05) were determined using Student t-test. (C) (Left panel) Schematic representation of the reporter plasmids containing the MYB binding site (MBS) #1 (pGL3–prom1353) or its deletion mutant without the MBS#1 (ΔMBS#1-prom230). (Right panel) Dual luciferase assay performed in untreated or Doxy-treated BV173-ShMYB cells transfected with the pGL3–prom1353 or the ΔMBS#1-prom230 plasmid. Promoter activity of each plasmid was determined 48 hours (h) after transfection. Luciferase activity values were normalized for transfection efficiency according to the activity of a co-transfected Renilla luciferase plasmid. Data are the average of three independent experiments performed in duplicate; error bars indicate Standard Error of Mean (*P≤0.05).

haematologica | 2019; 104(1)

M. Spagnuolo et al.

Figure 3. Biological effects of over-expressed miR-17-92 cluster in MYB silenced Philadelphia-positive (Ph+) BV173 cells. (A) (Upper panel) Western blots of a representative experiment showing specific knockdown of MYB in Doxycycline (Doxy)-treated [24, 48 and 72 hours (h)] BV173-ShMYB cells; (lower panel) qRT-PCR of the indicated members of the miR-17-92 cluster in BV173-ShMYB-Empty Vector (EV) and the miR-17-92 over-expressing cells. Results are expressed as fold changes [mean±Standard Error of Mean (SEM) from three independent experiments] in miRNA expression in BV173-ShMYB-miR-17-92 cells as compared with values in BV173-ShMYB-EV cells. (B) MTT and ATPlite assays; data are the average of three independent experiments, and percentage of cell survival (left panel) and cell viability (right panel) were assessed at the indicated times of Doxy treatment. (C) Percentage of S-phase cells over control for untreated or Doxy-treated (48 h) BV173ShMYB-EV and derivative miR-17-92 over-expressing lines (**P≤0.01). (D) (Left panel) Percentage of Annexin V for untreated or Doxy-treated (96 h) BV173-ShMYBEV and derivative miR-17-92 over-expressing lines (*P≤0.05). (Middle panel) Western blot of a representative experiment of MYB, uncleaved PARP, BCL-2 and actin protein levels in BV173-ShMYB-EV and BV173-ShMYB-miR-17-92 over-expressing cells, 72 h after MYB silencing. (Right panel) Densitometric analysis by imageJ software. Actin was used as loading control within the same sample and expressed as fold changes compared to control.

that include putative MBS (Figure 2A). As a positive control, ChIP was performed using an MBS-containing segment of the adenosine deaminase gene (ADA), a known transcriptional target of c-MYB.30 MYB bound efficiently, in both untreated cell lines, to the promoter region that includes MBS#1, the site closest to the TSS of MIR17HG (Figure 2B); in contrast, reduced binding was detected at all other promoter segments (Figure 2B), especially in BV173 cells. Binding of MYB to the region of the miR-1792 promoter that includes MBS#1 was markedly decreased upon Doxy treatment (72 h) of BV173- and K562-ShMYB cells (Figure 2B). As expected, MYB binding to the ADA promoter was also decreased (Figure 2B). To further investigate whether the miR-17-92 cluster is directly regulated by MYB we carried out luciferase assay using reporter plasmids with or without MBS#1 (PGL3prom1353 and ΔMBS#1-prom230, respectively) (Figure 2C, left). We found that the luciferase activity of the ShMYB-BV173 cells transfected with the PGL3-prom1353 was decreased by approximately 33% after a 24 h Doxy treatment to silence MYB expression; in contrast, in cells transfected with the truncated ΔMBS#1-prom230 plasmid lacking MBS#1 there was only a 4% decrease of luciferase activity after Doxy treatment (Figure 2C, right). These data strongly suggest that MYB is important for the transcription of the MIR17HG locus. 86

Involvement of the miR-17-92 cluster in the “MYB addiction” of Ph+ leukemia cells To investigate whether restoring expression of the miR17-92 cluster affects the phenotype of MYB-silenced cells, we generated BV173-ShMYB cells over-expressing the miR-17-92 cluster and assessed proliferation and survival of these cells upon MYB silencing. These studies were not performed in K562 cells because the biological effects induced by MYB silencing in these cells were modest, compared to those in BV173 cells. Expression of MYB was suppressed in Doxy-treated BV173-ShMYB cells and in the miR-17-92 derivative line which exhibited increased expression of each member of the miR-17-92 cluster (Figure 3A). Compared to BV173-ShMYB-EV cells, the miR-17-92 over-expressing cell lines showed increased proliferation (P≤0.01) upon MYB silencing. This was evident after 24 h of Doxy treatment and persisted at 48 h and 72 h (Figure 3B, left). Likewise, viability of Doxytreated BV173-ShMYB cells over-expressing the miR-1792 cluster was significantly increased (P≤0.01) compared to that of Doxy-treated BV173-ShMYB-EV cells (Figure 3B, right). DNA content analysis revealed that Doxytreated BV173-ShMYB cells over-expressing miR-17-92 have a greater proportion of S-phase cells than Doxytreated BV173-ShMYB-EV cells (12% vs. 6% after 48 h Doxy treatment) (Figure 3C). In addition, cultures of haematologica | 2019; 104(1)

MYB effects on miRNA profile of Ph+ leukemia cells

Table 1. Predicted down-and up-regulated target genes in BV173-ShMYB and K562-ShMYB cells after gene expression and miR-17-92 cluster analyses.

Down-regulated genes

BV173 DFCtest

K562 DFCtest

Up-regulated genes

Reference

BV173 DFCtest

K562 DFCtest

BAZ1B BUB1 CASP6 CNOT6L CPT1A EFNB2 GBE1 HDAC4 HRH2 ID2 ITGA4 ITGA4 MAD2L1 MAP3K1 MYO10 NR3C1 NRP2 PDE3B PIBF1 PRKRA REST RFC3 RPS6KA5 SCML2 SDC2 SERPINB8 TFRC TLE4 TNFAIP3

2.26E-10 0.025 0.001 0.027 0.02 0.046 4.74E-22 0.004 5.14E-09 0.031 2.53E-09 2.53E-09 1.76E-08 1.71E-09 2.59E-10 0.03 0.002 0.014 0.047 0.017 0.027 8.99E-11 0.028 0.03 0.014 7.04E-05 5.72E-10 7.65E-14 7.89E-19

0.0239 0.0033 0.0043 4.65E-05 0.026 0.0188 0.0034 0.0014 0.0222 0.0053 0.0152 0.0004 0.0472 0.0346 0.0362 0.0109 0.0133 0.0059 0.0063 0.0087 0.0256 0.025 0.0372 0.0475 0.0012 0.0111 1.72E-05 0.0017 0.0026

ABCA1 ADARB1 ARHGAP1 BPNT1 CD22 COL1A1 FRZB KIAA0513 PBX2 PEX10 PTP4A3 RAB13 RPL19 SPIB THBS1

0.009 1.01E-19 0.008 0.013 4.97E-15 0.014 0.0001 1.55E-17 0.0003 0.032 8.36E-17 4.56E-14 0.005 0.003 0.001

0.006 0.026 0.034 0.002 0.003 0.018 0.036 0.007 0.012 0.016 0.041 0.032 0.007 0.01 0.023

Doxy-treated BV173-ShMYB over-expressing miR-17-92 cells had less apoptosis than Doxy-treated BV173ShMYB-EV cells, as indicated by the lower frequency of Annexin V-positive cells (9% vs. 15%, after 96 h Doxy treatment) (Figure 3D, left) and the increased expression of uncleaved PARP and BCL-2 (48% and 14%, respectively) (Figure 3D, middle and right panels).

17-92 cluster in MYB-silenced cells should increase the levels of its putative targets. Thus, we performed qRTPCR to assess the expression of two candidate targets, PBX2 and FRZB, involved in the regulation of proliferation and apoptosis.31,32 Such analysis revealed a statistically significant (P≤0.05) increase of PBX2 and FRZB expression in MYB-silenced Ph+ ALL BV173 and SUP-B15 or K562 cells (Figure 4B and C).

Integrative analysis of gene expression profiles of MYB-silenced cells and predicted miRNA-regulated genes identifies novel putative miR-17-92 targets

FRZB is a potential effector of the miR-17-92 cluster in the “MYB addiction” of Ph+ leukemia cells

We used gene expression profiling of MYB-silenced cells to identify MYB target genes potentially regulated by the miR-17-92 cluster. The miRWalk 2.0 database was used to investigate potential interactions of the miR-17-92 cluster with genes regulated by MYB silencing in BV173 and K562 cells. From this analysis, we found that 44 genes modulated by MYB silencing (15 up-regulated and 29 down-regulated) are predicted targets of the miR-17-92 cluster (Table 1 and Figure 4A). We focused on the up-regulated genes since the decreased expression of the miR-

The oncogenic effect of the miR-17-92 cluster is caused by the co-operation of its members in targeting tumorsuppressive pathways.28,33 Several studies have shown that the miR-17-92 cluster directly targets “pro-apoptotic” genes such as Phosphatase and tensin homolog (PTEN), the apoptosis facilitator BCL2L11 (BIM) and the antiangiogenic factor thrombospondin-1 (THBS1) in normal lymphopoiesis,34-37 in MYC-driven lymphomas38 and in immunodeficiency or lymphoproliferative states.39 To assess whether the expression of validated miR-17-92 tar-

haematologica | 2019; 104(1)

M. Spagnuolo et al.

gets is modulated by miR-17-92 overexpression, qRT-PCR experiments were performed in Doxy-treated BV173ShMYB-EV cells and in the miR-17-92 over-expressing line. After 24 h Doxy treatment, levels of BIM and PTEN mRNA were essentially identical in both BV173-ShMYB-

EV and BV173-ShMYB-miR-17-92 over-expressing cells compared to those in NT cells (Figure 4D). In contrast, THBS1 mRNA levels showed an increase (P≤0.05) in Doxy-treated BV173-ShMYB-EV cells compared to untreated cells, and such an increase was blocked by over-

Figure 4. Transcriptional analysis and evaluation of mRNA expression levels of miR-17-92 cluster target genes. (A) Unsupervised hierarchical clustering of common deregulated genes from gene expression analysis of parental and MYB-silenced BV173 and K562 cells. (B and C) qRT-PCR of PBX2 and FRZB expression levels upon MYB knockdown [24 hours (h)] of the indicated Philadelphia-positive (Ph+) ShMYB cell lines. Results are mean of three experiments. Error bars indicate Standard Error of Mean (SEM). (D) Analysis of mRNA expression levels, using SYBR Green-based qRT-PCR, of BIM, PTEN and THBS1 in untreated and Doxy-treated BV173ShMYB-Empty Vector (EV) and ShMYB-miR-17-92 cells. Results are mean of three experiments. Error bars indicate SEM. (E) Quantification by SYBR Green-based qRTPCR of PBX2 and FRZB mRNA in untreated and Doxy-treated BV173-ShMYB-miR-17-92 cells. Values are reported as 2-ΔCt. GAPDH gene expression was used as endogenous control. Error bars indicate SEM (n=3). (F) (Left panel) Schematic representation of 3’UTRs of FRZB gene with putative binding sites for miR-17-92 cluster. (Right panel) Schematic representation of reporter plasmids containing the wild-type (wt) or mutant (76-81 mut, 1091-1097 mut of miR-17-92-binding sequences) FRZB 3’UTR. Dual Luciferase assay in recipient cells co-transfected with luciferase reporter vectors containing the wt-3’UTR FRZB or the indicated FRZB mutant and either the hsa-miR-17, the hsa-miR-19a mimics or a control (Ctr)-mimic RNA. Firefly luciferase activity of each sample was normalized by Renilla luciferase activity. Results are expressed as fold activation relative to the basal activity of the control mimic (ctr-mimic). (*P≤0.05). The normalized luciferase activity, set as mean of at least three independent experiments performed in duplicate, is shown. Error bars represent the mean±SEM (n=3).

haematologica | 2019; 104(1)

MYB effects on miRNA profile of Ph+ leukemia cells

expression of the miR-17-92 cluster (Figure 4D). Expression levels of BIM, PTEN and THBS1 mRNA, after 24 h Doxy treatment, were assessed also in the SUP-B15 ShMYB cells analysis. This revealed a statistically significant (P≤0.05) increase of BIM and THBS1 in MYB-silenced SUP-B15 cells compared to untreated cells (Online Supplementary Figure S3). The expression of p21 and E2F1 genes, two other experimentally validated miR-17-92 targets,40 was also assessed in MYB-silenced BV173 cells. MYB silencing induced an increase in the expression of p21 but this increase was not blocked by overexpression of the miR-17-92 cluster. In contrast, expression of E2F1 was down-modulated after MYB silencing and was not affected by overexpression of the miR-17-92 cluster (Online Supplementary Figure S4). These results suggest that MYB silencing modulates p21 and E2F1 expression independently of its effect on the miR-17-92 cluster expression.

Since our goal was to investigate novel miR-17-92 targets, potentially involved in the “MYB addiction” of Ph+ leukemia cells, we focused on FRZB because ectopic expression of the miR-17-92 cluster blocked the increased expression of FRZB mRNA but not of PBX2 mRNA (Figure 4E) induced by MYB silencing in BV173-ShMYB cells (Figure 4B). FRZB is the founding member of the secreted Frizzled-related protein (SFRP) family of Wnt inhibitors32,41 and suppresses Wnt signaling thus preventing the accumulation of β-catenin into the nucleus.42 Then, we used miRwalk (http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk2/), TargetScan 5.2 (http://www.targetscan.org), and miRanda (http://www.microrna.org/microrna) algorithms to assess the presence of putative miR-17-92-binding sites within the 3’untranslated region (3’UTR) of FRZB-mRNA. This analysis identified one putative miR-17-92 binding site for miR-19a (seed sequences: 76-81 bp) and one for miR-17 and -20a (seed sequences:1091-1097 bp) (Figure 4F,

Figure 5. Expression of the miR-17-92 cluster and its target FRZB correlates with MYB levels in Philadelphia-positive (Ph+) acute lymphoblastic leukemia (ALL) cells. (A) MiR-17-92 expression levels evaluated by stem-loop qRT-PCR in primary leukemia cells (Patient 1: p210BCR/ABL chronic myeloid leukemia (CML)-myeloid blast crisis) compared to normal CD34+ cells from a healthy subject [Control (Ctrl) CD34+] and Patient 2 (p190BCR/ABL ALL) compared to normal peripheral blood mononuclear (PBMC) cells (Ctrl/PB). Samples were normalized for RNU44 small-nucleolar RNA expression using the comparative Ct method. Data are the average of three experiments; error bars indicate Standard Deviation (SD). (B) mRNA quantification of MYB and FRZB, by SYBR Green-based qRT-PCR, in Patient 1 (p210BCR/ABL CML-myeloid blast crisis) and Patient 2 (p190BCR-ABL ALL) compared to normal CD34+ cells and PBMC cells from healthy donors (Ctrl/CD34+ and Ctrl/PB), respectively. Values are reported as 2-ΔCt normalizing to GAPDH gene expression. (C) mRNA expression by microarray of MYB or FRZB in normal B cells or Ph+ ALL cells. (Values represent the sum of all probes signals for each gene and are derived from dataset GSE13159).

haematologica | 2019; 104(1)

M. Spagnuolo et al.

Figure 6. Effect of FRZB expression on leukemogenesis and β-catenin activity of Philadelphia-positive (Ph+) BV173 cells. (A) Survival of mice injected with 2x106 BV173-ShMYB 7TFP pUltra-Empty Vector (EV) or BV173-ShMYB 7TFP pUltra-hot-FRZB cells (FRZB). (B) Luciferase reporter assay for β-catenin activity in GFP+ cells isolated from the bone marrow (bm) or spleen (sp) of a NOD scid gamma (NSG) mouse injected with (EV)- or FRZB-BV173 cells and sacrificed when terminally ill.

left panel). To assess whether FRZB is a direct target of miR-17-92, a human FRZB 3’UTR fragment containing wild-type or mutated miR-17 or miR-19a seed sequences (Figure 4F, middle panel) was cloned downstream of the firefly luciferase reporter gene and co-transfected with miR-17 or miR-19a mimics in 293T cells. The relative luciferase activity of the reporter with wild-type 3’UTR was decreased by 27% upon expression of the miR-17 mimic and by 29% upon expression of the miR-19a mimic; in contrast, there was no decrease in luciferase activity of the mutant reporter (Figure 4F, right panel), suggesting that FRZB is a direct target of miR-17 as well as of miR-19a. To investigate whether FRZB has a role as a miR-17-92 target gene in the “MYB addiction” of BCR-ABL-transformed cells, we assessed the relative expression of FRZB, the miR-17-92 cluster and MYB in blast cells from 2 Ph+ leukemia patients (n=1: p210BCR/ABL CML-myeloid blast crisis; n=1: p190BCR/ABL ALL). High expression of the miR-17-92 cluster correlated with that of MYB and was more abundant than in CD34+ or peripheral blood mononuclear cells from healthy donors (Figure 5A and B, left panel). In contrast, levels of FRZB were much higher in cells from healthy donors than in blast cells from the Ph+ leukemia patients (Figure 5B, right panel). In agreement with these findings, we found that, in a microarray dataset of 122 Ph+ ALL samples, MYB mRNA levels were more abundant in Ph+ ALL cells compared to normal B cells, while the opposite was found for FRZB expression (Figure 5C). To investigate directly whether expression of FRZB has a negative effect for leukemia development, NOD scid gamma (NSG) mice were injected with EV-transduced or FRZB-expressing BV173 cells carrying the β-catenin-Luc reporter plasmid and assessed for overall survival. Survival of the two groups was identical (Figure 6A); however, β-catenin activity was markedly reduced in BV173 cells isolated from the bone marrow or spleen of a mouse injected with FRZB-expressing compared to EV-transduced cells (Figure 6B). These data suggest that leukemia induced by Ph+ BV173 cells is β-catenin-independent but do not exclude the possibility that FRZB-dependent regulation of β-catenin activity is important for leukemia induced by primary Ph+ ALL cells. 90

Discussion The expression of MYB is critical for the proliferation and survival of many leukemic cells, including BCR-ABL1transformed myeloid and lymphoid cells;6,12 however, the mechanisms responsible for the “MYB addiction” of these cells are only partially understood. In this study, we assessed the miRNA expression profile of MYB-silenced BV173 and K562 CML-blast crisis cells with the goal of identifying miRNAs whose modulation might explain the impaired proliferation and survival associated with MYB knockdown in BCR-ABL1-transformed lymphoid or myeloid precursors.6,12 Interestingly, MYB appears to have broad effects, directly or indirectly, on the levels of miRNAs since approximately 24% and 13% of those expressed in BV173 and K562 cells, respectively, were modulated by MYB silencing. Although many miRNAs regulated by MYB exhibited changes in both cell lines, a high number of the modulated miRNAs exhibited cell-type specificity. We speculated that those modulated by MYB in a celltype specific manner may regulate pathways required for more specialized cell functions, while those regulated in both cell lines may be involved in more general biological processes, such as cell proliferation and survival. Within the miRNAs regulated by MYB in both cell lines, we focused on the miR-17-92 cluster because of its oncogenic role in many tumors,28,43,44 its involvement in BCR-ABL1transformed cells,45 and its regulation by MYC,44 a known MYB target.46 We found that MYB bound directly to the miR-17-92 promoter, suggesting that its effects on the expression of several members of the miR-17-92 cluster are direct, although an indirect effect through other transcription factors (eg. c-Myc) and/or co-activators cannot be excluded.47 On the other hand, silencing MYB alone does not abolish expression of the miR-17-92 cluster, suggesting that other transcription factors also regulate the expression of the miR-17-92 cluster in BCR-ABL1-transformed cells.16 Compared to control cells, MYB-silenced BV173 cells exhibit a marked inhibition of cell growth which is due to cell-cycle arrest and induction of apoptosis.48 Thus, we asked whether restoring expression of the miR-17-92 cluster would rescue the impaired growth of haematologica | 2019; 104(1)

MYB effects on miRNA profile of Ph+ leukemia cells

MYB-silenced BV173 cells. Ectopic expression of the miR17-92 cluster caused an increase in the S phase fraction and a decrease in the apoptosis of MYB-silenced BV173 cells, but the effect was modest. This is not surprising, since silencing MYB expression induces global changes in miRNA and mRNA levels causing an impaired proliferation and survival that cannot be rescued by expression of the miR-17-92 cluster alone. The expression of some established targets of the miR-17-92 cluster (e.g. p21 and E2F1) was also markedly modulated by MYB silencing; however, restoring the targets of the miR-17-92 cluster did not change the effects on such expression induced by MYB silencing, strongly suggesting that the predominant mechanism of MYB regulation of these two genes is miR17-92-independent. In contrast, ectopic expression of miR-17-92 completely blocked the upregulation of THBS1, a known miR-17-92 target,37 and of FRZB, a novel candidate for miR-17-92 inhibition, which is induced by MYB silencing. FRZB functions as an inhibitor of the Wnt/β-catenin signaling pathway which is activated in CML stem cells/early progenitors and is important for their proliferation and survival.22,42 However, ectopic expression of FRZB in BV173 cells, when injected in NSG mice, had no effect on their survival, in spite of a marked inhibition of β-catenin activity. These data suggest that BV173 cells induce leukemia in mice through β-catenin-independent mechanisms but do

References 9. 1. Kang ZJ, Liu YF, Xu LZ, et al. The Philadelphia chromosome in leukemogenesis. Chin J Cancer. 2016;35:48. 2. Li S, Ilaria RL Jr, Million RP, Daley GQ, Van Etten RA. The P190, P210, and P230 forms of the BCR/ABL oncogene induce a similar chronic myeloid leukemia-like syndrome in mice but have different lymphoid leukemogenic activity. J Exp Med. 1999; 189(9):1399-1412. 3. Perrotti D, Cesi V, Trotta R, et al. BCR-ABL suppresses C/EBPalpha expression through inhibitory action of hnRNP E2. Nat Genet. 2002;30(1):48-58. 4. Soliera AR, Mariani SA, Audia A, et al. Gfi1 inhibits proliferation and colony formation of p210BCR/ABL-expressing cells via transcriptional repression of STAT 5 and Mcl-1. Leukemia. 2012;26(7):1555-1563 5. Soliera AR, Lidonnici MR, Ferrari-Amorotti G, et al. Transcriptional repression of c-Myb and GATA-2 is involved in the biologic effects of C/EBPalpha in p210BCR/ABLexpressing cells. Blood. 2008;112(5):19421950. 6. Lidonnici MR, Corradini F, Waldron T, Bender TP, Calabretta B. Requirement of cMyb for p210 (BCR/ABL)-dependent transformation of hematopoietic progenitors and leukemogenesis. Blood. 2008; 111(9):47714779. 7. Ramsay RG, Gonda TJ. MYB function in normal and cancer cells. Nat Rev Cancer. 2008; 8(7):523-534. 8. Lieu YK, Reddy EP. Conditional c-myb knockout in adult hematopoietic stem cells leads to loss of self-renewal due to impaired proliferation and accelerated differentiation.

haematologica | 2019; 104(1)

10.

11.

12.

13.

14.

15.

16.

not exclude the possibility that FRZB-dependent regulation of β-catenin activity is important for leukemia induced by primary Ph+ ALL cells. In summary, this study illustrates the global effects of MYB expression on the miRNA profile of Ph+ leukemic cells and supports the concept that the “MYB addiction” of Ph+ BV173 cells is, in part, caused by modulation of miRNA-regulated pathways affecting cell proliferation and survival. Acknowledgments This article is dedicated to the memory of Professor Franco Mandelli for his life-long dedication to research on hematological malignancies and patients’ care. “The results achieved are not a treasure to be defended but a wealth to be transmitted”. The authors would like to thank Dr Scott M. Hammond from University of North Carolina for a kind gift of the MIR17HG promoter plasmids and Dr. Thomas Gonda for the pLVTSH ShMYB lentivirus. Funding This work was supported, in part, by NCI grant CA167169 to BC, Italian Association for Cancer Research (AIRC) grant to GB and by Funds Celgene protocol (08/CE/R/15) to FP. GR is a PhD student at University of Rome “Sapienza” and an Intramural funds (Hematologic Tumors) fellow. MS is a recipient of an Intramural fellowship.

Proc Natl Acad Sci USA. 2009; 106(51): 21689-21694 Bianchi E, Bulgarelli J, Ruberti S, et al. MYB controls erythroid versus megakaryocyte lineage fate decision through the miR-4863p-mediated downregulation of MAF. Cell Death Differ. 2015;22(12):1906-1921. Calabretta B, Sims RB, Valtieri M, et al. Normal and leukemic hematopoietic cells manifest differential sensitivity to inhibitory effects of c-myb antisense oligodeoxynucleotides: an in vitro study relevant to bone marrow purging. Proc Natl Acad Sci USA. 1991;88(6):2351-2355. Lahortiga I, De Keersmaecker K, Van Vlierberghe P, et al. Duplication of the MYB oncogene in T cell acute lymphoblastic leukemia. Nat Genet. 2007;39(5):593-595. Waldron T, De Dominici M, Soliera AR, et al. c-Myb and its target Bmi1 are required for p190BCR/ABL leukemogenesis in mouse and human cells. Leukemia. 2012;26(4):644653 Ratajczak MZ, Hijiya N, Catani L, et al. Acute- and chronic-phase chronic myelogenous leukemia colony-forming units are highly sensitive to the growth inhibitory effects of c-myb antisense oligodeoxynucleotides. Blood. 1992;79(8):1956-1961. Zuber J, Rappaport AR, Luo W, et al. An integrated approach to dissecting oncogene addiction implicates a Myb-coordinated self-renewal program as essential for leukemia maintenance. Genes Dev. 2011;25(15):1628-1640. El Gazzar M, McCall CE. MicroRNAs regulatory networks in myeloid lineage development and differentiation: regulators of the regulators. Immunol Cell Biol. 2012;90(6): 587-593. Organista-Nava J, Gomez-Gomez Y, Illades-

17.

18.

19. 20.

21.

22.

23.

24.

25.

Aguiar B, Leyva-Vazquez MA. Regulation of the miRNA expression by TEL/AML1, BCR/ABL, MLL/AF4 and TCF3/PBX1 oncoproteins in acute lymphoblastic leukemia (Review). Oncol Rep. 2016;36(3):1226-1232. Zhao H, Kalota A, Jin S, Gewirtz AM. The cmyb proto-oncogene and microRNA-15a comprise an active autoregulatory feedback loop in human hematopoietic cells. Blood. 2009;113(3):505-516. Xiao C, Calado DP, Galler G, et al. MiR-150 controls B cell differentiation by targeting the transcription factor c-Myb. Cell. 2007; 131(1):146-159. Fahl SP, Crittenden RB, Allman D, Bender TP. c-Myb is required for pro-B cell differentiation. J Immunol. 2009;183(9):5582-5592. Garcia P, Frampton J. Hematopoietic lineage commitment: miRNAs add specificity to a widely expressed transcription factor. Dev Cell. 2008;14(6):815-816. Vargova K, Curik N, Burda P, et al. MYB transcriptionally regulates the miR-155 host gene in chronic lymphocytic leukemia. Blood. 2011;117(14):3816-3825. Hu Y, Chen Y, Douglas L, Li S. Beta-Catenin is essential for survival of leukemic stem cells insensitive to kinase inhibition in mice with BCR-ABL-induced chronic myeloid leukemia. Leukemia. 2009; 23(1):109-116. Drabsch Y, Hugo H, Zhang R, et al. Mechanism of and requirement for estrogen-regulated MYB expression in estrogenreceptor-positive breast cancer cells. Proc Natl Acad Sci USA. 2007;104(34):1376213767. Horsham JL, Kalinowski FC, Epis MR, Ganda C, Brown RA, Leedman PJ. Clinical Potential of microRNA-7 in Cancer. J Clin Med. 2015;4(9):1668-1687. Blume CJ, Hotz-Wagenblatt A, Hüllein J, et

M. Spagnuolo et al.

26.

27.

28.

29.

30.

31.

32.

al. p53-dependent non-coding RNA networks in chronic lymphocytic leukemia. Leukemia. 2015;29(10):2015-23. Kuo WT, Yu SY, Li SC, et al. MicroRNA-324 in Human Cancer: miR-324-5p and miR324-3p Have Distinct Biological Functions in Human Cancer. Anticancer Res. 2016;36(10):5189-5196. Tamaddon G, Geramizadeh B, Karimi MH, Mowla SJ, Abroun S. miR-4284 and miR4484 as Putative Biomarkers for Diffuse Large B-Cell Lymphoma. Iran J Med Sci. 2016;41(4):334-339. Mogilyansky E, Rigoutsos I. The miR-17/92 cluster: a comprehensive update on its genomics, genetics, functions and increasingly important and numerous roles in health and disease. Cell Death Differ. 2013;20(12):1603-1614. Scherr M, Elder A, Battmer K, et al. Differential expression of miR-17~92 identifies BCL2 as a therapeutic target in BCRABL-positive B-lineage acute lymphoblastic leukemia. Leukemia. 2014;28(3):554-565. Ess KC, Whitaker TL, Cost GJ, Witte DP, Hutton JJ, Aronow BJ. A central role for a single c-Myb binding site in a thymic locus control region. Mol Cell Biol. 1995; 15(10):5707-5715. Pelosi A, Careccia S, Lulli V, et al. miRNA let-7c promotes granulocytic differentiation in acute myeloid leukemia. Oncogene. 2013;32(31):3648-3654. Surana R, Sikka S, Cai W, et al. Secreted frizzled related proteins: Implications in cancers. Biochim Biophys Acta. 2014;

1845(1):53-65. 33. Ventura A, Young AG, Winslow MM, et al. Targeted deletion reveals essential and overlapping functions of the miR-17 through 92 family of miRNA clusters. Cell. 2008;132(5):875-886. 34. Mavrakis KJ, Wolfe AL, Oricchio E, et al. Genome-wide RNA-mediated interference screen identifies miR-19 targets in Notchinduced T-cell acute lymphoblastic leukaemia. Nat Cell Biol. 2010;12(4):372379. 35. Dews M, Homayouni A, Yu D, et al. Augmentation of tumor angiogenesis by a Myc-activated microRNA cluster. Nat Genet. 2006;38(9):1060-1065. 36. Mo W, Zhang J, Li X, et al. Identification of novel AR-targeted microRNAs mediating androgen signalling through critical pathways to regulate cell viability in prostate cancer. PLoS One. 2013;8(2):e56592. 37. Psathas JN, Doonan PJ, Raman P, Freedman BD, Minn AJ, Thomas-Tikhonenko A. The Myc-miR-17-92 axis amplifies B-cell receptor signaling via inhibition of ITIM proteins: a novel lymphomagenic feed-forward loop. Blood. 2013;122(26):4220-4229. 38. Mu P, Han YC, Betel D, et al. Genetic dissection of the miR-17~92 cluster of microRNAs in Myc-induced B-cell lymphomas. Genes Dev. 2009;23(24):2806-2811. 39. Xiao C, Srinivasan L, Calado DP, et al. Lymphoproliferative disease and autoimmunity in mice with increased miR-17-92 expression in lymphocytes. Nat Immunol. 2008;9(4):405-414.

40. Inomata M, Tagawa H, Guo YM, Kameoka Y, Takahashi N, Sawada K. MicroRNA-1792 down-regulates expression of distinct targets in different B-cell lymphoma subtypes. Blood. 2009;113(2):396-402. 41. Leyns L, Bouwmeester T, Kim SH, Piccolo S, De Robertis EM. Frzb-1 is a secreted antagonist of Wnt signaling expressed in the Spemann organizer. Cell. 1997; 88(6):747756. 42. Reya T, Clevers H. Wnt signalling in stem cells and cancer. Nature. 2005; 434(7035): 843-850. 43. He L, Thomson JM, Hemann MT, et al. A microRNA polycistron as a potential human oncogene. Nature. 2005; 435(7043):828-833. 44. Olive V, Jiang I, He L. mir-17-92, a cluster of miRNAs in the midst of the cancer network. Int J Biochem Cell Biol. 2010; 42(8):1348-54. 45. Venturini L, Battmer K, Castoldi M, et al. Expression of the miR-17-92 polycistron in chronic myeloid leukemia (CML) CD34+ cells. Blood. 2007;109(10):4399-4405. 46. Cogswell JP, Cogswell PC, Kuehl WM, et al. Mechanism of c-myc regulation by c-Myb in different cell lineages. Mol Cell Biol. 1993;13(5):2858-2869. 47. Ji M, Rao E, Ramachandrareddy H, et al. The miR-17-92 microRNA cluster is regulated by multiple mechanisms in B-cell malignancies. Am J Pathol. 2011; 179(4):1645-1656. 48. De Dominici M, Porazzi P, Soliera AR, et al. Targeting CDK6 and BCL2 exploits the "MYB addiction" of Ph+ acute lymphoblastic leukemia. Cancer Res. 2018;78(4):10971109.

haematologica | 2019; 104(1)

ARTICLE

Chronic Myeloid Leukemia

Incidence, outcomes, and risk factors of pleural effusion in patients receiving dasatinib therapy for Philadelphia chromosome-positive leukemia

Ferrata Storti Foundation

Timothy P. Hughes,1 Pierre Laneuville,2 Philippe Rousselot,3 David S. Snyder,4 Delphine Rea,5 Neil P. Shah,6 David Paar,7 Elisabetta Abruzzese,8 Andreas Hochhaus,9 Jeffrey H. Lipton10 and Jorge E. Cortes11

Cancer Theme, SAHMRI, Division of Haematology, SA Pathology, University of Adelaide, South Australia, Australia; 2McGill University Health Centre, Montreal, QC, Canada; 3 Hôpital Mignot, Université Versailles Saint-Quentin-en-Yvelines, Le Chesnay, France; 4 City of Hope, Duarte, CA, USA; 5Service d'Hématologie Adulte, Hôpital Saint-Louis, Paris, France; 6UCSF School of Medicine, San Francisco, CA, USA; 7Bristol-Myers Squibb, Princeton, NJ, USA; 8Hematology, S. Eugenio Hospital, Tor Vergata University, Rome, Italy; 9Hematology/Oncology, Universitätsklinikum Jena, Germany; 10Princess Margaret Cancer Centre, Toronto, ON, Canada and 11University of Texas MD Anderson Cancer Center, Houston, TX, USA 1

Haematologica 2019 Volume 104(1):93-101

ABSTRACT

asatinib, a second-generation BCR-ABL1 tyrosine kinase inhibitor, is approved for the treatment of chronic myeloid leukemia and Philadelphia chromosome-positive acute lymphoblastic leukemia, both as first-line therapy and after imatinib intolerance or resistance. While generally well tolerated, dasatinib has been associated with a higher risk for pleural effusions. Frequency, risk factors, and outcomes associated with pleural effusion were assessed in two phase 3 trials (DASISION and 034/Dose-optimization) and a pooled population of 11 trials that evaluated patients with chronic myeloid leukemia and Philadelphia chromosome-positive acute lymphoblastic leukemia treated with dasatinib (including DASISION and 034/Doseoptimization). In this largest assessment of patients across the dasatinib clinical trial program (N=2712), pleural effusion developed in 6-9% of patients at risk annually in DASISION, and in 5-15% of patients at risk annually in 034/Dose-optimization. With a minimum follow up of 5 and 7 years, drug-related pleural effusion occurred in 28% of patients in DASISION and in 33% of patients in 034/Dose-optimization, respectively. A significant risk factor identified for developing pleural effusion by a multivariate analysis was age. We found that overall responses to dasatinib, progression-free survival, and overall survival were similar in patients who developed pleural effusion and in patients who did not. clinicaltrials.gov identifier 00481247; 00123474. Introduction Dasatinib is a potent second-generation BCR-ABL1 tyrosine kinase inhibitor (TKI) approved at 100 mg once daily (QD) as first-line therapy in patients with chronic myeloid leukemia in chronic phase (CML-CP), and in patients with CMLCP who are resistant to or intolerant of prior therapy.1 Dasatinib is also approved at 140 mg QD in patients with accelerated or blast phase CML (CML-AP/BP) or Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ ALL) with resistance to or intolerance of prior therapy.2-4 Although fluid retention has been associated with TKIs used to treat CML, pleural effusions, specifically with exudate characteristics, have been reported more commonly with dasatinib.1,5,6 The exact mechanisms behind treatment-related pleural effusion remain to be elucidated; however, it has been suggested that immune mechanisms may play a role, based on reports of association with lymphocytosis and the presence of lymphocyte-dominant exudates and chyle accumulate.7-9 Alternatively, pleural effusion (with or without exudates) may also occur through inhibition of platelet-derived growth factor receptor-β or SRC-family kinases.10,11 In haematologica | 2019; 104(1)

Correspondence: jcortes@mdanderson.org

Received: January 18, 2018. Accepted: August 3, 2018. Pre-published: August 9, 2018. doi:10.3324/haematol.2018.188987 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/93 ©2019 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.

T.P. Hughes et al. Table 1. Summary of pleural effusion events in dasatinib-treated patients with CML-CP in DASISION and 034/Dose-optimization.

Dasatinib-treated patients, n (%) DASISION 034/Dose-optimization (n=258) (n=662) Patients with PE (any grade), n (%) Patients with drug-related PE (any grade), n (%) Grade 1 Grade 2 Grade 3 Grade 4 Patients with >1 drug-related PE (any grade), n (%) Median time to first drug-related PE (any grade), weeks (range) Grade 1 Grade 2 Grade 3 Median duration of first drug-related PE (any grade), weeks (range) Grade 1 Grade 2 Grade 3 First occurrences of PE (any grade) by patients at risk, n/n at risk (%)* Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7

74 (29) 73 (28) 21 (8) 45 (17) 7 (3) 0 45 (17) 114 (4-299) 132 (4-299) 114 (6-281) 175 (114-274) 4 (0-223) 7 (1-223) 3 (1-210) 4 (0-25)

227 (34) 220 (33) 28 (4) 144 (22) 44 (7) 4 (1) 134 (20) 60 (1-371) 60 (2-369) 83 (1-371) 105 (3-350) 4 (0-235) 7 (0-80) 3 (0-235) 3 (0-96)

20/258 (8) 13/209 (6) 11/184 (6) 14/160 (9) 12/141 (9) NA NA

100/662 (15) 37/430 (9) 30/305 (10) 15/230 (7) 17/182 (9) 7/144 (5) 9/133 (7)

*Two patients in 034/Dose-optimization have not been categorized into any year due to missing PE onset date. CML-CP: chronic myeloid leukemia in chronic phase; NA: not applicable; PE: pleural effusion.

clinical practice, pleural effusion observed with dasatinib therapy remains a concern for both patients and prescribers. Risk factors for pleural effusion in dasatinib-treated patients have been described in previous reports, including advanced age, advanced disease, heart disease, preexisting hypertension, hypercholesterolemia, autoimmune disease, rash, drug dose and schedule, and the presence of lymphocytosis.10,12-14 A correlation with plasma trough level, drug exposure, treatment duration, and depth of response to treatment has been suggested, but not confirmed.15 Here, we present an analysis of the proportion of patients with pleural effusion by grade, and efficacy outcomes in these patients, in dasatinib clinical trials. Additionally, we report the results of multivariate analyses of dasatinib clinical trial data, exploring factors associated with the development of pleural effusion in dasatinibtreated patients.

Methods Patient populations DASISION (CA180-056). In the phase 3 dasatinib versus imatinib study in treatment-naive CML patients (DASISION [CA180056]; clinicaltrials.gov identifier 00481247), 519 patients with newly diagnosed CML-CP were randomized to receive either 100 mg QD dasatinib (n=259) or 400 mg QD imatinib (n=260).16-18 The pri94

mary endpoint was the proportion of patients with confirmed complete cytogenetic responses (CCyR) by 12 months. Patients included in this report had a minimum of 5 years of follow up.19 034/Dose-optimization (CA180-034). In the phase 3 dose-optimization study in imatinib-resistant or -intolerant CML-CP patients (CA180-034; clinicaltrials.gov identifier 00123474), 670 patients with CML-CP intolerant of or resistant to imatinib were randomized to receive dasatinib 100 mg QD (n=167), 140 mg QD (n=167), 50 mg twice daily (BID; n=168), or 70 mg BID (n=168).20-22 A subset of

patients modified their dose over the course of the study; however, analyses of data were performed according to each patientâ&#x20AC;&#x2122;s initial randomization. The primary endpoint was the percentage of patients with major cytogenetic response (MCyR) after a minimum follow up of 6 months. Patients included in this report had a minimum of 7 years of follow up.23 Pooled population of patients with Ph+ leukemia. Patients (N=2712) with Ph+ leukemia who were treated with first- or second-line dasatinib 15 mg to 240 mg QD in 1 of 11 phase 1, 2, or 3 trials were included in these analyses (Online Supplementary Table S1).16,19,21,23-31 DASISION and 034/Dose-optimization were analyzed separately and as part of the pooled population for this report. In total, 324 newly diagnosed patients treated with dasatinib 100 mg QD (DASISION [n=258], CA180-363 [n=66]), and 2388 patients with CML (CML-CP [n=1294], advanced phases of CML [n=958]) or Ph+ ALL (n=136) previously treated with imatinib were included. Previously treated patients received dasatinib at daily doses ranging from 15 mg to 240 mg administered once or in divided doses daily. haematologica | 2019; 104(1)

Pleural effusion in dasatinib-treated CML patients

Assessments Pleural effusions were monitored continuously in treated patients who received at least 1 dose of study drug (DASISION [n=258], 034/Dose-optimization [n=662], pooled population

[n=2712]). Pleural effusion by first onset is presented by year for patients at risk (the number of patients who were treated within a year and did not have pleural effusion). Effusions were graded according to the Common Terminology Criteria for Adverse

Figure 1. Retrospective multivariate analysis to determine if there was an association between pleural effusion and potential risk factors. Odds ratios (ORs) with 95% confidence intervals (CIs) are shown for potential predictive variables listed for patients with CML-CP treated with 100 mg QD dasatinib from DASISION (A), 034/Dose-optimization (B), or both DASISION and 034/Dose-optimization (C). AFRI: Africa; CA: Caribbean; CML-CP: chronic myeloid leukemia in chronic phase; EU: European Union; MMR: major molecular response, BCR-ABL1 transcripts â&#x2030;¤0.1% on the International Scale (IS) corresponding to a 3-log reduction from a standardized baseline; MR4: BCR-ABL1 transcripts <0.01% (IS) corresponding to a 4-log reduction from a standardized baseline; MR4.5: BCR-ABL1 transcripts â&#x2030;¤0.0032% (IS) corresponding to a 4.5-log reduction from a standardized baseline; NA: North America; PE: pleural effusion; QD: once daily.

haematologica | 2019; 104(1)

T.P. Hughes et al. Table 2. Hypertension by occurrence of drug-related pleural effusion in dasatinib-treated patients with CML-CP in DASISION and 034/Dose-optimization.

Dasatinib-treated patients, n (%) DASISION 034/Dose-optimization (n=258) (n=662) Patients with baseline hypertension With PE Without PE With anti-hypertensive medication With PE Without PE Without anti-hypertensive medication With PE Without PE Patients without baseline hypertension With PE Without PE

28 (11) 13 (46) 15 (54) 20 (71) 9 (45) 11 (55) 8 (29) 4 (50) 4 (50) 230 (89) 60 (26) 170 (74)

61 (9) 24 (39) 37 (61) 40 (66) 18 (45) 22 (55) 21 (34) 6 (29) 15 (71) 601 (91) 196 (33) 405 (67)

CML-CP: chronic myeloid leukemia in chronic phase; PE: pleural effusion.

Table 3. Age by occurrence of drug-related pleural effusion in dasatinib-treated patients with CML-CP in DASISION and 034/Dose-optimization.

Age, years

Mean (standard deviation) 95% CI (mean) Median (range) 95% CI (median)

DASISION (n=258) No PE (n=185) 43 (14) (41-45) 41 (18-84) (41-45)

034/Dose-optimization (n=662) With PE No PE (n=73) (n=442) 55 (13) (52-58) 56 (24-82) (52-58)

52 (15) (50-53) 53 (18-84) (50-53)

With PE (n=220) 58 (13) (57-60) 60 (22-84) (57-60)

CI: confidence interval; CML-CP: chronic myeloid leukemia in chronic phase; PE: pleural effusion.

Events Version 3.032 in DASISION and 034/Dose-optimization. Additional description of assessments and statistical analyses can be found in the Online Supplementary Material. Each study protocol was approved by all institutional review boards, ethics committees, and national competent authorities at participating sites.

Results Incidence of pleural effusion DASISION. With a minimum of 5 years of follow up, 73 (28%) patients in DASISION reported drug-related pleural effusion (Table 1). One case of pleural effusion was not attributed to dasatinib by the investigator (Grade 2) and occurred >30 days after the last study dose was given. Most drug-related pleural effusions were Grade 1 (8%) or Grade 2 (17%), and no Grade 4 events were reported. The median duration of all first cases of drug-related pleural effusion was 4 weeks. The median daily dose of dasatinib for patients who developed drug-related pleural effusion was 100 mg, similar to the overall dasatinib-treated population.19 New cases of pleural effusion (any grade) occurred in 8% of patients receiving dasatinib in the first year of the study, and 6-9% each remaining year up to 5 years, suggesting a steady but continuous risk over time (Table 1). The proportion of patients with a recurrent (>1) drug96

related pleural effusion (any grade) was 62% (n=45/73). Of these, 16 patients had 2 separate events, 6 had 3 separate events, 4 had 4 separate events, and 19 had â&#x2030;Ľ5 separate events. Twelve patients discontinued due to recurrent pleural effusion. Of these patients, 9 had â&#x2030;Ľ3 recurring events. Dose interruptions and dose reductions due to pleural effusion occurred in 62% and 41% of patients, respectively. 034/Dose-optimization. With a minimum follow up of 7 years, drug-related pleural effusion (any grade) occurred in 33% of patients (Table 1). Although the median dose of dasatinib for patients with drug-related pleural effusion (any grade) was 100 mg daily (range 0-180 mg), pleural effusion rates varied across dosing groups in 034/Doseoptimization, with 28% (n=46/165) of patients experiencing any-grade pleural effusion in the 100 mg QD group and 35% (n=174/497) in the other groups. The median duration of first pleural effusion was 4 weeks. The majority of drug-related pleural effusions were Grade 1/2, and Grade 3/4 drug-related pleural effusions were reported in 48 (7%) patients in the total treated population. Of these, 8 occurred in the 100 mg QD group and 40 in the remaining groups. First occurrences of pleural effusion (any grade) occurred in 15% of dasatinib-treated patients in the first year of the study, and in 5-10% per year up to 7 years. Recurrent (>1) drug-related pleural effusion (any grade) occurred in 61% (n=134/220). Of these, 54 patients had 2 separate events, 31 had 3 separate events, 16 had 4 sepahaematologica | 2019; 104(1)

Pleural effusion in dasatinib-treated CML patients rate events, and 33 had â&#x2030;Ľ5 separate events. In all 034/Dose-optimization dose groups, discontinuation due to recurrent pleural effusion occurred in 46 (7%) patients. The cumulative incidence rate of pleural effusion was lower in the 100 mg QD group than in dose groups and increased over time (at 2 years 15% vs. 24%, respectively; at 7 years 27% vs. 36%, respectively). Similarly, the cumulative incidence rate of Grade 3/4 pleural effusion was lower in the 100 mg QD group than in the other dose groups (at 7 years 5% vs. 9%, respectively), and increased over time (at 2 years 2% vs. 4%, respectively; at 5 years 4% vs. 7%, respectively). Within year 7 of the study, new cases of pleural effusion occurred in 5% (2/42) of patients at risk treated with dasatinib 100 mg QD and in 8% (7/91) of patients at risk in the other treatment arms. The cumulative rates of drug-related pleural effusion over time for the 100 mg QD arm were 14% at 2 years, 24% at 5 years, and 28% at 7 years. For the other treatment arms, the cumulative rates of drug-related pleural effusion over time were 24% at 2 years, 32% at 5 years, and 35% at 7 years. The incidence of pleural effusion was also lower in imatinib-intolerant and imatinib-resistant patients receiving 100 mg QD than in the other dose groups. Pleural effusion (any grade) was reported in 19% of patients who were imatinib-intolerant in the 100 mg QD arm and in 43% of imatinib-intolerant patients in the other dose groups, while pleural effusion (any grade) was reported in 31% of imatinib-resistant patients in the 100 mg QD arm and in 35% in the other dose groups. Pleural effusions were managed with dose interruptions in 44% of patients. Pooled population of patients with Ph+ leukemia. Eleven dasatinib clinical trials of patients with Ph+ leukemia, including the DASISION and 034/Dose-optimization trials, were pooled for this analysis to include the largest number of dasatinib-treated patients possible. Pleural effusion of any grade from any cause occurred in 946 patients (35%), 553 (34%) with CML-CP and 393 (36%) with CML-AP/BP or Ph+ ALL. Grade 3/4 pleural effusions were reported in 223 (8.2%) patients, 119 (7%) with CML-CP, and 104 (10%) with CML-AP/BP or Ph+ ALL. Deaths due to pleural effusion were reported in 5 (<1%) patients, all with CML-AP/BP or Ph+ ALL (4 were not receiving the currently approved dose of 140 mg QD dasatinib1). Drugrelated pleural effusion of any grade occurred in 538 (33%) patients with CML-CP and 345 (32%) with CML-AP/BP or Ph+ ALL (883 [33%] patients total). Grade 3/4 drugrelated pleural effusion episodes were reported in 114 (7%) patients with CML-CP and 93 (9%) patients with CML-AP/BP or Ph+ ALL (207 [8%] patients in total). One drug-related death was reported (<1%) in a patient with CML-AP/BP or Ph+ ALL.

Risk factors for pleural effusion Based on risk factors for pleural effusion in dasatinibtreated patients described in literature,9,11-13 retrospective multivariate analyses were performed to investigate the association between pleural effusion and baseline hypertension, age, and lymphocytosis in patients treated with dasatinib, as well as additional variables of interest including sex, region, dosing schedule (034/Dose-optimization only), baseline Euro (Hasford) risk scores (DASISION only), exposure to interferon alpha (034/Dose-optimization only), BCR-ABL1 levels, baseline parameters, major molecular response (MMR) at 12 months, line of therapy, duration of prior TKI therapy, and depth of MR at any haematologica | 2019; 104(1)

time. Average daily dose, prior skin rash, and prior autoimmune or lung disease were assessed in the DASISION/034/Dose-optimization pooled population only. DASISION. In DASISION, 28 (11%) patients had hypertension, and 13 of 28 (46%) patients with hypertension developed drug-related pleural effusion (Table 2). Of the 13 patients who developed pleural effusion, 9 were taking antihypertensive medications. When the relation between hypertension and pleural effusion was assessed in a multivariate analysis, there was no significant association found (Figure 1A). Pulmonary hypertension was reported in 14 (5%) dasatinib-treated patients, 9 of whom had pleural effusion. One patient with pulmonary hypertension underwent right heart catheterization in order to confirm pulmonary arterial hypertension (PAH); however, the procedure did not support a diagnosis of PAH. Twelve of the 14 pulmonary hypertension diagnoses were drug related.19 In DASISION, the correlation between pulmonary hypertension and pleural effusion was not confirmed. The median age of patients who developed drug-related pleural effusion in DASISION was 56 years, whereas for patients who did not develop pleural effusions it was 41 years (Table 3). Of significance, older age was found to be a risk factor for developing pleural effusion (odds ratio [OR] 1.067; 95% confidence interval [CI] 1.041-1.094; Figure 1A). These results were confirmed using a reduced model multivariate analysis. In DASISION, 29 (11%) patients had lymphocytosis at any time prior to pleural effusion, and 0 patients developed lymphocytosis after. However, lymphocytosis was not found to be a significant risk factor for the development of pleural effusion by multivariate analysis (Figure 1A). Of the patients receiving 100 mg QD dasatinib in the DASISION trial, 86 (33%) patients had low-risk Euro scores, 124 (48%) had intermediate-risk Euro scores, and 49 (19%) had high-risk Euro scores.16 Using a reduced model multivariate analysis, patients with intermediaterisk Euro scores were not found to be at an increased risk of developing pleural effusion compared with patients with low-risk Euro scores. Similarly, no association was observed between pleural effusion and patients with highrisk Euro scores compared with patients with low-risk Euro scores. The remaining variables investigated via multivariate analysis were found not to have an association with pleural effusion (Figure 1A). 034/Dose-optimization. Similar to DASISION, no significant association between pleural effusion and hypertension was found for patients in 034/Dose-optimization treated with dasatinib 100 mg QD, the currently approved dose for CML-CP1 (Figure 1B). Twenty-four of 61 (39%) patients from any treatment arm with hypertension developed drug-related pleural effusion (Table 2). Of these 24 patients, 18 were taking antihypertensive medication. Pulmonary hypertension (any grade) was reported in 3 (2%) patients in the 100 mg QD dose group and in 13 (3%) patients in the other dose groups.23 One patient (<1%) in the 100 mg QD dose group reported severe drugrelated PAH, confirmed with a right heart catheterization procedure.23 Similar to DASISION, the association between pulmonary hypertension and pleural effusion was not confirmed in these patients. In 034/Dose-optimization, the median age of all patients who developed drug-related pleural effusion was 97

T.P. Hughes et al. Table 4. Efficacy of dasatinib by occurrence of drug-related pleural effusion in dasatinib-treated patients with CML-CP in DASISION and 034/Dose-optimization.

Dasatinib-treated patients, n (%) DASISION* (N=258) 034/Dose-optimization† (N=662) No PE With PE 1 PE >1 PE Total No PE With PE 1 PE >1 PE (n=185) (n=73) (n=28) (n=45) (n=258) (n=442) (n=220) (n=86) (n=134) Best response to dasatinib‡ MMR MR4.5 CCyR Median MMR duration, months (range)

138 (75) 78 (42) 155 (84) NA

60 (82) 37 (51) 71 (97) NA

24 (86) NA

36 (80) NA

39 (0-68)

45 (0-70)

198 (77) 115 (45) 226 (88) 48 (0-70)

162 (37) 57 (13) 193 (44) NA

117 (53) 40 (18) 146 (66) NA

37 (43) NA

80 (60) NA

34 (0-85)

42 (0-84)

Total (n=662) 279 (42) 97 (15) 339 (51) 43 (0-90)

*Both on-study and follow-up patients are included. †Only on-study patients are included. ‡All responses listed were those obtained at end of study. For DASISION, that was 60 months, and for 034/Dose-optimization, that was 84 months. CCyR: complete cytogenetic response; CML-CP: chronic myeloid leukemia in chronic phase; MMR: major molecular response, BCR-ABL1 transcripts ≤0.1% on the International Scale (IS) corresponding to a 3-log reduction from the standardized baseline; MR4.5: BCR-ABL1 transcripts ≤0.0032% (IS) corresponding to a 4.5-log reduction from the standardized baseline; NA: not applicable; PE: pleural effusion.

60 years; patients who did not develop pleural effusion had a median age of 53 years (Table 3). Advanced age was found to be a significant risk factor for pleural effusion in patients treated with 100 mg QD dasatinib in a multivariate model (OR 1.074; 95% CI 1.035-1.114; Figure 1B). In addition to age, a strong statistical correlation was observed between the depth of MR and the overall incidence of pleural effusion, for those treated with 100 mg QD dasatinib (OR 3.851; 95% CI 1.182-12.552) (Figure 1B). Both age and depth of MR were confirmed as risk factors by a reduced model multivariate analysis. Lymphocytosis was not found to be associated with the development of pleural effusion in a multivariate analysis of patients treated with 100 mg QD dasatinib (Figure 1B); overall, 17 (10%) patients had lymphocytosis before the onset of pleural effusion versus 0 after. No other associations were found between pleural effusion and the remaining variables analyzed via multivariate analysis (Figure 1B). Pooled DASISION and 034/Dose-optimization. To assess the relation between potential prognostic factors and pleural effusion in a larger population, patients treated with 100 mg QD dasatinib from both DASISION (n=258) and 034/Dose-optimization (n=165) were pooled (n=423), and a multivariate analysis was performed (Figure 1C). In this pooled population, patients in MMR (OR 2.482; 95% CI 1.191-5.174) and MR4.5 (OR 2.756; 95% CI 1.30-5.841) had a significantly increased risk of developing pleural effusion compared with patients not in MMR. Increased age also was found to be a significant risk factor (OR 1.069; 95% CI 1.048-1.091). In the pooled DASISION and 034/Dose-optimization patient population, 40 (9%) patients had a history of lung disease, and 13 (3%) patients had a history of autoimmune disease. When the relationship between pleural effusion and prior lung disease, autoimmune disease, or skin rash was assessed using a reduced model multivariate analysis, no association was found. Average daily dasatinib dose also was not found to be a risk factor using the reduced model. In the reduced multivariate model, imatinib-intolerant 98

patients receiving second-line dasatinib had a significantly increased risk of developing pleural effusion compared with first-line patients (OR 0.232; 95% CI 0.086-0.623).

Efficacy of patients with pleural effusion DASISION. Of 73 dasatinib-treated patients in the DASISION trial with drug-related pleural effusion, 97% achieved CCyR, 82% had MMR, and 51% had MR4.5 (Table 4). These results are comparable to those in patients who did not have drug-related pleural effusion, and reflective of the overall responses for the entire study population.17 Duration of MMR was 39 months for those with 1 drug-related pleural effusion and 45 months for those with >1 (Table 4). Five-year progression-free survival (PFS) was similar for patients with or without drug-related pleural effusion (Online Supplementary Figure S1A). Of patients who experienced drug-related pleural effusion events, 5/73 (7%) progressed, whereas 21/185 (11%) patients who did not experience pleural effusion events progressed. Overall survival (OS) was similar in patients who did or did not experience drug-related pleural effusion as well (Online Supplementary Figure S1B). Many patients did achieve cytogenetic or molecular responses prior to the first occurrence of drug-related pleural effusion, and these responses were often maintained or improved despite dose modifications required to manage the effusion (Online Supplementary Table S2). Sixty-five patients (89% of patients with a drug-related pleural effusion) had CCyR, 35 (48%) had MMR, and 12 (16%) had MR4.5 prior to experiencing their first pleural effusion event. Following the first pleural effusion, 55 patients maintained/improved to CCyR. Similarly, 19 patients maintained/improved to MMR, and 32 patients maintained/improved to MR4.5. One patient went from MMR to BCR-ABL1 0.1-≤1% following their first pleural effusion event. No patient had BCR-ABL1 >10% or lost any cytogenetic response after their first event. 034/Dose-optimization. Of 220 dasatinib-treated patients in 034/Dose-optimization who had drug-related pleural effusion, 66% achieved CCyR, 53% had MMR, and 18% had MR4.5 (Table 4). As seen in the DASISION trial, molechaematologica | 2019; 104(1)

Pleural effusion in dasatinib-treated CML patients

ular responses in the 034/Dose-optimization trial for patients who had drug-related pleural effusion events versus patients who did not were similar, although a trend showed slightly higher molecular responses for those who experienced pleural effusions. The duration of MMR was 34 months for those with 1 drug-related pleural effusion, and 42 months for those with >1 (Table 4). At the end of the 034/Dose-optimization 7-year study, PFS for dasatinib-treated patients who experienced drug-related pleural effusion events was similar to PFS for patients who did not experience effusion events (Online Supplementary Figure S2A). Among patients who had drug-related pleural effusion, 93/220 (42%) progressed, compared with 220/442 (50%) patients who did not have pleural effusion but did progress. Seven-year OS in 034/Dose-optimization was similar across patients with and without drugrelated pleural effusion (Online Supplementary Figure S2B). As seen in DASISION, patients in 034/Dose-optimization were able to achieve responses prior to their first event, and these responses were maintained or improved following the pleural effusion in some patients (Online Supplementary Table S3). One hundred and sixteen patients (53% of patients with a drug-related pleural effusion evaluable for efficacy endpoints) had CCyR, 66 (30%) had MMR, and 15 (7%) had MR4.5 prior to experiencing a first case of pleural effusion. Following the effusion, 55 patients maintained or improved to CCyR and 9 patients lost CCyR. Changes in the depth of molecular response were also similar: 55 patients maintained/improved to MMR, and 31 patients maintained/improved to MR4.5. Three patients went from MMR to BCR-ABL1 0.1-â&#x2030;¤1% following their first pleural effusion event. Of these patients, all had BCR-ABL1 >0.1% after their first event. Two patients in 034/Dose-optimization were excluded from the efficacy analysis because the date of onset of pleural effusion was not captured.

Discussion Pleural effusion has been reported in dasatinib-treated patients at any time during the course of treatment, though the severity is generally mild to moderate. Grade 1 effusions are often asymptomatic and may not have been picked up in the absence of routine chest X-rays (only required in the DASISION trial), potentially reducing the incidence of clinically significant effusions in this patient population. However, we cannot comment that Grade 1 pleural effusions would not progress with time. The risk of pleural effusion remains even after long-term dasatinib treatment in effusion-naive patients; thus, maintaining awareness of the risk is important. As fluid retention events have been reported with most BCR-ABL1-targeted TKIs, it is tempting to attribute the occurrence of pleural effusion to a class effect on fluid overload. However, an immune-mediated mechanism is more likely for dasatinib-related pleural effusion, as exudate containing high lymphocyte counts (predominantly natural killer cells) and chyle accumulate have been reported in pleural fluids and tissue from patients on dasatinib.15,33,34 Pleural effusion developed slightly more often in patients with lymphocytosis than in patients without lymphocytosis (33% with lymphocytosis vs. 26% without lymphocytosis) in the DASISION trial, although this difference was not statistically significant.35 We found that haematologica | 2019; 104(1)

lymphocytosis occurred during therapy, which appears to represent a risk factor for pleural effusion because it preceded pleural effusion events. Through multivariate analyses, we determined that race, sex, region, exposure to interferon, BCR-ABL1 levels at 3 months, lymphocytosis, colitis, history of autoimmune disease, history of lung disease, history of skin rash, baseline smoking history, MMR at 12 months, average daily dose, line of therapy, baseline Euro risk scores, and duration of prior TKI therapy were not associated with an increased risk of pleural effusion. Other risk factors for pleural effusion previously described include advanced disease, heart disease, and hypercholesterolemia.10,12-14 It is difficult to analyze the association between the incidence of pleural effusion and the depth of molecular response achieved without correcting for the time of dasatinib exposure, given that most patients achieving deep molecular responses are typically on dasatinib longer and would therefore be expected to have a greater risk of developing pleural effusion. To address this, we evaluated MMR at 12 months as a potential risk factor; however, no association was observed. We found second-line patients with previous intolerance of imatinib to be at an increased risk of developing pleural effusion compared with first-line patients, although no association was observed for second-line imatinib-resistant patients. Finally, although effusions can develop in adults at any age, we determined that advanced age was the only significant patient risk factor for pleural effusion, particularly in those treated with 100 mg/day of dasatinib. Additional potential risk factors, such as a history of hypertension, can develop during treatment with dasatinib, and should be considered when evaluating individual patients, though the association between pleural effusion and hypertension was not substantiated in this analysis. Dasatinib dose, as a potential risk factor for pleural effusion, is of special interest. Wang et al. noted that pleural effusion was associated with trough drug concentrations, indicating that dasatinib pharmacokinetics may play a role in the development of pleural effusion.13 We did not find dasatinib dose to be a risk factor for the development of pleural effusion in the pooled population of patients initially treated with 100 mg QD dasatinib. However, since patients with advanced disease in the pooled population of dasatinib-treated patients with Ph+ leukemia described here were treated with higher doses of dasatinib (up to 240 mg daily) than patients with CML-CP, dasatinib dose may still be associated with the increased rate of Grade 3/4 pleural effusions observed. Also, we reported that pleural effusion (any grade) was observed in a lower percentage of patients in the 100 mg QD arm versus the other dose groups in 034/Dose-optimization. A retrospective study evaluating the toxicity-guided administration of a reduced-dose dasatinib regimen in similar imatinib-resistant/intolerant patients revealed that on/off treatment significantly reduced pleural effusions.36 Furthermore, recent sub-analyses of DASISION revealed that dose reductions for adverse events, including pleural effusion, did not affect dasatinib efficacy,37,38 suggesting that it may be possible to administer lower doses to populations at higher risk for pleural effusion development. Moreover, we found that the achievement of MMR and MR4.5 was found to be correlated with a higher risk of developing pleural effusion. This may be because patients without MMR may have discontinued dasatinib treatment earlier than 99

T.P. Hughes et al.

those with a response. This supports a hypothesis that pleural effusion may be a marker for longevity of treatment: As patients do well on dasatinib and remain on treatment longer, they may be more likely to develop a pleural effusion. Further investigation into the relation between duration of treatment and pleural effusion is warranted. Clinical data on molecular responses for patients who experienced pleural effusions during treatment with dasatinib are limited, though our observations indicate that patients who experienced pleural effusions while on dasatinib had similar responses to treatment as those who did not develop pleural effusion. A retrospective study examining dasatinib-related pleural effusion in CML patients across 21 hematologic centers in Italy revealed that at the time of the first effusion, 28.6% were in MMR and 37.8% were in MR4.5.39 In summary, pleural effusion is an adverse event seen disproportionately in patients treated with dasatinib;

References 1. Sprycel (dasatinib) [prescribing information]. Princeton, NJ: Bristol-Myers Squibb Company; 2017. 2. Kantarjian H, Cortes J, Kim DW, et al. Phase 3 study of dasatinib 140 mg once daily versus 70 mg twice daily in patients with chronic myeloid leukemia in accelerated phase resistant or intolerant to imatinib: 15-month median follow-up. Blood. 2009;113(25):6322-6329. 3. Lilly MB, Ottmann OG, Shah NP, et al. Dasatinib 140 mg once daily versus 70 mg twice daily in patients with Ph-positive acute lymphoblastic leukemia who failed imatinib: results from a phase 3 study. Am J Hematol. 2010;85(3):164-170. 4. Saglio G, Hochhaus A, Goh YT, et al. Dasatinib in imatinib-resistant or imatinibintolerant chronic myeloid leukemia in blast phase after 2 years of follow-up in a phase 3 study: efficacy and tolerability of 140 milligrams once daily and 70 milligrams twice daily. Cancer. 2010; 116(16):3852-3861. 5. Gleevec (imatinib) [prescribing information]. East Hanover, NJ: Novartis Pharmaceuticals Corporation; 2017. 6. Tasigna (nilotinib) [prescribing information]. East Hanover, NJ: Novartis Pharmaceuticals Corporation; 2017. 7. Porkka K, Khoury HJ, Paquette RL, Matloub Y, Sinha R, Cortes JE. Dasatinib 100 mg once daily minimizes the occurrence of pleural effusion in patients with chronic myeloid leukemia in chronic phase and efficacy is unaffected in patients who develop pleural effusion. Cancer. 2010; 116(2):377-386. 8. Mustjoki S, Ekblom M, Arstila TP, et al. Clonal expansion of T/NK-cells during tyrosine kinase inhibitor dasatinib therapy. Leukemia. 2009;23(8):1398-1405. 9. Agrawal V, Doelken P, Sahn SA. Pleural fluid analysis in chylous pleural effusion. Chest. 2008;133(6):1436-1441. 10. Quintas-Cardama A, Kantarjian H, O'Brien S, et al. Pleural effusion in patients with chronic myelogenous leukemia treated with dasatinib after imatinib failure. J Clin

100

however, the management of pleural effusion by dose reductions does not negatively affect the response rate to dasatinib. Advanced age and longevity of treatment were found to be predictive risk factors for the development of pleural effusion. Acknowledgments The authors would like to thank the patients and families for making these Bristol-Myers Squibb-sponsored trials possible. Funding This analysis was supported by funding from Bristol-Myers Squibb. The Bristol-Myers Squibb policy on data sharing may be found at: https://www.bms.com/researchers-andpartners/independent-research/data-sharing-requestprocess.html. Professional medical writing and editorial assistance was provided by Samantha L. Dwyer, PhD, and Jessica Franciosi, PhD, of StemScientific, an Ashfield Company, part of UDG Healthcare plc, funded by Bristol-Myers Squibb.

Oncol. 2007;25(25):3908-3914. 11. Buettner R, Mesa T, Vultur A, Lee F, Jove R. Inhibition of Src family kinases with dasatinib blocks migration and invasion of human melanoma cells. Mol Cancer Res. 2008;6(11):1766-1774. 12. Wang X, Roy A, Hochhaus A, Kantarjian HM, Chen T-T, Shah NP. Differential effects of dosing regimen on the safety and efficacy of dasatinib: retrospective exposure–response analysis of a Phase III study. Clin Pharmacol. 2013;5:85-97. 13. Conchon M, Freitas CM, Rego MA, Braga Junior JW. Dasatinib—clinical trials and management of adverse events in imatinib resistant/intolerant chronic myeloid leukemia. Rev Bras Hematol Hemoter. 2011;33(2):131-139. 14. de Lavallade H, Punnialingam S, Milojkovic D, et al. Pleural effusions in patients with chronic myeloid leukaemia treated with dasatinib may have an immune-mediated pathogenesis. Br J Haematol. 2008; 141(5):745-747. 15. Wang X, Roy A, Hochhaus A, Kantarjian HM, Chen TT, Shah NP. Differential effects of dosing regimen on the safety and efficacy of dasatinib: retrospective exposureresponse analysis of phase III study. Clin Pharmacol. 2013;10(5):85-97. 16. Kantarjian H, Shah NP, Hochhaus A, et al. Dasatinib versus imatinib in newly diagnosed chronic-phase chronic myeloid leukemia. N Engl J Med. 2010; 362(24):2260-2270. 17. Kantarjian HM, Shah NP, Cortes JE, et al. Dasatinib or imatinib in newly diagnosed chronic-phase chronic myeloid leukemia: 2-year follow-up from a randomized phase 3 trial (DASISION). Blood. 2012; 119(5):1123-1129. 18. Jabbour E, Kantarjian HM, Saglio G, et al. Early response with dasatinib or imatinib in chronic myeloid leukemia: 3-year followup from a randomized phase 3 trial (DASISION). Blood. 2014;123(4):494-500. 19. Cortes JE, Saglio G, Kantarjian HM, et al. Final 5-year study results of DASISION: the Dasatinib Versus Imatinib Study in Treatment-Naïve Chronic Myeloid Leukemia Patients Trial. J Clin Oncol.

2016;34(20):2333-2340. 20. Shah NP, Guilhot F, Cortes JE, et al. Longterm outcome with dasatinib after imatinib failure in chronic-phase chronic myeloid leukemia: follow-up of a phase 3 study. Blood. 2014;123(15):2317-2324. 21. Shah NP, Kantarjian HM, Kim DW, et al. Intermittent target inhibition with dasatinib 100 mg once daily preserves efficacy and improves tolerability in imatinib-resistant and -intolerant chronic-phase chronic myeloid leukemia. J Clin Oncol. 2008;26(19):3204-3212. 22. 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 with resistance, suboptimal response or intolerance to imatinib. Haematologica. 2010;95(2):232-240. 23. Shah NP, Rousselot P, Schiffer CA, et al. Dasatinib in imatinib-resistant or -intolerant chronic-phase, chronic myeloid leukemia patients: 7-year follow-up of study CA180-034. Am J Hematol. 2016; 91(9):869-874. 24. Saglio G, le Coutre P, Cortes J, et al. Safety and tolerability of dasatinib in patients with chronic myeloid leukemia (CML) and Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ ALL): pooled analysis of over 2400 patients. Haematologica. 2014;99(suppl); abstr P884. 25. Guilhot F, Apperley J, Kim DW, et al. Dasatinib induces significant hematologic and cytogenetic responses in patients with imatinib-resistant or -intolerant chronic myeloid leukemia in accelerated phase. Blood. 2007;109(10):4143-4150. 26. Cortes J, Rousselot P, Kim DW, et al. Dasatinib induces complete hematologic and cytogenetic responses in patients with imatinib-resistant or -intolerant chronic myeloid leukemia in blast crisis. Blood. 2007;109(8):3207-3213. 27. Hochhaus A, Kantarjian HM, Baccarani M, et al. Dasatinib induces notable hematologic and cytogenetic responses in chronicphase chronic myeloid leukemia after fail-

haematologica | 2019; 104(1)

Pleural effusion in dasatinib-treated CML patients

28.

29.

30.

31.

ure of imatinib therapy. Blood. 2007; 109(6):2303-2309. Kantarjian H, Pasquini R, Hamerschlak N, et al. Dasatinib or high-dose imatinib for chronic-phase chronic myeloid leukemia after failure of first-line imatinib: a randomized phase 2 trial. Blood. 2007;109(12): 5143-5150. Dasatinib in Imatinib Resistant/Intolerant Chinese CML (Chronic and Advanced Phase) Subjects. [clinicaltrials.gov identifier: NCT00529763]. https: //clinicaltrials. gov/ct2/show/NCT00529763?term=00529 763&rank=1. Accessed October 11, 2017. Trudel GC, Paliwal P, Lainas I. Dasatinib plus SMO antagonist versus dasatinib alone for treating patients (pts) with newly diagnosed Philadelphia chromosome-positive (Ph+) chronic myeloid leukemia in chronic phase (CML-CP): design of CA180-363, a phase II open-label randomized trial. J Clin Oncol. 2012;20(15 suppl):abstr TSP6634. Talpaz M, Shah NP, Kantarjian H, et al. Dasatinib in imatinib-resistant Philadelphia

haematologica | 2019; 104(1)

32.

33.

34.

35.

36.

chromosome-positive leukemias. N Engl J Med. 2006;354(24):2531-2541. National Cancer Institute. Common Terminology Criteria for Adverse Events (CTCAE) 3.0; 2006. https:// ctep.cancer.gov/protocolDevelopment/elec tronic_applications/docs/ctcaev3.pdf. Accessed May 11, 2016. Bergeron A, Rea D, Levy V, et al. Lung abnormalities after dasatinib treatment for chronic myeloid leukemia: a case series. Am J Respir Crit Care Med. 2007; 176(8):814-818. Cortes JE, Jimenez CA, Mauro M, et al. Pleural effusion in dasatinib-treated patients with chronic myeloid leukemia in chronic phase: identification and management. Clin Lymphoma Myeloma. 2017; 17(2):78-82. Schiffer CA, Cortes J, Saglio G, et al. The association of dasatinib-induced lymphocytosis with treatment outcome in patients with chronic myeloid leukemia (CML). Blood. 2013;122(21):2741. La RosĂŠe P, Martiat P, Leitner A, et al.

Improved tolerability by a modified intermittent treatment schedule of dasatinib for patients with chronic myeloid leukemia resistant or intolerant to imatinib. Ann Hematol. 2013;92(10):1345-1350. 37. Cortes J, Hochhaus A, Kantarjian H, et al. Impact of dose reductions on 5-year efficacy in newly diagnosed patients with chronic myeloid leukemia in chronic phase (CML-CP) from DASISION. Presented at the American Society of Clinical Oncology 2017 Annual Meeting; June 2-6, 2017; Chicago, IL. 38. Santos FP, Kantarjian H, Fava C, et al. Clinical impact of dose reductions and interruptions of second-generation tyrosine kinase inhibitors in patients with chronic myeloid leukaemia. Br J Haematol. 2010; 150(3):303-312. 39. Iurlo A, Galimberti S, Abruzzese E, et al. Pleural effusion and molecular response in dasatinib-treated chronic myeloid leukemia patients in a real-life Italian multicenter series. Ann Hematol. 2018;97(1):95-100.

101

ARTICLE Ferrata Storti Foundation

Acute Myeloid Leukemia

2-Bromopalmitate targets retinoic acid receptor alpha and overcomes all-trans retinoic acid resistance of acute promyelocytic leukemia

Ying Lu,1 Jin-Song Yan,2 Li Xia,1 Kang Qin,1 Qian-Qian Yin,3 Hong-Tao Xu,3 Meng-Qing Gao,2 Xiao-Ning Qu,2 Yu-Ting Sun2 and Guo-Qiang Chen1

Department of Pathophysiology, Key Laboratory of Cell Differentiation and Apoptosis of the Chinese Ministry of Education, Shanghai Jiao Tong University School of Medicine (SJTU-SM); 2Department of Hematology, Dalian Key Laboratory of Hematology, Liaoning Medical Center for Hematopoietic Stem Cell Transplantation, the Second Hospital of Dalian Medical University and 3Shanghai Institute for Advanced Immunochemical Studies, ShanghaiTech University, China 1

Haematologica 2019 Volume 104(1):102-112

YL, JSY and LX contributed equally to this work.

ABSTRACT

Correspondence: stove@shsmu.edu.cn or chengq@shsmu.edu.cn

Received: February 25, 2018. Accepted: July 30, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.191916 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/102 ©2019 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.

102

atty acid oxidation dependency of leukemia cells has been documented in recent studies. Pharmacologic inhibition of fatty acid oxidation, thereby, displays significant effects in suppressing leukemia. 2-Bromopalmitate, a palmitate analogue, was initially identified as an inhibitor of fatty acid oxidation, and recently recognized as an inhibitor of protein palmitoylation. However, the effects of 2-Bromopalmitate on leukemia and its cellular targets remain obscure. Herein, we discover in cultured cell lines, a transplantable mouse model, and primary blasts that 2-Bromopalmitate presents synergistic differentiation induction with alltrans retinoic acid in acute promyelocytic leukemia. Moreover, 2Bromopalmitate overcomes all-trans retinoic acid resistance in all-trans retinoic acid-resistant cells and leukemic mice. Mechanistically, 2Bromopalmitate covalently binds at cysteine 105 and cysteine 174 of retinoic acid receptor alpha (RARα) and stabilizes RARα protein in the presence of all-trans retinoic acid which is known to induce RARα degradation, leading to enhanced transcription of RARα-target genes. Mutation of both cysteines largely abrogates the synergistic effect of 2Bromopalmitate on all-trans retinoic acid-induced differentiation, demonstrating that 2-Bromopalmitate promotes all-trans retinoic acidinduced differentiation through binding RARα. All-trans retinoic acidbased regimens including arsenic trioxide or chemotherapy, as preferred therapy for acute promyelocytic leukemia, induce adverse events and irreversible resistance. We expect that combining all-trans retinoic acid with 2-Bromopalmitate would be a promising therapeutic strategy for acute promyelocytic leukemia, especially for overcoming all-trans retinoic acid resistance of relapsed acute promyelocytic leukemia patients.

Introduction It is increasingly recognized that fatty acid oxidation (FAO) plays an important role in supporting cell growth of many cancers including leukemia.1,2 Accordingly, inhibition of FAO by chemical compounds has yielded remarkable effects in suppressing cell growth, inducing apoptosis and relieving chemo-resistance, and thus holds therapeutic potential for leukemia.3-6 2-Bromopalmitate (2BP), a palmitate analogue, was initially identified as an inhibitor of FAO around 50 years ago.7,8 Mechanistically, 2BP inhibits carnitine palmitoyltransferase-1(CPT1) and suppresses the transfer of fatty acyl into mitochondria for oxidation. 7 Over the past decade, 2BP has often been referenced as being a general inhibitor of protein palmitoylation through covalent binding to protein acyl transferases (PAT).9,10 More recently, 2BP was demonstrated to modulate differentiation of neural stem cell and osteoblast haematologica | 2019; 104(1)

2-Bromopalmitate promotes APL differentiation

which involved protein palmitoylation and histone acetylation.11-13 Overall, the effects of 2BP on leukemia and its cellular targets remain obscure. Acute promyelocytic leukemia (APL) is a M3 subtype of acute myeloid leukemia(AML) genetically characterized by chromosome translocations involving retinoic acid receptor α (RARα) on chromosome 17 and promyelocytic leukemia (PML) on chromosome 15, which generates the oncogenic PML-RARα fusion protein.14-17 RARα is a ligand-dependent transcription factor that binds as heterodimers with retinoid X receptor (RXR) to their target response elements and plays pivotal roles in a series of physiological processes including cell growth, differentiation, survival, and death. The PML-RARα fusion protein, however, acts as a transcriptional repressor of RARα-target genes and results in maturation arrest of myeloid progenitors at the promyelocytic stage.15,18 All-trans retinoic acid (ATRA), a natural ligand for RAR, is the first Food and Drug Administration-approved drug for APL differentiation, which degrades PML-RARα complex through caspases-dependent or proteasome pathway. Combinations of ATRA, arsenic trioxide (ATO) or chemotherapy were developed and have dramatically improved the complete remission (CR) rate and survival time of APL patients.15,19-21 However, current ATRA-based regimens may cause adverse events including fatal retinoic acid syndrome, systemic infection or secondary leukemia.15,22,23 In addition, 5-10% of APL patients fail to respond to the therapy targeting PML-RARα or relapse after CR.24 More recently, resistance to ATO in APL was reported by several groups.24-26 Therefore, it is essential to further optimize ATRA-based therapy for better prognosis of de novo or relapsed APL patients. In the present study, 2BP was identified to present synergistic differentiation induction with ATRA in APL cells and murine model. Moreover, 2BP overcomes ATRA resistance in ATRA-resistant cells and leukemic mice. We expected that 2BP would be a promising candidate for APL therapy, especially for overcoming ATRA resistance of relapsed APL patients.

Methods Patients and cells Bone marrow samples were collected from 11 cases of newly diagnosed APL patients at the Department of Hematology of the Second Hospital of Dalian Medical University. Patients were diagnosed according to the French-American-British classification. Detailed information of patients is listed in Table 1. Informed consent was obtained from all patients in accordance with the Declaration of Helsinki, and all manipulations were approved by the Medical Science Ethic Committee of Dalian Medical University. Mononuclear cells were isolated by density gradient centrifugation using Lymphoprep, and cryopreserved. In addition, 3 potential donors for allogeneic bone marrow transplantation were used to purify normal healthy hematopoietic cells. Human CD34+ cells were enriched from bone marrow mononuclear cells using MiniMACS (Miltenyi Biotech, Bergisch Gladbach, Germany) following the manufacturer’s instructions.27 Confirmation of CD34+ cells’ phenotype and purity was assessed by flow cytometry analysis using CD34-PE-Cy7 (BD Biosciences, San Diego, CA). Purified CD34+ cells were grown in serum-free hematopoietic growth medium (HPGM; Lonza) supplemented with 10 ng/mL recombinant human interleukin-3 (rhIL-3), 10 ng/mL rhIL-6 and 50 ng/mL recombinant human stem cell factor (PeproTech). The primary APL cells, AML cell lines NB4, HL60, NB4-MR2, NB4-LR1, and NB4-LR2 were maintained in RPMI 1640 medium (Sigma-Aldrich, St Louis, MO), supplemented with 10% heat-inactivated fetal bovine serum (FBS; Gibco BRL) in a humidified incubator at 37 °C and 5% CO2/95% air (v/v).

Reagents and antibodies ATRA, arsenic trioxide, 2-Bromopalmitate(2BP), palmitate acid (PA), 16BP and 12BP, DNase-free RNase A and propidium iodide were obtained from Sigma. Rabbit polyclonal antibodies against RARα, RXRα, Vinculin and PML were obtained from Santa Cruz Biotechnology (Santa Cruz). Rabbit polyclonal antibodies against pyruvate kinase M2(PKM2) and β-actin were obtained from Cell Signaling Technology. Anti-PML-RARα fusion antibody was from Abcam.

Table 1. Patient data and response to 2BP and/or ATRA.

No. Sex Age(years) Chromosome

Blast WBC(x109/L)

1 2 3 4

M F M M

26 33 40 43

60.00% 58.50% 92.00% 69.50%

5 6

F M

43 53

8 9 10 11

M F M M

17 26 63 67

ND ND ND 46,XY,t(15;17) (q22;q12) ND 46,XY,-8,+22,t(15;17) (q22;q12) 46,XY,+10,i(11q), -13,-14,t(15;17)(q22;q12) 46,XY,t(15;17)(q22;q12) 46,XX ND 46,XY,t(15;17)(q22;q12)

CD11b positive cells% ATRA ATRA/2BP

Vehicle

2BP

37.88 11.21 37.55 1.3

2.10±0.72 0.40±0.31 2.80±0.93 1.10±0.17

2.78±0.35 1.80±0.74 3.01±0.89 1.42±0.43

31.30±2.12 34.17±0.77 36.56±2.33 38.28±2.90

59.11±2.44 61.65±0.93 60.40±0.69 65.02±4.13

<0.001 <0.01 <0.05 <0.001

89.70% 77.00%

3.16 17.42

3.04±0.11 0.25±0.19

12.8±0.11 5.11±0.38

33.13±1.25 29.51±0.22

57.83±3.02 59.66±1.51

<0.05 <0.001

87.50%

4.73

19.8±2.22

19.44±2.30

61.55±4.13

75.18±5.50

<0.01

89.37% 88.50% 92.00% 56.00%

1.05 2.2 5.29 1.6

14.52±2.44 1.40±0.50 8.80±1.33 12.80±1.23

16.22±1.90 1.07±0.92 10.51±1.55 11.85±1.87

41.85±1.55 29.75±1.73 46.25±3.01 45.87±0.30

48.20±2.13 31.22±1.51 45.11±0.73 40.33±1.33

>0.05 >0.05 >0.05 >0.05

Mononuclear cells from bone marrow of 11 APL patients were isolated by density gradient centrifugation using Lymphoprep and maintained in RPMI 1640 medium supplemented with 10% FBS. The cells were treated with 10 mM 2BP and/or 10-7M ATRA for 3 days and CD11b-positive cells were counted by flow cytometry. ND indicates not done. P: P value between ATRA and ATRA/2BP.

haematologica | 2019; 104(1)

103

Y. Lu et al.

Wright–Giemsa staining Wright-Giemsa staining kit was from BASO Diagnostic (Zhuhai, China). Briefly, the cytospin slides were prepared and solution A was added onto cells for 3-5 mins followed by solution B for 1 min. The slide was then washed under running water. Images were taken under inverted microscope.28 The images were quantified according to the shape of the nuclei (0= round, 1= curveted, 2= polylobulated). The score of each figure was calculated and normalized by counted cell number. The counting of leukemic cells was made on at least 100 cells from three independent experiments.

Establishment and analysis of transplantation leukemic mice Splenocytes isolated from leukemic PML-RARα or mutant

PML-RARα transgenic mice were injected into 6- to 8-week-old female FVB/N mice intravenously after sublethal irradiation.29 Two days after transplantation, the mice were treated with vehicle, 2BP, ATRA, ATO, or combination of these compounds. The peripheral blood (PB) and bone marrow (BM) cells were collected for morphological analysis. The spleen and liver were isolated for hematoxylin and eosin staining. Animal handling was approved by the committee for humane treatment of animals at Shanghai Jiao Tong University School of Medicine.

Statistical analysis Statistical analyses between the control and treatment groups were performed by standard two-tailed Student's t-test. All experiments were repeated at least three times. A value of P<0.05 was considered to be statistically significant.

Figure 1. 2BP enhances ATRA-induced cell differentiation in APL cell lines. (A) Chemical structure of 2-Bromopalmitate (2BP). (B-C) NB4 cells were treated with different concentrations of 2BP for 24 and 48 hours, and growth inhibition and cell viability % were evaluated by trypan-blue exclusion assay. *P<0.05 against vehicletreated group. (D) Effects of 2BP on CD34+ bone marrow mononuclear cells isolated from 3 healthy donors (#1, #2 and #3) are shown. (E-F) NB4 cells were incubated with 5 or 10 mM 2BP and/or 10-8M ATRA for the indicated hours and growth inhibition (E) and cell viability % (F) were evaluated by trypan-blue exclusion assay. *P<0.05 between the line-pointed group. (G-K) NB4 cells were incubated with 5 or 10mM 2BP and/or 10-8M ATRA for the indicated days and Wright’s staining morphology (G), NBT reduction(H), CD11b-(I), CD11c-(J) and CD15-positive(K) cells counted by flow cytometry are shown. Scale bars are 20 mm. The images of Wright’s staining (G) were further quantified according to the shape of nuclei (0= round, 1= curveted, 2= polylobulated). The score of each figure was calculated, normalized by cell number (counted from three independent experiments) and shown at the bottom. The ratio shows the total score/number of cells counted in each condition. *P<0.05 against ATRA-treated group. All values for percentage of NBT-positive cells represent means ± s.d. of triplicate samples in an independent experiment. All experiments were repeated at least three times with the same results.

104

haematologica | 2019; 104(1)

2-Bromopalmitate promotes APL differentiation

Results 2BP enhances differentiation induction of ATRA in APL cell lines and primary blasts from APL patients ATRA-sensitive NB4 cells were treated with different concentrations of 2BP (Figure 1A) and cell growth was determined. 2BP inhibited cell growth in a concentrationdependent manner (Figure 1B), with an increased percentage of cells at the G1 stage (Online Supplementary Figure S1). 2BP treatment induced a slight decrease of cell viability at 40 mM in NB4 cells (Figure 1C). The effects of 2BP on normal CD34+ hematopoietic cells purified from bone marrow samples of 3 healthy donors were also assessed. The viability of CD34+ hematopoietic cells in the 2BPtreated group was comparable with that in the vehicletreated group (Figure 1D). Next, we assessed the cellular effects of 2BP in combination with ATRA. As depicted in Figure 1E and F, the 2BP and ATRA combination synergistically induced growth arrest without inducing apparent apoptosis (Online Supplementary Figure S2). Intriguingly, 2BP alone at nontoxic concentrations 5 mM or 10 mM did not induce apparent differentiation of NB4 cells, but it significantly increased ATRA-induced granulocytic differentiation, as evidenced by mature granulocytic morphologic features (such as smaller cell size, reduced nucleus-cytoplasm ratio, condensed chromatin, curveted or polylobulated nuclei) (Figure 1G), increased NBT reduction (Figure 1H), and percentage of CD11b (Figure 1I), CD11c (Figure 1J) and CD15 (Figure 1K) cells. Quantitative analysis based

on the shape of nuclei demonstrated more mature granulocytic cells upon combination of 2BP with ATRA (bottom panel, Figure 1G). Notably, cotreatment of 2BP with ATRA also increased the percentage of CD11b and CD11c positive cells in another APL cell line, HL60 (Online Supplementary Figure S3). More importantly, the synergistic effect exhibited by the use of 2BP and ATRA combination in differentiation could also be seen in primary blasts from APL patients. The percentage of CD11b-positive cells treated with ATRA and 2BP was significantly increased in 7 out of 11 samples compared with the ATRA-treated group (Table 1). Notably, two APL samples with complex chromosome abnormalities (No. 6 and No. 7) which usually do not respond well to ATRA-based therapy displayed significant differentiation under treatment with ATRA and 2BP. Collectively, these data demonstrated that 2BP presents a synergistic differentiation-enhancing effect in APL cells when used in combination with ATRA. As mentioned above, combination of ATO with ATRA have dramatically improved the CR rate of APL patients, and is now used as a frontline treatment of APL.15,19,20 Therefore, we further evaluated the effect of the combination of ATO and 2BP on NB4 cells by flow cytometry. As shown in Online Supplementary Figure S4A, upon treatment of ATO/2BP, the expression of CD11b and CD11c was significantly increased on day 2 and day 3 compared with ATO-treated cells. Moreover, addition of 2BP enhanced the granulocytic differentiation of NB4 cells induced by ATRA/ATO (Online Supplementary Figure S4A). These data

Figure 2. 2BP enhances ATRA-induced APL cell differentiation in vivo. ATRA-sensitive leukemic (leu) mice were treated with vehicle (5% DMSO, 5% cremophor, 90% saline), 2BP (5 mg per kg body weight, intraperitoneally), ATRA (10 mg per kg body weight, intraperitoneally) or ATRA/2BP daily for five continuous days a week. Normal FVB/N mice were taken as negative controls. When the first vehicle-treated leukemic mice were moribund, all mice were killed and analyzed. (A) The survival (%) and lifetime (day) of leukemic mice in each group were recorded and Kaplanâ&#x20AC;&#x201C;Meier survival analysis was shown. The numbers of mice are indicated in parentheses, and *P<0.05 against ATRA-treated mice. (B) Cytologic analysis of peripheral blood (PB) and bone marrow (BM) cells derived from different agent-treated mice by Wrightâ&#x20AC;&#x2122;s staining. Scale bars are 20 mm. The images were quantified as described in Figure 1G and shown at the bottom. *P<0.05 against ATRA-treated mice. Leu represents leukemia. (C) The macroscopic appearance/the weight (mg/g bw) of spleen (top panels) are shown. Each column represents the mean with bar as s.d. of 3 mice in an independent experiment, and *P<0.05 between the line-pointed group. (D) The leukemic invasions in spleen and liver were analyzed by hematoxylin and eosin (H&E) staining.

haematologica | 2019; 104(1)

105

Y. Lu et al. A

Figure 3. 2BP overcomes ATRA resistance in vitro and in vivo. (A-B) ATRA-resistant NB4-MR2 cells were treated with 5 mM or 10 mM 2BP in combination with ATRA for 3 days. The Wright’s staining morphology (A) and percentages of CD11b (upper panel, B) and CD11c (middle panel, B) and CD15 (bottom panel, B) expression are shown. Scale bars are 20 mm. The images were quantified as described in Figure 1G and shown at the bottom. *P<0.05 against ATRA-treated group. (C-D) ATRAresistant transplantable leukemic mice (leu) were treated with vehicle, 2BP (5 mg per kg body weight, intraperitoneally), ATRA (10 mg per kg body weight, intraperitoneally) or ATRA/2BP daily for five continuous days a week. The survival (%) and lifetime (day) of leukemic mice in each group were recorded and Kaplan–Meier survival analysis is shown (C). *P<0.05 against ATRA-treated group. (D) The macroscopic appearance/the weight (mg/g bw) of spleen (top panels) are shown. Each column represents the mean with bar as s.d. of 3 mice in an independent experiment, and *P<0.05 between the line-pointed group. (E) Cytologic analysis PB and BM cells derived from different agent-treated mice by Wright’s staining. The images were quantified as described in Figure 1G and shown on the bottom. *P<0.05 against ATRA-treated group.

indicated that 2BP could enhance the differentiation effect of ATO on APL cells in vitro.

2BP enhances differentiation induction of ATRA in vivo Subsequently, the effect of 2BP was further evaluated on transplantable APL mice. We intravenously transplanted a high dose (4×105) of ATRA-sensitive leukemic blasts from transgenic mice expressing human PMLRARα into sublethally irradiated isogenic FVB/N recipients to generate ATRA-sensitive leukemic mice.30 Leukemic mice were treated intraperitoneally with 5 mg/kg body weight of 2BP with or without ATRA on day 2 after transplantation. Twenty-five days after transplantation, mice in the vehicle group started to be frail and sluggish, and rapidly died off in the following 3 days. The lifetime of mice in ATRA-treated group lasted as long as 35 days. In contrast, 2BP combined with ATRA extended lifetime and survival of leukemic mice to 45 days (Figure 2A). Consistently, more morphologically differentiated cells were observed in peripheral blood (PB) and bone marrow (BM) of leukemic mice that received ATRA and ATRA/2BP (Figure 2B). Further quantitative analysis based on the shape of nuclei showed more mature granulocytic cells upon combination of 2BP with ATRA (bottom panel, Figure 2B). In addition, enlarged spleens were found in vehicle-treated mice, which could be alleviated when treated with ATRA or ATRA/2BP (Figure 2C). Histological examination revealed that ATRA/2BP treatment inhibited the infiltration of leukemia cells into the spleen and liver (Figure 2D). These results suggested that the leukemic infiltration to spleen was reduced in presence of 2BP. Taken 106

together, these data demonstrated the potential of 2BP to synergistically induce the maturation of promyelocytes with ATRA in vivo. In addition, we evaluated the effect of 2BP on ATO in APL murine model.31 Consistent with the in vitro data (Online Supplementary Figure S4A), administration of ATO/2BP extended the survival of leukemic mice compared to that in the ATO-treated group (Online Supplementary Figure S4B). These data supported that 2BP enhanced the effect of ATO on APL in vivo. In addition, the lifespan of leukemic mice that received ATRA/ATO/2BP is comparable to that when treated with ATRA/ATO (Online Supplementary Figure S4B). The effect of 2BP on the combination of ATRA/ATO deserves further evaluation. To evaluate the in vivo acute toxicity and identify a clinically relevant dose of 2BP in mice, the maximum tolerated dose (MTD) was determined.32 Athymic nude mice were injected intraperitoneally with a single dose of 2BP at 50,100,200 mg/kg/dose. At 100 mg/kg, treated mice were alive at day 14, whereas at 200 mg/kg, the mice died on day 2, suggesting that the best-tolerated concentration of 2BP is 100 mg/kg.

2BP induces the differentiation of ATRA-resistant APL in vitro and in vivo Since ATRA resistance has been a major obstacle in clinical APL therapy, the ATRA/2BP combination was also designed to be examined to relieve ATRA resistance. NB4derived subclones, including NB4-MR2, NB4-LR1 and NB4-LR2 were tested. In accordance with previous reports, these subclones were refractory to ATRA-induced maturahaematologica | 2019; 104(1)

2-Bromopalmitate promotes APL differentiation

Figure 4. 2BP stabilizes RARα protein and enhances transcriptional activity of RARα. (A-D) NB4(A-B) and HL60(C-D) cells were incubated with 5 mM of 2BP for indicated hours. The mRNA (B and D) and protein (A and C) of RARα were detected by quantitative real-time PCR and western blots. β-actin was detected as the loading control. P value between the line-pointed group is shown. (E) NB4 and HL60 cells were incubated with 5 mM of 2BP for indicated hours and protein of RXRα was detected by western blot with β-actin as the loading control. (F-G) ATRA-sensitive NB4 and HL60 (F) or ATRA-resistant NB4-MR2, NB4-LR1, and NB4-LR2 cells (G) were treated with 5 mM 2BP in the presence or absence of ATRA for 48 hours and RARα protein was detected by western blot. The protein bands on the gels were quantified by densitometry from three independent experiments and shown in the bottom panel (A,C,E,F and G). Scanning was performed at optimal exposure time where band intensity was proportional to the concentration of protein present. Gel photographic images were stored as GRAYSCALE pictures in TIF format and were processed using ImageJ Software. P value between the line-pointed group is shown. (H) NB4 cells were treated with 5 mM 2BP and/or 10-8M ATRA for 48 hours, and the expression of indicated genes was detected by quantitative real-time PCR with specific primers. ** and *** indicated P value between 2BP plus ATRA and ATRA was < 0.01 and < 0.001, respectively. Each experiment was repeated three times.

tion (Figure 3A and Online Supplementary Figure S5). In contrast, ATRA combined with 2BP could induce granulocytic differentiation of NB4-MR2 (Figure 3A-B), NB4-LR1 and NB4-LR2 (Online Supplementary Figure S5) cells as evidenced by mature granulocytic morphologic features and increased CD11b and CD11c expression, indicating that 2BP could overcome ATRA resistance of promyelocytes. In vivo, ATRA-resistant leukemic blasts from transgenic mice expressing human MRP8-PML-RARα mutant were intravenously transplanted into sublethally irradiated isogenic FVB/N recipients to generate ATRA-resistant leukemic mice.33 Twenty-six days later, all vehicle-treated leukemic mice died abruptly (Figure 3C). ATRA-treated leukemic mice died in 27 days, verifying that the ATRA resistance was lethal for leukemic mice. In contrast, administration of 2BP along with ATRA overcame the resistance and extended the survival of leukemic mice to 36 days (Figure 3C). It was discovered that 2BP with ATRA facilitated the remission of swollen spleen on ATRA-resistant leukemic mice (Figure 3D). Consistent with this, more morphologically differentiated cells were observed in PB and BM in the ATRA/2BP-treated group (Figure 3E). These data indicated that 2BP presents a potential to overcome ATRA resistance. haematologica | 2019; 104(1)

2BP accumulates RARα protein and enhances ATRA-dependent transcriptional activity of RARα As previously documented, administration of ATRA induced a progressive degradation of wild-type RARα as well as the PML-RARα chimeric protein.14,15 To elucidate the mechanism of the positive effect of 2BP on APL cell differentiation, RARα and PML-RARα fusion protein in NB4 cells under 2BP treatment were detected. Treatment of 2BP for 48 hours significantly increased RARα protein level (Figure 4A), but not its mRNA level in NB4 (Figure 4B) and HL60 cells (Figure 4C-D). In contrast, RXRα, the binding partner of RARα protein in transactivating target genes as heterodimers, was not affected by 2BP (Figure 4E). Furthermore, we detected the effects of 2BP on RARα protein in the presence of ATRA, which was documented to induce degradation of RARα.34 When NB4 and HL60 cells were cotreated with ATRA and 2BP, the protein level of RARα was increased compared to ATRA-treated group (Figure 4F). Moreover, accumulated RARα protein induced by 2BP in the absence (MR2 and LR1) or presence of ATRA (MR2 and LR2) could also be seen in ATRA resistance NB4-MR2, NB4-LR1 or NB4-LR2 cells (Figure 4G). In addition, using immunofluorescent staining with antiPML antibody, we observed that ATRA restored the PML107

Y. Lu et al. A

F E

Figure 5. 2BP directly binds with RARα. (A-B) Cellular thermal shift assay (CETSA) was performed on NB4 cells as described in the Methods section. The effects of 2BP on RARα, vinculin and PKM2 at different temperatures (A) and different doses (B) were evaluated by Western blot analysis. (C) SPR analysis of the binding between 2BP and RARα. The recombinant RARα protein was immobilized on an activated CM5 chip. 2BP was then flowed across the chip at increasing concentrations. (D) The recombinant RARα protein was incubated with biotin-2BP for the times indicated, and the mixtures were blotted with anti-biotin antibody. The membrane stained with Ponceau S is shown on the bottom. (E-F) NB4 cells were treated with PA, 2BP, 16BP or 12BP (chemical structure shown on E) for 3 days and CD11b-positive cells were counted by flow cytometry and RARα expression was detected by western blot are shown. ***P<0.001 against vehicle-treated group. All experiments were repeated three times with the same results.

RARα–disrupted PML nuclear body, which was decreased by the addition of 2BP (Online Supplementary Figure S6A). Western blot analysis using an antibody against PMLRARα fusion protein showed that 2BP could increase the amount of PML-RARα in NB4 cells (Online Supplementary Figure S6B), suggesting that 2BP may also stabilize PMLRARα fusion protein. Next, we asked whether the transcriptional activity of RARα protein was also enhanced by the ATRA/2BP combination following the accumulation of protein level. Consistent with previous reports, ATRA, but not 2BP, increased the expression of RARα target genes including RARβ, CCAAT/enhancer-binding protein α(C/EBPα), retinoic acid inducible gene-E (RIG-E), RIG-I, RIG-G, interferon regulatory factor 1 (IRF-1), transglutaminase 2(TGM2), and ubiquitin-like modifier activating enzyme 7(UBE1L) gene and decreased that of myeloperoxidase (MPO) in NB4 cells (Figure 4H).35-40 More intriguingly, the modulation of these expressions by ATRA was significantly enhanced by the cotreatment of 2BP with ATRA in NB4 cells (Figure 4H). Taken together, these data indicated that 2BP accumulates RARα protein and enhances ATRAdependent transcriptional activity of RARα. 108

To further confirm the role of the stabilized RARα protein in the synergistic differentiation-inducing effect of 2BP in combination with ATRA, AM580, a selective RARα agonist,41 was applied to NB4 cells. The results showed that, similar to that seen under ATRA treatment, 2BP could enhance AM580-induced differentiation of NB4 cells (Online Supplementary Figure S7A-B). Moreover, administration of AM580 induced a loss of RARα proteins, which could also be accumulated by addition of 2BP (Online Supplementary Figure S7C). On the other hand, Ro 41-5253, a highly specific RARα antagonist,42 could effectively antagonize the synergistic differentiation-inducing effect of the 2BP and ATRA combination in NB4 cells (Online Supplementary Figure S7A-B). These data indicated that maintenance of RARα protein contributed to the enhancing effect of 2BP on ATRA-induced differentiation.

2BP directly targets RARα 2BP has been reported to be active in covalently binding target proteins because it is from the α-halo-carbonyl group.10 To further elucidate the mechanism of 2BP on ATRA-induced APL differentiation, we detected the possibility of RARα protein as a cellular target of 2BP. To this haematologica | 2019; 104(1)

2-Bromopalmitate promotes APL differentiation

Figure 6. Identificaiton of 2BP binding sites within RARα. (A) Structure schematic diagram of RARα protein. AF: activation function domain; DBD: DNA binding domain; LBD: ligand binding domain. (B-C) MS/MS analysis of the Cys105-(B) or Cys174- containing(C) tryptic peptide for recombinant RARα incubated without (top) and with (bottom) 2BP for 30 mins. (D) Recombinant wild-type (WT) RARα and the Cys105/174 double mutants(DM) were incubated with biotin-2BP for 30 mins, followed by blotting with anti-RARα antibody. The membrane stained with Ponceau S is shown at the bottom. (E) NB4 cells were infected with shRNA specifically against RARα to deplete endogeneous RARα and then infected with plasmids encompassing WT or DM RARα. These cells were treated with 5 mM 2BP and/or 10-8M ATRA for 3 days and CD11b-positive cells were counted by flow cytometry. *P<0.05 against ATRA-treated group. (F) NB4 cells expressing DM RARα were incubated with 5 mM of 2BP and/or 10-8M ATRA for 48 hours and RARα protein was detected by western blot with β-actin as the loading control.

haematologica | 2019; 104(1)

109

Y. Lu et al.

end, cellular thermal shift assay (CETSA), a method used to evaluate the binding of compounds to target proteins in cells and tissue samples based on the biophysical principle of ligand-induced thermal stabilization of target proteins,43,44 was applied in NB4 cells. We observed that 2BP treatment markedly increased the thermal stability of RARα protein at temperatures examined compared to vehicle treatment (Figure 5A). Furthermore, RARα protein was accumulated by 2BP in a concentration-dependent manner (Figure 5B). Vinculin, a cytoskeletal protein associated with cell-cell and cell-matrix junctions45 and pyruvate kinase M2(PKM2), a rate-limiting enzyme in glycolysis46,47 were taken as negative controls (Figure 5A-B). These data suggested that 2BP interacts with RARα in APL cells. The binding between 2BP and RARα protein was further evaluated by surface plasmon resonance (SPR) assay using a biacore platform. The sensorgrams showed that 2BP rapidly associated with immobilized recombinant RARα protein at an equilibrium dissociation constant of 28.97 nM (Figure 5C). Moreover, the response signal during the dissociation phase did not return to the baseline level for 2BP, indicating that 2BP could not be completely eluted from RARα (Figure 5C). These data suggested that 2BP is covalently bound to RARα protein. Experiments with biotin-tagged 2BP (hereafter named biotin-2BP10) further supported that biotin-2BP could covalently bind with recombinant RARα protein, and this binding displayed a time-dependent saturation (Figure 5D). Mechanistically, covalent binding of 2BP with targets is more possible between the α-halo-carbonyl group and cysteines within proteins.10 Thereafter, we compared the effect of 2BP analogs, including palmitate acid (PA), 16BP and 12BP with 2BP on ATRA-induced differentiation of APL cells. These analogs either lack a bromine atom or have a bromine atom in a different position (Figure 5E). The results showed that, unlike 2BP, the three compounds did not present a synergistic effect with ATRA as evidenced by the percentage of CD11b-positive cells (upper panel, Figure 5F). In addition, the three compounds did not accumulate RARα, as 2BP did (bottom panel, Figure 5F). These data indicated that the α-halo-carbonyl group is essential for the binding of 2BP with RARα.

Cys105 and Cys174 of RARα is the binding site for 2BP To further determine the specific cysteine (Cys) residue that is modified by 2BP in RARα protein, we purified and incubated recombinant RARα protein with 2BP, followed by mass spectrometry (MS) analysis. There are 18 cysteines located in different domains within RARα protein (Figure 6A). We identified 90% of the RARα protein sequence and 15 cysteine-containing peptides of the recombinant RARα incubated with and without 2BP (data not shown). The m/z ratio of the Cys105-containing peptide SSGYHYGVSACEGCK (Figure 6B) and Cys174-containing peptide KKEVPKPECSESY (Figure 6C) was measured as 1,660.66 and 1,579.75 in the absence of 2BP and 1,857.86 and 1,776.96 in the presence of 2BP. The calculated mass shift was consistent with the addition of one molecule of 2BP. As for Cys105, MS/MS analysis of both unmodified and modified Cys105-containing peptide gave a partial series of y-ion fragments corresponding to the predicted sequence. Both MS/MS spectra had the same mass from y1 to y4, whereas the mass shifted 254.22 Da for the Cys105-containing fragment (from y5 to y12) in 110

the modified peptide spectra (Figure 6B). To verify the binding of 2BP with Cys105 and Cys174, we incubated synthesized biotin-2BP with purified wildtype (WT) or Cys105/Cys174 double-mutated (DM) RARα proteins. Mutation of Cys105/Cys174 remarkably diminished the binding of RARα with 2BP in vitro (Figure 6D), indicating that 2BP covalently modified Cys105 and Cys174 of RARα. Finally, we knocked down the endogenous RARα and re-expressed flag-tagged Cys105/Cys174 DM RARα into NB4 cells and evaluated the effects of 2BP. The results showed that 2BP enhanced the differentiation effect of ATRA in RARα-WT expressing but not RARα-DM expressing NB4 cells, indicating that 2BP presents a synergistic effect with ATRA through binding with Cys105/Cys174 (Figure 6E). Consistent with this, DM RARα could not be efficiently accumulated by 2BP in the presence of ATRA in NB4 cells (Figure 6F). Collectively, these data indicated that 2BP enhanced the ATRA-induced differentiation through binding to Cys105/Cys174 within RARα protein.

Discussion For the past decades, much effort has been devoted to identifying novel compounds with specific targets that would maximize the therapeutic effects of ATRA.28,29,48,49 For example, our group reported that pharicin B, a novel natural ent-kaurene diterpenoid derived from Isodon pharicus leaves, stabilizes RARα protein and presents synergistic differentiation induction with ATRA in AML cells. It can also overcome retinoid resistance in two ATRAresistant NB4 subclones.48 Wang et al. identified a novel synthetic small compound, named LG-362B, targeting PML-RARα and blocking ATRA resistance on cellular differentiation and transplantable murine models.28 More recently, Li et al. reported that pseudokinase Tribble 3 (TRIB3) promotes PML-RARα-driven APL by interacting with PML-RARα and disturbing the TRIB3/PML-RARα interaction through an α-helix peptide Pep2-S160 produced significant anti-APL effects.49 All these studies sought to elucidate APL pathogenesis and find more therapeutic options for APL patients. In the present work, we have demonstrated that 2BP presents synergistic differentiation induction with ATRA in APL cell lines, primary APL blasts and in an APL murine model. Moreover, 2BP overcomes ATRA resistance both in vitro and in vivo, demonstrating therapeutic potential in APL. The cellular target of 2BP here, differently to previously reported CPT1 in FAO and PAT in protein palmitoylation, is RARα protein which triggers differentiation of leukemia cells through transcriptional mechanism. The binding of 2BP with RARα prevented the degradation of RARα protein and sustained its transcriptional activity, leading to the differentiation-enhancing effect of 2BP. Recently, multiple investigations have indicated that RAR is a potential drug target for cancer and metabolic diseases.50 Thus, our data provides a new candidate to probe potential pathophysiological and therapeutic roles of RARα. Notably, 2BP also stabilized PML-RARα, which was well-documented to block hematopoietic differentiation through interfering transcriptional activity of RARα, and the therapeutic effects of both ATRA and ATO relied on the degradation of this fusion protein.15,18,51 However, haematologica | 2019; 104(1)

2-Bromopalmitate promotes APL differentiation

forced expression of PML-RARα could increase ATRA sensitivity in U937 cells52 and restore ATRA sensitivity in NB4.007/6 ATRA-resistant cells,53 demonstrating a dual role for the fusion protein in leukemogenesis. On the other hand, a number of compounds capable of restraining ATRA-dependent PML-RARα proteolysis have been shown to enhance ATRA-induced differentiation,54-57 indicating that a mechanism independent of PML-RARα degradation that drives granulocytic maturation does exist. Therefore, how stabilized PML-RARα protein contributes to 2BP-enhanced cell differentiation deserves further exploration. Combination of ATO with ATRA serves as a frontline treatment of APL. ATO is also used as the best salvage therapy agent for ATRA-resistant APL patients.15,19,20,31 Mechanistically, ATO induced PML-RARα degradation through direct binding to cysteine residues in PML moiety of the fusion protein.51 However, resistance to ATO in APL was reported by several groups.24-26 Here, we observed an increase of APL cell differentiation and leukemia mice survival upon cotreatment of ATO with 2BP, suggesting that an approach combining ATO with 2BP warrants further investigation as a therapeutic strategy for APL patients. The effects of 2BP on cell differentiation have been observed in neural stem cell and osteoblast.11-13 However, no direct differentiation-associated target of 2BP has been identified. Interestingly, different from our results, research from Chen et al. showed that 2BP treatment impaired ATRA-induced neuronal differentiation in vitro which involved the palmitoylation of P300 and acetylation of histones H3 and H4.13 Further investigation of 2BPRARα interaction in the context of neural cells may offer some clues for better understanding the controversial effect of 2BP towards ATRA. Most of the 2BP-targeted enzymes, whether in lipid or nonlipid processing, contain cysteine residues in or near the enzyme active site, suggesting α-halo-carbonyl electrophilic alkylation mediates the observed irreversible inhibition.10 In the present work, we identified that of all the 18 cysteines within RARα sequence, Cys105 and

References 1. German NJ, Yoon H, Yusuf RZ, et al. PHD3 Loss in cancer enables metabolic reliance on fatty acid oxidation via deactivation of ACC2. Mol Cell. 2016;63(6):1006-1020. 2. Carracedo A, Cantley LC, Pandolfi PP. Cancer metabolism: fatty acid oxidation in the limelight. Nat Rev Cancer. 2013; 13(4):227-232. 3. Samudio I, Harmancey R, Fiegl M, et al. Pharmacologic inhibition of fatty acid oxidation sensitizes human leukemia cells to apoptosis induction. J Clin Invest. 2010; 120(1):142-156. 4. Farge T, Saland E, de Toni F, et al. Chemotherapy-resistant human acute myeloid leukemia cells are not enriched for leukemic stem cells but require oxidative metabolism. Cancer Discov. 2017;7(7):716735. 5. Tabe Y, Yamamoto S, Saitoh K, et al. Bone marrow adipocytes facilitate fatty acid oxidation activating AMPK and a transcriptional network supporting survival of acute

haematologica | 2019; 104(1)

9. 10.

11.

Cys174 are the major residues for 2BP binding. Substitution of WT RARα with Cys105/Cys174 DM significantly decreased the synergistic differentiation activity of 2BP as well as the accumulation of RARα protein in the presence of ATRA, indicating that despite the potential promiscuous cellular targets of 2BP,10 the binding of 2BP with RARα at Cys105/Cys174 is responsible for, albeit partially, preventing ATRA-triggered degradation of RARα which helps differentiation. Previous studies have showed that ATRA-triggered degradation of RARα was mediated by the ubiquitin-proteasome pathway.34 Our data thus suggested that the binding of 2BP may influence the proteasomal degradation of RARα protein. Overall, RARα modulation in the treatment of APL has generated considerable interest in the development of RAR modulators and uncovered a promising strategy for these diseases. Our discoveries demonstrate that by targeting RARα, 2BP not only helps ATRA-induced APL differentiation, but also reverts ATRA resistance in vitro and in vivo. Combining ATRA with 2BP could be a potential candidate for ameliorating ATRA-associated adverse effects and clinical relapsed APL. Acknowledgments The authors would like to thank Dr. Ai-Wu Zhou for his kindness in providing us the psumo3 vector. We appreciate Dr. Weiwei Wang for his professional suggestions on identification of the binding site of 2BP. Funding This work was supported by National Natural Science Foundation (81370652, 81770146, 81721004, 81430061,81570124,31570824), National Basic Research Program of China (2015CB910403), Foundation for the author of National Excellent Doctoral Dissertation of China (201074), Grants from Science and Technology Committee of Shanghai (13431900501) and the Reformation Project in the Key Clinical Departments of Provincial Hospitals on Construction of Diagnosis and Treatment Capacity in Liaoning Province (LNCCC-A02-2015).

monocytic leukemia cells. Cancer Res. 2017;77(6):1453-1464. Spaner DE, Lee E, Shi Y, et al. PPAR-alpha is a therapeutic target for chronic lymphocytic leukemia. Leukemia. 2013; 27(5):1090-1099. Chase JF, Tubbs PK. Specific inhibition of mitochondrial fatty acid oxidation by 2bromopalmitate and its coenzyme A and carnitine esters. Biochem J. 1972;129(1):5565. Burges RA, Butt WD, Baggaley A. Some effects of alpha-bromopalmitate, an inhibitor of fatty acid oxidation, on carbohydrate metabolism in the rat. Biochem J. 1968;109(3):38P-39P. Martin BR, Cravatt BF. Large-scale profiling of protein palmitoylation in mammalian cells. Nat Methods. 2009;6(2):135-138. Davda D, El Azzouny MA, Tom CT, et al. Profiling targets of the irreversible palmitoylation inhibitor 2-bromopalmitate. ACS Chem Biol. 2013;8(9):1912-1917. Leong WF, Zhou T, Lim GL, Li B. Protein palmitoylation regulates osteoblast differ-

12.

13.

14.

15. 16.

17.

entiation through BMP-induced osterix expression. PLoS One. 2009;4(1):e4135. Chen X, Du Z, Li X, et al. Protein palmitoylation regulates neural stem cell differentiation by modulation of EID1 activity. Mol Neurobiol. 2016;53(8):5722-5736. Chen X, Du Z, Shi W, et al. 2Bromopalmitate modulates neuronal differentiation through the regulation of histone acetylation. Stem Cell Res. 2014; 12(2):481491. de The H, Pandolfi PP, Chen Z. Acute promyelocytic leukemia: a paradigm for oncoprotein-targeted cure. Cancer Cell. 2017;32(5):552-560. Wang ZY, Chen Z. Acute promyelocytic leukemia: from highly fatal to highly curable. Blood. 2008;111(5):2505-2515. de The H, Chen Z. Acute promyelocytic leukaemia: novel insights into the mechanisms of cure. Nat Rev Cancer. 2010; 10(11):775-783. de The H, Chomienne C, Lanotte M, Degos L, Dejean A. The t(15;17) translocation of acute promyelocytic leukaemia

111

Y. Lu et al.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

112

fuses the retinoic acid receptor alpha gene to a novel transcribed locus. Nature. 1990; 347(6293):558-561. Dos Santos GA, Kats L, Pandolfi PP. Synergy against PML-RARa: targeting transcription, proteolysis, differentiation, and self-renewal in acute promyelocytic leukemia. J Exp Med. 2013;210(13):27932802. Chen GQ, Shi XG, Tang W, et al. Use of arsenic trioxide (As2O3) in the treatment of acute promyelocytic leukemia (APL): I. As2O3 exerts dose-dependent dual effects on APL cells. Blood. 1997;89(9):3345-3353. Shen ZX, Chen GQ, Ni JH, et al. Use of arsenic trioxide (As2O3) in the treatment of acute promyelocytic leukemia (APL): II. Clinical efficacy and pharmacokinetics in relapsed patients. Blood. 1997;89(9):33543360. Sanz MA, Fenaux P, Lo Coco F. Arsenic trioxide in the treatment of acute promyelocytic leukemia. A review of current evidence. Haematologica. 2005;90(9):12311235. Girmenia C, Lo Coco F, Breccia M, et al. Infectious complications in patients with acute promyelocytic leukaemia treated with the AIDA regimen. Leukemia. 2003;17(5):925-930. Latagliata R, Petti MC, Fenu S, et al. Therapy-related myelodysplastic syndrome-acute myelogenous leukemia in patients treated for acute promyelocytic leukemia: an emerging problem. Blood. 2002;99(3):822-824. Zhu HH, Qin YZ, Huang XJ. Resistance to arsenic therapy in acute promyelocytic leukemia. N Engl J Med. 2014; 370(19):1864-1866. Lehmann-Che J, Bally C, de The H. Resistance to therapy in acute promyelocytic leukemia. N Engl J Med. 2014; 371(12):1170-1172. Chendamarai E, Ganesan S, Alex AA, et al. Comparison of newly diagnosed and relapsed patients with acute promyelocytic leukemia treated with arsenic trioxide: insight into mechanisms of resistance. PLoS One. 2015;10(3):e0121912. Lyu Y, Lou J, Yang Y, et al. Dysfunction of the WT1-MEG3 signaling promotes AML leukemogenesis via p53-dependent and independent pathways. Leukemia. 2017;31(12):2543-2551. Wang X, Lin Q, Lv F, et al. LG-362B targets PML-RARalpha and blocks ATRA resistance of acute promyelocytic leukemia. Leukemia. 2016;30(7):1465-1474. Liu CX, Yin QQ, Zhou HC, et al. Adenanthin targets peroxiredoxin I and II to induce differentiation of leukemic cells. Nat Chem Biol. 2012;8(5):486-493. Brown D, Kogan S, Lagasse E, et al. A PMLRARalpha transgene initiates murine acute promyelocytic leukemia. Proc Natl Acad Sci USA. 1997;94(6):2551-2556. Tong Q, You H, Chen X, et al. ZYH005, a novel DNA intercalator, overcomes alltrans retinoic acid resistance in acute promyelocytic leukemia. Nucleic Acids

Res. 2018;46(7):3284-3297. 32. Bachmann AS, Opoku-Ansah J, IbarraRivera TR, et al. Syrbactin structural analog TIR-199 blocks proteasome activity and induces tumor cell death. J Biol Chem. 2016;291(16):8350-8362. 33. Kogan SC, Hong SH, Shultz DB, Privalsky ML, Bishop JM. Leukemia initiated by PMLRARalpha: the PML domain plays a critical role while retinoic acid-mediated transactivation is dispensable. Blood. 2000; 95(5):1541-1550. 34. Zhu J, Gianni M, Kopf E, et al. Retinoic acid induces proteasome-dependent degradation of retinoic acid receptor alpha (RARalpha) and oncogenic RARalpha fusion proteins. Proc Natl Acad Sci USA. 1999;96(26):14807-14812. 35. de The H, Vivanco-Ruiz MM, Tiollais P, Stunnenberg H, Dejean A. Identification of a retinoic acid responsive element in the retinoic acid receptor beta gene. Nature. 1990;343(6254):177-180. 36. Matikainen S, Ronni T, Hurme M, Pine R, Julkunen I. Retinoic acid activates interferon regulatory factor-1 gene expression in myeloid cells. Blood. 1996;88(1):114-123. 37. Yu M, Tong JH, Mao M, et al. Cloning of a gene (RIG-G) associated with retinoic acidinduced differentiation of acute promyelocytic leukemia cells and representing a new member of a family of interferon-stimulated genes. Proc Natl Acad Sci USA. 1997;94(14):7406-7411. 38. Davies PJ, Murtaugh MP, Moore WT, Jr., Johnson GS, Lucas D. Retinoic acidinduced expression of tissue transglutaminase in human promyelocytic leukemia (HL-60) cells. J Biol Chem. 1985; 260(8):5166-5174. 39. Kitareewan S, Pitha-Rowe I, Sekula D, et al. UBE1L is a retinoid target that triggers PML/RARalpha degradation and apoptosis in acute promyelocytic leukemia. Proc Natl Acad Sci USA. 2002;99(6):3806-3811. 40. Koeffler HP, Ranyard J, Pertcheck M. Myeloperoxidase: its structure and expression during myeloid differentiation. Blood. 1985;65(2):484-491. 41. Gianni M, Li Calzi M, Terao M, et al. AM580, a stable benzoic derivative of retinoic acid, has powerful and selective cyto-differentiating effects on acute promyelocytic leukemia cells. Blood. 1996; 87(4):1520-1531. 42. Toma S, Isnardi L, Raffo P, et al. RARalpha antagonist Ro 41-5253 inhibits proliferation and induces apoptosis in breast-cancer cell lines. Int J Cancer. 1998;78(1):86-94. 43. Martinez Molina D, Jafari R, Ignatushchenko M, et al. Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science. 2013;341(6141):84-87. 44. Jafari R, Almqvist H, Axelsson H, et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Nat Protoc. 2014;9(9):2100-2122. 45. Huang DL, Bax NA, Buckley CD, Weis WI, Dunn AR. Vinculin forms a directionally asymmetric catch bond with F-actin.

Science. 2017;357(6352):703-706. 46. Xia L, Wang XR, Wang XL, et al. A novel role for pyruvate kinase M2 as a corepressor for P53 during the DNA damage response in human tumor cells. J Biol Chem. 2016;291(50):26138-26150. 47. Panchabhai S, Schlam I, Sebastian S, Fonseca R. PKM2 and other key regulators of Warburg effect positively correlate with CD147 (EMMPRIN) gene expression and predict survival in multiple myeloma. Leukemia. 2017;31(4):991-994. 48. Gu ZM, Wu YL, Zhou MY, et al. Pharicin B stabilizes retinoic acid receptor-alpha and presents synergistic differentiation induction with ATRA in myeloid leukemic cells. Blood. 2010;116(24):5289-5297. 49. Li K, Wang F, Cao WB, et al. TRIB3 promotes APL progression through stabilization of the oncoprotein PML-RARalpha and inhibition of p53-mediated senescence. Cancer Cell. 2017;31(5):697-710 e697. 50. Altucci L, Leibowitz MD, Ogilvie KM, de Lera AR, Gronemeyer H. RAR and RXR modulation in cancer and metabolic disease. Nat Rev Drug Discov. 2007;6(10):793-810. 51. Zhang XW, Yan XJ, Zhou ZR, et al. Arsenic trioxide controls the fate of the PMLRARalpha oncoprotein by directly binding PML. Science. 2010;328(5975):240-243. 52. Ruthardt M, Testa U, Nervi C, et al. Opposite effects of the acute promyelocytic leukemia PML-retinoic acid receptor alpha (RAR alpha) and PLZF-RAR alpha fusion proteins on retinoic acid signalling. Mol Cell Biol. 1997;17(8):4859-4869. 53. Fanelli M, Minucci S, Gelmetti V, Nervi C, Gambacorti-Passerini C, Pelicci PG. Constitutive degradation of PML/RARalpha through the proteasome pathway mediates retinoic acid resistance. Blood. 1999;93(5):1477-1481. 54. Gianni M, Kalac Y, Ponzanelli I, Rambaldi A, Terao M, Garattini E. Tyrosine kinase inhibitor STI571 potentiates the pharmacologic activity of retinoic acid in acute promyelocytic leukemia cells: effects on the degradation of RARalpha and PMLRARalpha. Blood. 2001;97(10):3234-3243. 55. Finch RA, Li J, Chou TC, Sartorelli AC. Maintenance of retinoic acid receptor alpha pools by granulocyte colony-stimulating factor and lithium chloride in all-trans retinoic acid-treated WEHI-3B leukemia cells: relevance to the synergistic induction of terminal differentiation. Blood. 2000; 96(6):2262-2268. 56. Launay S, Gianni M, Diomede L, Machesky LM, Enouf J, Papp B. Enhancement of ATRA-induced cell differentiation by inhibition of calcium accumulation into the endoplasmic reticulum: cross-talk between RAR alpha and calciumdependent signaling. Blood. 2003; 101(8):3220-3228. 57. Witcher M, Shiu HY, Guo Q, Miller WH, Jr. Combination of retinoic acid and tumor necrosis factor overcomes the maturation block in a variety of retinoic acid-resistant acute promyelocytic leukemia cells. Blood. 2004;104(10):3335-3342.

haematologica | 2019; 104(1)

ARTICLE

Acute Myeloid Leukemia

Gemtuzumab ozogamicin for de novo acute myeloid leukemia: final efficacy and safety updates from the open-label, phase III ALFA-0701 trial Juliette Lambert,1 Cécile Pautas,2 Christine Terré,3 Emmanuel Raffoux,4 Pascal Turlure,5 Denis Caillot,6 Ollivier Legrand,7 Xavier Thomas,8 Claude Gardin,9 Karïn Gogat-Marchant,10 Stephen D. Rubin,11 Rebecca J. Benner,12 Pierre Bousset,13 Claude Preudhomme,14 Sylvie Chevret,15 Herve Dombret16 and Sylvie Castaigne17

Service d'Hématologie et Oncologie, Centre Hospitalier de Versailles, Le Chesnay, France; 2Service d’Hématologie et de Thérapie Cellulaire, Hôpital Henri Mondor, Créteil, France; 3Laboratoire de Cytogénétique, Centre Hospitalier de Versailles, France; 4Hôpital Saint-Louis (AP-HP), Université Paris Diderot, France; 5Service d'Hématologie Clinique et Thérapie Cellulaire, Centre Hospitalier Universitaire, Limoges, France; 6Hematologie Clinique, Hôpital François Mitterrand, Centre Hospitalier Universitaire, Dijon, France; 7 Hôpital Saint-Antoine (AP-HP), Université Paris Pierre et Marie Curie, France; 8Centre Hospitalier Lyon Sud, Hospices Civils de Lyon, Université Lyon 1, Pierre Benite, France; 9 Hôpital Avicenne (AP-HP), Université Paris 13, Bobigny, France; 10Global Clinical Development Pfizer Inc., Paris, France; 11Global Product Development, Pfizer Inc., Collegeville, PA, USA; 12Global Product Development, Pfizer Inc., Groton, CT, USA; 13Pfizer Oncology, Pfizer Inc., Paris, France; 14Université Lille, INSERM, Centre Hospitalier Universitaire Lille, UMR-S 1172 - Jean-Pierre Aubert Center - Centre de Recherche, Lille, France; 15Departement de Biostatistique, Hôpital Saint-Louis (AP-HP), Universite Paris Diderot, INSERM S 717, France; 16Hopital Saint-Louis (AP-HP), Universite Paris Diderot, France and 17Service d'Hématologie et Oncologie, Centre Hospitalier de Versailles, Université de Versailles Saint Quentin, Le Chesnay, France 1

ABSTRACT

Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):113-119

Correspondence: scastaigne@ch-versailles.fr

he randomized, phase III ALFA-0701 trial showed that a reduced and fractionated dose of gemtuzumab ozogamicin added to standard front-line chemotherapy significantly improves event-free survival (EFS) in adults with de novo acute myeloid leukemia (AML). Here we report an independent review of EFS, final overall survival (OS), and additional safety results from ALFA-0701. Patients (n=271) aged 50-70 years with de novo AML were randomized to receive conventional frontline induction chemotherapy (3+7 daunorubicin+cytarabine) with/without gemtuzumab ozogamicin 3 mg/m2 on days 1, 4, and 7 during induction. Patients in remission following induction therapy received 2 courses of consolidation therapy (daunorubicin+cytarabine) with/without gemtuzumab ozogamicin (3 mg/m2/day on day 1) according to their initial randomization. The primary end point was investigator-assessed EFS. Secondary end points included OS and safety. A blinded independent review confirmed the investigator-assessed EFS results [August 1, 2011; hazard ratio (HR) 0.66; 95% Confidence Interval (CI): 0.49-0.89; 2sided P=0.006], corresponding to a 34% reduction in risk of events in the gemtuzumab ozogamicin versus control arm. Final OS at April 30, 2013 favored gemtuzumab ozogamicin but was not significant. No differences in early death rate were observed between arms. The main toxicity associated with gemtuzumab ozogamicin was prolonged thrombocytopenia. Veno-occlusive disease (including after transplant) was observed in 6 patients in the gemtuzumab ozogamicin arm and 2 in the control arm. In conclusion, gemtuzumab ozogamicin added to standard intensive chemotherapy has a favorable benefit/risk ratio. These results expand front-line treatment options for adult patients with previously untreated AML. (Trial registered at clinicaltrials.gov;identifier: 00927498.) haematologica | 2019; 104(1)

Received: January 17, 2018. Accepted: July 25, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.188888 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/113 ©2019 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.

113

J. Lambert et al.

Introduction Acute myeloid leukemia (AML) is a heterogeneous disease, with classification based on morphological, cytogenetic, molecular, and immunophenotypic characteristics,1 which, along with patients' characteristics, such as age and Eastern Cooperative Oncology Group Performance Status (ECOG PS), influence treatment recommendations and outcomes.2 Gemtuzumab ozogamicin (GO) is an antibodydrug conjugate composed of the CD33-directed monoclonal antibody that is covalently linked to the cytotoxic agent N-acetyl gamma calicheamicin.3 Efficacy with single-agent GO (9 mg/m2 for each of 2 doses administered 14 days apart) was initially established in patients with CD33-positive AML in first recurrence based on clinically meaningful response rates observed in 3 phase II studies in patients with AML in first relapse. However, liver toxicity and a long duration of cytopenia were observed.4 Additional in vitro study results, showing that CD33-expressing leukemic cells rapidly re-express CD33 molecules on their cell surface after treatment with an anti-CD33 antibody,5 led the Acute Leukemia French Association (ALFA) to investigate the use of a fractionated dosing regimen of GO that might enhance the internalization process while improving safety compared with higher unfractionated dosing.6 The randomized, phase III ALFA-0701 study compared the efficacy and safety of the standard 3+7 daunorubicin (DNR; days 1-3) and cytarabine (AraC; days 1-7) induction regimen (D+A), with or without fractionated dosing of GO (3 mg/m2 on days 1, 4, and 7), in patients aged 50-70 years with treatment-naive AML.6 Patients in remission following induction therapy received 2 courses of consolidation therapy consisting of D+A with or without GO (3 mg/mÂ˛/day on day 1) according to their initial randomization. The initial results as of the cut-off date of August 1, 2011, reported for the ALFA-0701 trial by Castaigne et al. in 2012 showed that the study met its primary end point, with a significant improvement in investigator-assessed event-free survival (EFS) without an increase in the risk of death from toxicities when GO was added to the standard chemotherapy.6 Following publication of the initial results, the Centre Hospitalier de Versailles, in collaboration with Pfizer, performed a retrospective collection of additional data to provide a more complete assessment of the safety profile of GO, and to conduct a retrospective, independent, blinded review of EFS. This report presents the final overall survival (OS) results from the ALFA-0701 study based on a longer follow-up date (cut-off date, April 30, 2013), the results of the independent review of EFS, as well as additional safety results focused on adverse events (AEs) of special interest considered the most important for understanding the safety profile of GO.

Methods Details of this randomized, open-label, multicenter, phase III ALFA-0701 study have been previously described.6

Patients and treatment A total of 280 patients were randomized 1:1 to receive conventional 3+7 D+A induction chemotherapy, with DNR 60 mg/m2/d on days 1 to 3 and AraC 200 mg/m2/d on days 1 to 7 without (control arm) or with GO (GO arm) 3 mg/m2/d on days 1, 4, and 7; the total dose of GO per infusion was not to exceed one 5 mg vial. 114

A second induction course, with DNR and AraC, was given if leukemic blasts persisted at day 15 bone marrow aspirate (BMA). Patients with a complete remission (CR) or CR with incomplete platelet recovery (CRp) after induction treatment received 2 courses of consolidation, including DNR and AraC with or without GO 3 mg/m2/d on day 1 according to their randomization, provided the platelet count was â&#x2030;Ľ50x109/L on the planned day 1 of the consolidation course. Patients who experienced CR could be considered for allogeneic transplant according to ECOG PS, age, if a donor had or had not been found, and cytogenetic and molecular risk categories. An interval of two months between the last dose of GO and transplantation was recommended. The study was approved by the Saint-Germain en Laye ethics committee in France and the institutional review board of the French Regulatory Agency. All procedures were conducted in compliance with the Declaration of Helsinki. Written informed consent was provided by all patients (EuduraCT n.: 2007-002933-36).

Efficacy analyses This report presents: 1) final results of the secondary end point of OS, defined as the time from date of randomization to date of death from any cause at the cut-off date of April 30, 2013; 2) results of a blinded and independent review of the EFS end point, defined as the time from randomization to relapse, death from any cause, or failure to achieve CR or CRp, performed by hematology experts to study the reproducibility of this clinically important end point in AML trials; 3) results of the secondary end point relapse-free survival for patients experiencing a response; and 4) hematologic response by investigator assessment. The independent review committee analysis was based on the retrospective collection of all data used for efficacy measurements, including reports of BMA, complete blood count, extramedullary disease, or molecular or cytogenetic relapse available at the site from screening until death, or up to 28 days after either induction failure or relapse as determined by the investigator (whichever happened first).

Safety analyses Safety data presented in this report were collected retrospectively and consist of events of special interest considered the most important for understanding the safety profile of GO and serious AEs (SAEs). This includes all grades of hemorrhage, all grades of veno-occlusive disease (VOD), severe (grade â&#x2030;Ľ3) infections, any adverse event (AE) that led to early permanent discontinuation of either GO or chemotherapy, and laboratory data. Serious AE reporting contains all SAEs reported to the Pfizer safety database throughout the study and was not restricted to causality or predefined categories.

Statistical analyses Sample size calculations have been reported previously.6 The modified intent-to-treat (mITT) population was the primary population for evaluating efficacy end points and included all patients who were randomized, unless consent was withdrawn before treatment initiation. Analyses were made according to the initial randomization arm, regardless of whether patients received the study drug to which they were randomized.

Results Patients Of the 280 patients randomized in this study, data from 9 patients were excluded from the analyses because no signed copy of the informed consent was available in the haematologica | 2019; 104(1)

Fractionated doses of gemtuzumab ozogamicin in AML

2-sided P=0.16

Figure 1. Overall survival. Control: daunorubicin + cytarabine (D+A); GO: gemtuzumab ozogamicin plus D+A; OS: overall survival; HR: hazard ratio; CI: Confidence Interval; n: number.

site file. Thus, the mITT population comprised 271 patients (GO arm, n=135; control arm, n=136). Characteristics of the patients in the mITT population were evenly balanced between treatment arms and were as expected in this AML population (Online Supplementary Table S1). Overall, 268 (98.9%) patients (GO arm, n=134; control arm, n=134) received study treatment; however, 3 patients randomized to the GO arm did not receive GO (Online Supplementary Figure S1). Of the 131 patients who received GO, 123 (93.9%) received all 3 fractionated GO doses during induction, and 91 (69.5%) and 64 (48.9%) patients received GO during consolidation 1 and 2, respectively. Exposure to DNR and AraC was similar between treatment arms over all study phases, with 62.6% and 65.0% of patients in the GO and control arms, respectively, completing study treatment.

Efficacy Response rate Overall, the rate of CR with or without CRp was similar to the results initially published.6 There was no significant difference in the rate of CR or CRp in the GO arm compared with the control arm (Online Supplementary Table S2). A CR was achieved in 95 patients (70.4%) in the GO arm and 95 patients (69.9%) in the control arm; CRp was achieved in 15 patients (11.1%) in the GO arm and 5 patients (3.7%) in the control arm. Overall survival At the time of the primary study analysis performed at the data cut-off date of August 1, 2011, the overall median follow up was 14.8 months; 20.0 months in alive patients. At the time of the final OS analysis performed at the data cut off of April 30, 2013, the median follow up was 47.6 months in the GO arm and 41.0 months in the control arm. The final OS analysis shows a numerically longer OS in the GO arm; however, this difference did not reach statistical significance. Median OS was 27.5 months [95% haematologica | 2019; 104(1)

Confidence Interval (CI): 21.4-45.6] in the GO arm and 21.8 months (95%CI: 15.5-27.4) in the control arm [Hazard Ratio (HR), 0.81; 95%CI: 0.60-1.09; 2-sided P=0.16). A total of 80 (59.3%) patients in the GO arm and 88 (64.7%) in the control arm died before April 30, 2013 (Figure 1). Event-free survival The primary end point of EFS derived from investigator assessment was significantly longer for patients in the GO arm [median 17.3 months (95%CI: 13.4-30.0)] than in the control arm [median 9.5 months (8.1-12.0); HR: 0.56; 95%CI: 0.42-0.76; 2-sided P=0.0002 by log-rank test], corresponding to a 44% reduction in the risk of an event for patients in the GO arm compared with those in the control arm (Figure 2). At year 1, the number needed to treat (NNT) with GO to prevent an event was 6; NNT at year 3 was 4. Multiple EFS sensitivity analyses demonstrate the robustness of the primary EFS results (Online Supplementary Table S3). Of note, subgroup analyses showed that patients with favorable or intermediate cytogenetic risk (classified according to the International System for Human Cytogenetic Nomenclature criteria7) at baseline had significantly longer EFS in the GO arm than in the control arm (HR: 0.46; 95%CI: 0.31-0.68; P<0.0001). This advantage in EFS with GO was not apparent for patients with poor cytogenetic risk (HR: 1.11; 95%CI: 0.63-1.95; P=0.72). Similarly, activity was more apparent with GO for patients in favorable/intermediate risk groups compared with poor risk group by National Comprehensive Cancer Network or European LeukemiaNet (ELN) risk classifications. Results of all the other subgroup analyses were consistent with the effect of GO on primary EFS (Online Supplementary Figure S2). In addition, the results of the blinded independent review support the primary EFS by investigator assessment, with a median EFS of 13.6 months (95%CI: 9.0-19.2) in the GO arm and 8.5 months (95%CI: 7.5-12.0) in the control arm (HR: 0.66; 95%CI: 0.49-0.89; P=0.006) (Table 1). 115

J. Lambert et al. Table 1. Event-free survival (EFS) results (mITT population) by the investigator-assessed and blinded independent review methods.

Investigator assessed GO Control n=135 n=136 EFS Events, n (%) Induction failure Relapse Death Censored patients Median time to event, months [95% CI]* HR† [95% CI] P-value‡ Probability of being event-free [95% CI]§ At 2 years At 3 years

Blinded independent review GO Control n=135 n=136

73 (54.1) 17 (12.6) 44 (32.6) 12 (8.9) 62 (45.9) 17.3 [13.4−30.0] 0.56 [0.42−0.76] 0.0002

102 (75.0) 29 (21.3) 58 (42.6) 15 (11.0) 34 (25.0) 9.5 [8.1−12.0]

78 (57.8) 25 (18.5) 43 (31.9) 10 (7.4) 57 (42.2) 13.6 [9.0−19.2] 0.66 [0.49−0.89] 0.006

100 (73.5) 34 (25.0) 50 (36.8) 16 (11.8) 36 (26.5) 8.5 [7.5−12.0]

42.1 [32.9−51.0] 39.8 [30.2−49.3]

18.2 [11.1−26.7] 13.6 [5.8−24.8]

38.5 [29.6−47.3] 36.5 [27.3–45.7]

18.1 [11.1−26.5] 13.6 [5.8–24.7]

Control: 3+7 daunorubicin + cytarabine (DA); GO: gemtuzumab ozogamicin + 3+7 DA; HR: hazard ratio; mITT: modified intent to treat; n: number. *Median estimated by KaplanMeier method; Confidence Interval (CI) based on the Brookmeyer-Crowley method with log-log transformation. †Based on the Cox proportional hazards model. ‡Two-sided P-value from the log-rank test. §Estimated from Kaplan-Meier curve. Probability (%) calculated by the product-limit method; CI calculated from the log-log transformation of survival probability using a normal approximation and back transformation, and two-sided CIs for the estimates were computed using the Greenwood formula.

Table 2. Post-study treatment (mITT population). Patients with ≥1 follow-up therapy, n (%) Patients receiving GO as a component of follow-up therapy, n (%) Patients with HSCT, n ( %) Timing of HSCT, n (%) In first remission for responder patients After induction failure After relapse

GO n=135

Control n=136

96 (71.1) 2 (1.5) 32 (23.7)

109 (80.1) 30 (22.1) 53 (39.0)

17 (12.6) 2 (1.5) 13 (9.6)

22 (16.2) 9 (6.6) 22 (16.2)

Data are number (n) (%) unless otherwise indicated. Control: 3+7 daunorubicin + cytarabine (DA); GO: gemtuzumab ozogamicin + 3+7 DA; HSCT: hematopoietic stem cell transplant; mITT: modified intent to treat.

Of interest, it did not appear that low CD33 expression (<30% of blasts positive) had an influence on the EFS benefit with GO; however, few enrolled patients (13.7%) had low CD33 expression. Similarly, analysis using a Cox proportional hazards model incuding treatment and CD33 expression as a continuous variable did not show any effect of CD33 expression on EFS. Relapse-free survival Relapse-free survival (RFS) was significantly longer for patients in the GO arm compared with control (Online Supplementary Figure S3). Median RFS was 28.0 months (95%CI: 16.3-not estimable) in the GO arm and 11.4 months (95%CI: 10.0-14.4) in the control arm, corresponding to a 47% reduction in the risk of an event for patients in the GO arm compared with those in the control arm.

Post-study treatment The majority of patients [96 (71.1%) in the GO arm and 109 (80.1%) in the control arm] (Table 2) received at least 1 subsequent therapy for AML following study treatment. Overall, 32 patients (23.7%) in the GO arm and 53 (39.0%) in the control arm received a hematopoietic stem cell transplant (HSCT) either in first remission or after induction failure or relapse. All transplants were allogene116

ic, except one autologous transplant in the control arm. This included 17 patients (12.6%) in the GO arm and 22 (16.2%) in the control arm who received HSCT in first remission. Additional analysis of OS censoring the patients at the time of transplant showed no impact on the OS results (P=0.240). The other patients received rescue therapy after either induction failure or relapse, including 30 patients (22.1%) in the chemotherapy arm who subsequently received GO as rescue therapy as part of a compassionate use program.

Safety A summary of treatment-emergent AEs of special interest is shown in Table 3. Overall, 208 (77.6%) patients experienced a severe (grade ≥3) infection; the incidence was similar between arms [GO arm, 102 (77.9%); control arm, 106 (77.4%)]. Hemorrhage of any grade occurred in the majority of patients in both treatment arms [225 (84.0%)]; the rate was significantly higher (P=0.008) among patients in the GO arm [118 (90.1%)] than the control arm [107 (78.1%)]. Grade ≥3 hemorrhages were reported in 30 (22.9%) patients in the GO arm and 13 (9.5%) patients in the control arm. Six (4.6%) patients in the GO arm and 2 (1.5%) patients in the control arm experienced veno-occlusive disease (VOD; P=0.165) (Table 3 haematologica | 2019; 104(1)

Fractionated doses of gemtuzumab ozogamicin in AML

Table 3. Summary of all-causality adverse events of special interest by maximum CTCAE grade (as-treated population*).

Retrospective data, n (%) Infections: severe (grade ≥3) Hemorrhage: all grades (grade ≥1), total† Grade 3 Grade 4 Grade 5 VOD: all grades (grade ≥1), total† Grade 3 Grade 4 Grade 5

GO (n=131)

Control (n=137)

Total (n=268)

102 (77.9) 118 (90.1) 23 (17.6) 4 (3.1) 3 (2.3) 6 (4.6) 2 (1.5) 1 (0.8) 2 (1.5)

106 (77.4) 107 (78.1) 12 (8.8) 0 1 (0.7) 2 (1.5) 1 (0.7) 1 (0.7) 0

208 (77.6) 225 (84.0) 35 (13.1) 4 (1.5) 4 (1.5) 8 (3.0) 3 (1.1) 2 (0.7) 2 (0.7)

Control: 3+7 daunorubicin + cytarabine (D+A); CTCAE: Common Terminology Criteria for Adverse Events; GO: gemtuzumab ozogamicin plus D+A; MedDRA: Medical Dictionary for Regulatory Activities; NCI: National Cancer Institute; VOD: veno-occlusive disease; n: number. *Defined as all patients who received at least 1 dose of study medication and reported according to whether or not GO was received. †Adverse events were graded in accordance with the NCI CTCAE v.3.0 and coded by MedDRA v.18.0.

Table 4. Summary of deaths (as-treated population*).

Deaths, n (%)

GO (n=131)

Control (n=137)

Deaths within 30 days of initiating study treatment Deaths during safety reporting period† Mechanism(s) of death‡ Disease progression or relapse Septic shock Infection Liver toxicity Hemorrhage Other

5 (3.8) 6 (4.6)

3 (2.2) 5 (3.6)

2 (1.5) 2 (1.5) 0 1 (0.8) 3 (2.3) 2 (1.5)

2 (1.5) 2 (1.5) 1 (0.7) 0 1 (0.7) 3 (2.2)

Control: daunorubicin + cytarabine (D+A); GO: gemtuzumab ozogamicin plus D+A; n: number. *Defined as all patients who received at least 1 dose of study medication and reported according to whether or not GO was received. †≤28 days after last dose of any study drug treatment. [GO: daunorubicin, cytarabine and idarubicin (a component of salvage therapy. )] ‡More than 1 mechanism of death could be selected.

and Online Supplementary Table S4); the 2 patients in the control arm received GO during the follow-up phase of the study, as part of the compassionate use program after having relapsed before developing VOD. All-causality SAEs were reported for 164 (61.2%) patients, including 88 (67.2%) in the GO arm and 76 (55.5%) in the control arm (Online Supplementary Table S5). Among the most commonly reported SAEs were those related to infection: 54 (41.2%) patients in the GO arm and 52 (38.0%) in the control arm (data not shown). The median number of intensive care unit hospitalizations was similar between the GO and control arms, affecting 25 patients in each arm; median time spent in the intensive care unit was 0.70 (range, 0.3-6.0) weeks and 0.60 (range, 0.1-7.6) weeks, respectively. A total of 11 (4.1%) patients died during the period in which safety was reported; this included the time from the first dose to 28 days after the last dose of study treatment. The number of patient deaths was similar between treatment arms [GO arm, 6 (4.6%); control arm, 5 (3.6%)] (Table 4). The mechanisms of death were mostly similar between the 2 treatment arms, with the largest difference noted for patients for whom the mechanism of death involved hemorrhage [GO arm, 3 (2.3%); control arm, 1 (0.7%)]. Permanent discontinuation of GO and/or chemotherapy due to treatment-emergent AEs occurred in 41 (31.3%) haematologica | 2019; 104(1)

patients in the GO arm and 10 (7.3%) in the control arm (Online Supplementary Table S6). The most common reasons for study drug discontinuation among the 131 patients in the GO arm were thrombocytopenia in 20 (15.3%) patients and hepatobiliary disorders in 8 (6.1%) patients; among the 137 patients in the control arm, no (0%) and 1 (0.7%) patient, respectively, discontinued for the same reasons. Of note, the protocol was amended to withhold GO to patients during consolidation for persisting thrombocytopenia below 100x109/L not recovered ≥14 days after the scheduled date of consolidation therapy. Further investigations of the 20 patients in the GO arm who discontinued study drug because of persistent thrombocytopenia showed that no patients had SAE of hemorrhage reported. While the majority of patients experienced severe myelosuppression, with similar rates in both treatment arms (Online Supplementary Table S7), the median time to recovery of platelets was longer for patients in the GO arm than in the control arm for each treatment course (Table 5). Additional analyses conducted to identify severe (grade 3 and 4) persistent thrombocytopenia (i.e. platelet count <50x109/L at 45 days after day 1 of the previous treatment phase in which a patient experienced CRp) showed that more patients had severe persistent thrombocytopenia in the GO arm (20.4%) than in the control arm (2.0%). 117

J. Lambert et al.

2-sided P=0.0002

Figure 2. Event-free survival (EFS) (investigator-assessed). Control: daunorubicin + cytarabine (D+A); GO: gemtuzumab ozogamicin plus D+A; EFS: event-free survival; HR: hazard ratio; CI: Confidence Interval; n: number.

Non-hematologic laboratory test results revealed that liver chemistry abnormalities were generally similar between treatment arms, except that GO treatment was associated with more frequent elevations of aspartate aminotransferase than the control treatment (all grades: 89.2% and 73.9%; grade 3 / 4 : 14% and 9.0%, respectively) and more frequent elevations of alkaline phosphatase (all grades: 79.7% and 68.9%; grades 3 / 4: 13.3% and 5.3%, respectively) (Online Supplementary Table S7).

Discussion A previously published analysis of the ALFA-0701 study was based on a cut-off date of August 1, 2011.6 In this report, we present the final analysis of OS based on a cutoff date of April 30, 2013, and the results of a blinded, independent analysis of the primary end point, EFS. The final results demonstrate a trend toward longer OS for patients in the GO arm than in the control arm, although this difference did not reach statistical significance. Nevertheless, the longer OS trend in the GO arm observed in ALFA-0701 is consistent with a significant improvement in OS demonstrated in an individual patient data meta-analysis (IPD-MA).8 This IPD-MA consisted of 5 randomized studies comprising more than 3300 patients and included the ALFA-0701 study, in which GO was used in intensive induction chemotherapy in adult patients with newly diagnosed AML. In this IPD-MA, the primary end point of OS showed a significant improvement at five years for patients who received GO [Odds Ratio (OR): 0.90; 95%CI: 0.82-0.98; log-rank 2-sided P=0.01], corresponding to a 10% reduction in risk of death in the GO arm.8 The OS benefit was the most apparent in patients with favorable (OR: 0.47; 95%CI: 0.31-0.73) and intermediate (OR: 0.84; 95%CI: 0.75-0.95) cytogenetic risk disease.8 However, as mentioned in the most recent version of the ELN guidelines,2 OS may not be the best indicator of the efficacy of a new drug because of confounding 118

effects of allogeneic transplant or rescue therapies given after relapse or failure to reach CR. However, EFS, which includes failure to reach CR, relapse, or death due to any cause, including drug toxicity, may better reflect the efficacy of a single treatment. In the ALFA-0701 study, we reported that the majority of patients did receive poststudy treatment, including allogeneic transplant and rescue therapy after induction failure or relapse. Of note, this includes 30 patients from the control arm who finally received GO as rescue therapy. Furthermore, in the ALFA0701 study, the sample size was calculated to show a difference in EFS; thus, the relatively low number of patients in the study was probably not appropriate to show an OS benefit. At the time of the final EFS analysis of ALFA-0701, the addition of GO to D+A approximately doubled the median EFS compared with chemotherapy alone, with a median EFS of 17.3 months for patients randomized to the GO arm and 9.5 months for patients in the chemotherapyalone arm. Multiple sensitivity analyses demonstrated the robustness of the EFS results, and in this report, we established the reproducibility of the EFS end point in this trial via a blinded, independent review. A significant prolongation of EFS is considered by hematologists to contribute to the clinical benefit received by patients, as this translates into a durable first remission, increasing the probability of a patient being cured. The subgroup analyses of the effect of GO on EFS in ALFA-0701 generally supports the potential benefit with GO for patients with AML independently of the degree of CD33 positivity, although few enrolled patients (13.7%) had low CD33 expression (<30% of blasts positive). The effect of GO on EFS in ALFA-0701 is consistent with the positive effect of GO on RFS, as well as with the previously reported significant effect of GO on minimal residual disease assessed by NPM1 status.9 This enhancement of the quality of the response and the prolongation of the response are the 2 factors that sustained the prolongation of EFS with GO. In ALFA-0701, there was no difference in early mortalhaematologica | 2019; 104(1)

Fractionated doses of gemtuzumab ozogamicin in AML

Table 5. Time to recovery of platelets and persistent thrombocytopenia (as-treated population*).

Control

†

Patient with persistent thrombocytopenia Overall, n‡/N (%)§ǁ By treatment phase

Time to platelet recovery to 50x109/L Patients recovered n‡/N (%)¶ Median recovery time,#d Time to platelet recovery to 100x109/L Patients recovered, n‡/N (%)¶ Median recovery time,#d

Induction Course 8/108 (7.4)

22/108 (20.4) Consolidation Course 1 Course 2 8/94 (8.5) 10/76 (13.2)

Induction Course 1/101 (1.0)

2/101 (2.0) Consolidation Course 1 Course 2 0/94 (0) 2/85 (2.4)

109/131 (83.2) 34.0

92/97 (94.8) 32.0

80/82 (97.6) 36.5

118/137 (86.1) 86/97 (88.7) 29.0 27.0

85/89 (95.5) 30.0

99/131 (75.6) 35.0

71/97 (73.2) 35.0

70/82 (85.4) 43.0

111/137 (81.0) 80/97 (82.5) 30.0 28.0

82/89 (92.1) 32.0

Control: daunorubicin + cytarabine (D+A); CRp: complete remission with incomplete platelet recovery; GO: gemtuzumab ozogamicin plus D+A. *Defined as all patients who received at least 1 dose of study medication and reported according to whether or not GO was received. †Persistent thrombocytopenia was defined as platelet count not recovered to 50x109/L at 45 days after day 1 of the respective treatment phase in patients experiencing CRp. ‡n: number of patients with events. §Overall number of patients with persistent thrombocytopenia after any phase. ǁN: number of patients evaluable for thrombocytopenia was used as the denominator for calculating the percentage of patients with persistent thrombocytopenia. Patients evaluable for thrombocytopenia were those who received each of the courses and had platelet laboratory results collected during the retrospective data collection. N: number of patients who received each of the courses of treatment. #Based on Kaplan-Meier estimate.

ity rate between the 2 arms. The rationale for fractionated GO dosed at 3 mg/m² on days 1, 4, and 7, as used in ALFA-0701, is further supported by results of the IPD-MA, which showed less early mortality with the 3 mg/m² dose compared to 6 mg/m².8 Furthermore, study NCRI AML17, in which a single dose of 6 mg/m² of GO was used in combination with induction chemotherapy, provided no advantage in response, disease-free survival, or OS compared with a 3-mg/m² dose; however, in this study, the 30and 60-day mortality was significantly higher in the patients receiving 6 mg/m².8,10 Regarding AEs of special interest for GO, in the GO arm, there was no increase in either the incidence of severe infection or the percentage of patients who died of infection. Hemorrhage, VOD, and the number of patients who discontinued the study drug(s) because of AEs were increased with GO. The most frequent AE that led to permanent discontinuation of study drug in the GO arm (in 20 patients) was persistent thrombocytopenia, influenced in part by a protocol amendment recommending discontinuation of GO in case of persistent thrombocytopenia. Thus, physicians should be aware of thrombocytopenia

References 1. 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. 2. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017; 129(4):424-447. 3. Ricart AD. Antibody-drug conjugates of calicheamicin derivative: gemtuzumab ozogamicin and inotuzumab ozogamicin. Clin Cancer Res. 2011;17(20):6417-6427. 4. Bross PF, Beitz J, Chen G, et al. Approval summary: gemtuzumab ozogamicin in relapsed acute myeloid leukemia. Clin

haematologica | 2019; 104(1)

associated with GO and provide supportive care as required. Similarly, physicians must be aware of the risk of VOD with GO, particularly the increased risk of VOD, either preceding or following HSCT, and closely monitor for clinical signs such as hepatomegaly, rapid weight gain, and ascites, elevations in alanine aminotransferase, aspartate aminotransferase, total bilirubin, and alkaline phosphatase. In conclusion, the final results of this study indicate that GO added to standard chemotherapy significantly prolongs EFS in patients with newly diagnosed de novo AML and has an acceptable safety profile. This combination may expand front-line treatment options for this difficultto-treat patient population. Funding The authors would like to thank the Acute Leukemia French Association, which sponsored this study in collaboration with the Centre Hospitalier de Versailles (CHV); Pfizer acquired the study data and usage rights from CHV in March 2013. Editorial support was provided by Susan Reinwald, PhD, of Complete Healthcare Communications, LLC, and was sponsored by Pfizer.

Cancer Res. 2001;7(6):1490-1496. 5. Caron PC, Jurcic JG, Scott AM, et al. A phase 1B trial of humanized monoclonal antibody M195 (anti-CD33) in myeloid leukemia: specific targeting without immunogenicity. Blood. 1994;83(7):17601768. 6. Castaigne S, Pautas C, Terre C, et al. Effect of gemtuzumab ozogamicin on survival of adult patients with de-novo acute myeloid leukaemia (ALFA-0701): a randomised, open-label, phase 3 study. Lancet. 2012; 379(9825):1508-1516. 7. International System for Human Cytogenetic Nomenclature. Guidelines for cancer cytogenetics. In: Mittelman F, ed. Supplement to an International System for Human Cytogenetic Nomenclature. Basel, Switzerland: Karger; 1991:1-53. 8. Hills RK, Castaigne S, Appelbaum FR, et al.

Addition of gemtuzumab ozogamicin to induction chemotherapy in adult patients with acute myeloid leukaemia: a metaanalysis of individual patient data from randomised controlled trials. Lancet Oncol. 2014;15(9):986-996. 9. Lambert J, Lambert J, Nibourel O, et al. MRD assessed by WT1 and NPM1 transcript levels identifies distinct outcomes in AML patients and is influenced by gemtuzumab ozogamicin. Oncotarget. 2014; 5(15):6280-6288. 10. Burnett A, Cavenagh J, Russell N, et al. Defining the dose of gemtuzumab ozogamicin in combination with induction chemotherapy in acute myeloid leukemia: a comparison of 3 mg/m2 with 6 mg/m2 in the NCRI AML17 Trial. Haematologica. 2016;101(6):724-731.

119

ARTICLE Ferrata Storti Foundation

Haematologica 2018 Volume 104(1):120-127

Acute Myeloid Leukemia

Gemtuzumab ozogamicin in children with relapsed or refractory acute myeloid leukemia: a report by Berlin-Frankfurt-Münster study group Naghmeh Niktoreh,1* Beate Lerius,1* Martin Zimmermann,2 Bernd Gruhn,3 Gabriele Escherich,4 Jean-Pierre Bourquin,5 Michael Dworzak,6 Lucie Sramkova,7 Claudia Rossig,8 Ursula Creutzig,2 Dirk Reinhardt1 and Mareike Rasche1

Department of Pediatric Hematology and Oncology, University Hospital Essen, Germany; 2Department of Pediatric Hematology and Oncology, Hannover Medical School, Germany; 3Department of Pediatrics, Jena University Hospital, Germany; 4 Department of Pediatric Hematology and Oncology, Eppendorf University Hospital, Hamburg, Germany; 5Division of Pediatric Hematology/Oncology, University Children’s Hospital Zurich, Switzerland; 6St. Anna Children's Hospital and Children's Cancer Research Institute, Department of Pediatrics, Medical University of Vienna, Austria; 7 Department of Pediatric Hematology and Oncology, 2nd Faculty of Medicine, Charles University and University Hospital Motol, Prague, Czech Republic and 8University Children’s Hospital Münster, Pediatric Hematology and Oncology, Germany 1

*NN and BL contributed equally to this work as first authors.

ABSTRACT

Correspondence: mareike.rasche@uk-essen.de

Received: February 23, 2018. Accepted: August 3, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2018.191841 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/120

espite intensified salvage treatments, children with relapsed/refractory acute myeloid leukemia (AML) have poor survival. We evaluated gemtuzumab ozogamicin (CD33-targeted drug) used on a compassionate basis in patients diagnosed from 1995 until 2014 within Acute Myeloid Leukemia Berlin-Frankfurt-Münster studies, and identified 76 patients (<18 years) with highly-advanced and pre-treated AML [refractory de novo acute myeloid leukemia (n=10), de novo AML refractory to relapse (1st early: n=41; 1st late: n=10; 2nd or more: n=10), and secondary AML (n=5)]. At doses of 2.5-10 mg/m2, gemtuzumab ozogamicin was administered in 1-4 cycles as single agent (47%), combined with cytarabine (47%), or others (6%). Most common grade 3/4 adverse events were infections or febrile neutropenia (78% of severe adverse events), infusion-related immunological reactions (6%), and gastrointestinal symptoms (5%). Three patients experienced venoocclusive disease (one fatal due to exacerbation of a pre-existing cardiomyopathy). Sixty-four percent received subsequent hematopoietic stem cell transplantation. Probability of 4-year overall survival was 18±5% in all, 27±7% in patients with and 0% in patients without hematopoietic stem cell transplantation (P<0.0001). Administration of gemtuzumab ozogamicin on a patient-specific, compassionate use basis was frequently considered in our study group and proved to be effective for bridging children with very advanced AML to hematopoietic stem cell transplantation. Uniform prospective studies for these patients are urgently needed.

©2019 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.

120

Introduction Treatment of acute myeloid leukemia (AML) in children has improved remarkably during the past decades; however, pediatric patients with relapsed and refractory AML still have poor outcomes.1-6 These outcomes rely on disease-dependent characteristics, such as initial cytogenetics, in addition to response-to-therapy-related factors like the interval between initial diagnosis and relapse.7-10 Considering poor outcome and high toxicity of current salvage therapies, new targeted molecular treatments are needed.11-13 Gemtuzumab ozogamicin (GO) is an immunotoxin consisting of a potent humanized monoclonal antibody against CD33, and targets CD33 positive cells haematologica | 2019; 104(1)

Gemtuzumab in pediatric acute myeloid leukemia

which are present in approximately 80-90% of childhood AML.14,15 Although CD33 has been considered as a specific marker for hematopoietic cells of the myeloid lineage for a long time, this surface marker is also found to be expressed in hepatocytes which may potentially cause some off-target effects.16 Treatment with GO in different settings has previously been shown to be of value in pediatric AML. The first international experience of treatment with GO as monotherapy for compassionate use for children with relapsed or refractory AML (n=15) in 2003 suggested the efficacy of this treatment with doses of 4-9 mg/m2 in up to 3 cycles.17 A later phase II study showed that treatment with two doses of 7.5 mg/m2 GO with 14-day intervals in children with advanced AML (refractory, first refractory relapse or ≥second relapse) led to significantly higher survival in patients who received GO compared to patients who did not receive this treatment (3-year probability of overall survival: 27% vs. 0%, respectively; P=0.001).18 In addition to monotherapy, a good response rate was also achieved when GO was administered in combination with cytarabine in 17 children with relapsed or refractory AML [overall complete remission (CR) rate: 53%].19 Currently there are at least 9 active clinical trials worldwide investigating the effect of GO in AML (de novo, relapsed, or refractory). One of these trials only recruits children and 4 are recruiting children in addition to adults or elderly patients (Online Supplementary Table S1). Of interest, there is one active clinical trial which studies treatment with GO in compassionate use in refractory or relapsed AML (clinicaltrials.gov identifier: 02312037), which recruits children, adult, and elderly groups of patients.20 Considering the above-mentioned challenges in the treatment of relapsed and refractory pediatric AML, the infrequency of new treatment options, and in addition, the restricted accessibility of GO, we aimed to identify patients treated with GO as compassionate use in the AML-Berlin-Frankfurt-Münster (AML-BFM) study group and to evaluate the efficacy and safety in this heavily pretreated group of patients.

Methods

1993 and June 1998, and was followed by AML-BFM 98 which was opened in July 1998 and closed in June 2003. Between July 2003 and April 2004, the AML-BFM 98 Interim Study continued treatment of recruited patients with the best arm of the AML-BFM 98 trial. The AML-BFM 04 study was opened in April 2004 and randomization continued until April 2010. After April 2010 and until February 2014, the AML-BFM 2004 Interim Study continued treatment of patients with the experimental arm of the AML-BFM 2004 trial. Second-line treatment of patients included the Relapse AML 2001/01 trial24 which recruited patients from November 2001 to April 2009 and the International Registry Relapsed AML 2009.7 Patients included in the current cohort had received intensive treatment and/or HSCT before administration of GO. Considering the first treatment after initial diagnosis as first attempt, and each following treatment block or HSCT as further individual attempts, most of the patients (n=35, 46%) in the total cohort received GO as their 3rd treatment attempt (Online Supplementary Figure S1).

Treatment with GO During the period of analysis, 217 patients with non-response (NR) and 654 patients with at least one event of relapse were documented (Online Supplementary Figure S2). Within these patient records, 98 (from 39 different centers) were found with a positive history of GO treatment, 10 of which were included in the previously mentioned phase II trial18 and who were, thus, excluded from the current study. The remaining 88 patients received GO on a compassionate use basis, after failure of their first- and/or second-line treatments and/or when the general condition of patients was so poor that further intensive chemotherapy was not possible. Treatment with GO in these patients was recommended by the centralized study co-ordination office, and finally prescribed by the patients’ treating physicians at each local center. Patients or guardians provided written informed consent. GO was provided by International Pharmacy (San Francisco, CA, USA) through Clinigen (London, UK) and, for some patients, by Pfizer (New York, NY, USA) on a compassionate use program. From these 88 patients, 76 had sufficient data to evaluate the outcomes of GO use and were included in this study (Online Supplementary Figure S2). Three of the 76 patients have been previously reported.17 Grading of toxicity and adverse events was carried out using the NCI Common Terminology Criteria for Adverse Events (CTCAE), version 4.03, revised in June 14, 2010.25 Records concerning the safety evaluation of GO were not available for 5 patients (Online Supplementary Figure S2).

Patients Between January 1995 and March 2014, 2601 children with initial diagnosis at the age of ≤18 years were documented within the AML-BFM study group. Patients or guardians provided written informed consent. The current analysis was performed in accordance with the Declaration of Helsinki and was approved by the local ethics committee. The AML-BFM Study Group centrally reviewed the diagnosis of initial disease or relapse via bone marrow morphology and flow cytometry. All patients’ records were evaluated retrospectively for the use of GO. Medical reports of the patients treated with GO were reviewed retrospectively for evaluation of treatment outcomes and adverse events (AE).

Treatment protocols Before administration of GO, patients were treated based on randomized, phase III studies AML-BFM 93, 98, and 2004 running in 75 centers in Germany, Austria, Switzerland, and the Czech Republic. These treatment protocols have been previously described in detail.21-23 All studies were performed after the approval by national ethics committees and institutional review boards. The AML-BFM 93 recruited patients between January haematologica | 2019; 104(1)

Definitions and statistical analysis Bone marrow (BM) relapse was defined by presence of >5% of leukemic cells in the BM. Relapse events during the first 12 months from diagnosis of AML were considered as early relapse.3 Complete remission (CR) was defined with the presence of less than 5% blasts in morphological examination of BM in addition to satisfactory hematologic recovery (absolute neutrophil count > 1.0 x109/L, platelet count >80x109/L in the peripheral blood).21 Reaching CR without complete blood recovery was defined as incomplete CR (CRi).21 Response to GO was evaluated in patients who received HSCT after treatment with GO and was defined as CR plus CRi (CR/CRi). Persistence of ≥5% BM leukemic blasts was categorized as NR. Overall survival (OS) after GO was defined as the time from the first dose until death by any cause or until the last follow up. Probability of OS was calculated according to Kaplan-Meier and compared by log-rank test. P<0.05 was considered significant. Data were analyzed using SAS (Statistical Analysis System Version 9.4; SAS Institute, Cary, NC, USA). Data acquisition was stopped on 1st June 2017. 121

N. Niktoreh et al.

Results Patients’ characteristics and treatment with GO The majority of patients were initially diagnosed with AML with the French-American-British (FAB) classification of M4 (without atypical eosinophils) or M5 (n=29,

39%) and 13 (17%) patients had white blood cell (WBC) counts higher than 100x109/L at the time of their diagnosis. In total, 67 (88%) patients had high-risk AML as defined by morphology, cytogenetics and response to treatment, retrospectively (Table 1). Most of the patients (n=43, 56%) had been previously treated with liposomal

Table 1. Patients’ characteristics at initial diagnosis.

Feature

Total number of patients Age, years Median (range) Categories 0-2 3-10 11-18 Sex Female Male FAB Classification M0 M1/M2 M3 M4Eo-/M5 M6 M7 AML-not classified WBC count at diagnosis Median x109/L(range) Patients with WBC ≤100x109/L Patients with WBC >100x109/L No data Previous treatments Pre-treatment with FLA/G Pre-treatment with FLA/G+DX (+/- FLA/G)a Pre-treatment without FLA/G+DX or FLA/G HSCT prior to GO treatment Yes No Disease status prior to GO Refractory de novo AMLb De novo AML- refractory to 1st early relapsec De novo AML-refractory to 1st late relapsec De novo AML- refractory to 2nd relapsec De novo AML-refractory to 3rd relapsec Secondary AML- refractory to 1st/2nd relapsec Risk groupd Standard risk High risk

100 9.3 (1 months - 17 years)

15 26 35

20 34 46

33 43

43 57

6 25 1 29 1 7 7

8 33 1 39 1 9 9 14.8 (0.39 - 324)

62 13 1

82 17 1

12 43 21

16 56 28

14 62

18 82

10 41 10 8 2 5

13 54 13 11 3 6

9 67

12 88

n: number; FAB classification: French-American-British classification; M4Eo-: AML M4 subtype without the presence of atypical eosinophils; AML: acute myeloid leukemia; HSCT: hematopoietic stem cell transplantation; FLA/G: fludarabine, cytarabine with or without granulocyte colony stimulating factor; DX: liposomal daunorubicin; GO: gemtuzumab ozogamicin; WBC: white blood cell. aSome of the patients in this category received FLA/G in addition to FLA/G+DX. bGO was administered after failure of first-line treatment(s). c GO was administered after failure of treatment for the relapse episode. dStandard risk indicates FAB M1/2 with Auer rods, FAB M4 with atypical eosinophils (M4Eo), FAB M3 and/or favorable cytogenetics, such as t(8;21) and/or AML1-ETO, t(15;17), and/or PML-RARA and inv(16) or t(16;16) and/or CBFB/MYH1, if there was no persistence of BM blasts (≥5%) on day 15. FLT3-ITD positivity was not considered. All others were classified as high-risk patients.

122

haematologica | 2019; 104(1)

Gemtuzumab in pediatric acute myeloid leukemia

daunorubicin and fludarabine in addition to cytarabine with or without granulocyte colony stimulating factor (FLA/G+DX). Fourteen (18%) patients received HSCT prior to treatment with GO (Table 1). GO was administered after failure of treatment attempts at different time points of therapy and most (n=41, 54%) of the patients were treated with GO after failure of treatment(s) for firstearly relapse of de novo AML (de novo AML, refractory to

first early relapse) (Table 1). GO was equally prescribed as either monotherapy (n=36, 47%) or in combination with cytarabine (n=36, 47%), and the remaining patients (n=4, 6%) received GO in combination with other agents (Table 2). Most patients (n=48, 63%) received one cycle of GO as monotherapy or combination therapy (Table 2). GO was frequently administered at doses of 3 mg/m² or lower (n=37, 49%) (Table 2) and most of these patients received

Table 2. Details and outcomes of treatment with gemtuzumab ozogamicin.

Refractory de novo AML

De novo AML, refractory to 1st early relapse

De novo AML, refractory to 1st late relapse

De novo AML, refractory to ≥2nd relapses/secondary AML

Total cohort

10 (100)

41 (100)

10 (100)

15 (100)

76 (100)

10 (2 months -14) 10.3 (7 months -15)

7.5 (4 months -17) 8.1 (1 - 18)

12.6 (8 months - 16) 15.7 (2 - 20)

10.3 (1 month -16) 12.2 (9 months -16)

9.3 (1 month -17) 10.4 (7 months - 20)

4 (40) 6 (60) 0 (0)

17 (41) 20 (49) 4 (10)

3 (30) 7 (70) 0 (0)

12 (80) 3 (20) 0 (0)

36 (47) 36 (47) 4 (6)

6 (60) 4 (40) 0 (0) 0 (0)

27 (66) 10 (24) 3 (8) 1 (2)

7 (70) 3 (30) 0 (0) 0 (0)

8 (53) 7 (47) 0 (0) 0 (0)

48 (63) 24 (32) 3 (4) 1 (1)

6 (60) 2 (20) 2 (20) 0 (0) 0 (0)

20 (49) 5 (12) 10 (24) 4 (10) 2 (5)

7 (70) 0 (0) 1 (10) 2 (20) 0 (0)

4 (27) 1 (7) 8 (53) 2 (13) 0 (0)

37 (49) 8 (10) 21 (28) 8 (10) 2 (3)

8 (80) 8 0 2 (20)

28 (68) 27 1 13 (32)

8 (80) 8 0 2 (20)

5 (33) 3 2 10 (67)

49 (64) 46 3 27 (36)

2 (25) 6 (75) 0 (0) 0 (0)

9 (32) 16 (57) 1 (4) 2 (7)

2 (25) 5 (63) 1 (12) 0 (0)

1 (20) 2 (40) 1 (20) 1 (20)

14 (28) 29 (60) 3 (6) 3 (6)

2 (25) 2 (25) 3 (38) 1 (12)

1 (4) 13 (46) 12 (43) 2 (7)

1 (12) 2 (25) 3 (37) 2 (26)

1 (20) 3 (60) 1 (20) 0 (0)

5 (10) 20 (41) 19 (39) 5 (10)

3 (30) 7 (70)

5 (12) 36 (88)

5 (50) 5 (50)

2 (13) 13 (87)

15 (20) 61 (80)

Disease status prior to GO Number (%) Age, median (range) at initial diagnosis (years) at treatment with GO (years)

Treatment with GO monotherapy GO + cytarabine other Number of cycles one two three four GO dosage at the first cycle ≤ 3 mg/m² 3.1 – 5.4 mg/m² 5.5 - 7.5 mg/m² > 7.5 mg/m² n.d. HSCT after GO treatment Yes 1st HSCT 2nd HSCT No Extra treatment prior to HSCTa Chemotherapy No-treatment Other n.d. Bone marrow response prior to HSCTa CR CRi NR n.d. Patient status Alive Deceased

GO: gemtuzumab ozogamicin; n.d.: no data; HSCT: hematopoietic stem cell transplantation; CR: complete remission; CRi: complete remission with incomplete hematologic recovery; NR: non-response. aPercentages in this category were calculated only for patients who received HSCT.

haematologica | 2019; 104(1)

123

N. Niktoreh et al.

GO in combination with cytarabine or other agents following the failure of treatment after first early relapse (n=20 of 37, 54%) (data not shown). When GO was administered as monotherapy, higher doses were used (Figure 1).

Safety and toxicities Among 71 patients who were evaluated for AEs during the first cycle of treatment, most common AEs were grade 3/4 infections or febrile neutropenia (61 events in 49 patients, 69%) (Table 3). Two patients suffered from sepsis and were successfully managed with supportive and anti-bacterial therapy. Gastrointestinal (GI) symptoms (11 grade 1/2 events in 11 patients and 4 grade 3/4 events in 4 patients) and immunologic reactions such as infusionrelated fever, chills, or hypotension (7 grade-1/2 events in 7 patients and 5 grade 3/4 events in 5 patients) were most frequently observed after infections. Due to interference with other treatments, GI events could not be assigned to GO with certainty. Among grade 3 and 4 AEs, febrile neutropenia was most frequently observed in both groups of patients with monotherapy and combination therapy as well as in all different groups of patients independently of their previous treatments (FLA/G or FLA/G+DX or patients with HSCT prior to GO) (data not shown). The majority (4 of 5) of patients with severe infusion-related immunological reactions to GO treatment, had previously received GO as monotherapy with higher doses compared to patients with combination therapy (data not shown). Veno-occlusive disease (VOD) occurred in 3 patients who received GO as monotherapy (one cycle) at doses of 6, 7.5, and 9 mg/m2. All 3 patients had FLA/G+DX as treatment before GO, and received defibrotide as prophylaxis against VOD. Two of these patients were treated with HSCT prior to treatment with GO (6 or 7.5 mg/m2). The patient treated with 6 mg/m2 GO had a previous history of VOD and developed a GO-related VOD before the scheduled HSCT that was successfully treated without late effects or events. The other 2 patients could not receive HSCT after treatment with GO and both died. Cause of death in one of them was disease (leukemic) pro-

gression, and the other patient experienced exacerbation of a pre-existing cardiomyopathy leading to death 24 days after GO treatment. In addition, treatment with GO resulted in exacerbation of previous symptoms in 2 patients (pulmonary aspergillosis infection or gastrointestinal toxicity) and, due to overlap with other treatments and HSCT, the consequences of these AEs could not be exclusively correlated to GO (Table 3). Similarly, the consequences of different AEs in one patient with GO-related cytokine syndrome and pre-existing respiratory distress could not be distinguished from each other (Table 3). The respiratory distress in this patient was related to an aplasia-associated pneumonia which was caused by teatments before GO.

Outcome With a median follow up time of 4.3 years (range: 1-5 years), the probability of 4-year OS after treatment with

Figure 1. Gemtuzumab ozogamicin (GO) dosage. GO dosage based on administration with cytarabine or other agents. AML: acute myeloid leukemia; n.d. : no data.

124

Figure 2. Survival after gemtuzumab ozogamicin (GO). (A) 4-year probability of overall survival (OS) in all patients. (B) 4-year probability of OS in different disease statuses. (C) 4-year probability of OS based on administration of hematopoietic stem cell transplantation (HSCT). pOS: probablity of OS; n: number of patients; AML: acute myeloid leukemia; SE: Standard Error.

haematologica | 2019; 104(1)

Gemtuzumab in pediatric acute myeloid leukemia

GO was 18±5% in the total cohort (Figure 2A). A comparison of disease status prior to treatment with GO between patient groups showed that patients who received GO after de novo AML refractory to first late relapse had a probability of 4-year OS of 48±16%; this is significantly higher than in patients with de novo AML refractory to first early relapse (probability of 4-year OS: 12±5%; P=0.03) and patients with de novo AML refractory to second or more relapses/secondary AML (probability of 4-year OS: 10±9%; P=0.02) (Figure 2B). Based on the current retrospective analysis, factors such as number of cycles and administration of GO as monotherapy or in combination with cytarabine or other agents did not have any significant influence on survival (Online Supplementary Figure S3A-C). In the total cohort of patients, 27 (36%) did not receive HSCT after treatment with GO (Table 2) and none of these patients survived (Figure 2C). Of note, most (10 of 15, 67%) patients with de novo AML refractory to second or more relapses/secondary AML, did not receive an HSCT after GO treatment; this is higher than in other groups of patients (Table 2). From the total group of 76 patients, 49 (64%) received HSCT after GO treatment and their probability of 4-year OS was 27±7%. Thirty (61%) patients with HSCT received GO in combination with cytarabine (data not shown). HSCT after GO was the first HSCT for most of the patients (n=46, 94%), and mean time to transplantation after GO treatment was 41±30 days (range 11-135 days). The probability of 4year OS in patients who received HSCT early during the first three weeks after GO treatment was 9±9% and patients who received HSCT between three and six weeks after GO administration had a 4-year OS of 40±11% (P=0.06) (Online Supplementary Figure S3D). Previous treatment with or without FLA/G+DX before GO administration had no influence on the percentage of

HSCT achieved after treatment with GO (FLA/G+DX group: 29 of 43, 67%; no FLA/G+DX or FL/A group: 14 of 21, 67%) (data not shown). However, among transplanted patients, fewer patients with previous FLA/G+DX (7 of 29, 24%) survived compared to the patients without FLA/G+DX or FL/A (6 of 14, 42%) (data not shown). In addition, one patient with de novo AML-M3 refractory to first late relapse is alive after treatment with GO followed by additional chemotherapy and HSCT. Among 49 patients with subsequent HSCT, 25 (51%) reached CR/CRi before HSCT (Table 2). Out of 19 patients with no response to GO before HSCT, 8 (42%) received additional chemotherapy before HSCT and 3 of them survived. The remaining 11 (58%) patients with no response, received HSCT with no further chemotherapy and 2 of these patients survived (data not shown).

Discussion The major goal in the treatment of relapsed/refractory pediatric AML is to develop new therapeutic options to achieve complete remission and proceed to HSCT thereafter.26 Considering the previous history of intensive chemotherapy and/or previous HSCT in these patients, novel treatment options should be efficient, but must also have acceptable toxicity profiles. However, the introduction of new drugs is challenging. Due to increasing regulatory challenges and requirements, as well as limited accessibility of drugs such as GO, it is more and more time consuming to open phase III trials. Hence a follow-up trial, that has been planned since 2009 after the end of the phase II trial,18 was not opened until September 2016 and recruited the first patient in August 2017 (EudraCT n.: 2010-018980-41). In the current analysis, we have evaluated the use of GO

Table 3. Non-hematologic adverse events in 71 patients after the first treatment cycle with gemtuzumab ozogamicin.

AE severity/ consequence

Gastrointestinal Pain Infection or fever in neutropenia VOD Infusion-related immunological reactions Exacerbation of the previous conditions Constitutional symptoms Metabolic/laboratory Total, nf (%)

All gradesa Total AEs,

Grades 1/2 reversible

All outcomes n (%)

AEs Patients

AEs Patients AEs Patients AEs

15 (15) 4 (4) 61 (60)

11 1 0

4 3 58

4 3 46

0 0 2

0 0 1b

0 0 1

0 0 0

3 (3) 12 (12)

0 7

1 4

1 0

0 0

1c 0

1 0

0 1d

0 1

2 (2)

2 (2) 2 (2) 101 (100)

2 2

0 0

0 0 70 (69)

0 0

0 0 0 0 1 (1)

0 0

23 (23)

reversible

Grades 3 and 4 prolonged progressive reversible

3 (3)

Patients

1 (1)

fatal

n.a.

AEs Patients

3 (3)

AE: adverse events; n: number; n.a.: not applicable; VOD: veno-occlusive disease. aGrading of toxicity and adverse events based on the NCI Common Terminology Criteria for Adverse Events (CTCAE), version 4.03, revised June 14, 2010. bProgressive fever partly resulted from limiting its treatment due to joint decision to terminate the “active” therapy of the patient. cGO-related VOD led to exacerbation of a pre-existing cardiomyopathy causing death 24 days after the treatment with GO. dTreatment with GO resulted in occurrence of cytokine syndrome in a patient with pre-existing respiratory distress due to aplasia-associated pneumonia. Following a decrease in level of consciousness, this patient died five days after occurrence of the cytokine syndrome. eTreatment with GO resulted in worsening of previously present pulmonary aspergillosis infection or in worsening of previously observed gastrointestinal toxicity. fAll adverse events in 71 patients were evaluated.

haematologica | 2019; 104(1)

125

N. Niktoreh et al.

on a compassionate use basis in a cohort of children with refractory de novo AML and de novo or secondary AML refractory to relapse therapy. The use of GO in heavily pre-treated patients translated into a substantial rate of subsequent HSCT of 64% and a probability of 4-year OS of 27% in these patients. The safety profile in our cohort was tolerable, which may allow the use of GO in patients with a history of intensive therapies. We have identified 88 and analyzed 76 children with relapsed/refractory AML who were treated with GO on a compassionate use basis. Compared to other studies which evaluated the outcomes of treatment with GO in patients with relapsed/refractory AML either on a compassionate use basis (Zwaan et al., 15 patients;17 Brethon et al., 17 patients)19 or with other treatment strategies (29 patients with relapsed/refractory AML in an open-label, dose-escalation study;27 45 children with relapsed/refractory AML included in the AAML00P2 randomized clinical trial;28 30 children with advanced relapsed/refractory AML in an investigator-initiated phase II study)18 the current cohort contains the largest number of patients. With GO doses of 2.5-10 mg/m2, 64% of the patients included in our cohort subsequently received HSCT, and 51% of these patients reached complete remission with or without hematologic recovery before receiving HSCT. In 2005, Arceci et al. studied the effect of treatment with GO as a single agent with doses of 6-9 mg/m2 (2 doses, 2-week intervals) in 10 children with refractory AML and 19 with relapsed AML.27 They defined response to GO as achieving CR (presence of ≤5% blasts in BM with full hematologic recovery: hemoglobin level ≥9 g/dL, absolute neutrophil count ≥1.5x109/L, and platelet count ≥100x109/L) and showed a response rate of 30% and 26% in patients with refractory and relapsed AML, respectively, accompanied by acceptable safety profiles.27 In addition, in a study by Brethon et al., in 2006, outcome of monotherapy with GO on a compassionate basis (single dose: 3-9 mg/m2) in children with refractory (3 patients) and relapsed (9 patients) AML was investigated.29 The results of this study showed a CR (plus CR without hematologic recovery) rate of 25% in patients with subsequent HSCT.29 Of note, a subsequent study by the same group in 2008 showed that the response rate was higher (overall CR: 53%) when GO was administered in combination with cytarabine in children with relapsed or refractory AML.19 We could not confirm the benefit of this combination treatment in our cohort since combination of GO with cytarabine alone, or with cytarabine along with other agents or combination of GO with vincristine had no survival advantage compared to single agent therapy. Myelosuppression accompanied by fever was the most

References 1. Zwaan CM, Kolb EA, Reinhardt D, et al. Collaborative Efforts Driving Progress in Pediatric Acute Myeloid Leukemia. J Clin Oncol. 2015;33(27):2949-2962. 2. Kaspers G. How I treat paediatric relapsed acute myeloid leukaemia. Br J Haematol. 2014;166(5):636-645. 3. Sander A, Zimmermann M, Dworzak M, et al. Consequent and intensified relapse therapy improved survival in pediatric AML:

126

common GO-related AE in our patients (49 events in 58% of patients) (data not shown) that appeared to be independent of GO dosage, combination/monotherapy, previous treatments, or disease status prior to GO. Rates of myelosuppression by GO were previously reported at a range of 25-100% in different studies.19,28,29 In our cohort, the rate of VOD after treatment with GO was 4%. Frequency of VOD related to monotherapy or combination GO therapy in children has been previously reported with a wide range across different studies from no events29,30 to 10-24%.27,28,31 However, taken together, the results of previous studies show that the incidence of VOD is directly associated with the absolute single dose of GO, which is higher when it is administered as monotherapy.32 These findings are in accordance with the results of the current cohort. Considering the suggested increased risk of VOD in cases of monotherapy with GO, and the lack of a survival advantage between monotherapy and combination therapy in our current cohort, administration of GO in combination with other agents, such as cytarabine, may be useful to prevent this complication. It should be noted that the current analysis is limited by its retrospective study design and especially by the lack of a control group. These restrictions have hindered our efforts to identify the contribution of compassionate treatment with GO as a single factor to the outcomes of the current study. In conclusion, we have shown in a large cohort of patients with relapsed/refractory de novo and secondary AML, with a history of very intensive treatments including chemotherapy and/or HSCT, that administration of GO on a compassionate use basis was frequently considered. The study provides evidence that GO can enable a subsequent blast reduction that allowed HSCT in these patients and survival without imposing major adverse events. Since addition of GO showed the potential to improve treatment outcomes in the current cohort of patients with relapsed/refractory pediatric AML, the role of GO in these patient groups should ideally be proven in large prospective randomized clinical trials. Therefore, we have now included GO as front-line treatment of relapsed/refractory pediatric AML in our ongoing phase III multicenter clinical trial (EudraCT number: 2010-01898041, recruiting since August 2017). Acknowledgments We are grateful to all patients who recruited in the corresponding AML-BFM trials and physicians and study staff involved in patient care and studies maintenance. We thank Mr. Jans Enno Mueller for his valuable contribution to the study data-base. Supplementary information is available at Hematologica’s website.

results of relapse treatment in 379 patients of three consecutive AML-BFM trials. Leukemia. 2010;24(8):1422-1428. 4. Horton TM, Perentesis JP, Gamis AS, et al. A Phase 2 study of bortezomib combined with either idarubicin/cytarabine or cytarabine/etoposide in children with relapsed, refractory or secondary acute myeloid leukemia: a report from the Children's Oncology Group. Pediatr Blood Cancer. 2014;61(10):1754-1760. 5. Nakayama H, Tabuchi K, Tawa A, et al.

Outcome of children with relapsed acute myeloid leukemia following initial therapy under the AML99 protocol. Int J Hematol. 2014;100(2):171-179. 6. Cooper TM, Alonzo TA, Gerbing RB, et al. AAML0523: a report from the Children's Oncology Group on the efficacy of clofarabine in combination with cytarabine in pediatric patients with recurrent acute myeloid leukemia. Cancer. 2014; 120(16):2482-2489. 7. Kaspers GJ, Zimmermann M, Reinhardt D,

haematologica | 2019; 104(1)

Gemtuzumab in pediatric acute myeloid leukemia

9. 10.

11.

12.

13. 14.

15.

16.

et al. Improved outcome in pediatric relapsed acute myeloid leukemia: results of a randomized trial on liposomal daunorubicin by the International BFM Study Group. J Clin Oncol. 2013;31(5):599-607. Gorman MF, Ji L, Ko RH, et al. Outcome for children treated for relapsed or refractory acute myelogenous leukemia (rAML): a Therapeutic Advances in Childhood Leukemia (TACL) Consortium study. Pediatr Blood Cancer. 2010;55(3):421-429. Webb DK. Management of relapsed acute myeloid leukaemia. Br J Haematol. 1999;106(4):851-859. Rubnitz JE, Razzouk BI, Lensing S, Pounds S, Pui CH, Ribeiro RC. Prognostic factors and outcome of recurrence in childhood acute myeloid leukemia. Cancer. 2007;109(1):157-163. Bose P, Vachhani P, Cortes JE. Treatment of Relapsed/Refractory Acute Myeloid Leukemia. Curr Treat Options Oncol. 2017;18(3):17. Moore AS, Kearns PR, Knapper S, Pearson AD, Zwaan CM. Novel therapies for children with acute myeloid leukaemia. Leukemia. 2013;27(7):1451-1460. Tasian SK, Pollard JA, Aplenc R. Molecular therapeutic approaches for pediatric acute myeloid leukemia. Front Oncol. 2014;4:55. Hinman LM, Hamann PR, Wallace R, Menendez AT, Durr FE, Upeslacis J. Preparation and characterization of monoclonal antibody conjugates of the calicheamicins: a novel and potent family of antitumor antibiotics. Cancer Res. 1993;53(14):3336-3342. Sievers EL, Appelbaum FR, Spielberger RT, et al. Selective ablation of acute myeloid leukemia using antibody-targeted chemotherapy: a phase I study of an antiCD33 calicheamicin immunoconjugate. Blood. 1999;93(11):3678-3684. Maniecki MB, Hasle H, Bendix K, Moller HJ. Is hepatotoxicity in patients treated with gemtuzumabozogamicin due to specific targeting of hepatocytes? Leuk Res. 2011;35(6):e84-e86.

haematologica | 2019; 104(1)

17. Zwaan CM, Reinhardt D, Corbacioglu S, et al. Gemtuzumab ozogamicin: first clinical experiences in children with relapsed/refractory acute myeloid leukemia treated on compassionate-use basis. Blood. 2003;101(10):3868-3871. 18. Zwaan CM, Reinhardt D, Zimmerman M, et al. Salvage treatment for children with refractory first or second relapse of acute myeloid leukaemia with gemtuzumab ozogamicin: results of a phase II study. Br J Haematol. 2010;148(5):768-776. 19. Brethon B, Yakouben K, Oudot C, et al. Efficacy of fractionated gemtuzumab ozogamicin combined with cytarabine in advanced childhood myeloid leukaemia. Br J Haematol. 2008;143(4):541-547. 20. ClinicalTrials.gov. Expanded Access/ Compassionate Use Protocol For Relapsed Or Refractory CD33 Positive AML Patients In The USA Without Access To Comparable Or Alternative Therapy (AML). 2017 [cited Jan 2018]; Available from: https://clinicaltrials.gov. 21. Creutzig U, Zimmermann M, Ritter J, et al. Treatment strategies and long-term results in paediatric patients treated in four consecutive AML-BFM trials. Leukemia. 2005;19(12):2030-2042. 22. Creutzig U, Zimmermann M, Lehrnbecher T, et al. Less toxicity by optimizing chemotherapy, but not by addition of granulocyte colony-stimulating factor in children and adolescents with acute myeloid leukemia: results of AML-BFM 98. J Clin Oncol. 2006;24(27):4499-4506. 23. Creutzig U, Zimmermann M, Dworzak M, et al. Study AML-BFM 2004: Improved Survival In Childhood Acute Myeloid Leukemia without Increased Toxicity. Blood. 2010;116(21):181. 24. Creutzig U, Zimmermann M, Dworzak MN, et al. The prognostic significance of early treatment response in pediatric relapsed acute myeloid leukemia: results of the international study Relapsed AML 2001/01. Haematologica. 2014;99(9):14721478.

25. US Department of Health and Human Services. Common Terminology Criteria for Adverse Events (CTCAE) Version 4.03. 2010 [cited August 2016]; Available from: http://ctep.cancer.gov. 26. Davila J, Slotkin E, Renaud T. Relapsed and refractory pediatric acute myeloid leukemia: current and emerging treatments. Paediatr Drugs. 2014;16(2):151-168. 27. Arceci RJ, Sande J, Lange B, et al. Safety and efficacy of gemtuzumab ozogamicin in pediatric patients with advanced CD33+ acute myeloid leukemia. Blood. 2005; 106(4):1183-1188. 28. Aplenc R, Alonzo TA, Gerbing RB, et al. Safety and efficacy of gemtuzumab ozogamicin in combination with chemotherapy for pediatric acute myeloid leukemia: a report from the Children's Oncology Group. J Clin Oncol. 2008;26(14):2390-3295. 29. Brethon B, Auvrignon A, Galambrun C, et al. Efficacy and tolerability of gemtuzumab ozogamicin (anti-CD33 monoclonal antibody, CMA-676, Mylotarg) in children with relapsed/refractory myeloid leukemia. BMC Cancer. 2006;6:172. 30. Roman E, Cooney E, Harrison L, et al. Preliminary results of the safety of immunotherapy with gemtuzumab ozogamicin following reduced intensity allogeneic stem cell transplant in children with CD33+ acute myeloid leukemia. Clin Cancer Res. 2005;11(19):7164-7170. 31. Satwani P, Bhatia M, Garvin JH, Jr., et al. A Phase I study of gemtuzumab ozogamicin (GO) in combination with busulfan and cyclophosphamide (Bu/Cy) and allogeneic stem cell transplantation in children with poor-risk CD33+ AML: a new targeted immunochemotherapy myeloablative conditioning (MAC) regimen. Biol Blood Marrow Transplant. 2012;18(2):324-329. 32. Parigger J, Zwaan CM, Reinhardt D, Kaspers GJ. Dose-related efficacy and toxicity of gemtuzumab ozogamicin in pediatric acute myeloid leukemia. Expert Rev Anticancer Ther. 2016;16(2):137-146.

127

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):128-137

Acute Lymphoblastic Leukemia

Clinical and molecular characteristics of MEF2D fusion-positive B-cell precursor acute lymphoblastic leukemia in childhood, including a novel translocation resulting in MEF2D-HNRNPH1 gene fusion

Kentaro Ohki,1 Nobutaka Kiyokawa,1 Yuya Saito,1,2 Shinsuke Hirabayashi,1,3 Kazuhiko Nakabayashi,4 Hitoshi Ichikawa,5 Yukihide Momozawa,6 Kohji Okamura,7 Ai Yoshimi,1,8 Hiroko Ogata-Kawata,4 Hiromi Sakamoto,5 Motohiro Kato,1 Keitaro Fukushima,9 Daisuke Hasegawa,3 Hiroko Fukushima,10 Masako Imai,11 Ryosuke Kajiwara,12 Takashi Koike,13 Isao Komori,14 Atsushi Matsui,15 Makiko Mori,16 Koichi Moriwaki,17 Yasushi Noguchi,18 Myoung-ja Park,19 Takahiro Ueda,20 Shohei Yamamoto,21 Koichi Matsuda,22 Teruhiko Yoshida,5 Kenji Matsumoto,23 Kenichiro Hata,4 Michiaki Kubo,6 Yoichi Matsubara,24 Hiroyuki Takahashi,25 Takashi Fukushima,26 Yasuhide Hayashi,27 Katsuyoshi Koh,16 Atsushi Manabe3 and Akira Ohara25 for the Tokyo Children’s Cancer Study Group (TCCSG)

Department of Pediatric Hematology and Oncology Research, National Research Institute for Child Health and Development, Setagaya-ku, Tokyo; 2Department of Hematology/Oncology, Tokyo Metropolitan Children’s Medical Center, Fuchu-shi; 3 Department of Pediatrics, St. Luke's International Hospital, Chuo-ku, Tokyo; 4 Department of Maternal-Fetal Biology, National Research Institute for Child Health and Development, Setagaya-ku, Tokyo; 5Fundamental Innovative Oncology Core, National Cancer Center Research Institute, Chuo-ku, Tokyo;6Laboratory for Genotyping Development, RIKEN Center for Integrative Medical Sciences (IMS), Yokohama-shi, Kanagawa; 7Department of Systems BioMedicine, National Research Institute for Child Health and Development, Setagaya-ku, Tokyo; 8Division of Pediatric Hematology and Oncology, Ibaraki Children’s Hospital, Mito-shi; 9Department of Pediatrics, Dokkyo Medical University, Tochigi; 10Department of Pediatrics, University of Tsukuba Hospital, Ibaraki; 11Department of Pediatrics, Japanese Red Cross Musashino Hospital, Tokyo; 12 Department of Pediatrics, Yokohama City University Hospital, Kanagawa; 13Department of Pediatrics, Tokai University School of Medicine, Kanagawa; 14Department of Pediatrics, Matsudo City Hospital, Chiba; 15Department of Pediatrics, Japanese Red Cross Maebashi Hospital, Gunma; 16Department of Hematology/Oncology, Saitama Children’s Medical Center; 17Department of Pediatrics, Saitama Medical Center, Saitama Medical University; 18Department of Pediatrics, Japanese Red Cross Narita Hospital, Chiba; 19Department of Hematology/Oncology, Gunma Children’s Medical Center, Shibukawa-shi; 20Department of Pediatrics, Nippon Medical School, Bunkyo-ku, Tokyo; 21Department of Pediatrics, Showa University Fujigaoka Hospital, Yokohama-shi, Kanagawa; 22Laboratory of Clinical Genome Sequencing Department of Computational Biology and Medical Sciences Graduate School of Frontier Sciences, The University of Tokyo, Minato-ku; 23Department of Allergy and Clinical Immunology, National Research Institute for Child Health and Development, Setagaya-ku, Tokyo; 24Director, National Research Institute for Child Health and Development, Setagaya-ku, Tokyo; 25Department of Pediatrics, Toho University Omori Medical Center, Tokyo; 26Department of Child Health, Faculty of Medicine, University of Tsukuba, Ibaraki and 27Institute of Physiology and Medicine, Jobu University, Takasaki-shi, Gunma, Japan 1

Correspondence: oki-kn@ncchd.go.jp

Received: December 11, 2017. Accepted: August 29, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2017.186320 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/128

ABSTRACT ©2019 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.

128

usion genes involving MEF2D have recently been identified in precursor B-cell acute lymphoblastic leukemia, mutually exclusive of the common risk stratifying genetic abnormalities, although their true incidence and associated clinical characteristics remain unknown. We identified 16 cases of acute lymphoblastic leukemia and 1 of lymphoma harboring MEF2D fusions, including MEF2D-BCL9 (n=10), MEF2D-HNRNPUL1 (n=6), and one novel MEF2D-HNRNPH1 fusion. The incidence of MEF2D fusions overall was 2.4% among consecutive precursor B-cell acute lymphoblastic leukemia patients enrolled onto a single clinical trial. They frequently showed a cytoplasmic m chain-positive pre-B immunophenotype, and often expressed an aberrant CD5 antigen. Besides up- and down-regulation of HDAC9 and MEF2C, elevated GATA3 expression was also a characteristic feature of MEF2D haematologica | 2019; 104(1)

Fusion genes involving MEF2D in B-ALL

fusion-positive patients. Mutations of PHF6, recurrent in T-cell acute lymphoblastic leukemia, also showed an unexpectedly high frequency (50%) in these patients. MEF2D fusion-positive patients were older (median age 9 years) with elevated WBC counts (median: 27,300/ml) at presentation and, as a result, were mostly classified as NCI high risk. Although they responded well to steroid treatment, MEF2D fusion-positive patients showed a significantly worse outcome, with 53.3% relapse and subsequent death. Stem cell transplantation was ineffective as salvage therapy. Interestingly, relapse was frequently associated with the presence of CDKN2A/CDKN2B gene deletions. Our observations indicate that MEF2D fusions comprise a distinct subgroup of precursor B-cell acute lymphoblastic leukemia with a characteristic immunophenotype and gene expression signature, associated with distinct clinical features.

Introduction Precursor B-cell acute lymphoblastic leukemia (B-ALL) is a heterogeneous disease characterized by a variety of genetic abnormalities.1 In approximately one-quarter of B-ALL patients, known as the B-other-subgroup, the known major risk-stratifying cytogenetic abnormalities are absent.2 However, recent studies using advanced analytical approaches have described a range of novel genetic subgroups among B-other-ALL.3-11 The myocyte enhancer factor 2D (MEF2D) gene, located at 1q22, is present among these newly identified rearrangements in B-other-ALL.6,7,10,12-14 Seven known fusion partners: B-cell CLL/lymphoma 9 (BCL9, 1q21), heterogeneous nuclear ribonucleoprotein U-like 1 (HNRNPUL1, 19q13.2), deleted in azoospermia-associated protein 1 (DAZAP1, 19p13.3), colony stimulating factor 1 receptor (CSF1R, 5q32), synovial sarcoma translocation, chromosome 18 (SS18, 18q11.2), signal transducer and activator of transcription 6 (STAT6, 12q13.3), and Forkhead Box J2 (FOXJ2, 12p13.31) have been described, mostly among childhood and young adult B-ALL. The MEF2D gene encodes a member of the transcription factor family involved in the control of muscle and neuronal cell differentiation and development, which is regulated by class II histone deacetylases.15-17 It has been reported that rearrangements result in enhanced MEF2D transcriptional activity and lymphoid transformation, thus contributing to the development of a distinct subtype of high-risk leukemia.7,10 However, the true incidence and clinical characteristics, including outcome, of patients with B-ALL harboring MEF2D fusion genes remains unknown. In this study, we report the detailed analysis of a subgroup of B-ALL with MEF2D fusions within a Japanese pediatric ALL cohort. The heterogeneous nuclear ribonucleoprotein H1 gene (HNRNPH1), encoding another family member of heterogeneous nuclear ribonucleoprotein,18 was identified as a new fusion partner of the MEF2D gene. Novel immunophenotypic characteristics and accompanying genetic abnormalities as well as distinctive clinical features of B-ALL harboring MEF2D fusions are evaluated and discussed.

Methods

TCCSG L04-16 study19 while others, including 2 B-lymphoblastic lymphoma (LBL) patients, originated from different cohorts. All investigations were approved by the institutional review boards and informed consent or assent was obtained from parents or guardians based on their age and level of understanding, as described previously.5,11 Online Supplementary Figure S1 shows the analysis carried out on each case. Total RNA and genomic DNA were extracted from bone marrow or peripheral blood of patients using the miRNeasy Mini Kit and the QIAamp DNA Mini Kit (Qiagen, Inc., Valencia, CA, USA), respectively. In this paper, B-other-ALL is defined as B-ALL lacking the major risk stratifying genetic abnormalities, including high hyperdiploidy (≥ 51 chromosomes or DNA index ≥ 1.16), low hypodiploidy/near haploidy (≤ 44 chromosomes), fusions of ETV6-RUNX1, TCF3-PBX1, TCF3-HLF, BCR-ABL1, and MLL rearrangements as well as more recently identified genetic abnormalities, including rearrangements of CRLF2 and ZNF384, Ph-like ALL-related tyrosine kinase fusions as well as MEF2D fusions (Online Supplementary Table S1).

Whole transcriptome sequencing and RT-PCR From previous whole transcriptome sequencing (WTS) of Bother-ALL, we identified cases with ZNF384 fusions11 as well as other abnormalities (Online Supplementary Figure S1, Table S1).1921 We re-analyzed remaining 153 WTS data manipulated by “deFuse”,22 an algorithm for gene fusion discovery, and investigated the presence of MEF2D fusions. Details of this data analysis have been described previously.11 RT-PCR followed by Sanger sequencing was performed to confirm and screen for fusion transcripts, as described previously,5,11 using the primers listed in Online Supplementary Table S2.

Multiplex Ligation-dependent Probe Amplification (MLPA) MLPA analyses were performed on genomic DNA using two types of SALSA Reference Kits, P335 and P383 (MRC Holland, Amsterdam, the Netherlands), according to the manufacturer's instructions. After separation of amplified products, using the ABI3130 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA), the results were analyzed using Gene Mapper Software (Applied Biosystems) and data, including informative headers, electropherograms, ratio plots, validation boxes, and report tables were obtained. In this study, we present only results of deletions of the exons targeted in these kits.

Patient selection and sample preparation RNA and DNA samples, obtained from pediatric B-ALL patients and stored in the Tokyo Children’s Cancer Study Group (TCCSG) biobank5,11 were used in this study (Online Supplementary Figure S1, Table S1). As indicated in Online Supplementary Table S1, the majority of cases originated from the haematologica | 2019; 104(1)

Whole exome sequencing (WES) Exome libraries prepared from 100 ng of genomic DNA were sequenced using SBS v.4 reagents with the HiSeq2500 sequencing system. Details of whole exome data analyses have been described previously.11 129

K. Ohki et al.

Microarray and data analyses

Statistical analysis

The gene expression signature for MEF2D fusion-positive BALL was investigated by DNA microarray-based expression profiling using Human Genome U133 Plus 2.0 Arrays (Affymetrix, Santa Clara, CA, USA). Data were normalized and filtered as described previously.11 Further details are provided in Online Supplementary Methods.

Mutual univariate analyses of characteristics were conducted using Fisherâ&#x20AC;&#x2122;s exact test or the c2 test for qualitative variables. Overall survival (OS) and event-free survival (EFS) were estimated by the Kaplan-Meier method and compared by the log-rank test. Analyses were performed using Prism software, version 6.0 (GraphPad Software, Inc., La Jolla, CA, USA).

Table 1. Characteristics of MEF2D fusion-positive cases.

Case Fusion partner

Identified Age Sex Diagnosis Initial Karyotype (years) WBC (/ L)

BCL9

RT-PCR

BCL9

RT-PCR

BCL9

WTS

BCL9

Yes (BM) 1.0

WTS/RT-PCR 15

B-ALL 63,200 46,XX[20]

Yes (CNS) 1.1

WTS/RT-PCR 10

B-ALL 124,100 NA

240 Yes (BM) 2.0

RT-PCR

B-ALL 77,300 45,XY,t(1;2)(q21;p13),-9, i(9)(q10),add(16)(q13)[20] B-ALL 41,400 NA

30 Yes (CNS) 0.8

RT-PCR

B-ALL 46,200 NA

WTS/RT-PCR 10

B-ALL

5,400

WTS/RT-PCR 7

B-LBL

Stage IVNA

11 HNRNPUL1 WTS/RT-PCR 9

B-ALL

5,300

12 HNRNPUL1

B-ALL

8,200

13 HNRNPUL1 WTS/RT-PCR 5

B-ALL 27,300 46,XX[18]

736 Yes (BM) 1.3

14 HNRNPUL1 WTS/RT-PCR 15

B-ALL 6,200 fail

15 HNRNPUL1 WTS/RT-PCR 10

B-ALL 21,900 47,XX,t(3;9)

16 HNRNPUL1

B-ALL 10,800 46,XY,?ins(9)(q13p22p24)[20]

HR 1,134

17 HNRNPH1 WTS/RT-PCR 7

B-ALL 76,100 46,XX,t(1;5)(q21;q35) [8]/47,idem,+8[8]/46,XX[2]

RT-PCR

3,400

46,XX[4]

46,XX,-X,add(3)(q27), IR add(9)(q13),add(9)(q13), add(10)(q22),-12,del(12)(q13), del(13)(q12q14) -20,+22,+2mar[ cp3]/46,XX[2] B-ALL 29,400 46,XX[20] IR

RT-PCR

B-ALL

Initial Day8 Relapse Relapse Salvage FUP Current Samples Fusion risk blasts (site) date therapy (years) status obtained points (NCI) (/ L) (years) after at relapse

Yes (BM) 1.6

750 Yes (BM) 1.9

Yes (BM) 1.2

10.9+

1st CR

Chemo

1.2

Dead

Newly Ex4-4bp* diagnosed 1 ins-Ex9

Newly Ex6-Ex10 diagnosed SCT 2.0 Dead Newly Ex9-Ex9, diagnosed Ex5-Ex9 9.1+ 1st CR Newly Ex6-Ex10 diagnosed SCT 2.6 Dead Newly Ex5-Ex9 diagnosed Chemo 1.9 Dead Newly Ex6-Ex10 diagnosed SCT 4.2 Dead Newly Ex5-Ex10, diagnosed, Ex6-Ex10 1st relapse NA NA NA Newly Ex5-Ex9, diagnosed Ex6-Ex9*2 13.1+ 1st CR Newly Ex5-Ex9 diagnosed 5.4+ 1st CR Newly Ex6-Ex10 diagnosed 8.6+ 1st CR Newly Ex9-Ex12 diagnosed 9.9+ 1st CR Newly Ex9-Ex12 diagnosed SCT 1.9 Dead Newly Ex9-Ex12 diagnosed 4.5+ 1st CR Newly Ex9-Ex12 diagnosed Chemo, SCT 2.4 Dead Newly Ex9-Ex12 diagnosed 7.5+ 1st CR Newly Ex9-Ex12 diagnosed Chemo 1.8 Dead Newly Ex4-Ex5 diagnosedEx7-21bp*4 ins-Ex5

RT-PCR: Reverse Transcription Polymerase Chain Reaction; WTS: Whole Transcriptome Sequencing; WBC: white blood cells; CNS: central nervous system; FUP: follow up; B-ALL: precursor Bcell acute lymphoblastic leukemia; B-LBL: precursor B-cell lymphoblastic lymphoma; NCI: National Cancer Institute; SR: standard risk; IR: intermediate risk; HR: high risk; Ex: exon; BM: bone marrow; CR: complete remission; NA: data not available; SCT: stem-cell transplantation; Chemo: chemotherapy. *1: (TGTC). *2: Ex6-Ex9 fusion was detected only by WTS. *3: The PAX5-FOXP1 translocation was not detected by whole transcriptome sequencing. *4: (CCCGACCGACTTGTGTTCCGC).

130

haematologica | 2019; 104(1)

Fusion genes involving MEF2D in B-ALL

Results Detection of MEF2D fusions in pediatric B-ALL patients Among the 328 selected RNA samples from B-ALL patients (Online Supplementary Figure S1) analyzed by WTS and/or RT-PCR followed by Sanger sequencing, we iden-

tified 9 and 6 patients with MEF2D-BCL9 and MEF2DHNRNPUL1, respectively (Table 1, Online Supplementary Table S1 and S3, Figure 1 and Online Supplementary Figure S1). Of note, one case of each abnormality was identified by gene expression profiling (details in Online Supplementary Information). We also identified an additional case with MEF2D-BCL9 (Table 1, Case 10) in B-LBL. As

Figure 1. Structures of the MEF2D fusions. Structures of fusion proteins and nucleotide sequences of (a) - (e) MEF2D-BCL9, (f) MEF2D-HNRNPUL1, (g) and (h) MEF2D-HNRNPH1. The red arrowhead shows the donor and acceptor site breakpoint. The number of patients in whom a particular fusion isoform was found is indicated on the right.

haematologica | 2019; 104(1)

131

K. Ohki et al.

well as the known MEF2D fusions, we identified MEF2DHNRNPH1 as a novel fusion in 1 patient (Figure 1, Table 1, Case 17). Among the L04-16/L06-16 cohort,11,19 comprising a consecutive series of 290 B-ALL patients, including 126 classified as B-other-ALL, 5 MEF2D-BCL9 and 2 MEF2DHNRNPUL1 patients were identified (Online Supplementary Table S1 and Figure S1). The incidence of MEF2D fusions in childhood ALL, calculated from this

cohort, was 5.6% in B-other-ALL and 2.4% in B-ALL overall. MEF2D-BCL9 was the most recurrent, at a frequency of 4.0% in B-other-ALL and 1.7% in B-ALL overall.

Structure of MEF2D fusions The structure and sequences of MEF2D-BCL9 as well as a schematic representation of the predicted fusion proteins, are depicted in Figure 1, Table 1 and Online

P<0.01 P<0.05

Figure 2. Immunophenotypic characteristics of B-ALL patients with MEF2D fusions. (A) The positivity (percentage) of CD10, cytoplasmic m, CD5, CD38, and CD33 of MEF2D fusion-positive, TCF3-PBX1-positive, and B-other patients are plotted as a scattergram. A detailed list of positivity for each immunophenotypic marker of the patients is presented in Online Supplementary Table S4. (B) Typical histograms of CD10, CD19, cytoplasmic m, aberrant CD5, CD33, and CD38 of MEF2D-BCL9, MEF2D-HNRNPUL1, TCF3-PBX1, and B-other patients are indicated with a positive rate (%). X-axis, fluorescence intensity; Y-axis, relative cell number.

132

haematologica | 2019; 104(1)

Fusion genes involving MEF2D in B-ALL

Supplementary Table S3. Five isoforms of MEF2D-BCL9 fusion were identified in 10 patients, including one case in which different isoforms were present together. Most frequently, exon 6 of MEF2D was fused in-frame to exon 10 of BCL9 in 5 patients (Figure 2A, Type (a)), in accordance with previous reports.10 Exons 5 and 6 of MEF2D were also fused to exon 9 of BCL9 in 4 and 1 patient, respectively (Types (b) and (c)). In addition, we identified two novel breakpoints (Types (d) and (e)). The predicted protein from all fusions retains the DNA-binding MADS domain of the MEF2D protein, while it lacks most of the c-terminal portion of MEF2D as well as most of the functional domains of BCL9. The structure and sequences of MEF2D-HNRNPUL1 are also presented in Figure 1. Distinct from previous reports, only one isoform joining exon 9 of MEF2D to exon 12 of HNPNPUL1 (Type (f)) was observed among our 6 patients. In the case of MEF2D-HNRNPH1, two isoforms, in-frame fusions joining exons 4 and 7 of MEF2D to exon 5 of HNPNPH1 (Type (g) and (h)), respectively, were identified in the same patient.

Immunophenotypic characteristics of B-ALL patients with MEF2D fusions It has been reported that B-ALL patients with MEF2D fusions show dull or negative expression of CD10 and overexpression of CD38 antigens, based on immunophenotypic examination.10 Among our MEF2D fusion-positive cases, CD10 expression was lower than that seen in TCF3-PBX1-positive patients, but not B-others-ALL, as shown in Figure 2, Online Supplementary Figure S3 and Table S4. On the other hand, CD38 expression in MEF2D fusion-positive cases was higher than B-other-ALL patients, but not TCF3-PBX1-positive patients. We also observed that expression of the cytoplasmic m chain was higher in MEF2D fusion-positive cases than that in B-

other-ALL patients, with more than 10% of blasts positive for cytoplasmic m in 8/12 patients (ranging from 5.50 to 99.74%, mean: 54.14 ± 41.38%). Moreover, B-ALL patients with MEF2D fusions frequently exhibited aberrant CD5 expression, with more than 10% of the blasts positive for CD5 in 7/12 patients (ranging from 0.60 to 73.30%, mean: 27.01 ± 26.67%).

Additional genetic abnormalities in MEF2D fusion-positive patients To elucidate the frequency of additional genetic abnormalities in B-ALL with MEF2D fusions, we initially performed MLPA on 16 DNA samples from patients with MEF2D fusions. As shown in Online Supplementary Table S5, MLPA analysis revealed that deleted or amplified regions of IKZF1, CRLF2, EBF1, BTG1, PHF6, NF1, EZH2, SUZ12, or PTEN were absent from MEF2D fusion-positive cases, although 11/16 exhibited CDKN2A/CDKN2B deletions at a frequency significantly higher than that among the B-other-ALL patients enrolled on L04-16/L0616 study (P<0.001, data not shown). Heterozygous deletions of LEF1, PAX5, and ETV6 were detected in 1 case each. To further investigate the presence of additional genetic abnormalities affecting coding sequences in B-ALL with MEF2D fusions, we carried out WES on 3 DNA samples. We identified 16 mutations within genes that had been previously recognized within cancer: ALK, ARFGER3, BRCA1, BRMS1, C8orf4, ITIH1, MAPK13, NCOR2, NLE, NOTCH1, PHF3, PHF6, PHF10, PIK3R5, RB1, and TET1 (Online Supplementary Table S6). As mutations involving NOTCH1 and PHF6 are known to be recurrent in T-ALL23-26 and NLE encodes the modulator of NOTCH1,27 we were encouraged to investigate abnormalities in NOTCH1 signaling pathway genes and other genes reported as targets of recurrent genetic abnormalities in T-ALL. Thus, we performed RT-PCR and

Table 2. Additional genetic abnormalities of MEF2D fusion-positive cases.

Case Fusion partner 1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17

Gene Chromosome

FBXW7 4q31.3

NLE 17q12

NOTCH1 9q34.3

PHF6 Xq26.2

PTEN 10q23.31

WT1 11p13

BCL9 BCL9 BCL9 BCL9 BCL9 BCL9 BCL9 BCL9 BCL9 HNRNPUL1 HNRNPUL1 HNRNPUL1 HNRNPUL1 HNRNPUL1 HNRNPUL1 HNRNPH1

WT WT 6764 T C, M2255T WT WT WT WT WT WT WT WT WT 1282 C G, Q428E WT WT R15 fs WT WT WT WT 6920 A G, Q2307R, S66C WT WT WT WT WT WT WT WT WT WT WT K16I 692 C→T, P231L WT WT WT WT WT WT 1192 T→A, C398S WT WT WT WT WT WT WT WT WT D309H WT WT WT WT WT ex2 del (no start codon), C82Y WT WT WT WT WT ex2 del (no start codon) WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT WT D309H WT WT WT 904 C G, R302G WT 955 C→T, R319X WT WT Frequency 1/16 1/16 2/16 8/16 1/16 1/16 ex: exon; WT: wild-type; fs: frame shift; del: deletion. haematologica | 2019; 104(1)

133

K. Ohki et al.

Sanger sequencing of selected genes, including FBXW7, NLE, NOTCH1, PHF6, and PTEN, and combined these data with the results of WES. As shown in Table 2, mutations in NOTCH1 signaling pathway genes, including FBXW7, NLE, and NOTCH1, were detected in 4/16 patients. In addition, 8 and 1 of 16 patients exhibited mutations in PHF6 and PTEN, respectively.

Gene expression signature of MEF2D fusion-positive patients We have microarray data of B-ALL as indicated in Online Supplementary Figure S1,11 and, in addition, we performed WTS to search for new fusion genes specifically among cases of B-other-ALL. To further assess the func-

tional aspects of MEF2D fusions, we performed DNA microarray-based expression profiling. Upon supervised hierarchical clustering analysis using selected gene probe sets (Online Supplementary Table S7 and S8), we observed distinct clustering of MEF2D fusion cases, with clear separation from cases with other genetic abnormalities, indicating that MEF2D fusion cases have a distinct gene expression signature (Figure 3A). Of note was that this cluster of MEF2D fusion cases was close to that of TCF3PBX1-positive cases. Principal component analysis (PCA) also revealed clear clustering of MEF2D fusion cases separate from those cases with other genetic abnormalities, but close to the cluster of TCF3-PBX1-positive cases (Figure 3B).

Figure 3. Characteristics of gene expression profile in MEF2D-fusion-positive ALL. (A) Two-way hierarchical clustering and (B) principal component analysis (PCA) were performed on the microarray data, including B-ALLs with MEF2D fusion-positive and other types of genetic abnormalities, using the probe sets of differentially expressed genes between B-ALL with MEF2D fusions and other types of genetic abnormalities selected from filtered microarray probes (presented in Online Supplementary Tables S7 and S8). The results of clustering analysis are displayed using a heat map as a dendrogram.

134

haematologica | 2019; 104(1)

Fusion genes involving MEF2D in B-ALL

As TCF3-PBX1-positive cases also express the cytoplasmic m chain-positive pre-B ALL immunophenotype,28,29 we subsequently compared the gene expression of MEF2D fusion cases with TCF3-PBX1-positive cases as well as Bother-ALL. As shown in Online Supplementary Figure S4, MEF2D fusion patients were clearly separate, within a distinct cluster from TCF3-PBX1-positive and B-other-ALL cases, based on unsupervised hierarchical clustering. Upregulated genes common to both MEF2D fusion and TCF3-PBX1-positive cases compared to B-other-ALL included: IKZF2, IRF4, TCL6, IGHG, IGHV5-78, IGLL1, VPREB3, and BCL2L11, while RUNX2, IRF9, CRLF2, CD34, and CCND2 were common down-regulated genes (summarized in Online Supplementary Figure S5). HDAC9 was also identified as a common up-regulated gene in both MEF2D fusion and TCF3-PBX1-positive cases, whereas its expression was significantly higher in MEF2D fusion than TCF3-PBX1-positive cases. On the other hand, MEF2D, MME (coding CD10), and RAG1 were down-regulated in MEF2D fusion, but not in TCF3-PBX1-positive cases, thus the low-level expression of CD10 seen in MEF2D fusion cases was identified at the gene-expression level. Interestingly, GATA3 was identified as a highly expressed gene in MEF2D fusion cases, yet it was significantly down-regulated in TCF3-PBX1-positive cases. To explore the gene expression characteristics of MEF2D fusion ALL connected with B-cell differentiation, we performed GSEA using 18 curated gene sets of B lymphocytes at various differentiation stages, as well as 7 early hematopoietic stages including stem cells. Firstly, we compared the gene expression signatures of B-ALL cases with different types of genetic abnormalities. Using Bother-ALL as a reference control, we observed that the majority of gene expression signatures found in B lymphocytes at various differentiation stages were enriched in ALL cases with TCF3-PBX1 (Online Supplementary Table S10 and S11). In contrast, only three gene sets were enriched in MEF2D fusion ALL. We further examined the gene expression characteristics of MEF2D fusion ALL by direct comparison with TCF3-PBX1, and observed that most of the signatures of differentiation stage-specific B lymphocytes as well as early hematopoietic populations were enriched in TCF3-PBX1, but not MEF2D fusion ALL (Online Supplementary Table S10).

Clinical characteristics and outcomes of MEF2D fusion-positive patients The clinical findings and outcome of patients with MEF2D fusions are summarized in Table 1. MEF2D fusion B-ALL patients were aged between 3 and 15 years (median: 9 years) at presentation and comprised 6 males and 10 females. Their initial white blood cell (WBC) counts at presentation ranged from 3,400 to 124,100 (median: 27,300/ml). Analysis of fluids obtained by lumbar puncture revealed no indication of central nervous system (CNS) involvement (data not shown). Among 13 patients, 10 (62.5%) and 4 (25.0%) were classified with an intermediate risk (IR) and high risk (HR), respectively, based on advanced age, elevated WBC counts, or both, and thus standard risk (SR) was assigned to only 2 patients. The response to steroid monotherapy, using the cutoff of 1,000/mL for the blast count in peripheral blood on day 8 was not poor in MEF2D fusion patients; however, among 15 patients, 8 (53.3%) showed bone marrow or CNS relapse, and all of the relapsed patients died. As shown in Figure 4, both the 5year EFS and OS rates for MEF2D fusion patients were significantly lower compared with a consecutive series of Bother-ALL patients enrolled onto the L04-16/L06-16 study (P=0.0306 and P=0.0013, respectively), indicating that outcomes for MEF2D fusion-positive patients were significantly less favorable than those with B-other-ALL.

Discussion In this study, we identified MEF2D fusions, including a novel fusion gene, MEF2D-HNRNPH1, at an incidence of approximately 2% in our B-ALL cohort. It was notable that we identified MEF2D fusions in B-LBL. B-ALL patients with MEF2D fusions showed unique clinical and biological characteristics. They exhibited an older age at presentation and elevated WBC counts, thus were mostly classified into the IR or HR groups. Although their response to steroid treatment was not poor, MEF2D fusion patients showed a significantly worse prognosis with more than half of them relapsing and dying within 1 year. It is noteworthy that stem cell transplantation was not effective in any of the five cases where it was administered as a salvage therapy for relapsed patients in our cohort. Therefore, the establish-

Figure 4. Outcomes of patients with MEF2D fusions. (A) Kaplan-Meier estimates of eventfree survival (EFS) of patients with MEF2D fusions, and B-other, log-rank P=0.306). (B) Overall survival (OS) for the same as above (log-rank P=0.0013).

haematologica | 2019; 104(1)

P=0.0306

P=0.0013 135

K. Ohki et al.

ment of an early diagnostic method and a new therapeutic strategy are necessary for this type of B-ALL. As ex vivo sensitivity of xenografted leukemic cells harboring MEF2D fusions to HDAC inhibitors has been recently reported,10 it may offer a plausible therapeutic option for this type of BALL. In relation to diagnosis, we have already established a qPCR based detection assay for the frequent MEF2D fusions, which may be useful for rapid diagnosis in combination with FISH. Another interesting observation is that B-ALL cases with MEF2D fusions have a characteristic immunophenotype, most typically presenting as CD5- and cytoplasmic µ chainpositive. Although it was previously reported that their immunophenotypes were characterized by weak or absent expression of CD10 and high expression of CD38,10 we found that CD10 expression was not significantly lower than B-other-ALL, and that CD38 expression level was similar to that of TCF3-PBX1-positive cases. Therefore, B-ALL patients with MEF2D fusions could be more effectively diagnosed by combining genetic testing with immunophenotyping. As shown in Table 1, MEF2D fusion cases were mutually exclusive of the known risk stratifying chromosomal translocations, at least at the cytogenetic level. Among additional genetic alterations in MEF2D fusion cases, we identified a significantly higher frequency of CDKN2A/ CDKN2B deletions. At initial diagnosis, 7/8 (87.5%) relapsed patients had CDKN2A/CDKN2B deletions, while they were seen in only 2/7 (28.6%) non-relapse patients. Thus, the prognostic impact of CDKN2A/CDKN2B deletions in MEF2D fusion B-ALL needs to be assessed in larger patient cohorts. Interestingly, we also identified that MEF2D fusion patients had mutations in genes known to be recurrent in T-ALL; in particular, there was an unexpectedly high frequency of mutations in PHF6 (8/16). PHF6 encodes a plant homeodomain (PHD) factor with a proposed role in gene expression control.30 As PHF6 has been proposed to play a role as a tumor suppressor gene,24 it may participate in the pathogenesis of MEF2D fusion B-ALL. Thus, further investigation is needed to clarify this role. Additionally, mutations of genes in the RAS pathway (NRAS, KRAS, NF1, and PTPN11) or IKZF1 alterations were not observed in our cases, in conflict with previous reports. Our data also showed that MEF2D fusion B-ALL has a gene expression signature significantly distinct from other genetic subtypes of B-ALL. Supervised analysis led to separation of MEF2D fusion cases into isolated clusters both in PCA and hierarchical clustering, and these clusters were the closest neighbors to TCF3-PBX1-positive cases. This observation is consistent with the knowledge that both types of B-ALL tend to show a pre-B ALL immunophenotype, and high-level expression of pre-BCR components, including IGHG/IGHV5-78 and IGLL1, were commonly observed in both subtypes.31 However, when we examined the gene expression signatures closely, there were significant differences between them. It has recently been shown that MEF2 family proteins, including MEF2C and MEF2D, play a critical role in early Bcell differentiation. It was demonstrated that B-cell development was blocked at the pre-B-cell stage in Mef2c/d-deficient mice, indicating that they are essential for progression of B-cell precursors (large to small pre-B-cell transition).32 It was further shown that, upon activation via pre-B-cell receptor signaling, Mef2c/d induces target genes, including interferon regulatory factor 4 (Irf4; stimulates the expression of 136

Ikzf1/3) and histone deacetylase 5/9 (Hdac5/9; encoding the protein known to repress Mef2c activity) as we have summarized schematically in Online Supplementary Figure S4. On the other hand, MEF2D fusion B-ALL cases exhibited elevated expression of IRF4 and HDAC9 with down-regulation of MEF2C, although it has been suggested that deregulated expression of the N terminus of MEF2D should induce the up-regulation of the target genes of MEF2C/D, whereas subsequent excess of HDAC9 activity represses MEF2C and its downstream gene expression cascade.10 Our existing data further revealed that up-regulation of GATA3 is another gene expression characteristic of MEF2D fusion ALL. GATA3 is a critical transcription factor in early T-cell development33-35 and its transcriptional repression is essential for early B-cell commitment,36-37 while it has also been reported that GATA3 exhibits myeloid-inducing activity in committed B-lymphocytes under the defect of PAX5 function.38-41 Furthermore, significantly increased GATA3 mRNA levels were associated with ZNF384 fusion-positive cases and a higher risk of relapse in childhood B-ALL with the Ph-like phenotype.42-43 As our data also indicated downregulation of GATA3 expression in TCF3-PBX1-positive cases, ectopic overexpression of GATA3 appears to participate in establishment of a characteristic gene expression signature as well as the biological signature of MEF2D fusion ALL distinct from that of other genetic subtypes of B-ALL. The roles of the MEF2D fusion partners in terms of their biological effects within the fusion molecules of MEF2D fusion ALL are largely unknown. The genes known to be fused to MEF2D in B-ALL have a variety of biological functions. For example, BCL9 is known to be related to WNT/βcatenin signaling, whereas the MEF2D-BCL9 fusion retains only the last one or two exons of the BCL9 gene, thus lacking the functional domains required for WNT/β-catenin signaling, suggesting a role other than deregulation of WNT/βcatenin signaling.10 HNRNPUL1 encodes a nuclear RNAbinding protein of the heterogeneous nuclear ribonucleoprotein family that may play a role in nucleocytoplasmic RNA transport and DNA repair,18 whereas its role as a portion of the fusion molecule in the pathogenesis of B-ALL remains unclear. As HNRNPH1 is a member of the same protein family as HNRNPUL1, both MEF2D-NRNPUL1 and MEF2D-HNRNPH1 likely share the same unknown function. In conclusion, we have shown that ALL patients harboring MEF2D fusion genes possess a characteristic immunophenotype and gene expression signature as well as distinct clinical features, defining them as a distinct genetic subtype among B-other-ALL. Although additional studies are required to elucidate the biological function of the MEF2D fusion protein in leukemogenesis, our data has allowed improved characterization of this new B-ALL subtype. Acknowledgments The authors would like to thank K. Itagaki, H. Yagi, Y. Katayama, A. Tamura, K. Takeda, K. Hayashi, and the staff of the Laboratory for Genotyping Development, Riken Center for Integrative Medical Sciences for their excellent data management and experimental assistance. We thank all members of the Committees of ALL and of Research and Diagnosis of TCCSG. Funding This work was supported in part by a Health and Labour Sciences Research Grant (3rd-term comprehensive 10-year haematologica | 2019; 104(1)

Fusion genes involving MEF2D in B-ALL

strategy for cancer control H22-011), the Grant of the National Center for Child Health and Development (26-20), and the Advanced Research for Medical Products Mining Programme of the National Institute of Biomedical Innovation (NIBIO, 10-41, -42, -43, -44, -45), and Biobank Japan project funded by the Ministry of Education, Culture, Sports, Science and Technology

References 1. Hunger SP, Mullighan CG. Acute lymphoblastic leukemia in children. N Engl J Med. 2015;373(16):1541-1552. 2. Schwab C, Harrison CJ. Advances in B-cell precursor acute lymphoblastic leukemia genonics. HemaSphere. 2018;2(4):e53 3. Roberts KG, Morin RD, Zhang J, et al. Genetic alterations activating kinase and cytokine receptor signaling in high-risk acute lymphoblastic leukemia. Cancer Cell. 2012;22(2):153-166. 4. Roberts KG, Li Y, Payne-Turner D, et al. Targetable kinase-activating lesions in Phlike acute lymphoblastic leukemia. N Engl J Med. 2014;371(11):1005-1015. 5. Gocho Y, Kiyokawa N, Ichikawa H, et al. A novel recurrent EP300-ZNF384 gene fusion in B-cell precursor acute lymphoblastic leukemia. Leukemia. 2015;29(12):24452448. 6. Liu YF, Wang BY, Zhang WN, et al. Genomic profiling of adult and pediatric B-cell acute lymphoblastic leukemia. EBioMedicine. 2016;8:173-183. 7. Yasuda T, Tsuzuki S, Kawazu M, et al. Recurrent DUX4 fusions in B cell acute lymphoblastic leukemia of adolescents and young adults. Nat Genet. 2016;48(5):569574. 8. Lilljebjörn H, Henningsson R, HyreniusWittsten A, et al. Identification of ETV6RUNX1-like and DUX4-rearranged subtypes in paediatric B-cell precursor acute lymphoblastic leukaemia. Nat Commun. 2016;7:11790. 9. Zhang J, McCastlain K, Yoshihara H, et al. Deregulation of DUX4 and ERG in acute lymphoblastic leukemia. Nat Genet. 2016; 48(12):1481-1489. 10. Gu Z, Churchman M, Roberts K, et al. Genomic analyses identify recurrent MEF2D fusions in acute lymphoblastic leukaemia. Nat Commun. 2016;7:13331. 11. Hirabayashi S, Ohki K, Nakabayashi K, et al. ZNF384-related fusion genes define a subgroup of childhood B-cell precursor acute lymphoblastic leukemia with a characteristic immunotype. Haematologica. 2017;102 (1):118-129. 12. Prima V, Gore L, Caires A, et al. Cloning and functional characterization of MEF2D/DAZAP1 and DAZAP1/MEF2D fusion proteins created by a variant t(1;19)(q23;p13.3) in acute lymphoblastic leukemia. Leukemia. 2005;19(5):806-813. 13. Lilljebjörn H, Ågerstam H, Orsmark-Pietras C, et al. RNA-seq identifies clinically relevant fusion genes in leukemia including a novel MEF2D/CSF1R fusion responsive to imatinib. Leukemia. 2014;28(4):977-979. 14. Suzuki K, Okuno Y, Kawashima N, et al. MEF2D-BCL9 fusion gene is associated with high-risk acute B-cell precursor lymphoblastic leukemia in adolescents. J Clin Oncol. 2016;34(28):3451-3459. 15. Breitbart RE, Liang CS, Smoot LB, Laheru

haematologica | 2019; 104(1)

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

(MEXT) and the Japan Agency for Medical Research and Development (AMED), and the Practical Research for Innovative Cancer Control from AMED. These funding sources played no role in the collection, analysis, or interpretation of the results, or in the writing of the manuscript and decision to submit it.

DA, Mahdavi V, Nadal-Ginard B. A fourth human MEF2 transcription factor, hMEF2D, is an early marker of the myogenic lineage. Development. 1993; 118(4):1095-1106. Mao Z, Bonni A, Xia F, Nadal-Vicens M, Greenberg M E. (1999) Neuronal activitydependent cell survival mediated by transcription factor MEF2. Science. 1999;286 (5440):785-790. Miska EA, Karlsson C, Langley E, Nielsen SJ, Pines J, Kouzarides T. HDAC4 deacetylase associates with and represses the MEF2 transcription factor. EMBO J. 1999; 18(18): 5099-5107. Honoré B, Rasmussen HH, Vorum H, et al. Heterogeneous nuclear ribonucleoproteins H, H', and F are members of a ubiquitously expressed subfamily of related but distinct proteins encoded by genes mapping to different chromosomes. J Biol Chem. 1995; 270(48):28780-28789. Takahashi H, Kajiwara R, Kato M, et al. Treatment outcome of children with acute lymphoblastic leukemia: the Tokyo Children's Cancer Study Group (TCCSG) Study L04-16. Int J Hematol. 2018;108:98108. Masuzawa A, Kiyotani C, Osumi T, et al. Poor responses to tyrosine kinase inhibitors in a child with precursor B-cell acute lymphoblastic leukemia with SNX2-ABL1 chimeric transcript. Eur J Haematol. 2014; 92(3):263-267. Kobayashi K, Mitsui K, Ichikawa H, et al. ATF7IP as a novel PDGFRB fusion partner in acute lymphoblastic leukaemia in children. Br J Haematol. 2014;165(6):836-841. McPherson A, Hormozdiari F, Zayed A, et al. deFuse: an algorithm for gene fusion discovery in tumor RNA-Seq data. PLoS Comput Biol. 2011;7(5):e1001138. Weng AP, Ferrando AA, Lee W, et al. Activating mutations of NOTCH1 in human T cell acute lymphoblastic leukemia. Science. 2004;306(5694):269-271. Van Vlierberghe P, Palomero T, Khiabanian H, et al. PHF6 mutations in T-cell acute lymphoblastic leukemia. Nat Genet. 2010; 42(4):338-342. Wang Q, Qiu H, Jiang H, et al. Mutations of PHF6 are associated with mutations of NOTCH1, JAK1 and rearrangement of SETNUP214 in T-cell acute lymphoblastic leukemia. Haematologica. 2011;96(12): 1808-1814. Spinella JF, Cassart P, Richer C, et al. Genomic characterization of pediatric T-cell acute lymphoblastic leukemia reveals novel recurrent driver mutations. Oncotarget. 2016;7(40):65485-65503. Royet J, Bouwmeester T, Cohen SM. Notchless encodes a novel WD40-repeatcontaining protein that modulates Notch signaling activity. EMBO J. 1998; 17(24):7351-7360. Carroll AJ, Crist WM, Parmley RT, Roper M, Cooper MD, Finley WH. Pre-B cell leukemia associated with chromosome translocation 1;19. Blood. 1984;63(3):721-724.

29. Borowitz MJ, Hunger SP, Carroll AJ, et al. Predictability of the t(1;19)(q23;p13) from surface antigen phenotype: implications for screening cases of childhood acute lymphoblastic leukemia for molecular analysis: a Pediatric Oncology Group study. Blood. 1993;82(4):1086-1091. 30. Lower KM, Turner G, Kerr BA, et al. Mutations in PHF6 are associated with Börjeson-Forssman-Lehmann syndrome. Nat Genet. 2002;32(4):661-665. 31. Chen D, Zheng J, Gerasimcik N, et al. The expression pattern of the Pre-B cell receptor components correlates with cellular stage and clinical outcome in acute lymphoblastic leukemia. PLoS One. 2016; 11(9):e0162638. 32. Herglotz J, Unrau L, Hauschildt F, et al. Essential control of early B-cell development by Mef2 transcription factors. Blood. 2016;127(5):572-581. 33. Landry DB, Engel JD, Sen R. Functional GATA 3 binding sites within murine CD8 alpha upstream regulatory sequences. J Exp Med. 1993;178(3):941-949. 34. Ting CN, Olson MC, Barton KP, Leiden JM. Transcription factor GATA-3 is required for development of the T-cell lineage. Nature. 1996;384(6608):474-478. 35. Gao J, Chen YH, Peterson LC. GATA family transcriptional factors: emerging suspects in hematologic disorders. Exp Hematol Oncol. 2015;4:28. 36. Banerjee A, Northrup D, Boukarabila H, Jacobsen SE, Allman D. Transcriptional repression of Gata3 is essential for early B cell commitment. Immunity. 2013; 38(5):930-42. 37. Somasundaram R, Prasad MA, Ungerbäck J, Sigvardsson M. Transcription factor networks in B-cell differentiation link development to acute lymphoid leukemia. Blood. 2015;126(2):144-152. 38. Chen D, Zhang G. Enforced expression of the GATA-3 transcription factor affects cell fate decisions in hematopoiesis. Exp Hematol. 2001;29(8):971-980. 39. Ku CJ, Hosoya T, Maillard I, Engel JD. GATA 3 regulates hematopoietic stem cell maintenance and cell cycle entry. Blood. 2012;119(10):2242-2251. 40. Frelin C, Herrington R, Janmohamed S, et al. GATA 3 regulates the self renewal of long term hematopoietic stem cells. Nat Immunol. 2013;14(10):1037-1044. 41. Heavey B, Charalambous C, Cobaleda C, Busslinger M. Myeloid lineage switch of Pax5 mutant but not wild-type B cell progenitors by C/EBPalpha and GATA factors. EMBO J. 2003;22(15):3887-3897. 42. Perez-Andreu V, Roberts KG, Harvey RC, et al. Inherited GATA3 variants are associated with Ph-like childhood acute lymphoblastic leukemia and risk of relapse. Nat Genet. 2013;45(12):1494-1498. 43. Perez-Andreu V, Roberts KG, Xu H, et al. A genome-wide association study of susceptibility to acute lymphoblastic leukemia in adolescents and young adults. Blood. 2015; 125(4):680-686.

137

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):138-146

Non-Hodgkin Lymphoma

A phase 2 study of rituximab, bendamustine, bortezomib and dexamethasone for first-line treatment of older patients with mantle cell lymphoma

Rémy Gressin 1,2,Nicolas Daguindau,3 Adrian Tempescul,4 Anne Moreau,5 Sylvain Carras,1 Emmanuelle Tchernonog,6 Anna Schmitt,7 Roch Houot,8 Caroline Dartigeas,9 Jean Michel Pignon,10 Selim Corm,11 Anne Banos,12 Christiane Mounier,13 Jehan Dupuis,14 Margaret Macro,15 Joel Fleury,16 Fabrice Jardin,17 Clementine Sarkozy,18 Ghandi Damaj,19 Pierre Feugier,20 Luc Matthieu Fornecker,21 Cecile Chabrot,22 Veronique Dorvaux,23 Krimo Bouadallah,24 Sandy Amorin,25 Reda Garidi,26 Laurent Voillat,27 Bertrand Joly,28 Philippe Solal Celigny,29 Nadine Morineau,30 Marie Pierre Moles,31 Hacene Zerazhi,32 Jean Fontan,33 Yazid Arkam,34 Magda Alexis,35 Vincent Delwail,36 Jean Pierre Vilque,37 Loic Ysebaert,38 Steven Le Gouill,39 Mary B. Callanan,2 40for the Lymphoma Study Association

Onco-Hematology Department, Grenoble University Hospital; 2INSERM 1209, CNRS UMR 5309, Faculté de Médecine, Université Grenoble-Alpes, Institute for Advanced Biosciences, Grenoble; 3Hematology Department, Annecy Hospital; 4Hematology Department, Brest University Hospital; 5Pathology Department, Nantes University Hospital; 6Hematology Department, Montpellier University Hospital; 7Hematology Department, Cancer Institute Bergonie Bordeaux; 8Hematology Department, Rennes University Hospital; 9Hematology Department, Tours University Hospital; 10Hematology Department, Dunkerque Hospital; 11Hematology Department, Chambery Hospital; 12 Hematology Department, Bayonne Hospital; 13Hematology Department, Loire Cancer Institute, Saint Etienne; 14Lymphoid Malignancies Unit, Henri Mondor University Hospital, Assistance Publique-Hôpitaux de Paris, Créteil; 15IHBN - Hematology Department, Caen University Hospital; 16Hematology Department, Clermont-Ferrand Cancer Institute; 17Hematology Department, Rouen University Hospital; 18Hematology Department, Hospices Civils de Lyon, Centre Hospitalier Lyon Sud. INSERM 1052; 19 Hematology Department, Amiens University Hospital; 20Hematology Department, Nancy University Hospital; 21Hematology Department, University Hospital Strasbourg; 22 Hematology Department, University Clermont-Ferrand Hospital; 23Hematology Department, Metz University Hospital; 24Hematology Department, Bordeaux University Hospital; 25Hematology Department, University Hospital Paris Saint-Louis; 26Hematology Department, Saint Quentin Hospital; 27Hematology Department, Chalon Hospital; 28 Hematology Department, Corbeil Hospital; 29Hematology Department, Victor Hugo Clinic, Le Mans; 30Hematology Department, Catherine de Sienne Clinic, Nantes; 31 Hematology Department, Angers University Hospital; 32Hematology Department, Avignon Hospital; 33Hematology Department, Besançon University Hospital; 34 Hematology Department, Mulhouse Hospital; 35Hematology Department, Orleans Hospital; 36Onco-Hematology Department, University Hospital Poitiers and INSERM, CIC 1402, Poitiers University; 37Hematology Department, Baclesse Caen Cancer Center; 38 Hematology Department, Toulouse University Hospital; 39Hematology Department, Nantes University Hospital and 40Unit for Innovation in Genetics and Epigenetics in Oncology, Dijon University Hospital, France 1

Parts of this study were presented at the 2014 ASH meeting, the 2017 French Hematology Association meeting and the 2017 International Conference of Malignant Lymphoma.

Correspondence: rgressin@chu-grenoble.fr or mary.callanan@chu-dijon.fr Received: February 23, 2018. Accepted: August 23, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2018.191429 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/138

RG, SL and MBC contributed equally to this work.

ABSTRACT

138

e present results of a prospective, multicenter, phase II study evaluating rituximab, bendamustine, bortezomib and dexamethasone as first-line treatment for patients with mantle cell lymphoma aged 65 years or older. A total of 74 patients were enrolled (median age, 73 years). Patients received a maximum of six cycles of treatment at 28-day intervals. The primary objective was to achieve an 18-month progression-free survival rate of 65% or higher. Secondary objectives were to evaluate toxicity and the prognostic impact of mantle cell lymphoma prognostic index, Ki67 expression, [18F]fluorodeoxyglucose-positron emission tomography and molecular minimal residual disease, in peripheral blood or bone marrow. With a median follow-up of haematologica | 2019; 104(1)

RiBVD regimen as first-line treatment for older MCL patients

52 months, the 24-month progression-free survival rate was 70%, hence the primary objective was reached. After six cycles of treatment, 91% (54/59) of responding patients were analyzed for peripheral blood residual disease and 87% of these (47/54) were negative. Four-year overall survival rates of the patients who did not have or had detectable molecular residual disease in the blood at completion of treatment were 86.6% and 28.6%, respectively (P<0.0001). Neither the mantle cell lymphoma index, nor fluorodeoxyglucose-positron emission tomography nor Ki67 positivity (cut off of â&#x2030;Ľ30%) showed a prognostic impact for survival. Hematologic grade 3-4 toxicities were mainly neutropenia (51%), thrombocytopenia (35%) and lymphopenia (65%). Grade 3-4 non-hematologic toxicities were mainly fatigue (18.5%), neuropathy (15%) and infections. In conclusion, the tested treatment regimen is active as frontline therapy in older patients with mantle cell lymphoma, with manageable toxicity. Minimal residual disease status after induction could serve as an early predictor of survival in mantle cell lymphoma. ClinicalTrials.gov: NCT 01457144.

Introduction Mantle cell lymphoma (MCL) is a rare subtype of B-cell non-Hodgkin lymphoma characterized by the genetic hallmark t(11;14)(q13;q32) chromosomal translocation which leads to overexpression of cyclin D1.1 The standard-of-care for the treatment of older MCL patients (>65 years), has been eight cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) given at 21-day intervals (R-CHOP-21), followed by maintenance therapy which has been shown to improve response duration and overall survival (OS) in patients who reach the maintenance phase.2 Complete response rates do, however, remain low with R-CHOP (30-35%) and the median progression-free survival (PFS) is in the range of 14-18 months.3,4 After R-CHOP and maintenance therapy, the 4-year OS rate was 87%.2 Although dose-intensive and high-dose cytarabine-containing regimens, with or without autologous stem cell transplantation consolidation in younger patients, has improved outcomes (the median PFS is now well in excess of 5 years), such approaches are frequently not feasible, given that the median age at diagnosis of MCL is in the mid to late 60s.1 Bortezomib was the first novel agent to be approved for the treatment of patients with relapsed/refractory MCL.5,6 The addition of bortezomib to rituximab-anthracyclinebased regimens has improved the results, compared to those achieved by R-CHOP, for frontline therapy in MCL, leading to a complete response rate of 50% and a median PFS of 25 months albeit with increased hematologic toxicity.7,8 Two phase III trials have shown the superiority of bendamustine-rituximab combination therapy over RCHOP or R-CHOP/R-CVP (rituximab, cyclophosphamide, vincristine, prednisolone) with respect to overall and complete response rates and reduced toxicity.9,10 However, superior PFS was observed in only one of the latter phase III studies.9 More recently, combining genotoxic agents (such as cytarabine) or targeted agents (such as bortezomib or lenalidomide) with bendamustine and rituximab (BR) has shown efficacy in both first-line and salvage therapy in MCL.1,11-13 In anticipation of the above findings, our group initiated a phase II trial to assess the efficacy of a new regimen combining rituximab, bortezomib, bendamustine and dexamethasone (RiBVD) for first-line therapy of older MCL patients. Specifically, for the trial design, associating RiBVD, we took into account the interim results of the BR regimen, for which the reported overall response rates haematologica | 2019; 104(1)

were 90% in relapsing MCL patients,14,15 and the promising results (32% overall response rate) of bortezomib monotherapy in relapsing MCL patients (PINNACLE study).6 Pre-defined secondary objectives of our study included assessment of molecular complete response rates in blood and bone marrow and evaluation of their prognostic impact on survival.

Methods Study design and patients The RiBVD multicenter phase II trial enrolled newly diagnosed MCL patients â&#x2030;Ľ65 years or <65 years if ineligible or unwilling to undergo autologous stem cell transplantation. The study was conducted in 37 centers of the Lymphoma Study Association (LYSA) (NCT 01457144) and was approved by institutional review boards and ethics committees at all sites, and conducted according to the Declaration of Helsinki. The diagnosis of MCL was established according to the World Health Organization (WHO) 2008 criteria. Ki67 staining and scoring were performed centrally, according to European MCL Network recommendations.16 All pathology results were reviewed centrally by the LYSA pathology commission. Eligible patients gave written, informed consent, as per standard guidelines. Inclusion and exclusion criteria are summarized in Online Supplementary Table S1. The RiBVD regimen consisted of a maximum of six cycles of 28 days each for all enrolled patients, as described in the Online Supplementary Methods and Online Supplementary Table S2.

Response and safety assessments The International Working Group (IWG) 1999 and 2007 criteria were used to define responses after four and six cycles, respectively. [18F]fluorodeoxyglucose (FDG)-positron emission tomography (FDG-PET) responses were evaluated in each center with the fivepoint scale, visual method of Deauville.17 Hematologic and nonhematologic toxicity was monitored continuously during treatment and at follow-up visits and graded according to the National Cancer Institute criteria (Common Terminology Criteria for Adverse Events, version 3.0) (see the Online Supplementary Methods for details).

Molecular minimal residual disease Molecular responses were evaluated centrally by real-time quantitative polymerase chain reaction targeted to patient-specific, IGH V(D)J clono-specific rearrangements, to quantify tumor B cells, according to EURO-MRD guidelines, as previously described.18 Minimal residual disease (MRD) analysis was per139

R. Gressin et al.

formed before treatment (baseline), after four courses of treatment (mid-term MRD), and at the end of treatment (after 6 courses of RiBVD) in peripheral blood and bone marrow until progression or relapse, for a maximum follow-up period of 3 years.18,19 During the follow-up, MRD was evaluated in the blood at 3 monthly intervals for 1 year and every 6 months thereafter while bone marrow MRD monitoring was performed at yearly intervals. A description of additional methods, the MRD study cohort, sample source and numbers is given in the Online Supplementary Methods and illustrated in Online Supplementary Figure S1.

Sample size calculation and statistical analysis The primary objective of the study was to prolong PFS by 6 months compared to the 18-month median PFS reported for patients treated with R-CHOP-21.3 The number of patients to be enrolled was calculated by a one-step Fleming method. In order to define superiority of the RiBVD regimen over R-CHOP, a PFS rate of 65% or more (H1) was required at 18 months. The treatment was to be considered a failure if the PFS rate at 18 months was ≤50%. Taking into account alpha and beta risks of 5% and 20%, respectively, 69 patients needed to be enrolled. Based on a maximum 10% error in diagnosis, 76 patients had to be enrolled. Additional details are given in the Online Supplementary Methods.

Results Patients A total of 76 MCL patients were enrolled between November 2011 and December 2012 (Figure 1). All patients were monitored for 3 years after their last cycle of

Figure 1. Consort diagram for the RiBVD phase 2 trial. MCL: mantle cell lymphoma; DLBCL: diffuse large B-cell lymphoma; HBV: hepatitis B virus; MRD: minimal residual disease; BM: bone marrow; PML: progressive multifocal leukoencephalopathy.

140

therapy. Two patients were excluded - one because of a misdiagnosis of MCL (diffuse large B-cell lymphoma) and one because of exclusion criteria (hepatitis B) – leaving 74 patients for data analyses (Figure 1). Seventy-one patients had MCL confirmed by central review. The diagnosis was made on tumor biopsies (45 on lymph nodes and 26 on extra-nodal tissue). Due to unsuccessful tissue biopsy in three patients, a diagnosis of MCL was made by flow cytometry in peripheral blood (1 patient) or bone marrow (2 patients). Ki67 staining was performed in 56 patients, and was found ≥30% positive in 59% of these patients (31 of 56 patients) (Table 2).

Treatment response Seventy-four patients initiated therapy. Sixty-seven patients received at least four cycles (90.5%) of treatment Table 1. Patients’ demographics and clinical characteristics.

Characteristics Age (years) median range Sex male female WHO Performance Status 0-1 2-4 Lactate dehydrogenase normal >normal B symptoms no yes Ann Arbor stage II III-IV Bulky tumor no yes Extranodal involvement no yes Bone marrow involvement no yes Spleen involvement no yes MIPI score low intermediate high MIB1 / ki67 prolferation index <30% ≥30% Pathology classic blastoid

73 64-83 49 25

66 34

73 11

85 15

44 28

61 39

56 17

76 24

4 70

6 94

52 21

71 29

7 67

9 91

24 46

34 66

38 35

52 48

2 12 58

3 17 80

21 30

41 59

61 10

86 14

WHO: World Heath Organization; MIPI: Mantle Cell Lymphoma International Prognostic Index.

haematologica | 2019; 104(1)

RiBVD regimen as first-line treatment for older MCL patients

and 59 (80%) received all six planned cycles (Figure 1). Of the planned 444 cycles, 406 (91.5%) were administered. Fifteen patients stopped therapy before receiving all six cycles (Figure 1). After four cycles, the overall response rate was 86.5% (64/74) and the complete response rate (confirmed and unconfirmed complete responses) was 56.5% (42/74). At the end of treatment, the overall response rate was 84% (62/74) and the complete response rate was 75.5% (56/74). FDG-PET evaluations were performed after four cycles of treatment in 64 patients (100% of the 64 responders) and after cycle 6 in 59 patients (95% of the 62 responders). Interim and final FDG-PET were negative in 64% (41/64) and in 78% of evaluated patients (46/59), respectively.

Molecular minimal residual disease in blood and bone marrow Molecular MRD was assessed in a total of 58 of the 74 patients eligible for MRD analysis (in all, 732 samples were assessed, see Online Supplementary Figure S1). Molecular MRD analysis was not possible in 16 of 74 MRD-eligible patients because of a lack of MRD target (n=6), missing follow-up samples (n=9) or because an MRD target reference sample was not available (n=1) (Online Supplementary Figure S1). After four cycles (midterm), 57 patients were analyzed for molecular MRD (57 peripheral blood samples; 48 bone marrow samples, of which 48 patients with paired bone marrow and peripheral blood MRD samples, at the mid-term analysis). Of these, 50 patients were negative for molecular MRD (32 in complete remission, 18 in partial remission) and seven were positive (2 in complete remission, 4 in partial remission and 1 with stable disease) in the blood and/or bone marrow, for a molecular response rate of 79% (defined by a quantitative polymerase chain reaction assay with a sensitivity of 10-5). After six cycles of treatment, 54 patients were analyzed for molecular MRD (54 peripheral blood samples; 46 bone marrow samples, of which 46 patients with paired bone marrow and peripheral blood samples for MRD analysis at the end of treatment). Of these 54 patients, 41 were MRD-negative (39 in complete remis-

sion, 1 in partial remission and 1 with stable disease) and 13 were MRD-positive (8 in complete remission, 4 in partial remission and 1 with progressive disease) in blood and/or bone marrow (76% molecular response rate) (Figure 2 and Online Supplementary Figure S1). Molecular MRD response rates were then analyzed separately in the peripheral blood versus bone marrow at the mid-term follow-up time-point (after 4 treatment cycles) and at the end of treatment (after 6 cycles). Blood samples were molecular MRD-negative from 88% (50/57) of patients after four cycles and 87% (47/54 patients) after six cycles (Figure 2C, left panel). The corresponding percentages for bone marrow samples were 77% (37/48 patients) after four cycles of treatment and 76% (35/46 patients) at the end of treatment (after 6 cycles) (Online Supplementary Figure S3).

Survival analyses and prognostic factors With a median follow-up time of 52 months, 74 patients were evaluable. Overall, 24 patients died, four during treatment (2 from cardiac arrest, 1 with pneumonia and 1 with progressive multifocal leukoencephalopathy, after cycle 3) and 20 during follow-up (16 due to progressive disease, 1 from pancreatic adenocarcinoma, 1 from cardiac arrest and 2 from unknown causes). The 2-year PFS was 70.3% compared to 57.6% at 4 years. The 2-year OS was 81.1% compared to 71.3% at 4 years. The Mantle Cell Lymphoma International Prognostic Index (MIPI) score was not predictive for PFS or OS perhaps due to the small number of patients because a trend could be observed (Table 2 and Online Supplementary Figure S2A). Indeed, the 4-year OS for the 58 MIPI high-risk patients was 66.8% compared to 85.7% for the 14 MIPI low- or intermediate-risk patients (P=0.13). Neither histology (classical subtype versus blastoid subtype) nor Ki67 expression (<30% versus â&#x2030;Ľ30% of positive MCL cells in the tumor biopsy) was predictive for OS (P=0.10 and P=0.24, respectively) or PFS (P=0.08 and P=0.13, respectively) (Table 2 and Online Supplementary Figure S2B). Clinical responses (complete or partial response versus no response), as assessed by the Cheson 1999 criteria, were

Table 2. Prognostic factors for progression-free survival and overall survival.

Prognostic factors

P for PFS

P for OS

Pathology (classic vs. blastoid form) MIPI score (high vs. low/Intermediate) Ki67 (< vs. â&#x2030;Ľ 30%) Response IWC 1999 (CR vs. PR vs. failure) FDG-PET midterm FDG-PET treatment end. MRD blood and/or bone marrow at mid-term (neg 45; pos 12) MRD blood and/or bone marrow at treatment end (neg 41; pos 13) MRD blood mid-term (neg 50; pos 7) MRD blood treatment end (neg 47; pos 7) MRD bone marrow midterm (neg 37; pos 11) MRD bone marrow treatment end (neg 35; pos 11)

71 72 51 74 64 59 57 54 57 54 48 46

0.08 0.18 0.35 <0.0001 0.19 0.48 0.20 0.04 0.01 <0.0001 0.24 0.20

0.10 0.13 0.24 <0.0001 0.57 0.98 0.33 0.02 0.047 <0.0001 0.41 0.19

N: number of patients who could be evaluated ; MIPI score,: Mantle-Cell Lymphoma International Prognosis Index; Ki67/Mib1, proliferation index score; Response IWC 1999, response according to the 1999 International Workshop Criteria; PF: progression free survival; OS: overall survival (at 4 years), respectively; CR: complete response; PR: partial response; FDG-PET, [18F]fluorodeoxyglucose positron emission tomography; mid-term, analysis after four cycles; treatment end, analysis after six cycles; MRD, (molecular) minimal residual disease.

haematologica | 2019; 104(1)

141

R. Gressin et al.

highly predictive for PFS or OS whether determined at the mid-term staging or at the end of treatment. There were no survival differences (PFS or OS) between patients in partial or complete remission. Neither mid-term nor final FDGPET scan responses were predictive for PFS or OS. The most highly predictive factor for PFS and OS (P<0.0001) was MRD status in peripheral blood at the end of treatment (Figure 2C, right panel and Table 2). Molecular blood MRD status at mid-term was also significant for PFS (P=0.01) and weakly significant for OS (P=0.047) (Table 2). By contrast, MRD status in the bone marrow after four cycles of treatment (mid-term) or at the end of treatment

was not predictive for either PFS or OS (see Online Supplementary Figure S3 for end-of-treatment data). The 4year OS for patients who were MRD-negative in blood at the end of treatment (n=47/54) was 86.6% compared to 28.6% for blood MRD-positive patients (n=7/54). Continued molecular remission status in the peripheral blood after therapy (at the 12-month follow-up) was significantly associated with longer PFS (33 patients; 4-year PFS 97%). By contrast, the median PFS for patients who remained MRD-positive (n=6) or who had converted to an MRD-positive status in the peripheral blood by the 12month follow-up (n=7) was 11 and 26 months, respectively.

Figure 2. Survival of patients with mantle cell lymphoma following frontline treatment with the RiBVD regimen. (A) Progression-free survival of the 74 patients. (B) Overall survival of the 74 patients (C) Molecular response rates and overall survival according to molecular residual disease (MRD) status in peripheral blood after six cycles of RiBVD (treatment end).

142

haematologica | 2019; 104(1)

RiBVD regimen as first-line treatment for older MCL patients

Toxicity Fifteen patients of 74 (20%) stopped treatment before the sixth cycle, four because of death (1 case each of pneumonia and progressive multifocal leukoencephalopathy and 2 cardiac arrests), five because of grade 3-4 toxicity [septicemia (n=1), neuropathy (n=2), digestive tract toxicity (n=1) and pleural effusion (n=1)], three because of progression or stable disease and three for other causes (Figure 1). During treatment, 49 of 74 patients (66.2%) developed grade 3-4 hematologic toxicities (51% neutropenia, 35% thrombocytopenia and 19% anemia) (Table 3). Neutropenia translated into febrile neutropenia in 11 patients (11%), which was grade 3-4 in six. Lymphopenia at the end of treatment was reported in 65% of the patients (48/70) and was mainly grade 3-4 (lymphocytes <0.5x109/L) (Table 3). Persistent grade 3-4 lymphopenia was seen in 28.8% of patients at 1 year after the completion of treatment (17 of 59 surviving patients who could be evaluated). Forty-two patients (56.7%) had non-hematologic grade 3-4 toxicities at the end of treatment (Table 3). The most frequent non-hematologic toxicities (seen in more than 10% of patients) were fatigue, peripheral neuropathy and fever with or without neutropenia, which occurred in 18.5% (n=14), 15% (n=11) and 15 (n=11) of cases, respectively (Table 3). Other toxicities were reported in four or fewer patients (i.e. less than 6%) and were as follows: pulmonary toxicity, cardiac toxicity, hyperglycemia, elevated transaminases, digestive tract toxicity, cutaneous rash, allergy and fever without neutropenia. No patient experienced cytomegalovirus reactivation or pneumocystis infection.

Twenty-four episodes of infection were declared as serious adverse events. These represented one-third of the 76 serious adverse events reported during the 406 cycles of therapy, or during follow up. They included seven cases of opportunistic infection (4 cases during treatment and 3 further cases 1 year after the end of treatment), which were as follows; herpes zoster (n=3), progressive multifocal leukoencephalopathy (n=1), cytomegalovirus colitis (n=1), listeriosis (n=1) and oral candidosis (n=1). Additional infections were pneumonia (n=9), staphylococcal infection (n=2), followed by non-recurring infections of various types (no more than 1 case each, as follows; acute pyelonephritis, bronchitis, catheter site infection, upper aero-digestive tract infection and Clostridium difficileinduced colitis). Regarding neurotoxicity, grade 2 to 4 neuropathy was observed in 21.5% of patients (16 of 74 patients) (Table 3). Neuropathy was generally reported after cycle 3 of treatment. Bortezomib was stopped indefinitely in ten of the 11 patients with grade 3-4 neurotoxicity but not in cases of grade 2 toxicity. Partial reversibility of neuropathy was reported in 13 of the 16 patients (81%) with grade 2 to 4 neurotoxicity.

Discussion We report the results of a prospective, phase II study by the French LYSA group. The study aimed to test the efficacy of six cycles of RiBVD, without maintenance therapy, for first-line treatment of MCL patients aged â&#x2030;Ľ65

Table 3. Hematologic and non-hematologic toxicity.

All Grades

Grade 3

Grade 4

Events

Neutropenia Thrombocytopenia Lymphopenia Anemia Fatigue Neuropathy Fever Febrile neutropenia Lung Cardiac Hyperglycemia Rash GOT/GPT Digestive tract Allergy Weight loss Nausea Bilirubin Creatinine Ear Infusion-related reaction Calcium

52 67 67 70 56 32 35 8 26 16 23 25 29 35 24 25 25 12 24 2 18 4

70 90.5 95.7 94.5 75.5 43* 47 11 35 21.5 31 34 39 47 32 34 34 16 32.5 3 24 5

17 18 38 6 12 10 5 4 4 4 4 3 3 3 1 0 0 0 0 0 0 0

23 24 54 8 16 13.5 6.5 5.5 5.5 5.5 5.5 4 4 4 1 0 0 0 0 0 0 0

21 8 8 0 2 1 2 2 2 1 0 0 0 0 0 0 0 0 0 0 0 0

28.5 11 11 0 2.5 1.5 2.5 2.5 2.5 1.5 0 0 0 0 0 0 0 0 0 0 0 0

GOT/GPT: glutamic oxaloacetic transaminase/glutamic-pyruvic transaminase; CMV: cytomegalovirus. *includes grade 2 = 6.8% (n=5/74).

haematologica | 2019; 104(1)

143

R. Gressin et al.

years. With a median follow up of 52 months, the 2-year PFS of the 74 patients with analyzable data was 70%, thus reaching the primary objective of the study which was to improve median PFS by 6 months compared to the reported 18-month PFS for patients treated with R-CHOP.3,4 The 4-year PFS (57.6%) observed here for RiBVD-treated patients is in line with the median PFS reported for the BR regimen (35.4 months) and other bortezomib-containing regimens such as 24.7 months for VR-CAP (bortezomib, rituximab, cyclophosphamide, doxorubicin, and prednisone) and 26 months for the RiPAD+C regimen (rituximab, bortezomib, doxorubicin, dexamethasone, and chlorambucil).7-9 The favorable PFS with the RiBVD regimen may be related to the marked depth of response (75.7% rate of confirmed and unconfirmed complete responses according to the Cheson 1999 criteria; 78% rate of complete responses according to the Cheson 2007 criteria). Although not strictly comparable outside of a randomized trial, it is worth noting that complete response rates with other regimens are lower: R-CHOP (34%), BR (40%), VR-CAP (53%) and RiPAD+C (51%).2,3,7,8 The higher rate of complete responses (confirmed and unconfirmed) with the RiBVD regimen translated into higher molecular response rates [76% of patients (41/54) were MRD-negative in blood and/or bone marrow at the end of treatment] compared to published data for R-CHOP in older MCL patients [67% MRD negativity (54/81 patients)].20 It is worth noting that 80% of patients in our study had high-risk MIPI scores and that 59% had ≥30% Ki67 positivity (range, 5% to 95% positivity) which appears high compared to the percentages in other studies in MCL patients over 65 years old (see Online Supplementary Table S4). In keeping with results of high-dose cytarabine treatment in younger MCL patients,21-23 a recent phase II trial from an Italian group (FIL) confirmed the efficacy of the RBAC500 regimen (which associates rituximab, bendamustine and cytarabine) for treatment of older MCL patients.12 In updated clinical results for the R-BAC500 regimen, complete response rates of 93% and molecular response rates of 78% (35/45) in the blood have been reported.24 The 2-year PFS and OS were estimated as 83% and 86%, respectively.24 However, it is worth noting that the baseline characteristics of the MCL cohort treated with RBAC500, the patients in our study and those in other published series of MCL cases differ quite widely (Online Supplementary Table S4). The rate of treatment discontinuation in our study was 20% (15/74 patients). This is broadly in line with rates reported for R-CHOP [17% (43/242)] and VR-CAP [18.8% (45/240)] in older MCL patients treated in first line,7 and is lower than that reported with the R-BAC500 regimen (33%)24 (Online Supplementary Table S4). The reported rates of premature therapy cessation for the BR and R-CHOP regimens are 8% and 5%, respectively.10 The 5.4% rate of toxic deaths reported here with the RiBVD regimen is in line with that observed for other first-line regimens used in older MCL patients. For instance, in a phase III study comparing the R-CHOP regimen to VR-CAP, Robak and colleagues reported 14/242 deaths (6%) and 11/240 deaths (5%), in the R-CHOP and VR-CAP arms, respectively.7 Of these deaths, a total of six were due to infection and three to cardiac failure. A trial of lenalidomide, bendamustine and rituximab recently documented a complete response rate of 64%, 144

with molecular MRD negativity reached in 34% patients. Toxicity, however, was greater than with the RiBVD regimen, with infectious grade 3-5 toxicities seen in 42% of the 51 recruited patients.13 Two large phase III trials of BR ± ibrutinib or BR ± ACP196 (acalabrutinib) are still ongoing and results are pending. Grade 3-4 hematologic toxicities observed with the RiBVD regimen (51% neutropenia and 35% thrombocytopenia) were in line with those seen in patients treated with other regimens such as R-CHOP (60% neutropenia and 18% thrombocytopenia), VR-CAP (85% and 57%, respectively) or R-BAC (rituximab, bendamustine and cytarabine; 49% and 52%, respectively).2,7,24 Lymphopenia (65% of grade 3-4), which is known to occur with the BR regimen, may contribute, with neutropenia, to the relatively high number of infectious episodes seen in this study.9,10 The rate of lymphopenia at 1 year was 32.5%, which is indicative of longer-term immunosuppression with the RiBVD regimen. Whether this is related to the use of dexamethasone or to the immunosuppressive effects of bendamustine remains to be established but indicates that precautionary measures to control infection are advisable. Contrary to published results for subcutaneous administration of bortezomib, we noted a relatively high incidence of grade 3-4 neurotoxicity, which is a limiting factor for the RiBVD regimen.25 Although the incidence was comparable to that observed in our previous RiPAD+C trial (18% grade 3-4 toxicity, 7/39 patients), in which bortezomib was administered intravenously at the same dose,8 it is higher than that observed in other studies using comparable intravenous doses in which grade 3-4 toxicity was reported in 7% to 8% of patients treated with R-BV (rituximab, bendamustine, and bortezomib) and VR-CAP regimens.7,11 Further investigations will be required to understand the reason for this. Of note in this respect is the discovery of genetic risk loci for severe peripheral neuropathy in European patients with multiple myeloma treated with bortezomib.26 In this study, molecular response in peripheral blood at the end of treatment (after 6 RiBVD cycles) was identified as a major predictive factor for PFS and OS, thus further emphasizing the importance of the depth of response, beyond standard clinical complete response, in MCL.20 Indeed, there was no difference in OS between patients in complete or partial remission at the end of treatment, as defined by the IWG criteria with or without FDG-PET. This finding supports the notion that PET and molecular MRD provide different prognostic information in MCL, probably because they are measuring different types of disease activity, in different disease compartments, with differing sensitivities. The maximum standardized uptake value (SUVmax) defined by FDG-PET, also described as an independent prognosis factor, was not analyzed in our cohort.27 Unexpectedly, neither the MIPI nor Ki67 scores (30% cut-off) had any impact on PFS or OS with the RiBVD regimen. This may reflect differences in treatment efficacy by RiBVD in patients with high-risk MIPI scores (70% OS at 36 months), compared to the efficacy of historical treatment controls in the original patient cohort that was used to define the high-risk MIPI score (40% OS at 36 months).28 Peripheral blood, but not bone marrow-based MRD status, was highly predictive of PFS and OS in this study (4-year OS of 86.6% for MRD-negative patients comhaematologica | 2019; 104(1)

RiBVD regimen as first-line treatment for older MCL patients

pared to 28.6% for MRD-positive patients; (P<0.0001). While for peripheral blood this is broadly in keeping with findings of the EU-MCL network for other treatment regimens,20 results concerning the prognostic impact of bone marrow molecular MRD, in patients treated with RiBVD, differ.20 The prognostic impact of MRD in patients treated with the R-BAC500 regimen has not been reported as yet.24 One avenue of investigation for clarification of these issues will be testing of ‘next generation’ cellular and molecular methods of MRD detection. Multi-parametric flow cytometry, although requiring very high levels of expertise, has been shown to be feasible and provide satisfactory sensitivity, when compared to highly standardized quantitative polymerase chain reaction methods in MCL.29 For molecular MRD, droplet digital polymerase chain reaction analysis is gaining interest30 as is molecular MRD assessment in circulating cell-free DNA in B-cell non-Hodgkin lymphoma.31-33 Combinatorial approaches (metabolic, cellular/molecular) as reported here for MCL, and in follicular lymphoma,34 will also be useful. Ultimately, careful investigation of residual disease, particularly by combinatorial approaches, will be needed to further refine MRD-driven precision medicine approaches in lymphoma, and other cancers.35 MRD positivity at the end of treatment or at 1 year of follow-up was found to be highly predictive for early relapse (at 11 and 26 months, respectively) in patients treated with the RiBVD regimen. Although numbers of MRD-negative patients were small, these findings add further weight to the notion that achieving durable molecular remission is an important goal in MCL. Indeed, maintenance therapy and/or pre-emptive treatment directed to patients in molecular relapse or remaining MRD-positive after treatment has been shown to play a significant role in prolonging clinical response in MCL.2,23,36,37 The choice of maintenance or a pre-emptive therapy strategy may depend on the nature of initial therapy, as highlighted by a

References 1. Campo E, Rule S. Mantle cell lymphoma: evolving management strategies. Blood. 2015;125(1):48-55. 2. Kluin-Nelemans HC, Hoster E, Hermine O, et al. Treatment of older patients with mantle-cell lymphoma. N Engl J Med. 2012;367(6):520-531. 3. Lenz G, Dreyling M, Hoster E, et al. Immunochemotherapy with rituximab and cyclophosphamide, doxorubicin, vincristine, and prednisone significantly improves response and time to treatment failure, but not long-term outcome in patients with previously untreated mantle cell lymphoma: results of a prospective randomized trial of the German Low Grade Lymphoma Study Group (GLSG). J Clin Oncol. 2005;23(9):1984-1992. 4. Howard OM, Gribben JG, Neuberg DS, et al. Rituximab and CHOP induction therapy for newly diagnosed mantle-cell lymphoma: molecular complete responses are not predictive of progression-free survival. J Clin Oncol. 2002;20(5):1288-1294. 5. Baumann U, Fernandez-Saiz V, Rudelius M, et al. Disruption of the PRKCD-FBXO25-

haematologica | 2019; 104(1)

10.

recent study that failed to show the benefit of rituximab maintenance after bendamustine.38 In conclusion, our results identify the combination of rituximab, bendamustine bortezomib and dexamethasone, without maintenance therapy, as a promising treatment option in MCL patients ≥65 years old. The RiBVD regimen compares favorably with other treatment strategies used in this setting, although randomized trials are still lacking. Prolonged PFS appears to result from rapid clearance of (re)circulating tumor B cells in the post-induction phase. Continued molecular remission in the blood was predictive of prolonged survival, indicating that molecular MRD monitoring and molecular response offer significant potential as precision medicine tools for early and late clinical decision-making in MCL. Acknowledgments We thank Dr Jean Marie Quésada, biostatistician at the Grenoble Clinical Investigation Center (INSERM CIC 1406), for his help in data analysis, Valérie Rolland-Neyret and Roseline Delepine, data coordinators of the trial, for their excellent work, and all of the clinicians, not mentioned in the author list, who enrolled patients. We also thank the hematopathologists of the LYSA Pathology Commission for their pathology review (Dr. Danielle Canioni, Dr. Barbara Burroni, Prof. Alexandra TraverseGlehen, Prof. Antoine Martin) and staff of the LYSA-Pathology core facility for providing their expert assistance for immunohistochemistry. We thank Estelle Gimenez for excellent technical coordination of the MRD work. We also thank our partners Mundipharma, Jannsen-Cilag, and Roche for funding. MBC acknowledges additional institutional support from the Université Grenoble-Alpes, Université Bourgogne-Franche Comté, INSERM and CNRS. Research funding via the French National Cancer Institute and the ITMO Cancer ‘Epigenetics and Cancer programme’ is also acknowledged. Finally, we thank Professor André Goy, chairman and executive director of the John Theurer Cancer Center, Hackensack for his critical review of the manuscript.

HAX-1 axis attenuates the apoptotic response and drives lymphomagenesis. Nat Med. 2014;20(12):1401-1409. Goy A, Bernstein SH, Kahl BS, et al. Bortezomib in patients with relapsed or refractory mantle cell lymphoma: updated time-to-event analyses of the multicenter phase 2 PINNACLE study. Ann Oncol. 2009;20(3):520-525. Robak T, Huang H, Jin J, et al. Bortezomibbased therapy for newly diagnosed mantlecell lymphoma. N Engl J Med. 2015;372(10):944-953. Houot R, Le Gouill S, Ojeda Uribe M, et al. Combination of rituximab, bortezomib, doxorubicin, dexamethasone and chlorambucil (RiPAD+C) as first-line therapy for elderly mantle cell lymphoma patients: results of a phase II trial from the GOELAMS. Ann Oncol. 2012;23(6):1555-1561. Rummel MJ, Niederle N, Maschmeyer G, et al. Bendamustine plus rituximab versus CHOP plus rituximab as first-line treatment for patients with indolent and mantle-cell lymphomas: an open-label, multicentre, randomised, phase 3 non-inferiority trial. Lancet. 2013;381(9873):1203-1210. Flinn IW, van der Jagt R, Kahl BS, et al. Randomized trial of bendamustine-ritux-

11.

12.

13.

14.

15.

imab or R-CHOP/R-CVP in first-line treatment of indolent NHL or MCL: the BRIGHT study. Blood. 2014;123(19):29442952. Friedberg JW, Vose JM, Kelly JL, et al. The combination of bendamustine, bortezomib, and rituximab for patients with relapsed/refractory indolent and mantle cell non-Hodgkin lymphoma. Blood. 2011;117 (10):2807-2812. Visco C, Finotto S, Zambello R, et al. Combination of rituximab, bendamustine, and cytarabine for patients with mantle-cell non-Hodgkin lymphoma ineligible for intensive regimens or autologous transplantation. J Clin Oncol. 2013;31(11):14421449. Albertsson-Lindblad A, Kolstad A, Laurell A, et al. Lenalidomide-bendamustine-rituximab in patients older than 65 years with untreated mantle cell lymphoma. Blood. 2016;128(14):1814-1820. Rummel MJ, Al-Batran SE, Kim SZ, et al. Bendamustine plus rituximab is effective and has a favorable toxicity profile in the treatment of mantle cell and low-grade non-Hodgkin's lymphoma. J Clin Oncol. 2005;23(15):3383-3389. Robinson KS, Williams ME, van der Jagt

145

R. Gressin et al.

16.

17.

18.

19.

20.

21.

22.

146

RH, et al. Phase II multicenter study of bendamustine plus rituximab in patients with relapsed indolent B-cell and mantle cell non-Hodgkin's lymphoma. J Clin Oncol. 2008;26(27):4473-4479. Klapper W, Hoster E, Determann O, et al. Ki-67 as a prognostic marker in mantle cell lymphoma-consensus guidelines of the pathology panel of the European MCL Network. J Hematop. 2009;2(2):103-111. 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. Gimenez E, Chauvet M, Rabin L, et al. Cloned IGH VDJ targets as tools for personalized minimal residual disease monitoring in mature lymphoid malignancies; a feasibility study in mantle cell lymphoma by the Groupe Ouest Est d'Etude des Leucemies et Autres Maladies du Sang. Br J Haematol. 2012;158(2):186-197. Pott C, Schrader C, Gesk S, et al. Quantitative assessment of molecular remission after high-dose therapy with autologous stem cell transplantation predicts long-term remission in mantle cell lymphoma. Blood. 2006;107(6):2271-2278. Pott C, Hoster E, Delfau-Larue MH, et al. Molecular remission is an independent predictor of clinical outcome in patients with mantle cell lymphoma after combined immunochemotherapy: a European MCL intergroup study. Blood. 2010;115(16): 3215-3223. Geisler CH, Kolstad A, Laurell A, et al. Nordic MCL2 trial update: six-year followup after intensive immunochemotherapy for untreated mantle cell lymphoma followed by BEAM or BEAC + autologous stem-cell support: still very long survival but late relapses do occur. Br J Haematol. 2012;158(3):355-362. Hermine O, Hoster E, Walewski J, et al. Addition of high-dose cytarabine to immunochemotherapy before autologous stem-cell transplantation in patients aged

23.

24.

25.

26.

27.

28.

29.

30.

65 years or younger with mantle cell lymphoma (MCL younger): a randomised, open-label, phase 3 trial of the European Mantle Cell Lymphoma Network. Lancet. 2016;388(10044):565-575. Le Gouill S, Thieblemont C, Oberic L, et al. Rituximab after autologous stem-cell transplantation in mantle-cell lymphoma. N Engl J Med. 2017;377(13):1250-1260. Visco C, Chiappella A, Nassi L, et al. Rituximab, bendamustine, and low-dose cytarabine as induction therapy in elderly patients with mantle cell lymphoma: a multicentre, phase 2 trial from Fondazione Italiana Linfomi. Lancet Haematol. 2017;4(1):e15-e23. Moreau P, Pylypenko H, Grosicki S, et al. Subcutaneous versus intravenous administration of bortezomib in patients with relapsed multiple myeloma: a randomised, phase 3, non-inferiority study. Lancet Oncol. 2011;12(5):431-440. Magrangeas F, Kuiper R, Avet-Loiseau H, et al. A genome-wide association study identifies a novel locus for bortezomib-induced peripheral neuropathy in european patients with multiple myeloma. Clin Cancer Res. 2016;22(17):4350-4355. Bodet-Milin C, Touzeau C, Leux C, et al. Prognostic impact of 18F-fluoro-deoxyglucose positron emission tomography in untreated mantle cell lymphoma: a retrospective study from the GOELAMS group. Eur J Nucl Med Mol Imaging. 2010;37(9): 1633-1642. Hoster E, Dreyling M, Klapper W, et al. A new prognostic index (MIPI) for patients with advanced-stage mantle cell lymphoma. Blood. 2008;111(2):558-565. Cheminant M, Derrieux C, Touzart A, et al. Minimal residual disease monitoring by 8color flow cytometry in mantle cell lymphoma: an EU-MCL and LYSA study. Haematologica. 2016;101(3):336-345. Drandi D, Ferrero S, Ladetto M. Droplet digital PCR for minimal residual disease detection in mature lymphoproliferative

31.

32.

33.

34.

35.

36.

37.

38.

disorders. Methods Mol Biol. 2018;1768: 229-256. Scherer F, Kurtz DM, Newman AM, et al. Distinct biological subtypes and patterns of genome evolution in lymphoma revealed by circulating tumor DNA. Sci Transl Med. 2016;8(364):364ra155. Scherer F, Kurtz DM, Diehn M, Alizadeh AA. High-throughput sequencing for noninvasive disease detection in hematologic malignancies. Blood. 2017;130(4):440-452. Camus V, Jardin F, Tilly H. The value of liquid biopsy in diagnosis and monitoring of diffuse large B-cell lymphoma: recent developments and future potential. Expert Rev Mol Diagn. 2017;17(6):557-566. Delfau-Larue MH, van der Gucht A, Dupuis J, et al. Total metabolic tumor volume, circulating tumor cells, cell-free DNA: distinct prognostic value in follicular lymphoma. Blood Adv. 2018;2(7):807-816. Luskin MR, Murakami MA, Manalis SR, Weinstock DM. Targeting minimal residual disease: a path to cure? Nat Rev Cancer. 2018;18(4):255-263. Kolstad A, Pedersen LB, Eskelund CW, et al. Molecular monitoring after autologous stem cell transplantation and preemptive rituximab treatment of molecular relapse; results from the mordic mantle cell lymphoma studies (MCL2 and MCL3) with median follow-up of 8.5 years. Biol Blood Marrow Transplant. 2017;23(3):428-435. Andersen NS, Pedersen LB, Laurell A, et al. Pre-emptive treatment with rituximab of molecular relapse after autologous stem cell transplantation in mantle cell lymphoma. J Clin Oncol. 2009;27(26):4365-4370. Rummel MJ, Knauf W, Goener M, et al. Two years rituximab maintenance vs. observation after first line treatment with bendamustine plus rituximab (B-R) in patients with mantle cell lymphoma: first results of a prospective, randomized, multicenter phase II study (a subgroup study of the StiL NHL7-2008 MAINTAIN trial). J Clin Oncol. 2016;34(15_suppl):7503.

haematologica | 2019; 104(1)

ARTICLE

Chronic Lymphocytic Leukemia

Feasibility and efficacy of addition of individualized-dose lenalidomide to chlorambucil and rituximab as first-line treatment in elderly and FCR-unfit patients with advanced chronic lymphocytic leukemia Arnon P. Kater,1 Marinus H.J. van Oers,1 Yvette van Norden,2 Lina van der Straten,3 Julia Driessen,1 Ward F.M. Posthuma,4,5 Martin Schipperus,6 Martine E.D. Chamuleau,7 Marcel Nijland,8 Jeanette K. Doorduijn,9 Michel Van Gelder,10 Mels Hoogendoorn,11 Francien De Croon,12 Shulamiet Wittebol,13 J. Martijn Kerst,14 Erik W.A. Marijt,5 Reinier A.P. Raymakers,15 Martijn R. Schaafsma,16 Johan A. Dobber,17 Sabina Kersting6 and Mark-David Levin3 on behalf of the HOVON CLL study group

Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):147-154

Department of Hematology and Lymphoma and Myeloma Center Amsterdam, Academic Medical Center, Amsterdam; 2Department of Hematology - HOVON Data Center, Erasmus MC Cancer Institute, Rotterdam; 3Department of Internal Medicine, Albert Schweitzer Hospital, Dordrecht; 4Department of Internal Medicine, Reinier de Graaf Hospital, Delft; 5 Department of Hematology, Leiden University Medical Center; 6Department of Hematology, Haga Hospital, the Hague; 7Department of Hematology, VU University Medical Center, Amsterdam; 8Department of Hematology, University Medical Center, Groningen; 9 Department of Hematology, Erasmus MC Cancer Institute, Rotterdam; 10Department of Hematology, Maastricht University Medical Center; 11Department of Internal Medicine, Medical Center, Leeuwarden; 12Department of Internal Medicine, Ikazia Hospital, Rotterdam; 13Department of Internal Medicine, Gelderland Valley Hospital, Ede; 14 Department of Medical Oncology, Antoni van Leeuwenhoek Hospital, Amsterdam; 15 Department of Hematology, University Medical Center, Utrecht; 16Department of Hematology, Medical Spectrum Twente, Enschede and 17Laboratory Special Hematology, Academic Medical Center, Amsterdam, the Netherlands 1

ABSTRACT

enalidomide has been proven to be effective but with a distinct and difficult to manage toxicity profile in the context of chronic lymphocytic leukemia, potentially hampering combination treatment with this drug. We conducted a phase 1-2 study to evaluate the efficacy and safety of six cycles of chlorambucil (7 mg/m2 daily), rituximab (375 mg/m2 cycle 1 and 500 mg/m2 cycles 2-6) and individually-dosed lenalidomide (escalated from 2.5 mg to 10 mg) (induction-I) in first-line treatment of patients with chronic lymphocytic leukemia unfit for treatment with fludarabine, cyclophosphamide and rituximab. This was followed by 6 months of 10 mg lenalidomide monotherapy (induction-II). Of 53 evaluable patients in phase 2 of the study, 47 (89%) completed induction-I and 36 (68%) completed induction-II. In an intention-to-treat analysis, the overall response rate was 83%. The median progressionfree survival was 49 months, after a median follow-up time of 27 months. The 2- and 3-year progression-free survival rates were 58% and 54%, respectively. The corresponding rates for overall survival were 98% and 95%. No tumor lysis syndrome was observed, while tumor flair reaction occurred in five patients (9%, 1 grade 3). The most common hematologic toxicity was grade 3-4 neutropenia, which occurred in 73% of the patients. In conclusion, addition of lenalidomide to a chemotherapy backbone followed by a fixed duration of lenalidomide monotherapy resulted in high remission rates and progression-free survival rates, which seem comparable to those observed with novel drug combinations including novel CD20 monoclonal antibodies or kinase inhibitors. Although lenalidomide-specific toxicity remains a concern, an individualized dose-escalation schedule is feasible and results in an acceptable toxicity profile. EuraCT number: 2010-022294-34. haematologica | 2019; 104(1)

Correspondence: a.p.kater@amc.uva.nl

Received: March 28, 2018. Accepted: August 9, 2018. Pre-published: August 16, 2018. doi:10.3324/haematol.2018.193854 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/147 ÂŠ2019 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.

147

A.P. Kater et al.

Introduction Although important progress has been made in the management of chronic lymphocytic leukemia (CLL) in the last decades, treatment for elderly and unfit patients is still not optimized. Despite clear advantages of the chemotherapeutic drug chlorambucil in elderly patients, with respect to toxicity, oral administration and low costs, its efficacy as a single agent is low in CLL. The CLL11 trial found that the progression-free survival of patients given combination treatment with chlorambucil and rituximab or chlorambucil and obinutuzumab was longer than that of patients given chlorambucil monotherapy (median progression-free survival: 15.4 months and 29.2 months, respectively, versus 11.1 months).1 Since the publication of the findings of this trial, treatment with chlorambucil and an anti-CD20 monoclonal antibody has become the standard, first-line therapy for elderly patients and those unfit for treatment with fludarabine, cyclophosphamide and rituximab (FCR).2 Nevertheless, relapses occur in virtually all patients within 3.5 years.3 A more recent trial compared chlorambucil monotherapy with continuous treatment with the Bruton tyrosine kinase inhibitor ibrutinib and showed improved progression-free survival and overall survival in patients treated in the ibrutinib arm.4 Based on this study, ibrutinib also acquired a label for first-line treatment for previously untreated FCR-unfit patients. However, treatment with targeted inhibitors is not considered curative, most likely because of the pronounced evolutionary capacity CLL cells resulting in the emergence of drug-resistant clones.5 Moreover, prolonged treatment with targeted inhibitors has significant medical, social, and economic costs. It is, therefore, necessary to optimize therapy for elderly and FCR-unfit patients and combine therapies with other mechanisms of action. Lenalidomide is an oral immunomodulatory drug that has multiple mechanisms of action on the immune system. It alters the interaction between CLL cells and the protective microenvironment and stimulates the cytotoxicity of natural killer cells against CLL cells. Lenalidomide restores the immunological synapse between T cells and CLL cells, reversing T-cell dysfunction and enhancing the ability of immune recognition of tumor cells. In addition, lenalidomide directly affects cell proliferation through upregulation

of p21 activity. This effect is independent of the TP53 pathway and could thus be applicable for patients with TP53 dysfunction too.6-8 The overall response rates to lenalidomide as a single agent are low with few complete remissions (overall response rate range: 32%-72%, complete remission rate range: 0%-10%) in previously untreated or treated CLL patients.9-13 Addition of rituximab to lenalidomide resulted in an increased overall response rate of 78% with complete remissions in 11% of patients with relapsed/refractory CLL.14-16 Moreover, addition of rituximab seems to diminish tumor lysis syndrome and tumor flare reaction, which are distinct and difficult to manage toxicity profiles reported in CLL patients treated with lenalidomide.16 Given the key role of the microenvironment in chemoresistance, addition of lenalidomide to chlorambucil and rituximab may result in further improvement of response rates.6 Furthermore, extended duration of treatment has been reported to improve both the overall response rate and quality of responses.16 Currently it is not known whether combination treatment with lenalidomide, rituximab and chlorambucil is feasible in terms of safety and efficacy. As such, we conducted a phase 1-2 study in which six cycles of triple therapy followed by six cycles of lenalidomide monotherapy were tested in elderly and FCR-unfit patients with advanced, previously untreated CLL.

Methods Study design and patients This prospective, open-label study consisted of a phase 1 dose-finding part and a phase 2 efficacy part. Treatment-naïve patients diagnosed with immunophenotypically confirmed CLL, aged 65 – 80 years or 18 – 64 years with a Cumulative Illness Rating Scale score ≥7, in Binet C (Rai III-IV) stage, or with confirmed active disease in Binet A or B (Rai 0-II) stage were enrolled.17 A complete list of inclusion and exclusion criteria is presented in Online Supplementary Table S1.

Treatment All patients were treated with six cycles (every 28 days) of a combination of chlorambucil (p.o. cycles 1-6, days 1 – 7), ritux-

Figure 1. Schedule of phase 2 of the study. *or maximum tolerated dose

148

haematologica | 2019; 104(1)

Chlorambucil, rituximab and lenalidomide in CLL

imab (375 mg/m2 i.v. cycle 1, day 1; 500 mg/m2 i.v. cycles 2-6, day 1) and lenalidomide (induction-I). The dose of lenalidomide was escalated from 2.5 mg to 10 mg during induction-I. Subsequently patients were treated with six cycles of lenalidomide 10 mg p.o. daily (induction-II) (Figure 1). The criteria for discontinuation and restarting lenalidomide and prophylactic treatment are presented in Online Supplementary Tables S2, S3, and S4, respectively.

Study phase 1 This phase of the study focused on determining the maximum tolerated dose and the recommended dose level of chlorambucil in combination with rituximab and lenalidomide. Six patients started combination treatment with chlorambucil at dose level 1 (7 mg). If no more than one dose-limiting toxicity occurred (Online Supplementary Table S5), the dose was escalated to 10 mg (dose level 2) for the subsequent patients, on the basis of which the recommended dose level was established.

Study phase 2 The aim of this phase was to evaluate the efficacy of chlorambucil, at the recommended dose level, in combination with rituximab and lenalidomide, in terms of overall response rate.

Endpoints The protocol endpoints are presented in Online Supplementary Table S6. Responses were determined according to the 2008 International Workshop on CLL criteria17 and evaluated after cycle 3 (clinically), after cycle 6, at the end of the study treatment and during follow-up. The presence of minimal residual disease, as measured in the peripheral blood, was determined centrally using six-color flow cytometry.18 Toxicity was reported according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 4.0. The Cairo-Bishop grading classification19 and CTCAE version 3.020 were used to grade tumor lysis syndrome and tumor flare reaction, respectively.20

Statistical analysis All analyses were performed according to the intention-totreat principle but, in agreement with the protocol, excluding patients not considered eligible in hindsight. The patientsâ&#x20AC;&#x2122; characteristics and treatment toxicity were summarized by descriptive cross-tabulations. Responses were tabulated according to the fraction of the optimal dose of lenalidomide in each cycle (<90% versus â&#x2030;Ľ90%). The Kaplan-Meier method was used for time-to-event analysis. All statistical analyses were performed using STATA Statistical Software version 14.

Ethics Written informed consent was obtained before enrollment in the trial. The study was approved by an accredited Ethical Committee and Institutional Review Board and was performed according to the Declaration of Helsinki, the International Conference on Harmonization Good Clinical Practice Guidelines and the European Union Clinical Trial Directive (2001/20/EG). The study was registered with EuraCT number 2010-022294-34.

Results Between September 20, 2011 and October 18, 2015, 63 previously untreated patients with CLL from 26 centers in the Netherlands were enrolled in this study.

Maximum tolerated dose and recommended dose level of chlorambucil given in combination with rituximab and lenalidomide

Figure 2. Flowchart of number of patients through the study protocol and off protocol, with reasons. Chlor 7: chlorambucil 7 mg/m2; Chlor 10: chlorambucil 10 mg/m2; SLL: small lymphocytic leukemia.

haematologica | 2019; 104(1)

Twelve patients were included in phase 1 of this study aimed at determining the recommended dose level of chlorambucil in combination with rituximab and lenalidomide. In the first dose level group 7 mg/m2 chlorambucil), no dose-limiting toxicity was observed. Subsequently, six patients were included in the second dose level, of 10 mg/m2, of chlorambucil. Again, no doselimiting toxicities were observed at this dose level. There were no differences in the proportions of adverse or severe adverse events between the groups treated with the two different doses. Lenalidomide dose modifications were more frequently applied in the second dose level group. Despite the lack of significant differences in toxicity between the two groups treated with different chlorambucil dosages, the principal investigators, with sup149

A.P. Kater et al.

port of the data safety and monitoring board, decided to continue part 2 of the trial with the recommended dose level of 7 mg/m2, day 1-7, of chlorambucil (dose level 1) based on toxicity reports of an international phase 3 study.9

Phase 2 patients and study treatment For the phase 2 part of the trial, 57 patients were included of whom four were subsequently excluded because they were not eligible in hindsight having been found to have small lymphocytic lymphoma. Combination treatment with chlorambucil, rituximab and lenalidomide was started in 53 patients (induction-I). The patients’ disposition through the trial is summarized in Figure 2 and the clinical, biological and cytogenetic characteristics of the patients are reported in Table 1. The median age of the patients was 71 years (range, 60 – 80). Mutational status could be assessed in 39 patients. The IGVH genes were mutated in 20 patients (51%) and unmutated in 19 patients (49%). Deletion of chromosome 17p was found in eight of 51 (17%) patients. Eight of 50 (15%) and 23 of 50 (43%) patients had deletion of chromosome 11q and 13q, respectively. Forty-seven (89%) patients completed the six planned courses of chlorambucil, rituximab and lenalidomide. Treatment was prematurely discontinued in 11 patients (21%). Reasons for discontinuation were excessive toxic-

Table 1. Patients’ characteristics.

Patients’ characteristics

Number of patients (n=53)

Median age [range], years 71 [60-80] Age, n. (%) <65 years 1 (2) ≥65 to ≤70 years 25 (47) >70 years 27 (51) Male sex, n. (%) 29 (55) RAI stage, n. (%) 0-I 11 (21) II 10 (19) III 22 (42) IV 10 (19) CIRS score − n. (%) ≤6 47 (87) >6 4 (8) Unknown 2 (4) Median CIRS [range] 1 [0-9] Lactate dehydrogenase, n. (%) Lactate dehydrogenase ≤ upper level of normal 33 (62) Lactate dehydrogenase > upper level of normal 18 (34) Unknown 2 (4) Median β2-microglobulin [range] (n=43) 3.8 [1.6-10.4] IGVH mutational status − n. (%) Mutated 20 (38) Unmutated 19 (36) Unknown 14 (26) Cytogenetic abnormalities − n. (%) Del17p 8 out of 51 (17%) Del11q 8 out of 51 (17%) Trisomy 12 12 out of 50 (24%) Del 13q 23 out of 50 (46%) CRS: Cumulative Illness Rating Scale; IGVH: immunoglobulin variable region heavy chain; del: deletion.

150

ity (n=8: 4 cases of skin toxicity, 1 grade 3 allergic reaction, 1 case of neuropathy, 1 acute coronary syndrome and 1 case of mucositis), refusal (n=2) and progression (n=1). Following combination treatment, lenalidomide monotherapy was initiated in 42 patients (79%) (induction-II). Treatment was prematurely discontinued during induction-II in six patients, due to excessive toxicity (n=5: 3 cases of hematologic toxicity in the form of persistent neutropenia and 2 cases of diarrhea with no improvement following dose reductions.) or refusal (n=1). Treatment was completed according to the protocol in 36 (68%) patients. The full dose of lenalidomide during induction-I was given to 76% of the patients in cycle 1, to 57% of the patients in cycle 2, to 50% in cycle 3, 57% in cycle 4, 53% in cycle 5 and 51% in cycle 6 (Figure 3). During cycle 6 of induction-II, 25 (69%) patients received lenalidomide at the full dose. The median dose of lenalidomide according to the prescribed protocol dosing was 86.7% (range, 10%-101%) and 99.7% (range, 25%-104%) in inductionI and induction-II, respectively.

Response evaluation On an intention-to-treat basis, response was analyzed in 53 patients at the end of induction-I. The overall response rate was 83% (95% confidence interval: 72%92%), which resulted in a positive trial based on the phase 2 design as stated in the protocol. The responding patients all achieved a partial response and no complete responses were observed on induction-I (Figure 4). Responses at the end of induction-II were evaluated in all patients (n=42) who started this phase of induction. The overall response rate was 93% (95% confidence interval: 79%-98%) and consisted of 14% complete responses (n=6) and 79% partial responses (n=33). The disease remained stable in two (5%) patients and pro-

Figure 3. Lenalidomide dosing: cumulative dose of lenalidomide compared to optimal dose per treatment cycle during induction-I. Lena: Lenalidomide.

haematologica | 2019; 104(1)

Chlorambucil, rituximab and lenalidomide in CLL

gressed in one (2%) patient. Improvement of the response after induction-II was observed in eight (15%) of all patients who started induction-I: six patients (11%) had an improvement from a partial to complete response and two patients (4%) had an improvement from stable disease to a partial response in induction-II. Flow-based minimal residual disease analysis was performed on peripheral blood in 41 patients. Of these patients, four (8%) achieved minimal residual disease negativity after induction-I and an additional two (4%) after induction-II.

Survival After a median follow-up of 27 months, the median progression-free survival was 49 months (Figure 5A). At 2 and 3 years, 58% [standard error (SE)=8%] and 54% (SE=8%) of the patients, respectively, were alive without progression. The 3-year progression-free survival rate of patients with a deletion of chromosome 17p [del(17p); n=8] was lower than that of patients without this deletion (38% versus 59%, respectively). For patients who started inductionII (n=42), the subsequent progression-free survival was 41 months and the 2-year progression-free survival rate was 56% (SE=9%). The median event-free survival was 49 months and the event-free survival rate at 3 years was 53% (SE=8%) with 13% (SE=5%) non-responders and 34% (SE=8%) with progressive disease. Of the patients with progressive disease, one patient started the next treatment before progressing. During follow-up the median overall survival was not reached with 2- and 3-year overall survival rates of 98% (SE=2%) and 95% (SE=3%), respectively (Figure 5B). With regard to overall survival following induction-II, no deaths had occurred among the patients who started induction-II.

Safety Two patients included in the phase 2 part of the trial developed a grade 4 adverse event, consisting of neutropenic sepsis. These severe adverse events occurred during induction-I in cycle 2 and cycle 5. No grade 4 adverse events were observed during induction-II. Tumor lysis syndrome did not occur. A tumor flare reaction was reported in five patients (9%) and was â&#x2030;¤ grade 2 in four patients and grade 3 in one patient. Six patients developed a secondary malignancy, which was localized skin cancer in all but one. One patient had a solid tumor. Grade 3-4 neutropenia occurred in 39 (73%) and 27 (64%) patients during induction-I and induction-II, respectively, which prompted granulocyte colony-stimulating factor administration as shown per cycle in Online

Figure 4. Responses after induction-I and -II. PD: progressive disease; SD: stable disease; PR: partial response; CR: complete response.

Table 2. Grade 3-4 toxicities.

Number of patients (%) Induction I Grade 3

Grade 4

Grade 3

Induction II Grade 4

Hematologic toxicity Neutropenia Thrombocytopenia Anemia

14 (26%) 5 (9%) 1 (2%)

25 (47%) 3 (6%) -

14 (33%) 5 (12%) -

13 (31%) 2 (5%) -

Other adverse events Any Infections and infestations General disorders and administration site conditions Skin and subcutaneous tissue disorders Respiratory, thoracic and mediastinal disorders Musculoskeletal and connective tissue disorders Nervous system disorders Blood and lymphatic disorders Cardiac disorders Immune system disorders Metabolism and nutrition disorders Renal and urinary disorders Investigations Psychiatric disorders

21 (40%) 5 (9%) 6 (11%) 6 (11%) 3 (6%) 1 (2%) 1 (2%) 1 (2%) 1 (2%) 2 (4%) 2 (4%) 1 (2%) 1 (2%) 1 (2%)

2 (4%) 1 (2%) 1 (2%) -

8 (19%) 4 (10%) 3 (7%) 1 (2%) -

haematologica | 2019; 104(1)

151

A.P. Kater et al. A

Figure 5. Survival outcomes after registration. (A) Progression-free survival and (B) overall survival, with the numbers of patients at risk at 1, 2 and 3 years.

Supplementary Table S7. Neutropenic sepsis occurred in two (4%) patients. Grade 3-4 thrombocytopenia was recorded in 15% and 17% of the patients after inductionI and induction-II, respectively. Despite this, no grade 3-4 bleeds occurred. Grade 3-4 anemia was recorded in 2% of the patients during induction-I. Although all patients received thromboembolic prophylaxis, two patients (4%) had a thromboembolic event (i.e. deep vein thrombosis and thrombophlebitis despite prophylactic aspirin). Grade 3-4 skin and subcutaneous tissue disorders occurred in six patients (11%). Thirteen patients (25%) developed grade 2 skin toxicity. Other grade 3-4 toxicities occurred in 26 patients (49%), of which 23 (44%) during induction-I and eight (19%) during induction-II. The majority of these toxic events were infections (15%), gastrointestinal disorders (15%), or general disorders and administration site conditions (11%) (Table 2). Two patients (4%) have died during disease progression. These deaths occurred 19 and 24 months after registration in the study.

Discussion In this prospective, open label, phase 1/2 study, we investigated the activity of lenalidomide in combination with chlorambucil and rituximab for elderly and FCRunfit patients with previously untreated CLL. After a median follow-up of 27 months, the median progression-free survival in our study was 49 months. The 2- and 3-year progression-free survival rates were 58% and 54%, respectively. The corresponding overall survival rates were 98% and 95%. Since this study is the first to investigate triple treatment with chlorambucil, rituximab and lenalidomide, direct comparison with other clinical trials is limited. Recently, novel first-line regimens have been tested in frail, elderly patients: e.g., chlorambucil as a backbone with novel CD20 monoclonal antibodies and novel chemotherapy-free combinations. The CLL11 trial compared treatment with chlorambucil monotherapy, chlorambucil and rituximab or chlorambucil and obinutuzumab.1 The median progression-free survival was 16.3 months for patients 152

treated with chlorambucil and rituximab and 26.7 months for those treated with chlorambucil and obinutuzumab. The COMPLEMENT-1 trial investigated treatment with chlorambucil and ofatumumab, comparing this with chlorambucil monotherapy.21 After a median follow-up of 29 months, the median progression-free survival was 22 months and the overall response rate was 82% for patients treated with chlorambucil and ofatumumab. With regards to chemotherapy-free combinations of lenalidomide with rituximab, an overall response rate of 88% and a median progression-free survival of 20 months were observed in a phase 2 study.14 In the phase 1b/2 PCYC-1102/1103 studies, treatment with ibrutinib resulted in overall response rates of 87% and 89% in treatment-naĂŻve and relapsed/refractory patients, respectively. The overall response rates were 97%, 89% and 79% in relapsed/refractor patients with a del(11q), complex karyotype and del(17p), respectively. After a median followup of 61.5 months, the median progression-free survival had not been reached in treatment-naĂŻve patients and was 51 months in relapsed/refractory patients.22,23 With all the caveats necessary when comparing different trials, the progression-free survival and overall response rates in the current study seem at least comparable to those observed with chlorambucil in combination with novel CD20 monoclonal antibodies and those observed with kinase inhibitors. The dose of chlorambucil used in our study was 7 mg/m2 on days 1-7 of cycles 1-6. Chlorambucil was administered in combination with rituximab and lenalidomide during six cycles. In the COMPLEMENT-1 trial patients were treated with 10 mg/m2 of chlorambucil on days 1-7 during a minimum of three and a maximum of 12 cycles. In the CLL11 trial, chlorambucil was administered at a dose of 0.5 mg/kg on days 1 and 15 of cycles 1-6. Lenalidomide was started at a dose of 2.5 mg/day, and was steadily increased to 10 mg/day which was maintained until cycle 6, provided there were no dose-limiting toxicities. The escalation scheme of lenalidomide was previously described in combination with rituximab in a phase 2 trial.14 However, in that study lenalidomide was administered for 21 days, followed by a period of rest each cycle. The use of a continuous dose of lenalidomide haematologica | 2019; 104(1)

Chlorambucil, rituximab and lenalidomide in CLL

in combination with rituximab was previously described and was proven to be safe and effective.15,24 Following induction-I, patients received an additional six cycles of monotherapy with lenalidomide 10 mg/day. As previously reported, responses can improve with lenalidomide consolidation treatment.11,25-27 In addition, the CLLM1 study showed that lenalidomide (5-10-15 mg/day) can be efficaciously used as maintenance treatment, prolonging the time to progression as compared with placebo, in first-line patients with CLL who do not achieve minimal residual disease negativity following chemoimmunotherapy induction.25 Although lenalidomide-specific toxicity remains a concern, an individualized dose-escalation schedule is feasible and results in an acceptable toxicity profile and less frequent occurrence of tumor lysis syndrome and tumor flare reactions. Grade 3-4 toxicities were reported in 26 (49%) patients. The most frequently reported toxicities were infections and gastrointestinal disorders. Despite prophylaxis with granulocyte colony-stimulating factor, 64% and 73% of patients developed grade 3-4 neutropenia. Similar percentages were observed in the GIMEMA trial, in which patients were treated with a combination of fludarabine, lenalidomide and rituximab.9 The rate of grade 3-4 toxicities reported with ofatumumab and chlorambucil in the COMPLEMENT-1 trial was 50%,21 which is comparable to the rate in our study. Treatment with ibrutinib, as described in the PCYC-1102/1103 studies,4 was less toxic than treatment with chlorambucil, rituximab and lenalidomide. The ORIGIN study, in which lenalidomide monotherapy was compared with chlorambucil, was stopped prematurely after an imbalance in deaths and treatment-

References 8. 1. Goede V, Fischer K, Busch R, et al. Obinutuzumab plus chlorambucil in patients with CLL and coexisting conditions. N Eng J Med. 2014;370(12):11011110. 2. Hallek M. Chronic lymphocytic leukemia: 2017 update on diagnosis, risk stratification, and treatment. Am J Hematol. 2017;92(9):946-965. 3. Goede V, Fischer K, Engelke A, et al. Obinutuzumab as frontline treatment of chronic lymphocytic leukemia: updated results of the CLL11 study. Leukemia. 2015;29:16024. 4. Burger JA, Tedeschi A, Barr PM, et al. Ibrutinib as initial therapy for patients with chronic lymphocytic leukemia. N Eng J Med. 2015;373(25):2425-2437. 5. Landau DA, Sun C, Rosebrock D, et al. The evolutionary landscape of chronic lymphocytic leukemia treated with ibrutinib targeted therapy. Nat Commun. 2017;8(1): 2185. 6. Kater AP, Tonino SH, Egle A, Ramsay AG. How does lenalidomide target the chronic lymphocytic leukemia microenvironment? Blood. 2014;124(14):2184-2189. 7. Itchaki G, Brown JR. Lenalidomide in the treatment of chronic lymphocytic

haematologica | 2019; 104(1)

10.

11.

12.

13.

emergent adverse events in the lenalidomide arm.10 Based on these observations, lenalidomide monotherapy was not recommended as first-line treatment in CLL patients, particularly in those who are elderly and/or frail. Although our study has shown that adverse events and deaths can be reduced by using individualized dose schedules, intensive monitoring is required. The high death rate observed in the ORIGIN study in lenalidomide-treated patients10 was not replicated in our study. Due to rapid developments, the clinical impact of our study might be limited. Although chemo-immunotherapy is still considered a standard first-line option in CLL patients without del17p/TP53 mutation, phase 2 as well as phase 3 studies examining new chemotherapy-free regimens show a high proportion of minimal residual disease-negative responses.28 Nevertheless, data currently available imply that patients will relapse even following treatment with novel agents. Based on the findings of this study and on lenalidomide’s unique mechanisms of action,6 there might be a role for this drug or for the newer immunomodulatory treatments either in combination with novel agents, or in patients who are not eligible for novel therapies such as ibrutinib. In conclusion, our study showed that triple treatment with chlorambucil, rituximab and lenalidomide is an effective therapy in previously untreated elderly and FCR-unfit patients with CLL. Intensive monitoring is of paramount importance to minimize toxicity and ensure safety. Acknowledgments The authors would like to thank the HOVON109-trial team of the Hovon Data Centre for their help with the trial management and central data management.

leukemia. Expert Opin Investig Drugs. 2017;26(5):633-650. Browning RL, Byrd WH, Gupta N, et al. Lenalidomide induces interleukin-21 production by T cells and enhances IL21-mediated cytotoxicity in chronic lymphocytic leukemia B cells. Cancer Immunol Res. 2016;4(8):698-707. Mauro FR, Carella AM, Molica S, et al. Fludarabine, cyclophosphamide and lenalidomide in patients with relapsed/refractory chronic lymphocytic leukemia. A multicenter phase I–II GIMEMA trial. Leuk Lymphoma. 2017;58(7):1640-1647. Chanan-Khan A, Egyed M, Robak T, et al. Randomized phase 3 study of lenalidomide versus chlorambucil as first-line therapy for older patients with chronic lymphocytic leukemia (the ORIGIN trial). Leukemia. 2017;31(5):1240. Chen CI, Paul H, Wang T, et al. Long term follow up of a phase 2 trial of single agent lenalidomide in previously untreated patients with chronic lymphocytic leukaemia. Br J Haematol. 2014;165(5):731-733. Badoux XC, Keating MJ, Wen S, et al. Lenalidomide as initial therapy of elderly patients with chronic lymphocytic leukemia. Blood. 2011;118(13):3489-3498. Ferrajoli A, Lee B-N, Schlette EJ, et al. Lenalidomide induces complete and partial

14.

15.

16.

17.

remissions in patients with relapsed and refractory chronic lymphocytic leukemia. Blood. 2008;111(11):5291-5297. James DF, Werner L, Brown JR, et al. Lenalidomide and rituximab for the initial treatment of patients with chronic lymphocytic leukemia: a multicenter clinical-translational study from the chronic lymphocytic leukemia research consortium. J Clin Oncol. 2014;32(19):2067-2073. Thompson PA, Rozovski U, Keating MJ, et al. The addition of CD20 monoclonal antibodies to lenalidomide improves response rates and survival in relapsed/refractory patients with chronic lymphocytic leukaemia relative to lenalidomide monotherapy–the MD Anderson Cancer Center experience. Br J Haematol. 2015;171(2):281-284. Ferrajoli A, Badoux XC, O'Brien S, et al. Combination therapy with lenalidomide and rituximab in patients with relapsed chronic lymphocytic leukemia (CLL). Blood. 2009;114(22):206-206. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute–Working Group 1996 guidelines. Blood. 2008;111 (12):5446-5456.

153

A.P. Kater et al. 18. Rawstron A, Böttcher S, Letestu R, et al. Improving efficiency and sensitivity: European Research Initiative in CLL (ERIC) update on the international harmonised approach for flow cytometric residual disease monitoring in CLL. Leukemia. 2013;27(1):142. 19. Cairo MS, Bishop M. Tumour lysis syndrome: new therapeutic strategies and classification. Br J Haematol. 2004;127(1):3-11. 20. Trotti A, Colevas AD, Setser A, et al. CTCAE v3. 0: development of a comprehensive grading system for the adverse effects of cancer treatment. Semin Radiat Oncol. 2003;13(3):176-181. 21. Hillmen P, Robak T, Janssens A, et al. Chlorambucil plus ofatumumab versus chlorambucil alone in previously untreated patients with chronic lymphocytic leukaemia (COMPLEMENT 1): a ran-

154

22.

23.

24.

25.

domised, multicentre, open-label phase 3 trial. Lancet. 2015;385(9980):1873-1883. Cheson B, Bennett J, Grever M, et al. National Cancer Institute-sponsored Working Group guidelines for chronic lymphocytic leukemia: revised guidelines for diagnosis and treatment. Blood. 1996;87(12): 4990-4997. O’Brien S, Furman RR, Coutre S, et al. Single-agent ibrutinib in treatment-naïve and relapsed/refractory chronic lymphocytic leukemia: a 5-year experience. Blood. 2018;131(17):1910-1919. Badoux XC, Keating MJ, Wen S, et al. Phase II study of lenalidomide and rituximab as salvage therapy for patients with relapsed or refractory chronic lymphocytic leukemia. J Clin Oncol. 2013;31(5):584. Fink AM, Bahlo J, Robrecht S, et al. Lenalidomide maintenance after first-line

therapy for high-risk chronic lymphocytic leukaemia (CLLM1): final results from a randomised, double-blind, phase 3 study. Lancet Haematol. 2017;4(10):e475-e486. 26. Shanafelt TD, Ramsay AG, Zent CS, et al. Long-term repair of T-cell synapse activity in a phase II trial of chemoimmunotherapy followed by lenalidomide consolidation in previously untreated chronic lymphocytic leukemia (CLL). Blood. 2013;121(20):41374141. 27. Strati P, Keating MJ, Wierda WG, et al. Lenalidomide induces long-lasting responses in elderly patients with chronic lymphocytic leukemia. Blood. 2013;122(5):734-737. 28. Fischer K, Al-Sawaf O, Fink A-M, et al. Venetoclax and obinutuzumab in chronic lymphocytic leukemia. Blood. 2017;129 (19):2702-2705.

haematologica | 2019; 104(1)

ARTICLE

Plasma Cell Disorders

DOT1L inhibition blocks multiple myeloma cell proliferation by suppressing IRF4-MYC signaling

Kazuya Ishiguro,1,2 Hiroshi Kitajima,2 Takeshi Niinuma,2 Tadao Ishida,3 Reo Maruyama,4 Hiroshi Ikeda,1 Toshiaki Hayashi,1 Hajime Sasaki,1 Hideki Wakasugi,1 Koyo Nishiyama,2 Tetsuya Shindo,2 Eiichiro Yamamoto,1,2 Masahiro Kai,2 Yasushi Sasaki,5 Takashi Tokino,5 Hiroshi Nakase1 and Hiromu Suzuki2

Department of Gastroenterology and Hepatology, Sapporo Medical University School of Medicine; 2Department of Molecular Biology, Sapporo Medical University School of Medicine; 3Department of Hematology, Japanese Red Cross Medical Center, Tokyo; 4 Project for Cancer Epigenomics, Cancer Institute, Japanese Foundation for Cancer Research, Tokyo; 5Department of Medical Genome Sciences, Research Institute for Frontier Medicine and Sapporo Medical University School of Medicine, Japan 1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):155-165

ABSTRACT

pigenetic alterations play an important role in the pathogenesis in multiple myeloma, but their biological and clinical relevance is not fully understood. Here, we show that DOT1L, which catalyzes methylation of histone H3 lysine 79, is required for myeloma cell survival. DOT1L expression levels were higher in monoclonal gammopathy of undetermined significance and smoldering multiple myeloma than in normal plasma cells. Treatment with a DOT1L inhibitor induced cell cycle arrest and apoptosis in myeloma cells, and strongly suppressed cell proliferation in vitro. The anti-myeloma effect of DOT1L inhibition was confirmed in a mouse xenograft model. Chromatin immunoprecipitation-sequencing and microarray analysis revealed that DOT1L inhibition downregulated histone H3 lysine 79 dimethylation and expression of IRF4-MYC signaling genes in myeloma cells. In addition, DOT1L inhibition upregulated genes associated with immune responses and interferon signaling. Myeloma cells with histone modifier mutations or lower IRF4/MYC expression were less sensitive to DOT1L inhibition, but with prolonged treatment, anti-proliferative effects were achieved in these cells. Our data suggest that DOT1L plays an essential role in the development of multiple myeloma and that DOT1L inhibition may provide new therapies for myeloma treatment.

Correspondence: hsuzuki@sapmed.ac.jp

Received: February 15, 2018. Accepted: August 29, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2018.191262

Introduction Multiple myeloma (MM) is a genetically complex disorder caused by monoclonal proliferation of abnormal plasma cells. MM accounts for 1% of all cancers and 10% of hematologic malignancies in the United States, and there are 101,000 deaths per year caused by MM around the world.1 Despite development of a variety of new therapeutic agents, including proteasome inhibitors, immunomodulatory drugs, monoclonal antibodies and histone deacetylase inhibitors, MM remains an incurable disorder.2 Epigenetic alterations such as aberrant DNA methylation and histone modification play key roles in the pathogenesis of MM and are thought to be potential therapeutic targets.3,4 For instance, the histone deacetylase (HDAC) inhibitor panobinostat reportedly exerts synergistic anti-myeloma effects when combined with bortezomib and dexamethasone, yielding a complete or near complete response in 27.6% of patients with relapsed or relapsed and refractory MM.5 Notably, HDAC inhibitors appear to affect a wide variety of non-histone proteins in addition to histones, exerting anti-myeloma effects that include upregulation of CDKN1A and disruption of aggresomes.6 Methylation of histone lysine residues is a major epigenetic mechanism by which chromatin organization and gene expression are regulated.7 For instance, methylation of histone H3 lysine 4 (H3K4), H3K36 and H3K79 is assohaematologica | 2019; 104(1)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/155 ÂŠ2019 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.

155

K. Ishiguro et al.

ciated with active transcription, while methylation of H3K9 and H3K27 are well known to be repressive marks.7,8 Moreover, dysregulation of histone methylation appears to be involved in the pathogenesis of MM. Mutations in genes encoding the histone modifiers H3K27 demethylase UTX (also known as KDM6A); H3K4 methyltransferases MLL, MLL2, and MLL3; H3K9 methyltransferase G9a (also known as EHMT2); and H3K36 methyltransferase MMSET (also known as WHSC1 or NSD2) have been detected in MM.9,10 MMSET is overexpressed in MM with t(4;14), which leads to a global accumulation of H3K36 dimethylation (H3K36me2) and reduction of H3K27me3.11 EZH2 is also reportedly overexpressed in MM, is associated with a poor prognosis, and is considered a potential therapeutic target.12,13 In the present study, we aimed to examine the pathological and therapeutic implications of histone methylation in MM.

plate) were treated with the respective inhibitors at 0.25-1 mM or with DMSO for up to 18 days, refreshing the medium and drug every 3 days.

Xenograft studies For xenograft studies, we used the ex vivo drug pre-treatment method.15,16 RPMI-8226 cells were pre-treated for 3 days with 1 mM SGC0946 or EPZ-5676 or with DMSO, after which 1×107 cells were suspended in 200 ml of RPMI-1640 medium and subcutaneously injected into the bilateral thighs of 6-week-old C.B-17 SCID mice. Tumor size was measured every 3 days using digital calipers, and tumor volume was calculated using the formula, length × width2/2. All animal experiments were conducted in compliance with the protocol approved by the Institutional Animal Care and Use Committee of Sapporo Medical University.

Results Methods Cell lines and clinical specimens MM cell lines (RPMI-8226, MM.1S, KMS-11, KMS-12BM, KMS-12PE and U-266) were obtained and cultured as described previously.14 All cell lines were authenticated using short tandem repeat analysis performed by JCRB (Tokyo, Japan) or BEX (Tokyo, Japan) between 2015 and 2017. Total RNA and genomic DNA were extracted using RNeasy Mini Kits (Qiagen, Hilden, Germany) and QIAamp DNA Mini Kits (Qiagen) according to the manufacturer’s instructions. Specimens of bone marrow or peripheral blood were respectively collected from MM or plasma cell leukemia (PCL) patients, after which CD138-positive cells were isolated using a MACS manual cell separator (Miltenyi Biotec, Bergisch Gladbach, Germany). CD138-positive cells were cultured for 24 hours in RPMI-1640 medium supplemented with 20% fetal bovine serum (FBS) and 1% penicillin/streptomycin/amphotericin B, after which drug treatment and cell viability assays were performed. This study was performed in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Sapporo Medical University. Informed consent was obtained from all patients before specimen collection.

Reagents The H3K4 methyltransferase LSD1 inhibitor S2101 was purchased from Merck Millipore (Burlington, MA, USA). The LSD1 inhibitor GSK2879552, H3K27 methyltransferase EZH2 inhibitor GSK126, and H3K79 methyltransferase DOT1L inhibitor EPZ5676 were all purchased from Chemietek (Indianapolis, IN, USA). The H3K9 methyltransferase G9a inhibitor UNC0638, H3K27 demethylase JMJD3/UTX inhibitor GSKJ1, DOT1L inhibitor SGC0946, and MYC inhibitor 10058-F4 were all purchased from Sigma-Aldrich (St. Louis, MO, USA).

Drug treatment and cell viability assay To screen for anti-proliferative effects of histone methyltransferase or demethylase inhibitors, MM cell lines (3×104 to 1×105 cells/well in 6-well plate) were treated with the respective drugs at a concentration of 1 mM or with DMSO for up to 14 days, refreshing the medium and drugs every 3 to 4 days. Cell viabilities were assessed on days 3-4 and 11-14 using a Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) and a microplate reader (Model 680; Bio-Rad, Hercules, CA, USA) according to the manufacturer’s instructions. To further analyze the effect of DOT1L inhibitors, MM cell lines (2×104 to 8×104 cells/well in 6-well plate) or patientderived CD138-positive cells (1.3×105 to 3×105 cells/well in 6-well 156

DOT1L is a potential therapeutic target in MM To determine whether histone methylation modifiers could be useful therapeutic targets in MM, we first tested the effects of the following compounds on proliferation of MM cell lines: the LSD1 inhibitors S2101 and GSK2879552, the G9a inhibitor UNC0638, the EZH2 inhibitor GSK126, the JMJD3 inhibitor GSKJ1 and the DOT1L inhibitor SGC0946 (Figure 1A). Five MM cell lines were treated with the drugs (1 µM) for up to 14 days, and cell viabilities were assessed early (days 3-4) and late (days 11-14) during the treatment. We found that inhibitors of G9a, EZH2 and DOT1L each moderately suppressed proliferation of more than 2 MM cell lines at the early times (Figure 1A). Longer treatment with these drugs led to stronger growth suppressive effects in most of the MM cell lines, though not in KMS-12PE cells (Figure 1A). By contrast, the LSD1 inhibitors were less effective. We tested 2 LSD1 inhibitors (S2101 for KMS12BM and MM.1S cells and GSK2879552 for RPMI-8226, KMS-12PE and U-266 cells), but neither suppressed MM cell proliferation and appeared to even promote it in most cases (Figure 1A). Among them, we selected DOT1L for further analysis, because its inhibition had a strong antiproliferative effect. Analysis using published data sets revealed that expression of DOT1L is increased during the progression from normal plasma cells (NPCs) to monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SmMM) (Figure 1B). By contrast, we detected no significant difference in the levels of DOT1L expression between SmMM and symptomatic MM (SyMM) (Figure 1B). qRT-PCR showed that DOT1L is expressed at various levels in the MM cell lines tested, irrespective of their sensitivity to the DOT1L inhibitor (Figure 1C).

DOT1L inhibitors induce growth suppression, cell cycle arrest and apoptosis in MM cells To further evaluate the therapeutic potential of DOT1L inhibition in MM, we treated MM cell lines with two DOT1L inhibitors, SGC0946 and EPZ-5676. Western blot analysis using an antibody specific for mono-, di-, and trimethylated H3K79 (H3K79me1/me2/me3) showed that treatment with either drug (1 mM, 3 days) significantly reduced levels of H3K79me1/me2/me3 in RPMI-8226 cells (Online haematologica | 2019; 104(1)

DOT1L is a therapeutic target in myeloma

Supplementary Figure S1) and strongly suppressed proliferation of RPMI-8226, MM.1S, KMS-11 and KMS-12BM cells (Figure 2A). That the drugs were acting through DOT1L inhibition was confirmed by the finding that shRNA-mediated DOT1L knockdown moderately suppressed MM cell proliferation (Online Supplementary Figure S2). On the other hand, DOT1L inhibitors were less effective or ineffective in KMS-12PE and U-266 cells (Figure 2A). Moreover, ex vivo treatment with SGC0946 or EPZ-5676 strongly suppressed tumor formation by MM cells in SCID mice (Figure 2B, Online Supplementary Figure S3). To evaluate the effects of DOT1L inhibitors in

primary tumors, we isolated CD138-positive cells from MM and PCL patients. We found that both drugs moderately suppressed the viability of primary tumor cells (Figure 2C). Cell cycle analysis using flow cytometry revealed that treatment with SGC0946 (1 mM, 6 days) or EPZ-5676 (1 mM, 6 days) led to increases in sub-G1 and G0-G1 phase populations and decreases in S phase populations in RPMI-8226 and MM.1S cells, which suggests DOT1L inhibition induces G1-S arrest and apoptosis (Figure 3A). Induction of apoptosis by DOT1L inhibitors was confirmed using Annexin V staining assays (Figure 3B, Online

haematologica | 2019; 104(1)

Figure 1. Identification of DOT1L as a potential therapeutic target in MM. (A) Effects of inhibitors of histone methylation modifiers on MM cell proliferation. Shown are summarized results of cell viability assays in MM cell lines treated with the indicated drugs (1 mM) at early and late times. Results are normalized to cells treated with DMSO. The data are presented as means of 5 replications; error bars represent standard errors of means (SEMs). Statistical analyses compared cells treated with DMSO and those treated with the indicated drugs using t-tests (unpaired, two-sided). * P<0.05. (B) Comparison of DOT1L mRNA expression among normal plasma cells (NPC, n=22), monoclonal gammopathy of undetermined significance (MGUS, n=44), and smoldering multiple myeloma (SmMM, n=12) (left) and between SmMM (n=24) and symptomatic MM (SyMM, n=69) (right) using the indicated datasets. (C) qRT-PCR analysis of DOT1L in the indicated MM cell lines. Results are normalized to ACTB expression. Shown are means of 3 replications; error bars represent standard errors of means (SEMs).

157

K. Ishiguro et al.

Supplementary Figure S4). Cell cycle and apoptosis were also analyzed at an earlier time point (3 days), and similar results were observed in RPMI-8226 cells (Online Supplementary Figure S5). By contrast, induction of cell cycle arrest and apoptosis were relatively limited in MM.1S cells at this time point, which is consistent with the results of the cell viability assays (Online Supplementary Figure S5, Figure S2A).

DOT1L inhibitors suppress IRF4 and MYC signaling in MM cells To clarify the molecular mechanism underlying the antitumor effects of DOT1L inhibition, we next analyzed H3K79me2 levels and gene expression status in MM cells. Earlier studies showed that 2 to 3 days of DOT1L inhibition resulted in evident depletion of H3K79me2 in tumor cells, while mRNA expression of target genes was signifi-

C Patient Age Sex

Disease

Previous treatment

Days

Days Figure 2. Antitumor effects of DOT1L inhibitors in MM. (A) Results of cell viability assays in MM cell lines treated with the indicated concentrations of DOT1L inhibitors. Results are normalized to untreated cells. Shown are means of 3 replications; error bars represent SEMs. (B) Tumor growth in mice injected with RPMI8226 cells pretreated with SGC0946 or EPZ-5676 (left thigh) or DMSO (right thigh). Growth curves are means of 5 replicates; error bars represent SEMs. (C) Results of cell viability assays in primary tumor cells. CD138-positive cells isolated from MM or PCL patients were treated with DOT1L inhibitors (1 ÂľM) for the indicated periods. A summary of the patients is at the top. Shown are means of 3-6 replications; error bars represent SEMs. sPCL: secondary plasma cell leukemia; MPB: Melphalan + Prednisolone + Bortezomib; Bd: Bortezomib + Dexamethasone; Ld: Lenalidomide + Dexamethasone; MP: Melphalan + Prednisolone.

158

haematologica | 2019; 104(1)

DOT1L is a therapeutic target in myeloma

cantly reduced after 6 to 7 days of treatment.17,18 We therefore performed ChIP-seq analysis of H3K79me2 in RPMI8226 and MM.1S cells treated for 3 days with 1 mM SGC0946 or with DMSO, and gene expression microarray analysis with cells treated with the drug for 6 days. ChIPseq analyses of RPMI-8226 and MM.1S cells respectively identified 4483 and 1590 genes at which H3K79me2 levels were significantly reduced by DOT1L inhibition (Figure 4A, Online Supplementary Tables S1 and S2). Microarray analysis revealed that expression of 912 and 390 genes were downregulated (> 1.5-fold) by SGC0946 in these cells (Online Supplementary Tables S3 and S4). Collectively, we identified 249 genes in RPMI-8226 cells in which both

H3K79me2 and expression levels were significantly decreased by SGC0946, while 67 genes were similarly affected in MM.1S cells (Figure 4A). We also identified 13 genes in which both H3K79me2 and expression levels were decreased in both cell lines and 123 genes in which either H3K79me2 levels or their expression levels were decreased in these cell lines (Figure 4A). Among these, we noted that 4 genes associated with IRF4 and MYC signaling (MYC, IRF4, PRDM1 and KLF2) were affected in both cell lines (Figure 4A). Visualization of the ChIP-seq data clearly revealed decreased H3K79me2 levels in those 4 genes in both cell lines treated with SGC0946, though differential peak call analysis failed to detect KLF2 in RPMI-

Figure 3. Effects of DOT1L inhibitors on cell cycle and apoptosis in MM cells. (A) Results of cell cycle analysis in MM cells treated with the indicated DOT1L inhibitors (1mM, 6 days). Representative results are shown on the left. Summarized results of 3 replications are shown on the right; error bars represent SEMs. (B) Results of apoptosis assays in MM cell lines treated with the indicated DOT1L inhibitors (1mM, 6 days). The results were confirmed in at least 3 independent experiments, and representative results are shown (also see Online Supplementary Figure S4).

haematologica | 2019; 104(1)

159

K. Ishiguro et al. A

Figure 4. Analysis of H3K79me2 and gene expression levels in MM cells treated with DOT1L inhibitors. (A) Integrated analysis of H3K79me2 and gene expression levels in RPMI-8226 and MM.1S cells treated with SGC0946. Left; Venn diagrams of genes whose H3K79me2 or expression levels were suppressed (> 1.5-fold) by SGC0946. Right; Genes indicated in the Venn diagram shown in boxes of their respective colors. Red letters indicate IRF4-MYC signaling genes. Gene expression was assessed in 2 replicates of microarray analyses, and H3K79me2 was assessed in a single ChIP seq analysis. (B) Representative results of ChIP-seq analyses showing decreased H3K79me2 levels at IRF4-MYC signaling genes induced by SGC0946. Results of input DNA are shown below. The numbers on the vertical axis indicate the numbers of sequence reads. Diff Peak: differential peak; TSS: transcription start site. Regions analyzed by ChIP-PCR are indicated by red arrows. (C) ChIP-qPCR analysis of IRF4-MYC signaling genes in MM cells treated with the indicated DOT1L inhibitors (1 mM, 3 days). Results are normalized to respective input DNAs. Shown are means of 3 replications; error bars represent SEMs. An intergenic region located 28 kb upstream of KLF2 was used as a negative control. *P<0.05; NS: not significant. (D) qRT-PCR analysis of IRF4-MYC signaling genes in MM cells treated with the indicated DOT1L inhibitors (1 mM, 6 days). Results are normalized to ACTB expression. Shown are means of 3 replications; error bars represent SEMs. *P<0.05.

160

haematologica | 2019; 104(1)

DOT1L is a therapeutic target in myeloma

8226 cells (Figure 4B). The decreased H3K79me2 levels in these genes in MM cells treated with DOT1L inhibitors were further confirmed by a ChIP-qPCR analysis (Figure 4C). To validate binding of DOT1L to genes marked by H3K79me2, we performed ChIP qPCR and ChIP-seq analysis using an anti-DOT1L antibody. We observed enrichment of DOT1L in the IRF4-MYC signaling genes in both cell lines (Online Supplementary Figure S6). In addition, qRT-PCR analysis confirmed decreased expression of the 4 genes in MM cell lines treated with DOT1L inhibitors, though our microarray analysis failed

to detect suppression of MYC by SGC0946 in MM.1S cells (Figure 4D). We also confirmed decreased levels of MYC protein in MM cell lines treated with DOT1L inhibitors (Online Supplementary Figure S7), and we found that IRF4 expression was downregulated by DOT1L knockdown in MM cells (Online Supplementary Figure S2A).

DOT1L inhibitors affect immune responses and interferon signaling in MM cells In addition to their suppressive effects on H3K79me2 and gene expression, DOT1L inhibitors also increased

Figure 5, DOT1L inhibition affects immune responses and interferon signaling in MM cells. (A) GO analysis of genes of upregulated (> 1.5-fold) by SGC0946 (1 mM, 6 days) in RPMI-8226 and MM.1S cells. (B) Pathway analysis of genes upregulated by SGC0946 in the indicated MM cell lines. (C) qRT-PCR analysis of interferonstimulated genes in MM cells treated with the indicated DOT1L inhibitors. Results are normalized to ACTB expression. Shown are means of 3 replications; error bars represent SEMs. *P<0.05; NS: not significant.

haematologica | 2019; 104(1)

161

K. Ishiguro et al. A

Figure 6. Factors potentially associated with the sensitivity of MM cells to DOT1L inhibitors. (A) Mutations of cancer-related genes detected in KMS-12BM and KMS12PE. Genes shared by both cell lines are indicated by gray letters. fsDel: frameshift deletion. (B) qRT-PCR of MYC and IRF4. Results are normalized to ACTB expression. Shown are means of 3 replications; error bars represent SEMs. (C) Results of cell viability assays in KMS-12BM and KMS-12PE cells treated for 2 days with the indicated concentrations of the MYC inhibitor 10058-F4. Results are normalized to cells treated with DMSO. Shown are means of 3 replications; error bars represent SEMs. (D) qRT-PCR analysis of MYC and IRF4 in KMS-12BM and KMS-12PE cells treated with the indicated DOT1L inhibitors (1 mM, 6 days). Shown are means of 3 replications; error bars represent SEMs. (E) Results of cell viability assays in KMS-12PE and U-266 cells subjected to extended DOT1L inhibitor treatment. Cells were treated with SGC0946 (1 mM) or DMSO for 9 days, after which 3x104 cells were placed in a new 6-well plate and cell viabilities were assessed at the indicated times. Results are normalized to cells treated with DMSO. Shown are means of 3 replications; error bars represent SEMs. (F) Heat map showing genes downregulated (> 1.5-fold) by SGC0946 (1 mM, 12 days) in KMS-12PE cells. (G, H) GO (G) and pathway (H) analyses of genes upregulated (> 1.5-fold) by SGC0946 (1 mM, 12 days) in KMS-12PE cells. (I) Hypothesized mechanism underlying the antitumor effect of DOT1L inhibitors in MM cells.

162

haematologica | 2019; 104(1)

DOT1L is a therapeutic target in myeloma

expression of a number of genes in MM cells. Microarray analysis showed that 1255 probe sets were upregulated (> 1.5-fold) by SGC0946 in RPMI-8226 cells, as were 492 probe sets in MM.1S cells (Online Supplementary Tables S5 and S6). Among them, 143 probe sets (including 125 protein coding genes) were upregulated in both cell lines (Online Supplementary Figure S8A). Gene ontology (GO) analysis suggested that genes involved in “immune system” and “immune response” were significantly enriched among the upregulated genes in both cell lines (Figure 5A), while pathway analysis revealed that genes associated with “interferon (IFN) signaling” were significantly enriched among the upregulated genes (Figure 5B). qRTPCR analysis confirmed that a series of IFN-stimulated genes were upregulated by SGC0946 and EPZ-5676 in MM cells (Figure 5C). However, ChIP-seq analysis showed that H3K79me2 levels were not significantly altered at these genes (Online Supplementary Figure S8B). This suggests DOT1L inhibitors may affect immune responses and IFN signaling through a H3K79me2-independent mechanism in MM cells.

Gene mutations are potentially associated with the DOT1L sensitivity in MM cells Although the KMS-12BM and KMS-12PE cell lines were established from the same patient,19 KMS-12PE cells were less sensitive to DOT1L inhibitors and other epigenetic drugs than KMS-12BM cells (Figures 1A and 2A). To determine the molecular mechanism underlying this difference in drug sensitivity, we carried out targeted sequencing of a panel of cancer-related genes in these cell lines. In the 409 genes analyzed, mutations were detected in 12 genes in KMS-12BM cells, while 14 mutations in 13 genes were detected in KMS-12PE cells (Figure 6A, Online Supplementary Table S7). Among these mutations, 8 were found in both cell lines (Figure 6A). Notably, KMS-12PE cells exhibited mutations in multiple histone modifier genes (EP300, KMT2C and KMT2D), but KMS-12BM cells did not (Figure 6A).

Extended treatment enhances the effect of DOT1L inhibitors in MM cells We next clarified whether IRF4-MYC signaling is associated with the different sensitivities of KMS-12BM and KMS-12PE cells to DOT1L inhibitors. qRT-PCR revealed that MYC and IRF4 were expressed at lower levels in KMS-12PE than KMS-12BM cells, suggesting lower IRF4MYC signaling may be associated with the impaired antitumor effect of DOT1L inhibitors (Figure 6B). Consistent with that idea, KMS-12PE cells were also less sensitive to the MYC inhibitor 10058-F4 than were KMS-12BM cells (Figure 6C). By contrast, RPMI-8226 cells, which were sensitive to DOT1L inhibitors, were also highly sensitive to 10058-F4 (Online Supplementary Figure S9). It is noteworthy, however, that DOT1L inhibitors suppressed expression of MYC and IRF4 in KMS-12PE cells (Figure 6D). Moreover, extended treatment with DOT1L inhibitors for up to 18 days led to strong growth suppression in the less sensitive KMS-12PE and U-266 cell lines (Figure 6E). To clarify the mechanism underlying antitumor effect of the extended treatment, we performed gene expression microarray analysis with KMS-12PE cells treated with SGC0946 or DMSO for 12 days. This analysis identified 509 (401 unique genes) probe sets that were downregulated (> 1.5-fold) and 865 (739 unique genes) haematologica | 2019; 104(1)

that were upregulated (> 1.5-fold) by SGC0946 in KMS12PE cells (Online Supplementary Tables S8 and S9). Among them, multiple IRF4-MYC signaling genes were downregulated by SGC0946 (Figure 6F). Moreover, gene ontology and pathway analyses showed that genes associated with “immune response” and “IFN signaling” were significantly enriched among the upregulated genes (Figure 6G and H). These results demonstrate that the time required for DOT1L inhibitors to exert their effects varies among MM cells.

Discussion In the present study, we show that DOT1L inhibitors exerted a strong anti-myeloma effect. Expression of DOT1L is significantly higher in SmMM than NPC, suggesting DOT1L may be causally associated with myelomagenesis. DOT1L is the only known histone methyltransferase that catalyzes mono-, di- and trimethylation at H3K79.20,21 In mammals, most of H3K79 is unmethylated, and H3K79 methylation is linked to active transcription.20,22 DOT1L is a component of large transcription complexes that also include transcription factors such as AF4, AF9, AF10, ENL and P-TEFb.20,23-26 Within these complexes, DOT1L may initiate or sustain active transcription by mediating H3K79 methylation. DOT1L is also a potential drug target in mixed lineage leukemia (MLL) gene rearranged leukemia. DOT1L forms a complex with MLL fusion proteins, and DOT1L-mediated H3K79 methylation leads to enhanced expression of target oncogenes, including HOXA9 and MEIS1.26,27 Recent studies have demonstrated the selective and strong antitumor effects of DOT1L inhibitors against MLL-rearranged leukemia.17,18,28 Similarly, DOT1L is a potential therapeutic target in lung and breast cancer with high DOT1L expression and neuroblastoma with MYCN amplification.29-31 We found that DOT1L inhibition targets the IRF4-MYC axis in MM cells (Figure 6I). Aberrant activation of several transcription factors, including MYC, MAF, NF-κB and IRF4, is involved in the development of MM.4 MM cell survival is strongly dependent on IRF4 and MYC, and MYC is a direct target gene of IRF4 transactivation, while IRF4 is a direct target of MYC.32,33 The IRF4-MYC axis is thus considered to be an important therapeutic target in MM, and a recent study showed that CBP/EP300 bromodomain inhibitors directly suppress IRF4 expression and inhibit MM cell viability.34 Moreover, dependence on the KDM3A-KLF2-IRF4 axis was recently reported in MM.35 KDM3A maintains KLF2 and IRF4 expression via H3K9 demethylation, and KLF2 directly targets IRF4 while IRF4 reciprocally activates KLF2, forming a positive autoregulatory circuit.35 We found that DOT1L inhibition leads to decreased levels of H3K79me2 and repression of IRF4 and its target genes, including MYC, PRDM1 (also known as BLIMP1) and KLF2 in MM cells.32 As previously shown, H3K79me2 peaked just behind the transcriptional start site of the active genes and gradually declined over the course of the gene body,20,36 and it was significantly depleted in MM cells treated with a DOT1L inhibitor. We found that genes associated with immune responses and IFN signaling were significantly upregulated by DOT1L inhibition in MM cells. The potential of IFN in the clinical treatment of MM has long been recognized. 163

K. Ishiguro et al.

Interferon-alpha (IFN-α) reportedly induces apoptosis and inhibits growth in MM cell lines.37 In the 1980s, IFN-α was used as monotherapy in MM, with an overall response rate of 15-20%.38 In this study, we found that a series of INF-stimulated genes were upregulated in MM cells by DOT1L inhibitors, which may contribute to the drugs’ anti-myeloma effects. The mechanism underlying the activation of immune response genes and INF signaling is unclear. One possible mechanism is that DOT1L inhibition causes DNA damage that leads to stimulation of IFN signaling.39 An earlier study reported that IRF4 functions as a repressor by binding to the IFN-stimulated response elements of several genes, including IFN-stimulated gene 15 (ISG15), which suggests downregulation of IRF4 by DOT1L inhibition may have caused upregulation of ISG15.40 More recent studies showed that DNA methyltransferase inhibitors stimulate an interferon response by inducing endogenous double strand RNAs, which contribute to the antitumor effect through DNA demethylation.41,42 It is presently unclear whether DOT1L inhibition exerts similar effects in MM cells, however. Finally, we investigated the mechanism determining DOT1L sensitivity by focusing on two MM cell lines with different sensitivities to DOT1L inhibitors (KMS-12BM and KMS-12PE). These cell lines were both established from a 64-year-old female MM patient, KMS-12BM from bone marrow and KMS-12PE from pleural effusion.19 Targeted sequencing of a panel of cancer-related genes revealed that, although many of the mutations were present in both cell lines, the less sensitive KMS-12PE cells har-

References 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin. 2017; 67(1):730. 2. Dingli D, Ailawadhi S, Bergsagel PL, et al. Therapy for relapsed multiple myeloma: Guidelines from the Mayo stratification for myeloma and risk-adapted therapy. Mayo Clin Proc. 2017;92(4):578-598. 3. Dimopoulos K, Gimsing P, Gronbaek K. The role of epigenetics in the biology of multiple myeloma. Blood Cancer J. 2014; 4:e207. 4. Morgan GJ, Walker BA, Davies FE. The genetic architecture of multiple myeloma. Nat Rev Cancer. 2012;12(5):335-348. 5. San-Miguel JF, Hungria VT, Yoon SS, et al. Panobinostat plus bortezomib and dexamethasone versus placebo plus bortezomib and dexamethasone in patients with relapsed or relapsed and refractory multiple myeloma: a multicentre, randomised, double-blind phase 3 trial. Lancet Oncol. 2014; 15(11):1195-1206. 6. Hideshima T, Richardson PG, Anderson KC. Mechanism of action of proteasome inhibitors and deacetylase inhibitors and the biological basis of synergy in multiple myeloma. Mol Cancer Ther. 2011;10(11):2034-2042. 7. Mozzetta C, Boyarchuk E, Pontis J, Ait-SiAli S. Sound of silence: the properties and functions of repressive Lys methyltransferases. Nat Rev Mol Cell Biol. 2015; 16(8):499513. 8. Black JC, Van Rechem C, Whetstine JR. Histone lysine methylation dynamics: establishment, regulation, and biological impact.

164

bored mutations in the histone modifier genes EP300, KMT2C and KMT2D, which were absent in KMS-12BM cells. A recent study reported that mutations in epigenetic modifier genes were more frequently found in previously treated MM patients than in newly diagnosed patients, suggesting these mutations are associated with disease progression and chemoresistance.10 Thus, the mutation of epigenetic modifier genes in KMS-12PE cells may be associated with their reduced sensitivity to DOT1L inhibitors. In addition, we also noted that KMS-12PE cells express MYC and IRF4 at lower levels than KMS-12BM cells, suggesting decreased dependence on the IRF4-MYC axis may lead to lower sensitivity to DOT1L inhibitors in KMS-12PE cells. It is noteworthy, however, that prolonged treatment with DOT1L inhibitors exerted suppressive effects on IRF4/MYC expression and cell viability in the less sensitive MM cell lines. Our results suggest that DOT1L is a promising therapeutic target in MM, and further exploration of DOT1L inhibitors for MM treatment is warranted. Acknowledgments The authors thank Dr. William F. Goldman for editing the manuscript and Ms. Mutsumi Toyota and Ms. Tomo Hatahira for technical assistance. Funding This study was supported in part by Grant-in-Aid for Scientific Research (C) from the Japan Society for Promotion of Science (JSPS KAKENHI 15K09456, T. Ishida) and Takeda Science Foundation (2018, T. Niinuma).

Mol Cell. 2012;48(4):491-507. 9. Chapman MA, Lawrence MS, Keats JJ, et al. Initial genome sequencing and analysis of multiple myeloma. Nature. 2011; 471(7339):467-472. 10. Pawlyn C, Kaiser MF, Heuck C, et al. The spectrum and clinical impact of epigenetic modifier mutations in myeloma. Clin Cancer Res. 2016;22(23):5783-5794. 11. Martinez-Garcia E, Popovic R, Min DJ, et al. The MMSET histone methyl transferase switches global histone methylation and alters gene expression in t(4;14) multiple myeloma cells. Blood. 2011;117(1):211-220. 12. Hernando H, Gelato KA, Lesche R, et al. EZH2 inhibition blocks multiple myeloma cell growth through upregulation of epithelial tumor suppressor genes. Mol Cancer Ther. 2016;15(2):287-298. 13. Pawlyn C, Bright MD, Buros AF, et al. Overexpression of EZH2 in multiple myeloma is associated with poor prognosis and dysregulation of cell cycle control. Blood Cancer J. 2017;7(3):e549. 14. Nojima M, Maruyama R, Yasui H, et al. Genomic screening for genes silenced by DNA methylation revealed an association between RASD1 inactivation and dexamethasone resistance in multiple myeloma. Clin Cancer Res. 2009;15(13):4356-4364. 15. Chen CW, Koche RP, Sinha AU, et al. DOT1L inhibits SIRT1-mediated epigenetic silencing to maintain leukemic gene expression in MLL-rearranged leukemia. Nat Med. 2015;21(4):335-343. 16. Lu R, Wang P, Parton T, et al. Epigenetic perturbations by Arg882-mutated DNMT3A potentiate aberrant stem cell gene-expres-

17.

18.

19.

20.

21.

22.

23.

24.

25.

sion program and acute leukemia development. Cancer Cell. 2016; 30(1):92-107. Daigle SR, Olhava EJ, Therkelsen CA, et al. Potent inhibition of DOT1L as treatment of MLL-fusion leukemia. Blood. 2013; 122(6):1017-1025. Yu W, Chory EJ, Wernimont AK, et al. Catalytic site remodelling of the DOT1L methyltransferase by selective inhibitors. Nat Commun. 2012;3:1288. Namba M, Ohtsuki T, Mori M, et al. Establishment of five human myeloma cell lines. In Vitro Cell Dev Biol. 1989; 25(8):723729. Vlaming H, van Leeuwen F. The upstreams and downstreams of H3K79 methylation by DOT1L. Chromosoma. 2016; 125(4):593605. Feng Q, Wang H, Ng HH, et al. Methylation of H3-lysine 79 is mediated by a new family of HMTases without a SET domain. Curr Biol. 2002;12(12):1052-1058. Steger DJ, Lefterova MI, Ying L, et al. DOT1L/KMT4 recruitment and H3K79 methylation are ubiquitously coupled with gene transcription in mammalian cells. Mol Cell Biol. 2008;28(8):2825-2839. Bitoun E, Oliver PL, Davies KE. The mixedlineage leukemia fusion partner AF4 stimulates RNA polymerase II transcriptional elongation and mediates coordinated chromatin remodeling. Hum Mol Genet. 2007; 16(1):92-106. Mueller D, Bach C, Zeisig D, et al. A role for the MLL fusion partner ENL in transcriptional elongation and chromatin modification. Blood. 2007;110(13):4445-4454. Mueller D, Garcia-Cuellar MP, Bach C, Buhl

haematologica | 2019; 104(1)

DOT1L is a therapeutic target in myeloma

26. 27.

28.

29.

30.

31.

S, Maethner E, Slany RK. Misguided transcriptional elongation causes mixed lineage leukemia. PLoS Biol. 2009;7(11):e1000249. Okada Y, Feng Q, Lin Y, et al. hDOT1L links histone methylation to leukemogenesis. Cell. 2005;121(2):167-178. Krivtsov AV, Feng Z, Lemieux ME, et al. H3K79 methylation profiles define murine and human MLL-AF4 leukemias. Cancer Cell. 2008;14(5):355-368. Daigle SR, Olhava EJ, Therkelsen CA, et al. Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor. Cancer Cell. 2011;20(1):53-65. Kim W, Kim R, Park G, Park JW, Kim JE. Deficiency of H3K79 histone methyltransferase Dot1-like protein (DOT1L) inhibits cell proliferation. J Biol Chem. 2012; 287(8):5588-5599. Zhang L, Deng L, Chen F, et al. Inhibition of histone H3K79 methylation selectively inhibits proliferation, self-renewal and metastatic potential of breast cancer. Oncotarget. 2014;5(21):10665-10677. Wong M, Tee AEL, Milazzo G, et al. The histone methyltransferase DOT1L promotes neuroblastoma by regulating gene

haematologica | 2019; 104(1)

32. 33. 34.

35.

36.

37.

transcription. Cancer Res. 2017;77(9):25222533. Shaffer AL, Emre NC, Lamy L, et al. IRF4 addiction in multiple myeloma. Nature. 2008;454(7201):226-231. Holien T, Vatsveen TK, Hella H, Waage A, Sundan A. Addiction to c-MYC in multiple myeloma. Blood. 2012;120(12):2450-2453. Conery AR, Centore RC, Neiss A, et al. Bromodomain inhibition of the transcriptional coactivators CBP/EP300 as a therapeutic strategy to target the IRF4 network in multiple myeloma. Elife. 2016;5. Ohguchi H, Hideshima T, Bhasin MK, et al. The KDM3A-KLF2-IRF4 axis maintains myeloma cell survival. Nat Commun. 2016; 7:10258. Suzuki H, Takatsuka S, Akashi H, et al. Genome-wide profiling of chromatin signatures reveals epigenetic regulation of MicroRNA genes in colorectal cancer. Cancer Res. 2011;71(17):5646-5658. Minami R, Muta K, Ilseung C, Abe Y, Nishimura J, Nawata H. Interleukin-6 sensitizes multiple myeloma cell lines for apoptosis induced by interferon-alpha. Exp Hematol. 2000;28(3):244-255.

38. Zhang L, Tai YT, Ho MZG, Qiu L, Anderson KC. Interferon-alpha-based immunotherapies in the treatment of B cell-derived hematologic neoplasms in today's treat-to-target era. Exp Hematol Oncol. 2017;6:20. 39. Cheon H, Borden EC, Stark GR. Interferons and their stimulated genes in the tumor microenvironment. Semin Oncol. 2014; 41(2):156-173. 40. Rosenbauer F, Waring JF, Foerster J, Wietstruk M, Philipp D, Horak I. Interferon consensus sequence binding protein and interferon regulatory factor-4/Pip form a complex that represses the expression of the interferon-stimulated gene-15 in macrophages. Blood. 1999;94(12):42744281. 41. Chiappinelli KB, Strissel PL, Desrichard A, et al. Inhibiting DNA methylation causes an interferon response in cancer via dsRNA including endogenous retroviruses. Cell. 2015;162(5):974-986. 42. Roulois D, Loo Yau H, Singhania R, et al. DNA-demethylating agents target colorectal cancer cells by inducing viral mimicry by endogenous transcripts. Cell. 2015; 162(5):961-973.

165

ARTICLE Ferrata Storti Foundation

Platelet Biology & its Disorders

Clinical factors and biomarkers predict outcome in patients with immune-mediated thrombotic thrombocytopenic purpura Elizabeth M. Staley,†1 Wenjing Cao,†1 Huy P. Pham,2 Chong H. Kim,3 Nicole K. Kocher,1 Lucy Zheng,1 Radhika Gangaraju,4 Robin G. Lorenz,1 Lance A. Williams,1 Marisa B. Marques1 and X. Long Zheng1§

Division of Laboratory Medicine, Department of Pathology, The University of Alabama at Birmingham, AL; 2Department of Pathology, Keck School of Medicine of the University of Southern California, Los Angeles, CA; 3Department of Clinical Pharmacy, University of Colorado Anschutz Medical Campus, Aurora, CO and 4Division of Hematology and Oncology, Department of Medicine, The University of Alabama at Birmingham, AL, USA 1

Haematologica 2019 Volume 104(1):166-175

EMS and WC Contributed equally to this work.

ABSTRACT

Correspondence: xzheng@uabmc.edu or longzheng01@gmail.com Received: May 23, 2018. Accepted: August 23, 2018. Pre-published: August 31, 2018. doi:10.3324/haematol.2018.198275 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/166 ©2019 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.

166

mmune-mediated thrombotic thrombocytopenic purpura is characterized by severe thrombocytopenia and microangiopathic hemolytic anemia. It is primarily caused by immunoglobin G type autoantibodies against ADAMTS13, a plasma metalloprotease that cleaves von Willebrand factor. However, reliable markers predictive of patient outcomes are yet to be identified. Seventy-three unique patients with a confirmed diagnosis of immune-mediated thrombotic thrombocytopenic purpura between April 2006 and December 2017 were enrolled from the Univeristy of Alabama at Birmingham Medical Center. Clinical information, laboratory values, and a panel of special biomarkers were collected and/or determined. The results demonstrated that the biomarkers associated with endothelial injury (e.g., von Willebrand factor antigen and collagen-binding activity), acute inflammation (e.g., human neutrophil peptides 1-3 and histone/deoxyribonucleic acid complexes), and activation of the complement alternative pathway (e.g., factors Bb and iC3b) were all significantly increased in patients with acute immune-mediated thrombotic thrombocytopenic purpura compared to those in the healthy controls. Moreover, failure to normalize platelet counts within 7 days or failure to markedly reduce serum lactate dehydrogenase by day 5, low total serum protein or albumin, and high serum troponin levels were also predictive of mortality, as were the prolonged activated partial thromboplastin time, high fibrinogen, and elevated serum lactate dehydrogenase, Bb, and sC5b-9 on admission. These results may help to stratify patients for more intensive management. The findings may also provide a framework for future multicenter studies to identify valuable prognostic markers for immune-mediated thrombotic thrombocytopenic purpura.

Introduction Immune-mediated thrombotic thrombocytopenic purpura (iTTP) is a rare, but lifethreatening, hematologic disorder.1,2 It is characterized by severe thrombocytopenia and microangiopathic hemolytic anemia (MAHA), with or without end organ damage. The underlying pathophysiology of iTTP is a functional deficiency of plasma ADAMTS13 activity, resulting from autoantibodies targeting plasma ADAMTS13, a metalloprotease that cleaves von Willebrand factor (VWF).3-5 Therapeutic plasma exchange (TPE) remains the standard of care, in conjunction with immunosuppressive therapies that include corticosteroids and rituximab to inhibit acute inflammation and autoantibody production.1,6 However, an in-hospital mortality rate remains as high as ~20%7,8 or less than 10% following the introduction of a novel therapy caplacizumab, an anti-VWF nanobody;9 nearly 30% of surviving patients may experience disease exacerbation10 and/or relapse.7 Currently, clinical factors and biomarkers predictive of clinical course/outcome are scanty, and their predictive values have yet to be established in diverse patient populations. haematologica | 2019; 104(1)

Prognostic markers in autoimmune TTP

Demographic features such as race, gender and age are shown to associate with disease prevalence and severity. For instance, iTTP occurs more commonly in AfricanAmerican females11,12 and, perhaps not surprisingly, older age (>60 years) is associated with an increased mortality.12,13 Additionally, serum levels of creatine kinase-muscle/brain (CK-MB), troponin I,14 lactate dehydrogenase (LDH), ADAMTS13 antigen or activity levels, antiADAMTS13 antibody levels13 and, more recently, the platelet recovery rate15 are shown to be associated with increased mortality. In this study, we describe the Alabama cohort of 73 unique patients with confirmed diagnosis of iTTP selected from a total of 142 admissions. This cohort of patients was primarily from the Southeastern United States. Clinical information, laboratory values, and various biomarkers were collected and analysed with respect to their associations with admission type, disease severity, and mortality.

Methods Patients The Institutional Review Board (IRB) of the University of Alabama at Birmingham (UAB) has approved the study protocol. UAB medical center serves as a referral center for the diagnosis and management of patients with thrombotic microangiopathy (TMA) for the state of Alabama and several neighboring states in the Southeast United State of America. Some patients were initially seen by a primary care physician, local internist, or hematologist. If TMA was suspected, patients were referred to the UAB Medical Center for further evaluation and treatment, which may have involved a delay in diagnosis and treatment of one to several days. There were also patients who came directly to the UAB Emergency Department (ED). Within hours of arrival at UAB, a central intravenous catheter was inserted, blood samples were collected for laboratory tests including ADAMTS13 activity and inhibitors, and therapeutic plasma exchange (TPE) was urgently initiated. Seventy-three patients at the UAB Medical Center, between April 2006 and December 2017, were included in this study. Control samples were collected from healthy individuals (age 27-69 years), both male (1/3) and female (2/3), representing the local ethnic population, who did not have a history of hematological diseases, malignancy, and acute inflammatory disorders. Whole blood was anticoagulated with 3.2% sodium citrate; plasma was separated within four hours of collection, and stored at -80Â°C prior to analysis. Clinical data pertinent to each patient, including demographic information, past and current medical history, signs and symptoms on admission, laboratory test results, presumptive and final diagnosis, hospital-course, outcome and long-term follow up, were collected by a physician and maintained in the Alabama Registry Database.

Exclusion and inclusion criteria Patients were excluded from analysis if their final diagnosis was determined to be an alternative TMA, for example: atypical hemolytic uremic syndrome (aHUS), congenital TTP, HIV-related thrombocytopenia, HELLP syndrome, a life-threatening condition during pregnancy with clinical features of hemolysis, elevated liver enzymes, and low platelet count,16 TMA following solid organ or hematopoietic stem cell transplantation, druginduced TTP (i.e., clopidogrel, ticlopidine or gemcitabine) and/or sepsis. Additionally, we excluded patients who were treated haematologica | 2019; 104(1)

Table 1. Demographics, clinical presentations, and comorbidities in 73 unique patients with iTTP.

Parameters Age (Years) Sex Female Male Ethnicity African American Caucasian Afro-hispanic Presenting symptoms CNS symptoms Abdominal pain Chest pain Comorbidities Hypertension Diabetes mellitus Systemic lupus erythematosus Pregnancy Smoking Drug use

Values* 41 (32, 52) 41 (56%) 32 (44%) 56 (76.7%) 16 (21.9%) 1 (1.4%) 40 (54.7%) 28 (38.4%) 7 (9.6%) 38 (52.1%) 13 (17.8%) 8 (11.0%) 3 (4.1%) 36 (49.3%) 15 (20.5%)

*All values are expressed as the number and percentage of patients (in parenthesis) in each category except for age, which is expressed as the median and 95% confidence interval.

prior to sample drawn with plasma infusion (>3L) and/or TPE and those in remission. Thus, this cohort includes patients experiencing their first episode or an exacerbation, or a relapse (only if the sample from the initial episode was not available). Confirmatory tests for ADAMTS13 activity and inhibitors were performed at the Blood Center of Wisconsin (Milwaukee, WI, USA).

Assays for ADAMTS13 activity, inhibitors, and anti-ADAMTS13 IgG Plasma ADAMTS13 activity and inhibitor titers were determined using a commerical FRETS-VWF73 assay17 and an inhouse FRETS-based assay as previously described.18 Plasma antiADAMTS13 IgG was determined by an enzyme-linked immunosorbent assay (ELISA) (Diapharma, West Chester, Ohio) in accordance with the manufacturerâ&#x20AC;&#x2122;s recommendations.

Assays for plasma VWF antigen and collagen-binding activity Plasma VWF antigen (VWF-Ag) and collagen-binding activity (VWF-CBA) levels were determined using in-house ELISA-based assays as previously described.19

Assays for complement activation and inflammatory markers Plasma levels of complement activation markers including iC3b, sC5b-9, Bb, and C4d were determined using a commercial ELISA assays (MicroVue, San Diego, CA, USA) following manufacturer's instructions.8 Plasma HNP1-3 levels were also determined by an ELISA assay, which recognizes all HNP1-3 (Hycult Biotech, Plymouth Meeting, PA).8 Finally, plasma histone-DNA complexes were quantified by the ELISA assay previously described (Roche, Indianapolis, IN, USA).20 167

E.M. Staley et al.

Statistical Analysis The data were expressed as the medians and 95% confidence intervals (95% CI) for most parameters unless specified otherwise. Mann-Whitney (for two groups) and Kruskal-Willis tests (for multiple groups) for continuous variables and Fisher exact test for categorical variables were performed, with descriptive statistics to summarize both quantitative and categorical variables. Association between various demographic, clinical, and laboratory parameters with clinical outcomes such as mortality, remission, exacerbation, relapse, etc. were also determined using univariate analysis and spearman correlation tests. Furthermore, Cox proportional hazard

regression was used to determine the hazard ratios for the predictive variables. P-values of <0.05 and <0.01 were deemed statistically significant and highly significantly, respectively.

Results Clinical and admission laboratory characteristics of the Alabama iTTP cohort The majority of our 73 iTTP patients came from the surrounding areas of the cities of Montgomery (n=16, 22%),

Figure 1. The Alabama TTP Cohort. A. A map of the state of Alabama denotes the geographic distribution of the 73 unique patients with iTTP that were included in the study. Five patients residing in Oregon, Ohio, New York, Georgia, and Mississippi are not shown on the map. The location of the UAB Medical Center is marked with a blue star. B. Algorithm demonstrates the number of admissions, patients being excluded, and the final cohort of patients in the study.

Table 2. Therapeutic interventions and observed long-term outcomes in 73 patients with iTTP.

Outcome Rem. Exac/Rel.-Rem. Exac.& Rel.-Rem. Death Total

N (%) 22 (30.1) 24 (32.9) 18 (24.7) 9 (12.3) 73 (100)

#TPE (95% CI)$ 8 (6-12) 16.5 (13-21)* 31.5 (18-46)**** 6 (1-12)n.s. 15 (12-17)

Steriods N (%)

Rituxan N (%)

Vincr. N (%)

Splen. N (%)

Follow up$ (days) (95% CI)

19 (26.2) 19 (26.2) 17 (23.3) 4 (5.4) 69 (94.5)

3 (4.1) 11 (15.1) 10 (13.7) 4 (5.4) 28 (38.4)

0 (0) 2 (2.7) 2 (2.7) 1 (1.4) 5 (6.8)

1 (1.4) 0 (0) 3 (4.1) 0 (0) 4 (5.4)

1090 (73-2631) 686 (211-1690) 1587 (115-2390) 817 (118-2390) 828 (326-1720)

N: number of patients; $: the number of therapeutuc plasma exchange (TPE) expressed as the median and 95% confidence interval (CI). * and **** stars indicate the P valules <0.05 and P<0.0001, respectively. n.s. refers to not statistically signficant or P value >0.05. Rem., exac., and rel., are remission, exacerbation, and relapse, respectively. Steroids: corticosteriods; Rituxan: rituximab; Vincr.: vincristine; Splen.: splenectomy.

168

haematologica | 2019; 104(1)

Prognostic markers in autoimmune TTP

Birmingham (n=13, 17.8%), and Tuscaloosa (n=9, 12.3%). The remaining patients were from other locations in the state of Alabama with a small number coming from other states including Mississippi, Georgia, New York, Oregon, and Ohio (Figure 1A). Thus, this cohort represents a

patient population primarily from the southeastern part of the United States of America. Since inclusion criteria were designed to collect only unique admission events for a patient experiencing an acute iTTP, only 73 unique patients of 142 admissions for the diagnosis and treatment

A A

B B

Figure 2. Plasma ADAMTS13 activity and autoantibodies in 73 unique patients with iTTP. Plasma levels of ADAMTS13 activity (A) and functional inhibitor titers (B) in patients with iTTP compared with those in the healthy controls. Additionally, plasma anti-ADAMTS13 IgG levels in iTTP patients with negative (<0.4 U/mL) and positive (>=0.4 U/mL) inhibitors are shown (C). Each individual dot represents a single patient with the median Âą 95% confidence intervals (solid red lines). Mann-Whitney was used to determine the statistical significance between two groups. Here *, **, ***, and **** indicate the P values of <0.05, <0.001, 0.0005, and <0.0001, respectively; n.s. stands for no statistical difference between two groups.

haematologica | 2019; 104(1)

Figure 3. Plasma levels of VWF antigen and collagen-binding activitity in patients with iTTP. Plasma VWF antigen (VWF-Ag) (A), collagen-binding activity (VWF-CBA) (B), and the ratio of VWF-CBA to VWF-Ag (C) were determined in patients with iTTP (initial vs. exacerbated or relapsed) and the healthy controls. Each dot represents a single patient and solid lines are the median Âą 95% confidence intervals. Kruskal-Willis analysis was used to determine the statistical significance among three different groups. Here *, **, ***, and **** indicate the P values of <0.05, <0.001, 0.0005, and <0.0001, respectively; n.s. stands for no statistical difference.

169

E.M. Staley et al.

Table 3. Univariate analysis identifies the important laboratory parameters that predict mortality in 73 unique patients with iTTP.

Laboratory values 9

Platelet counts (x10 /L) ΔD2-D0 ΔD5-D0 Hemoglobin (g/dL) Hematocrit (%) WBC (x109/L) PT (sec) aPTT (sec) Fibrinogen (mg/dL) D-dimer (mg/L) LDH (U/L) ΔD2-D0 ΔD5-D0 Troponin I (ng/mL) At Emergency Department Prior to TPE Creatinine (mg/dL) Total bilirubin (mg/dL) Indirect bilirubin (mg/dL) Total protein (g/L) Albumin (g/L) ADAMTS13 inhibitor (U/mL) Anti-ADAMTS13 IgG (U/mL) VWF-Ag (%) VWF-CBA (%) HNP1-3 (ng/mL) Histone/DNA (U/mL) Bb (mg/mL) C4d (mg/mL) iC3b (mg/mL) sC5b (mg/mL)

Survived (n=64)

Died (n=9)

12.8 (11-15)# 2.4 (1-4.5) 25.4 (21.1-33.7) 8.6 (8-9.4) 24.5 (23-26) 10.2 (8.8-12.2) 14.6 (14.4-15) 30 (29-33) 421 (398-480) 2062 (1370-3497) 1053 (912-7332) -366 (-444 to -275) -180 (-213 to -132)

15.5 (7.9-24) 3.7 (-5-16.9) 1.9 (-4.3-53.3) 9.8 (6.8-11) 29 (18-34) 13.3 (4.9-21.7) 15.5 (13.4-20.1) 34 (27-63) 488 (353-804) 7,083 (708-11871) 1546 (787-7332) -187 (-310 to 283) -109 (-282 to 43)

0.55 0.92 0.09 0.49 0.54 0.54 0.38 0.10 0.18 0.09 0.16 0.08 0.03*

0.4 (0.1-1.0) 0.1 (0.1-0.4) 1.2 (1.1-1.5) 2.3 (1.8-2.8) 1.7 (1.3-2.1) 6.5 (6.2-6.8) 3.5 (3.4-3.8) 1.3 (1.1-1.8) 4695 (3939-3317) 276 (225-327) 199 (173-256) 33 (27-41) 59 (47-87) 2.5 (2-3) 2.7 (2-3.1) 14.8 (12.7-17.8) 1.2 (0.5-6.2)

3.3 (0.8-197) 0.9 (0.1-97) 1.7 (1-3.6) 2.9 (1.4-10.2) 2.3 (1.2-3.7) 5.9 (4.4-7.9) 3.2 (1.6-3.7) 0.9 (0.5-3.6) 2881(653-11103) 353 (165-742) 284 (133-926) 33 (16-239) 126 (27-257) 3.8 (1.5-8) 3 (1.4-9.1) 14.7 (7.5-30.8) 1.2 (0.5-6.2)

0.008** 0.01* 0.15 0.49 0.30 0.13 0.03* 1.00 0.39 0.25 0.19 0.86 0.07 0.09 0.86 0.59 0.59

WBC: white blood cell; ΔLDH: change in lactate dehydrogenase; PT: prothrombin time; aPTT: activated partial thromboplastin time; VWF-Ag: von Willebrand factor antigen; VWFCBA: von Willebrand factor collagen-binding activity; IgG: immunoglobulin G; #All data are expressed as the median values and 95% confidential interval in parenthesis. MannWhitney test was performed to determine the statistical significance. Here, * and ** indicate P values <0.05 and <0.01, respectively.

of iTTP were included in this study (Figure 1B). The median age of the patients was 41 years, with 56% being female and 44% male. Seventy-seven percent of patients were of African-American decent, while Caucasian and Hispanic patients comprised only 22% and <1%, respectively. A substantial number of patients had one or several comorbidities, including hypertension (52%), diabetes mellitus (18%), and systemic lupus erythematosus (SLE) (11%). Only 3 patients (4%) were pregnant at the time of the diagnosis. In addition, 49% of patients reported a history of smoking, and 21% reported use of recreational drugs (Table 1). Our blood samples were primarily obtained from patients at their initial presentation (75%) and during a relapse (23%); only one sample was obtained from an exacerbation. This bias is the result of our inclusion criteria designed to prioritize the collection and study of unique patients. The clinical presentation on admission primarily included signs and symptoms of the central 170

nervous system (CNS) (54.7%), abdominal pain (38.4%), and chest pain (9.6%) (Table 1). The most common blood type in our patient population was type O (63%), followed by A (22%), B (14%), and AB (1%), with 93% being Rh positive. The frequency of type O was statistically higher than that expected (47.9%) (Online Supplementary Table S1), suggesting that patients with type O blood group are not protected from developing iTTP. All iTTP patients received TPE as primary therapy. The median number of TPEs for all patients was 15 (95% confidence interval [CI] of 12-17). Patients also received corticosteroids (94%), rituximab (38.4%), and vincristine (6.8%) at various times during hospitalization. Only 5.4% of patients underwent splenectomy as part of their treatment. The outcomes of these patients after treatment were classified as the following: single episode to remission (30.1%), exacerbation or relapse to remission (32.9%), haematologica | 2019; 104(1)

Prognostic markers in autoimmune TTP

exacerbation and relapse to remission (24.7%), and death (12.3%). Eight out of 9 patients who died within 35 days following admission were unresponsive and/or refractory to TPE; 1 patient died of unrelated disease one year after the initial therapy. Therefore, the overall remission rate was 87.5% despite exacerbation and/or relapse after a median follow up of 828 days (or 2.3 years) (95% CI of 326-1,720 days or 1-4.7 years) (Table 2).

2B). All 8 patients who tested negative (<0.4 U/mL) for ADAMTS13 inhibitors had significantly increased levels of anti-ADAMTS13 IgG (Figure 2C). Therefore, all patients were confirmed as having severe ADAMTS13 deficiency and positive antibody against ADAMTS13, hence the diagnosis of iTTP.

ADAMTS13 activity and inhibitors in patients with iTTP

Plasma VWF-Ag, VWF-CBA, and the ratio of VWFCBA/VWF-Ag were determined in plasma samples obtained from iTTP patients and healthy controls. Plasma levels of VWF-Ag (P<0.0001) (Figure 3A) and VWF-CBA (P<0.005) (Figure 3B) on admission were significantly higher in patients with either an initial or a relapse episode of iTTP than those in the healthy controls. However, the ratio of VWF-CBA to VWF-Ag, a quantitative measurement of VWF multimer size,21 was only significantly different between the initial episodes and controls (P<0.01), but not between the relapse episodes and controls (P>0.05) (Figure 3C). Plasma levels of VWF-Ag and VWFCBA were moderately associated with each other with a Spearman correlation coefficient (rho) of 0.592 (P<0.0001). These results indicated that despite increased expression and/or release of VWF, the ultra-large VWF multimers may be selectively removed from the circulation in some cases as a result of ongoing microvascular thrombosis during the acute episode. A positive association between plasma levels of VWF-Ag and serum LDH (rho=0.36, P=0.004) supports our hypothesis.

All patients except for 2 having ADAMTS13 activity of 13 and 26 U/dL, respectively) had plasma ADAMTS13 activity <5 U/dL (or <5% normal) (Figure 2A). However, these two samples were drawn by ordering physicians following TPE or a large volume of plasma transfusion. However, the ADAMTS13 activity in pre-treatment samples stored separately for research purposes was <5 U/dL. Therefore, all patients included in the study had severe deficiency of ADAMTS13 activity. ADAMTS13 inhibitors (>0.4 U/mL), measured by the 50/50 mixing study, were found to be present in 65 of 73 (89%) patients. There was no difference in the inhibitor titers in patients presenting during the initial and those with a relapse episode (Figure

Plasma VWF antigen and collagen-binding activity in patients with iTTP

Plasma levels of HNP1-3 and histone/DNA complexes in acute iTTP

Previous studies demonstrated that plasma levels of HNP1-38 and DNA/histone complexes20 were significantly elevated in patients with acute iTTP. However, their prognostic value in iTTP was not determined. Consistent with the results previously reported, plasma levels of HNP1-3 (P<0.0001) (Figure 4A) and histone/DNA complexes (P<0.0001) (Figure 4B) in patients with an initial or exacerbated or relapsed episode of acute iTTP were dramatically increased when compared to those in the healthy controls. Additionally, plasma HNP1-3 levels in iTTP patients were significantly correlated with the levels of histone/DNA complex (rho=0.50, P<0.001). While none of these two biomarkers was predictive of an adverse outcome in iTTP patients (e.g., mortality, exacerbation, and relapse), plasma levels of HNP1-3 appeared to significantly correlate with

Table 4. COX hadard regression analysis identifies the statistically significant parameters that predict mortality in iTTP patients.

Figure 4. Plasma levels of HNP1-3 and histone/DNA complexes in patients with iTTP. Plasma levels of HNP1-3 (A) and histone/DNA complexes (B) in patients with acute iTTP (initial vs. exacerbated or relapsed) and healthy controls are shown as the dot plots for individual patients with the median ± 95% confidence intervals (solid red lines). Kruskal-Willis analysis was used to determine the statistical significance. Here **** indicates P value <0.0001 when compared the values in the healthy controls.

haematologica | 2019; 104(1)

Parameters

HR* (95% CI)

Total protein (g/L) Albumin (g/L) aPTT (sec) Fibrinogen (mg/dL) ΔLDH (D5-D0) (U/L) Bb ( g/mL) sC5b-9 ( g/mL)

0.37 (0.15-0.92) 0.23 (0.09-0.60) 2.03 (1.91-2.16) 1.90 (1.89-1.91) 2.93 (2.9-2.93) 1.30 (1.01-1.68) 1.54 (1.00-2.36)

0.03 0.003 0.02 0.03 0.04 0.04 0.05

*The hazard ratio (HR) was calculated based on the change (Δ) of LDH by 100 U/L, aPTT by 10 sec, fibrinogen by 100 mg/dL, Bb, and sC5-9 by 1 mg/mL, and change in LDH level by 100 U/L.

171

E.M. Staley et al.

serum troponin (rho=0.37, P<0.05) and LDH (rho=0.48, P<0.001). The plasma levels of histone/DNA complex were also correlated with serum LDH (rho=0.66, P<0.0001), but not troponin. Noticeably, serum troponin was elevated in 35 of 37 (~95%) patients tested in the ED, and remained elevated in 59 of 68 (~87%) patients tested after being admitted to hospital (prior to TPE). These results suggest that plasma levels of HNP1-3 and histone/DNA complexes may also be markers of organ injury in addition to serum LDH and troponin in patients with acute iTTP.

Plasma levels of complement activation markers in acute iTTP While complement activation through an alternative pathway is the primary cause of aHUS,22 recent studies have suggested that complement activation may also occur in patients with acute iTTP.23-25 To assess the prognostic value of complement activation markers in these patients, we determined plasma levels of C4d (classical pathway), Bb (alternative pathway), and iC3b and sC5b-9

(common pathway) in iTTP patients and healthy controls. As shown, plasma C4d levels were not significantly increased in patients with acute iTTP (P>0.05) (Figure 5A). Plasma levels of Bb (P<0.0001) (Figure 5B), iC3b (P<0.01) (Figure 5C), and sC5b-9 (P<0.0001) (Figure 5D) were significantly elevated when compared to healthy controls. There was a modestly positive correlation between Bb and creatinine (rho=0.3, P<0.05), LDH (rho=0.65, P<0.0001), and troponin (rho=0.329, P<0.05) in iTTP patients. Additionally, there was a positive correlation between sC5b-9 and LDH (rho=0.4, P<001 ). These results suggest that complement over-activation through the alternative pathway may participate in the pathophysiology of acute iTTP, although a causative role of complement activation in iTTP is yet to be determined in animal models.

Predictive values of certain clinical factors and routine laboratory parameters The associations between demographics, clinical presentation, admission laboratory values, the aforemen-

Figure 5. Plasma levels of complement activation markers in patients with iTTP. Plasma levels of complement activation markers including C4d (A), Bb (B), iC3b (C), and sC5b-9 (D) in patients with acute iTTP (initial vs. exacterbated or relapsed) and the healthy controls. Each dot represents the value of each individual subject. The red solid lines are the medians Âą 95% confidence interval. Kruskal-Willis analysis was used to determine the statistical significance. Here *, **, and **** indicate the P values of <0.01, and <0.0001, respectively; n.s. stands for no statistical difference.

172

haematologica | 2019; 104(1)

Prognostic markers in autoimmune TTP

tioned biomarkers, and patient outcomes were determined. On univariate analysis, the following variables were significantly associated with in-patient mortality: an inability to normalize platelet count within seven days of TPE (i.e., platelet count of ≥150x109/L) (Online SupplementaryTable S2), failure to significantly reduce serum LDH after five days of treatment (P=0.03), low serum albumin (P=0.03), and high troponin levels at ED (P=0.008) and after being admitted to hospital prior to TPE (P=0.01) (Table 3). Additionally, Cox regression analysis demonstrated that high levels of total serum protein or albumin on admission were associated with a reduced risk of in-hospital mortality (HR, 0.37 or 0.21, P=0.032 or 0.003), while prolonged aPTT (HR 2.03, P=0.02), increased fibrinogen (HR 1.9, P=0.03), elevated LDH at day five (HR 2.93, P=0.04), high plasma Bb (HR 1.3, P=0.04), and sC5b9 (HR 1.54, P=0.05) were all found to be the significant markers of mortality (Table 4). When the achievement of clinical remission was used as an outcome, the normalization of platelet count within seven days (P<0.001) and absence of SLE diagnosis (P=0.038) were found to be the significant prognostic factors for remission (Online Supplementary Table S2). Together, our results demonstrate that lack of admission coagulopathy, low serum tropin level, absence of lupus, platelet recovery in seven days, and marked reduction of LDH in five days, patient overall well-being, and low levels of complement activation markers appear to associate with a good outcome in patients with acute iTTP.

Discussion The Alabama cohort represents a distinct, yet undescribed, iTTP patient population. Our patients were predominantly of African American descent (77%); currently, African Americans only make up 25% of the current population in Alabama (https://cber.cba.ua.edu/rbriefs/ black.html). This overrepresentation of African-Americans may be in part attributed to the strict inclusion criteria that required the presence of plasma ADAMTS13 activity <5 U/dL and either positive inhibitors or elevated antiADAMTS13 IgG in the appropriate clinical context. Patients of African American descent have been shown to have reduced frequency of a protective HLA DRB1*04 allele, which renders them more susceptible to the development of autoantibodies against ADAMTS13.26 Since 2015, we have seen approximately 17 new cases of iTTP per year at the UAB Medical Center, which serves as the major referral center for diagnosis and management of TMA in the Southeastern United States. With a population of ~4.9 millions in the State of Alabama, the incidence rate is estimated to be ~3.5 cases per million per year, fairly similar to that reported in the literature.27,28 This incidence rate is most likely underestimated due to patients who may expire prior to reaching our institution or being enrolled in the study. Data from the Ohio Registry suggest that there are differences in race, neurological symptoms, platelet counts, LDH, ADAMTS13 activity, and total number of TPE required between patients with an initial episode and those with relapses, despite no major difference in outcome.29 Our data do not show a difference in patients with an initial episode or relapse regarding admission neurologic symptoms, hemoglobin, platelet count, LDH, inhibitor, haematologica | 2019; 104(1)

anti-ADAMTS13 IgG, and/or any of the novel biomarkers evaluated. To date, the diagnosis of iTTP relies on laboratory findings of severe deficiency of plasma ADAMTS13 activity (i.e., <5 U/dL, or <10 U/dL depending on the lab cut-off) with a positive inhibitor, and/or elevated antiADAMTS13 IgG levels in an appropriate clinical context (i.e., thrombocytopenia and microangiopathic hemolytic anemia without other explanations).1,30 All patients in this study had an ADAMTS13 activity <5 U/dL when only the pre-treatment plasma samples were interrogated. Of note, all patients, including 5 with negative inhibitors (<0.4 U/mL), had significantly elevated levels of antiADAMTS13 IgG by ELISA . These results suggest that the ELISA-based binding assay may be more sensitive than the functional assays for the diagnosis of iTTP. Therefore, we recommend that any patient with severe deficiency of plasma ADAMTS13 activity (<10 U/dL), but a negative functional inhibitor, should undergo anti-ADAMTS13 IgG ELISA testing. While anti-ADAMTS13 IgG may be detected in plasma of healthy individuals, their levels are not sufficient to inhibit ADAMTS13 activity.31 Findings of high concentrations of anti-ADAMTS13 IgG in addition to low ADAMTS13 antigen levels appear to predict adverse outcomes in patients with iTTP according to the UK registry data.13 The hallmark of iTTP is the autoantibody-mediated inhibition of plasma ADAMTS13 activity. This leads to the inability to cleave newly released ultra-large VWF multimers released from and anchored on the endothelium,32 in the circulating blood,33 or at the site of thrombus formation.34 Subsequently, the accumulated ultra-large VWF multimers may spontaneously agglutinate platelets in small arterioles and capillaries, leading to end organ damage.2 Data supporting this hypothesis include significant elevation of plasma levels of VWF antigen and collagen-binding activity in acute iTTP patients despite the slightly reduced ratios of VWF activity to VWF antigen in these patients during the acute setting – consistent with the partial consumption of ultra large VWF multimers during active thrombus formation. In 1982, Moake et al. showed that ultralarge VWF multimers are only detectable in chronic and relapsing TTP patients during remission, not during the acute disease.33 Plasma from Adamts13–/– mice appear to have ultra-large VWF multimers when the animals are well or not stressed.35 Nevertheless, the increased levels of plasma VWF antigen, but not VWF activity, correlate with increased levels of serum LDH, suggesting that either plasma VWF antigen or serum LDH may be used as a biomarker of end-organ damage. Infection, inflammation, and pregnancy are known to be complement-amplifying conditions, presumably through activation of the alternative pathway.36 Consistent with our previous findings,8,20 plasma levels of HNP1-3, histone/DNA complexes, Bb, and iC3b were also significantly increased in this cohort of iTTP patients; plasma HNP1-3, histone/DNA complexes, Bb, and iC3b correlate with the markers of organ damage, including serum LDH, creatinine and/or troponin. There was a strong correlation between plasma sC5b and serum LDH. Rapid reduction in LDH was correlated with survival, consistent with a previous report.37 The elevated plasma levels of Bb and sC5b-9 were found to be predictive of mortality in the Cox regression analysis. These results suggest that both neutrophil and complement activation may participate in 173

E.M. Staley et al.

the pathogenesis of iTTP, although the causative role of HNP1-3, histone, and complement activation in iTTP are yet to be determined. Other clinical and laboratory factors such as the elevated troponin levels and reduced Glasgow Coma Score (GCS) confer a six-fold and nine-fold increase, respectively, in the mortality of patients with iTTP, reported from the UK TTP registry.13 Our univariate analysis also demonstrated the association between an increased troponin level and mortality. In summary, our study further demonstrates the utility of several clinical and laboratory markers including aPTT, fibrinogen, troponin, the rate of platelet and LDH normalization, and total protein/albumin etc. for predicting outcome in patients with iTTP. Moreover, we identified several novel biomarkers related to inflammation (e.g., VWF, HNP1-3, and histone/DNA complexes) and innate immunity (e.g., Bb and sC5b-9) that may be used to assess dis-

References 1. Saha M, McDaniel JK, Zheng XL. Thrombotic thrombocytopenic purpura: pathogenesis, diagnosis and potential novel therapeutics. J Thromb Haemost. 2017;15(10):1889-1900. 2. Moake JL. Thrombotic thrombocytopenic purpura: the systemic clumping "plague". Annu Rev Med. 2002;53:75-88. 3. Tsai HM, Lian EC. Antibodies to von Willebrand factor-cleaving protease in acute thrombotic thrombocytopenic purpura. N Eng J Med. 1998;339(22):15851594. 4. Zheng XL, Wu HM, Shang D, Falls E, Skipwith CG, Cataland SR, et al. Multiple domains of ADAMTS13 are targeted by autoantibodies against ADAMTS13 in patients with acquired idiopathic thrombotic thrombocytopenic purpura. Haematologica. 2010;95(9):1555-1562. 5. Casina V, Hu WB, Mao JH, et al. High resolution epitope mapping by HX MS reveals the pathogenic mechanism and a possible therapy for autoimmune TTP syndrome. Proc Natl Acad Sci USA. 2015;112(31):9620-9625. 6. George JN. How I treat patients with thrombotic thrombocytopenic purpura: 2010. Blood. 2010;116(20):4060-4069. 7. Kremer Hovinga JA, Vesely SK, Terrell DR, Lammle B, George JN. Survival and relapse in patients with thrombotic thrombocytopenic purpura. Blood. 2010;115(8):15001511. 8. Cao W, Pham HP, Williams LA, et al. Human neutrophil peptides and complement factor Bb in pathogenesis of acquired thrombotic thrombocytopenic purpura. Haematologica. 2016;101(11):1319-1326. 9. Peyvandi F, Scully M, Kremer Hovinga JA, et al. Caplacizumab reduces the frequency of major thromboembolic events, exacerbations and death in patients with acquired thrombotic thrombocytopenic purpura. J Thromb Haemost. 2017;15(7):1448-1452. 10. Cataland SR, Yang SB, Witkoff L, et al. Demographic and ADAMTS13 biomarker data as predictors of early recurrences of idiopathic thrombotic thrombocytopenic purpura. Eur J Haematol. 2009;83(6):559564.

174

ease severity and predict long-term outcome in patients with iTTP. Our findings may help stratify patients who may benefit from more intensive care and management including twice daily TPE, early rituximab, and eclulizumab etc. to reduce in-hospital mortality and long-term complications. Funding This work was partially supported by grants from the National Heart, Lung, and Blood Institute (HL-26724 and HL-115187) and the Answering T.T.P. Foundation, Canada. Acknowledgments The collection and storage of patient samples could not have occurred without the help of the UAB apheresis nurses, the UAB Coagulation Lab, as well as the UAB Pathology residents and Transfusion Medicine fellows.

11. Zheng XL, Kaufman RM, Goodnough LT, Sadler JE. Effect of plasma exchange on plasma ADAMTS13 metalloprotease activity, inhibitor level, and clinical outcome in patients with idiopathic and nonidiopathic thrombotic thrombocytopenic purpura. Blood. 2004;103(11):4043-4049. 12. Terrell DR, Vesely SK, Kremer Hovinga JA, Lammle B, George JN. Different disparities of gender and race among the thrombotic thrombocytopenic purpura and hemolyticuremic syndromes. Am J Hematol. 2010; 85(11):844-847. 13. Alwan F, Vendramin C, Vanhoorelbeke K, et al. Presenting ADAMTS13 antibody and antigen levels predict prognosis in immune-mediated thrombotic thrombocytopenic purpura. Blood. 2017;130(4):466471. 14. Brazelton J, Oster RA, McCleskey B, Fuller J, Adamski J, Marques MB. Increased troponin I is associated with fatal outcome in acquired thrombotic thrombocytopenic purpura. J Clin Apher. 2017;32(5):311-318. 15. Liu C, Kallogjeri D, Dynis M, Grossman BJ. Platelet recovery rate during plasma exchange predicts early and late responses in patients with thrombotic thrombocytopenic purpura. Transfusion. 2013;53(5):1096-1107. 16. Dusse LM, Alpoim PN, Silva JT, Rios DR, Brandao AH, Cabral AC. Revisiting HELLP syndrome. Clin Chim Acta. 2015;451(Pt B):117-120. 17. Kokame K, Nobe Y, Kokubo Y, Okayama A, Miyata T. FRETS-VWF73, a first fluorogenic substrate for ADAMTS13 assay. Br J Haematol. 2005;129(1):93-100. 18. Zhang L, Lawson HL, Harish VC, Huff JD, Knovich MA, Owen J. Creation of a recombinant peptide substrate for fluorescence resonance energy transfer-based protease assays. Anal Biochem. 2006;358(2):298-300. 19. Kumar M, Cao W, McDaniel JK, et al. Plasma ADAMTS13 activity and von Willebrand factor antigen and activity in patients with subarachnoid haemorrhage. Thromb Haemost. 2017;117(4):691-699. 20. Fuchs TA, Kremer Hovinga JA, Schatzberg D, Wagner DD, Lammle B. Circulating DNA and myeloperoxidase indicate disease activity in patients with thrombotic

21.

22.

23.

24.

25.

26.

27.

28.

29.

microangiopathies. Blood. 2012;120(6): 1157-1164. Gerritsen HE, Turecek PL, Schwarz HP, Lammle B, Furlan M. Assay of von Willebrand factor (vWF)-cleaving protease based on decreased collagen binding affinity of degraded vWF: a tool for the diagnosis of thrombotic thrombocytopenic purpura (TTP). Thromb Haemost. 1999;82(5):1386-1389. Afshar-Kharghan V. Atypical hemolytic uremic syndrome. Hematology Am Soc Hematology Am Soc Hematol Educ Program. 2016;2016(1):217-225. Westwood JP, Langley K, Heelas E, Machin SJ, Scully M. Complement and cytokine response in acute Thrombotic Thrombocytopenic Purpura. Br J Haematol. 2014;164(6):858-866. Tati R, Kristoffersson AC, Stahl AL, et al. Complement activation associated with ADAMTS13 deficiency in human and murine thrombotic microangiopathy. J Immunol. 2013;191(5):2184-2193. Westwood JP, Langley K, Heelas E, Machin SJ, Scully M. Complement and cytokine response in acute Thrombotic Thrombocytopenic Purpura. Br J Haematol. 2013;64(6):858-866. Martino S, Jamme M, Deligny C, et al. Thrombotic thrombocytopenic purpura in black people: impact of ethnicity on survival and genetic risk factors. PLoS One. 2016;11(7):e0156679. Terrell DR, Williams LA, Vesely SK, Lammle B, Hovinga JA, George JN. The incidence of thrombotic thrombocytopenic purpura-hemolytic uremic syndrome: all patients, idiopathic patients, and patients with severe ADAMTS-13 deficiency. J Thromb Haemost. 2005;3(7):1432-1436. Mariotte E, Azoulay E, Galicier L, et al. Epidemiology and pathophysiology of adulthood-onset thrombotic microangiopathy with severe ADAMTS13 deficiency (thrombotic thrombocytopenic purpura): a cross-sectional analysis of the French national registry for thrombotic microangiopathy. Lancet Haematol. 2016;3(5):e237-245. Masias C, Wu H, McGookey M, Jay L, Cataland S, Yang S. No major differences in outcomes between the initial and relapse

haematologica | 2019; 104(1)

Prognostic markers in autoimmune TTP

episodes in patients with thrombotic thrombocytopenic purpura: The experience from the Ohio State University Registry. Am J Hematol. 2018;93(3):E73E75. 30. Scully M, Cataland S, Coppo P, et al. Consensus on the standardization of terminology in thrombotic thrombocytopenic purpura and related thrombotic microangiopathies. J Thromb Haemost. 2016;15(2):312-322. 31. Grillberger R, Casina VC, Turecek PL, Zheng XL, Rottensteiner H, Scheiflinger F. Anti-ADAMTS13 IgG autoantibodies present in healthy individuals share linear epitopes with those in patients with throm-

haematologica | 2019; 104(1)

botic thrombocytopenic purpura. Haematologica. 2014;99(4):e58-60. 32. Dong JF, Moake JL, Nolasco L, et al. ADAMTS-13 rapidly cleaves newly secreted ultralarge von Willebrand factor multimers on the endothelial surface under flowing conditions. Blood. 2002;100(12):40334039. 33. Moake JL, Rudy CK, Troll JH, et al. Unusually large plasma factor VIII:von Willebrand factor multimers in chronic relapsing thrombotic thrombocytopenic purpura. N Engl J Med. 1982;307(23):14321435. 34. Sadler JE. A new name in thrombosis, ADAMTS13. Proc Natl Acad Sci USA.

2002;99(18):11552-11554. 35. Jin SY, Xiao J, Bao J, Zhou S, Wright JF, Zheng XL. AAV-mediated expression of an ADAMTS13 variant prevents shigatoxininduced thrombotic thrombocytopenic purpura. Blood. 2013;121(19):3825-3829, S1-3. 36. Noris M, Mescia F, Remuzzi G. STEC-HUS, atypical HUS and TTP are all diseases of complement activation. Nat Rev Nephrol. 2012;8(11):622-633. 37. Patton JF, Manning KR, Case D, Owen J. Serum lactate dehydrogenase and platelet count predict survival in thrombotic thrombocytopenic purpura. Am J Hematol. 1994; 47(2):94-99.

175

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):176-188

Coagulation & its Disorders

Computed tomography pulmonary angiography versus ventilation-perfusion lung scanning for diagnosing pulmonary embolism during pregnancy: a systematic review and meta-analysis Cécile Tromeur,1,2,3 Liselotte M. van der Pol,1,4 Pierre-Yves Le Roux,5 Yvonne Ende-Verhaar,1 Pierre-Yves Salaun,5 Christophe Leroyer,2,3 Francis Couturaud,2,3 Lucia J.M. Kroft,6 Menno V. Huisman1 and Frederikus A. Klok1

Department of Thrombosis and Hemostasis, Leiden University Medical Center, the Netherlands; 2Groupe d’Etude de la Thrombose de Bretagne Occidentale, University of Brest, Equipe d’Accueil 3878, Department of Internal Medicine and Chest Diseases, CHRU Brest, France; 3Centre d’Investigation Clinique INSERM 1412, University of Brest, France; 4 Department of Internal Medicine, Haga Teaching Hospital, the Hague, the Netherlands; 5 Département de Médecine Nucléaire, CHRU Brest, France and 6Department of Radiology, Leiden University Medical Center, the Netherlands 1

ABSTRACT

Correspondence: tromeurcecile@gmail.com

Received: April 23, 2017. Accepted: August 14, 2018. Pre-published: August 16, 2018. doi:10.3324/haematol.2018.196121 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/176 ©2019 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.

176

ifferences between computed tomography pulmonary angiography and ventilation-perfusion lung scanning in pregnant patients with suspected acute pulmonary embolism are not well-known, leading to ongoing debate on which test to choose. We searched in PubMed, EMBASE, Web of Science and the Cochrane Library databases and identified all relevant articles and abstracts published up to October 1, 2017. We assessed diagnostic efficiency, frequency of non-diagnostic results and maternal and fetal exposure to radiation exposure. We included 13 studies for the diagnostic efficiency analysis, 30 for the analysis of non-diagnostic results and 22 for the radiation exposure analysis. The pooled rate of false negative test results was 0% for both imaging strategies with overlapping confidence intervals. The pooled rates of non-diagnostic results with computed tomography pulmonary angiography and ventilation-perfusion lung scans were 12% (95% confidence interval: 817) and 14% (95% confidence interval: 10-18), respectively. Reported maternal and fetal radiation exposure doses were well below the safety threshold, but could not be compared between the two diagnostic methods given the lack of high quality data. Both imaging tests seem equally safe to rule out pulmonary embolism in pregnancy. We found no significant differences in efficiency and radiation exposures between computed tomography pulmonary angiography and ventilation-perfusion lung scanning although direct comparisons were not possible.

Introduction Pulmonary embolism (PE) is a major complication of pregnancy and responsible for 2% to 14% of all maternal deaths worldwide.1,2 Although accurate diagnostic tests for PE are essential for this specific population, high quality diagnostic studies are unavailable.3 Clinical decision rules, which are the cornerstone of PE diagnostic management in the non-pregnant population, were not developed for, nor validated in pregnant patients.4 Furthermore, considering the physiological increase of D-dimer levels throughout pregnancy, the optimal D-dimer threshold to rule out PE is unknown.5 The application of D-dimer tests and clinical decision rules as the initial step of the diagnostic algorithm for suspected PE cannot, therefore, be recommended in pregnant patients.3 Moreover, the optimal choice of imaging test to rule out or confirm acute PE in haematologica | 2019; 104(1)

CTPA versus V-Q lung scan in pregnant women

pregnant patients is highly debated. The two most used imaging tests for suspected acute PE in the non-pregnant population are computed tomography pulmonary angiography (CTPA) and ventilation-perfusion (V-Q) lung scanning, with CTPA being the imaging test of choice because of its high accuracy, wide availability, and ability to exclude other pathologies.6,7 As is generally the case with V-Q lung scans, the risk of non-diagnostic tests with CTPA is relatively high, in part because of the hemodynamic changes that occur during pregnancy, such as hemodilution and increased heart rate, which make it necessary to have a CTPA protocol specifically designed for pregnant patients. Additionally, elevation of the diaphragm, due to the enlarged uterus, accentuates the interruption of contrast by non-opacified blood from the inferior vena cava and may lead to decreased contrast attenuation in areas of the pulmonary arteries.6 Moreover, both CTPA and V-Q lung scanning involve exposure of the fetus and patientsâ&#x20AC;&#x2122; breasts to radiation. The lack of high quality management studies comparing both imaging tests fuels an ongoing debate in the literature on which of the two options should be preferred. We set out to perform a systematic review and metaanalysis of published literature to compare the diagnostic efficiency of CTPA versus V-Q lung scans in pregnant patients with suspected acute PE. We also aimed to compare the rate of non-diagnostic scan results and radiation exposure for both the mother and fetus.

Methods Search strategy For this meta-analysis, we conducted a search for all relevant full publications in PubMed, EMBASE, Web of Science and the Cochrane Library databases. We searched EMBASE, Web of Science and the Cochrane library databases for relevant meeting-abstracts as well. The complete search strategy is detailed in Online Supplementary Appendix A.

Selection of studies Search results were combined and duplicates were removed. Studies were screened for relevance by two independent reviewers (CT and LvdP) following a specific three-step program and applying Covidence software (www.covidence.org). Disagreements were resolved by a third investigator (FK) by majority rule. The first and second steps consisted of title and abstract screening followed by full text screening for the remaining articles. The final selection of the studies to include in the meta-analysis was based on assessment of relevance and study quality. The assessment of relevance was based on the following criteria: (i) prospective patient inclusion, (ii) inclusion of consecutive patients, (iii) reported rate of non-diagnostic test results, and (iv) reported incidence of PE at baseline. The assessment of bias was evaluated in accordance with the PRISMA criteria:8 (i) pre-specified study protocol, (ii) clear description of inclusion and exclusion criteria, (iii) inclusion of consecutive patients, (iv) objective diagnosis of PE, (v) reported losses to follow-up, (vi) clear distinction between pregnant and post-partum patients, and (vii) assessment of the primary endpoints in all patients. Studies were included in the meta-analyses according to the definition of each endpoint. The final step was data extraction. For each included study, we extracted the first authorâ&#x20AC;&#x2122;s name and year of publication, study design (prospective or retrospective), setting of the study haematologica | 2019; 104(1)

(single- or multicenter), number of patients in the index cohort, the baseline incidence of PE, the duration of follow up, and the predefined study endpoints.

Study outcomes and definitions We predefined three major study endpoints. The first was the diagnostic efficiency of both imaging tests as expressed by the number of false negative scans. This first outcome required a follow up of at least 3 months as well as reporting of the number of diagnosed PE events during this follow up. The second endpoint was the rate of non-diagnostic results with CTPA and V-Q lung scans. For CTPA, scan results were defined non-diagnostic when the radiologist was unable to confirm or exclude the diagnosis of PE, usually because of suboptimal contrast opacification and respiratory motion artifacts, or the need for an additional imaging test. For V-Q lung scanning, the definition of non-diagnostic results was based on the PIOPED criteria, i.e. intermediate and low probability scan results, since these require an additional diagnostic test to confirm or rule out PE with sufficient certainty. The third endpoint was fetal and maternal radiation exposure due to CTPA and V-Q lung scanning. The CTPA radiation exposure was collected for studies in real-life patients as well as with anthropometric phantom models simulating a gravid woman.

Statistical analysis The baseline incidence of PE and rate of false negative scans were calculated with corresponding 95% confidence intervals (95% CI). The number of non-diagnostic results from all studies was collected and the rate of non-diagnostic results was calculated using the number of non-diagnostic tests divided by the number of patients in each study. We applied a random effects model according to DerSimonian and Laird for the calculation of the pooled rates of the four study endpoints.9 We predefined that we would not undertake data pooling in case studies for any of the three endpoints because they were not comparable due to extensive differences in study design or imaging protocols, which do not allow for reliable statistics or data pooling. Heterogeneity across the various cohort studies was assessed by calculating the I2 statistic. Heterogeneity was defined as low when I2 was <25%, intermediate when I2 was 25-75% and high when I2 was >75%.10 All analyses were performed in Stata 14.0 (Stata Corp., College Station, TX, USA).

Results Study selection The initial search identified 303 records in PubMed, 318 articles in EMBASE, 76 articles in Web of Science, and three articles in the Cochrane Library. After a first screening of titles and abstracts, 565 articles were excluded. A further 78 articles were excluded based on the predefined inclusion criteria (Figure 1): 20 studies did not report the study outcomes of interest, two articles concerned thyroid function after CTPA, five articles involved surveys about clinical practice, two articles were duplicates, four were guidelines, five were letters to the editor and did not report the outcomes of interest, 37 were review articles and four were irrelevant case reports. Two additional relevant articles were identified after reviewing the references lists of the selected studies. A final 49 evidencebased studies were fully assessed for study quality6,7,11-57 (Table 1): 13 were included in the analysis of false negative scans7,14-17,20,33,35,37,45,51,53,55 (Table 2), 30 were included in the analysis of non-diagnostic results7,14-17,19-21,23-29,32-37,45,46,49,52-57 177

C. Tromeur et al.

(Table 3), and 11 were included in the radiation exposure analysis16,18,20,21,24,28,33,34,52,54,57 (Table 4). Finally, 11 studies involving anthropometric phantoms simulating pregnancy were also included58-68 (Table 5).

normal initial CTPA, for a pooled number of false negative scans of 0.0% (95% CI: 0.0-0.16; I2=5.7) in the CTPA group (Figure 2). The risk of bias was high in two studies,17,51 moderate in nine studies7,14-16,20,33,35,45,53 and low in only two studies37,55 (Table 1).

First study endpoint: diagnostic accuracy A total of 13 relevant studies were selected to study the rate of false negative CTPA and V-Q lung scan examinations.7,14-17,20,33,35,37,45,51,53,55 These studies were published between 199714 and 2017,53,55 and involved a total of 1270 patients investigated with V-Q lung scanning and 837 patients investigated with CTPA (Table 2). Data were extracted from ten full text articles7,14-17,20,33,35,37,55 and three meeting abstracts.45,51,53 Only one of these 13 studies was a prospective study in 143 patients investigated with CTPA.45 The prevalence of PE ranged between 0%20 and 22.2%,35 with the highest prevalences in the few smaller studies (median 4.1%). The duration of follow up varied from at least 3 months to 24 months.35 In two studies, the total duration of follow up was not reported.14,17 None of the 1270 patients investigated with V-Q lung scanning was diagnosed with recurrent PE or deep vein thrombosis (DVT) during follow up, resulting in a pooled number of false negative scans of 0% (95% CI: 0-0.04; I2=0.0). Three of 837 patients were diagnosed with non-fatal PE after a

Second study endpoint: non-diagnostic results A total of 30 relevant studies were selected to evaluate the rate of non-diagnostic or inconclusive results of V-Q lung scans or CTPA.7,14-17,19-21,23-29,32-37,45,46,49,52-57 These studies involved a total of 2535 patients investigated with V-Q lung scanning and 1774 patients assessed by CTPA (Table 3). The rate of non-diagnostic results with V-Q lung scanning ranged from 1.3%36 to 40%14 whereas the rate of non-diagnostic results with CTPA ranged from 0%19 to 57.1%.23,56 The rate of additional imaging tests after a first non-diagnostic V-Q lung scan ranged from 14%37 to 100%23,27 whereas it ranged from 0%35 to 62%15 after a first non-diagnostic CTPA. The pooled rates of non-diagnostic test results with V-Q lung scanning and with CTPA were 14% (95% CI: 1018, I2=90.30%) and 12% (95% CI: 6-17, I2=93.86%), respectively. The 95% confidence intervals of the non-diagnostic rate values overlap (Figure 3). The risk of bias was high in 16 studies,17,19,21,24-28,32,34,36,46,49,54,56,57 moderate in 12 studies7,1416,20,23,29,33,35,45,52,53 and low in only two studies37,55 (Table 1).

Figure 1. Flow chart of the systematic review. MA: meeting abstract; OA: original article; CUS: compression ultrasonography.

178

haematologica | 2019; 104(1)

CTPA versus V-Q lung scan in pregnant women Table 1. Assessment of relevance and bias of the included studies.

continued on the next page

haematologica | 2019; 104(1)

179

C. Tromeur et al. continued from the previous page

continued on the next page

180

haematologica | 2019; 104(1)

CTPA versus V-Q lung scan in pregnant women continued from the previous page

Third study endpoint: radiation exposure

Discussion

Eleven clinically based studies were selected to compare radiation exposure during CTPA and V-Q lung scanning.16,18,20,21,24,28,33,34,52,54,57 The mean maternal effective dose ranged from 0.9 to 5.85 milliSievert (mSv) with V-Q lung scanning and from 0.23 to 9.7 mSv with CTPA (Table 4). The fetal/uterus absorbed dose ranged from 0.2 to 0.7 milliGray (mGy) with V-Q lung scanning and from 0.002 to 0.51 mGy with CTPA.28 Direct comparisons between V-Q lung scanning and CTPA were not possible because of variations in the imaging protocols used and the methods of measuring or calculating radiation exposure. The dose-length product (DLP) was available in four studies:16,20,21,57 it ranged from 69.34Âą10.95 mGy/cm57 to 397.54Âą100.4 mGy/cm.16 Because of the large differences in the applied, mostly unstandardized CTPA protocols among these studies, we refrained from data pooling. A total of 11 relevant studies assessing CTPA radiation exposure in female phantoms showed that the mean maternal effective dose ranged from 2.5 mSv58 to 4.9 mSv59 (Table 5). The fetal/uterus absorbed dose ranged from 0.003 mGy66 to 0.73 mGy.67 These results from the phantom studies should be interpreted with caution and may not be directly extrapolated to clinical practice because of the wide variations in scan techniques and methods of measuring and/or calculating the radiation exposure. No phantom studies with V-Q lung scanning were available.

Our systematic review and meta-analysis provides an overview of all published literature on diagnostic accuracy, scan efficiency and radiation exposure dose of V-Q lung scans versus CTPA in pregnant patients with suspected acute PE. The negative predictive value and rates of non-diagnostic tests were comparable between V-Q lung scans and CTPA, although significant heterogeneity, overall high risk of bias and absence of direct comparisons prevent definite conclusions. Moreover and importantly, studies included in the meta-analysis are mostly outdated and none of the available studies evaluated state-of-theart imaging techniques as currently used in clinical practice. Maternal and fetal radiation exposure with CTPA and V-Q lung scanning could not be compared because of lack of homogeneity in radiation calculation methods and large differences between the scan protocols used. However, all reported radiation measurements for both imaging techniques were clearly below the established harmful threshold of 100 mGy.69 The pooled failure rate for both imaging modalities was negligible, suggesting that both CTPA and V-Q lung scanning can equally safely exclude PE during pregnancy. Our findings are concordant with those recently reported.70 Indeed, in the Cochrane review including 11 studies with 695 CTPA and 665 V-Q lung scan results, the median negative predictive value for both imaging techniques was

haematologica | 2019; 104(1)

181

C. Tromeur et al.

100%.70 The very high negative predictive values need to be interpreted on the background of the very low prevalence of PE, which varied between 1% and 7% in the studies evaluated, implying a very low post-test probability of PE even with less than optimal sensitivity of a diagnostic test.71 Only if current active trials confirm the safety of using the clinical decision rule and a D-dimer test to select patients with a higher pre-test probability of PE, could the diagnostic safety of CTPA and VQ-lung scanning be better tested and compared.3,72 Notably, increasing the level of suspicion of PE with a specific strategy during pregnancy may lead to a lower negative predictive value of both CTPA and V-Q lung scanning. It has been widely acknowledged that, in contrast to CTPA, the risk of a non-diagnostic test result with V-Q lung scanning is considerable. Importantly, we found that the pooled risks of a non-diagnostic test for both imaging tests in the setting of pregnant patients with suspected PE were comparable. These pooled risks need to be put in perspective. For CTPA, a non-conclusive result was defined as suboptimal contrast opacification and respiratory motion artifacts that did not allow for a certain inclusion or exclusion of PE. For V-Q lung scanning, we defined non-diagnostic or inconclusive results according to the PIOPED criteria as intermediate and low probability scan results.73 We found considerably higher rates of non-diagnostic results with CTPA and V-Q lung scanning than those reported in a recent Cochrane review.70

Notably, the definition of non-diagnostic tests was not provided in the Cochrane review and, based on our results, was probably underestimated. Indeed, most of the retrospective studies included in the Cochrane review used intermediate probability V-Q lung scan results as the definition of non-diagnostic results and low probability scans as normal scans whereas we classified low and intermediate probability scan results as non-conclusive. Importantly, clinical probability assessed by clinical judgement or a validated prediction rule is essential for the correct interpretation of a V-Q lung scan: a non-diagnostic V-Q lung scan may exclude PE when combined with negative proximal compression ultrasound sonography in patients with a low clinical probability of PE.73 Compression ultrasound sonography may also be helpful when combined with an intermediate V-Q lung scan probability to confirm or rule out acute PE. Unfortunately, such information was not provided by the studies identified. Therefore, the rate of non-diagnostic V-Q lung scans in our analysis may be biased towards overestimation. Again, the lack of direct comparisons and studies evaluating state-of-the art imaging protocols does not allow for definite conclusions. Of note, we cannot rule out the potential bias that while standard V-Q scan reporting involves a statement on non-diagnostic results, this is not the case for CTPA. It is generally known that CTPA results in relatively higher maternal radiation exposure but lower fetal

Table 2. Analysis of the rate of false negative test results after V-Q lung scans and CTPA.

Study

Number of patients subjected to imaging test (n)

Baseline PE prevalence

Balan et al. 1997 Chan et al. 2002 Scarsbook et al. 2007* Ezwawah et al. 2008 Shahir et al. 2010** Revelet al. 2011 Cutts et al. 2014 Sheen et al. 2017 Golfam et al. 2017

82 113 96 19 99 91 183 225 362

22% (18/82) 7.1% (8/113) 1.0% (1/96) NP 1% (1/99) 11% (10/91) 2.2% (4/183) 2.7% (6/225) 4.7% (17/363)

Scarsbook et al. 2007 Litmanovitch et al. 2009 Shahir et al. 2010 Revel et al. 2011 Bourjeily et al. 2012

9 26 106 43 343

22.2% (2/9) 0% (0/26) 3.7% (4/106) 16% (7/43) 2.6% (9/343)

Browne et al. 2014 Nijkeuter et al. 2013 Sheen et al.2017

70 143 97

1.4% (1/70) 4.2% (6/143) 4.1% (4/97)

Number of true negative test (n)

Number of VTE during follow-up (n)

NPV (%), 95% CI

Duration of follow-up (months)

0 0 0 0 0 0 0 0 0

100, (88.97-100) 100, (95.58-100) 100, (95.86-100) 100, (83.18-100) 100, (95.25-100) 100, (94.34-100) 100, (97.83-100) 100 (98.10-100) 100 (98.95-100)

NP 6 24.5 3 3 3 NP 3 3

6 26 95 28 335

0 0 1 0 0

100, (60.97-100) 100, (87.13-100) 98.96, (94.33-99.82) 100, (87.94-100) 100, (98.86-100)

69 129 84

0 0 2

100, (94.73-100) 100, (97.11-100) 97.94, (99.43-92.79)

24.5 18 3 3 3 months or 6 weeks postpartum 6 3 3

V-Q lung scanning 31 83 89 19 77 64 173 198 316 CTPA

PE: pulmonary embolism; VTE: venous tromboembolism; NPV: negative predictive value; CI: confidence intervals; NP: not provided; V-Q scanning: ventilation perfusion scanning. CTPA: computed tomography pulmonary angiography; *one PE was diagnosed after 3 months of follow-up. **very low PE probability V-Q lung scans are considered as normal V-Q lung scans.

182

haematologica | 2019; 104(1)

CTPA versus V-Q lung scan in pregnant women

absorbed doses than V-Q lung scanning. Importantly, most of the radiation exposures reported in the literature were not measured directly but were calculated and,

therefore, fully dependent on the scan techniques used, which were largely outdated compared to the ones currently used. The higher breast radiation exposure with

Table 3. Analysis of rate of non-diagnostic test results of V-Q lung scanning and CTPA.

Study

Number of Non-diagnostic Non-diagnostic Additional Additional Additional Additional Non-conclusive Anticoagulation patients imaging imaging imaging tests imaging test imaging test imaging test additional despite subjected test (n) test (%) in case of first confirming excluding imaging non-diagnostic to imaging non-diagnostic PE (n) PE (n) test (n) results test (n) test, n (%)

Balan et al. 1997 Chan et al. 2002 Scarsbook et al. 2007 Ridge et al. 2009 Shahir et al. 2010 ** Revel et al. 2011 Scott et al. 2011 Sellem et al. 2013 Abele et al. 2013‡ Astani et al. 2014 ** Cutts et al. 2014† Ramsay et al. 2015† Richard et al. 2015 Sheen et al. 2017 Golfam et al. 2017 Armstrong et al. 2017

82 113 96 25 99 91 73 116 74 23 183 127 77 225 362 769

33 28 7 1 22 17 1 22 13 5 6 37 7 21 29 74

40 24,8 7.3 4 21 18.7 1.3 18.9 16.2 21.7 3.3 29.1 9 9.3 8 9.1

Scarsbook et al. 2007 King-Im et al. 2008 Ridge et al. 2009

9 40 28

1 0 10

11 0 35.7

Bourjeily et al. 2012

343

20.7

Browne et al. 2014 Moradi et al. 2015 Shahir et al. 2015 Ridge et al. 2011

70 27 95 45

1 1 11 10

1.4 3.7 11.5 21.7

Bajc et al. 2015 Scott et al. 2011 Shahir et al. 2010 Revel et al. 2011 Nijkeuter et al. 2013 Tomas et al. 2013 Litmanovitch et al. 2009 Potton et al. 2009 Sheen et al. 2017 Armstrong et al. 2017 Yeo et al. 2017 Mitchell et al. 2017 Halpenny et al. 2017

61 18 106 43 143 10 26 34 97 269 7 99 204

6 2 6 8 8 3 1 7 9 23 4 12 62

9.8 11.1 5.7 18.6 5.5 30 3.8 20 9.3 8.9 57.1 12 30.4

V-Q lung scanning* NP NP NP NP NP NP NP NP NP NP 2 (29) CTPA 0 2 0 1 (100) CTPA NP NP NP 3 (14) CTPA 1 2 0 NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP 13 (100) CTPA 1 9 3 NA NA NA NA NA 2 (33) CTPA 0 0 2 19 (51) CTPA 1 8 10 1 CTPA 0 0 0 9 (43) CTPA 2 5 2 NP NP NP NP NP NP NP NP NP NP CTPA 0 (0) NA NA NA NA NP NP NP NP NP 5 (50) 3 CTPA 1 (V-Q lung scan) 1 (CTPA) 2 (CTPA) 2 V-Q lung scan 1 (V-Q lung scan) 44 (62) 5 CUS+V-Q lung 1 (CUS) NP NP scan or CTPA 39 CUS alone NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP 5 (50) 3 CTPA 1 (V-Q lung scan) 1 (CTPA) 2 (CTPA) 2 V-Q lung scan 1(V-Q lung scan) 1 (17) CTPA 0 0 1 NP NP NP NP NP 3 (50) Q lung scan 0 3 0 3 (37.5) CTPA 0 2 1 NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP 4 (57) NP NP NP NP 3 (33) Q lung scan 0 2 1 NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP NP

12 4 0 NP NP NP NP NP NP NA 2 4 2 NP NP NP NP NP NP NP

NP NP NP NP NP NP 0 NP 1 NP NP NP NP NP NP NP NP

CTPA: computed tomography pulmonary angiography; V-Q scanning: ventilation-perfusion scanning; NP: not provided; NA: not applicable. PE: pulmonary embolism; CUR: compression ultrasonography; *non diagnostic V-Q lung scans were defined by intermediate and low probability scan results. †89 low probability V-Q scans were considered as normal V-Q lung scans. ‡ non-diagnostic V-Q scans were defined as abnormal perfusion scans. ** very low PE probability V-Q lung scans were considered as normal V-Q lung scans.

haematologica | 2019; 104(1)

183

C. Tromeur et al. Table 4. Overview of studies on radiation exposure from CTPA or V-Q lung scanning in real-life patients.

Study

Number of imaging tests

Radiation exposure: real-life studies CTPA radiation exposure V-Q lung scanning radiation exposure 1st 2nd 3rd Average Q lung scanning V-Q lung scanning 1st

Browne et al.2014 Jordan et al.20151

70 CTPA 34 CTPA

Moradi et al. 2015 Ridge et al.20112

27 CTPA 28 CTPA 20 CTPA 77 V-Q lung scanning

Richard e et al.2015

Astani et al. 2014

Revel et al. 2011

Litmanovicth et al. 2009 Armstrong et al. 2017

Mitchell et al. 2017

Halpenny et al. 2017

9.0 9.5 mSv mSv

7.15 mSv* 9.7 9.4 mSv mSv 5.46 mSv* 4.8 mSv** 5.6 mSV** NP

MMED mSv**** BAD mGy FAB mGy 23 V-Q Maternal 21.07 21.26 20.74 21.02 lung scans effective 30 CTPA dose mSV**** BAD mGy 43.36 43.14 46.55 44.35 UFAD mGy 0.47 0.51 0.38 0.46 94 V-Q lung MMED 7.3 mSv* scans mSv 46 CTPA 26 CTPA 769 V-Q lung scans 269 CTPA 84 CTPA 120 kV 15 CTPA 80 kV 69A 135B

2nd

3rd Average

1st

2nd

DLP mGy/cm

3rd Average

NP NP

397.54±100.4 NP

NP NP NP

303.55±98.74 NP NP NP

2.18

5.82

0.19 1.04

0.27 0.24 0.19 1.00 1.07

0.21 1.04

0.81 1.22

0.28 0.24

0.27 0.29 0.27 0.24

0.28 0.25

0.35 0.40 0.9 mSv†

1.24 0.76 0.7 1.32 1.34

0.76 1.29

0.37 0.39 0.42 0.38

0.37 0.40

NP NP NP NP NP NP

MMED mSv BAD mGy

1.79 mSv

105.65±39.77

2-14

0.28

UFAD mGy MED mSv BAD mGy MMED mSv BAD mGy Mean effective mSv Mean effective dose mSv

0.002-0.02 0.23 2.24 0.04 0.25 1.66

0.2 NP

NP NP NP NP NP 118.48±20.05

0.97

69.34±10.95

CTPA: computed tomography pulmonary angiography; V-Q lung scanning: ventilation-perfusion lung scanning; Q: perfusion; MMED: mean maternal effective dose; BA: breast absorbed dose; FAD: fetal absorbed dose; UFAD: uterus/fetal absorbed dose. NP: not provided. *DLP :dose length product (image noise) ; mSv =DLP mGy/cm *0.018(standard conversion). ** The mean effective dose per patient. †88 MBq*11 * 10 - 3 ; each injected megabecquerel represents an effective dose of 11 *10 -3 mSv. 1Average radiation exposure in milliSieverts (k=18 mSv/mGy cm), Radiation dose in pregnant patients. 2Two different CTPA protocols were assessed. **** dose calculation method not provided.

CTPA partly explains the recommendation of V-Q lung scans by international guidelines for pregnant patients with suspected PE. The Society of Thoracic Radiology clinical practice guidelines have presented comparable radiation exposure doses to our findings.74 However, since the studies in our review did not provide all imaging protocol details or full disclosure of the mathematical formulas used, the reported radiation doses in Table 5 are neither comparable between studies nor reproducible. Moreover, mathematical body phantoms (Monte Carlo simulation) of pregnant patients were used instead of realistic physical phantoms in three of the CTPA phantom studies.65,66,68 The presented radiation exposure doses in both phantom and human studies should therefore be interpreted with great caution. Moreover, the risk of early breast cancer seems similar after VQ lung scanning and CTPA.75 184

State-of-the-art imaging techniques For the diagnosis of acute PE, accuracy and pulmonary arterial opacification are significantly improved by optimizing the CTPA protocol for the pregnant patient. This optimization includes a high flow rate (6 instead of 4 mL/s), a high volume (an approximately 25% increase) followed by saline flush, a high concentration of contrast medium (370 mg I/mL), and shallow held inspiration (to avoid the Valsalva maneuver).24 In the Leiden University Medical Center, the contrast volume and speed are titrated according to the patient’s weight. Advised measures to reduce radiation dose include using a 100 kV protocol76 and reduced z-axis technique with limited scan volume from just above the aorta to the basal lung fields (excluding the upper and lower marginal zones).77 For the diagnosis of acute PE with lung scintigraphy in pregnancy, a two-step protocol is suggested to minimize radiation. Initially, perhaematologica | 2019; 104(1)

CTPA versus V-Q lung scan in pregnant women Table 5. Overview of studies on radiation exposure from CTPA or V-Q lung scanning in phantom studies.

Study Chatterson et al. 2014 Chatterson et al. 2011 Doshi et al. 2008

100k Vp 100k Vp 100k Vp 120k Vp** Hurwitz et al. 2006 140k Vp Litmanovitch et al. 2011 100k Vp 120k Vpâ&#x20AC;Ą Winer-Muran et al. 2002 *** 120k Vp Perisinakis et al. 2014 *** 100k Vp 120k Vp Iball et al. 2008 NA Kennedy et al. 2007 NA Motavalli et al. 2017*** 80 kVp 100 kVp 120 kVp Isodoro et al. 2017 100 kVp

Phantom studies with CTPA Foetal/uterus absorbed dose (mGy) 1st trimester 2nd trimester 3rd trimester 0.05 0.11

0.024-0.07

0.003-0.020 NP NP NA NA < 0.01 0.02 0.09 0.28

NP 0.3 0.06* 0.10-0.23* NP 0.084* 0.023-0.140* 0.008-0.077 NP NP NA NA <0.02 0.08 0.2 0.73

Maternal effective dose (mSv) 1st trimester 2nd trimester 3rd trimester

0.13 0.5

2.5 4.9 NP NP NP NP NP

0.051-0.131 NP NP NA NA 0.04 0.18 0.47 0.57

NP NP NP NA NA NP NP NP NP

NP NP NA NA NP NP NP NP

CTPA: computed tomography pulmonary angiography: Vp: kilovolt protocol; NP: not provided; NA: not applicable. *mean fetal absorbed dose;** two different CTPA protocols with 120 kV were assessed; â&#x20AC;Ą three different CTPA protocols with 120kV were assessed; *** Monte Carlo simulation.

Figure 2. Meta-analysis of false negative tests after a first negative ventilation-perfusion lung scan and computed tomography pulmonary angiography in pregnant patients with suspected acute pulmonary embolism. A false negative test is defined by a first negative computed tomography pulmonary angiography (CTPA) or ventilation-perfusion (V-Q) lung scan in a woman who had a pulmonary embolism (PE) diagnosed during the 3 months of follow-up. Three patients had a PE during the follow up.37,55 The type of imaging test performed to diagnose the PE was not provided.

haematologica | 2019; 104(1)

185

C. Tromeur et al.

fusion-only scintigraphy should be performed using a reduced dose of 99mTc-MAA (approximately a quarter of the usual dose administrated for a one-step V/Q scan). Because of the low frequency of co-morbid pulmonary disorders, PE can be excluded in most cases on the basis of a normal perfusion pattern. Ventilation images should only be performed in the case of abnormal perfusion images.

Conclusion Based on the available data, direct comparisons of safety and efficiency between CTPA and V-Q lung scanning do not seem valid. The available studies are based mostly on techniques that are outdated with regard to the current and presently evolving techniques, for both CTPA and V-Q

Figure 3. Meta-analysis of non-diagnostic results of ventilation-perfusion lung scanning and computed tomography pulmonary angiography in pregnant patients with suspected acute pulmonary embolism. The number and type of additional imaging tests are provided in Table 3. V-Q: ventilation-perfusion; CTPA: computed tomography pulmonary angiography.

186

haematologica | 2019; 104(1)

CTPA versus V-Q lung scan in pregnant women

lung scanning. Our most important finding appears to be the very low rate of false negative test results for both imaging modalities, although the low disease prevalence among the studies prevents a solid evaluation of the sensitivity. Moreover, radiation doses associated with CTPA and V-Q lung scanning are well below the safety threshold.

References 1. Guimicheva B, Czuprynska J, Arya R. The prevention of pregnancy-related venous thromboembolism. Br J Haematol. 2015;168(2):163-174. 2. Say L, Chou D, Gemmill A, et al. Global causes of maternal death: a WHO systematic analysis. Lancet Glob Health. 2014;2(6): e323-333. 3. Tromeur C, van der Pol LM, Klok FA, Couturaud F, Huisman MV. Pitfalls in the diagnostic management of pulmonary embolism in pregnancy. Thromb Res. 2017;151 (Suppl 1):S86-S91. 4. Huisman MV, Klok FA. How I diagnose acute pulmonary embolism. Blood. 2013;121(22):4443-4448. 5. Van der Pol LM, Mairuhu AT, Tromeur C, Couturaud F, Huisman MV, Klok FA. Use of clinical prediction rules and D-dimer tests in the diagnostic management of pregnant patients with suspected acute pulmonary embolism. Blood Rev. 2017;31(2):31-36. 6. Cahill AG, Stout MJ, Macones GA, Bhalla S. Diagnosing pulmonary embolism in pregnancy using computed-tomographic angiography or ventilation-perfusion. Obstet Gynecol. 2009;114(1):124-129. 7. Chan WS, Ray JG, Murray S, Coady GE, Coates G, Ginsberg JS. Suspected pulmonary embolism in pregnancy: clinical presentation, results of lung scanning, and subsequent maternal and pediatric outcomes. Arch Intern Med. 2002;162(10):11701175. 8. PRISMA. The PRISMA statement http://www.prisma-statement.org/statement.htm. [cited Aug, 2015)]. 9. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7 (3):177-188. 10. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in metaanalyses. BMJ. 2003;327(7414):557-560. 11. Abujudeh HH, Kaewlai R, Farsad K, Orr E, Gilman M, Shepard JA. Computed tomography pulmonary angiography: an assessment of the radiology report. Acad Radiol. 2009;16(11):1309-1315. 12. Andreou AK, Curtin JJ, Wilde S, Clark A. Does pregnancy affect vascular enhancement in patients undergoing CT pulmonary angiography? Eur Radiol. 2008;18(12):27162722. 13. Angulo VJ, Ramirez GA, Torres Gomez LG, Vargas GA, Cortes SL. [Pulmonary thromboembolism in obstetrics]. Ginecol Obstet Mex. 2004;72:400-406. 14. Balan KK, Critchley M, Vedavathy KK, Smith ML, Vinjamuri S. The value of ventilation-perfusion imaging in pregnancy. Br J Radiol. 1997;70(832):338-340. 15. Bourjeily G, Khalil H, Raker C, et al. Outcomes of negative multidetector computed tomography with pulmonary angio-

haematologica | 2019; 104(1)

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

Depending on new developments and insights of pending studies, decisions regarding the imaging modality of choice should be based on local availability of techniques combined with use of optimal scan protocols tailored to the pregnant patient.

graphy in pregnant women suspected of pulmonary embolism. Lung. 2012;190(1): 105-111. Browne AM, Cronin CG, NiMhuircheartaigh J, et al. Evaluation of imaging quality of pulmonary 64-MDCT angiography in pregnancy and puerperium. AJR Am J Roentgenol. 2014;202(1):60-64. Cutts BA, Tran HA, Merriman E, et al. The utility of the Wells clinical prediction model and ventilation-perfusion scanning for pulmonary embolism diagnosis in pregnancy. Blood Coagul Fibrinolysis. 2014;25(4):375378. Jordan EJ, Godelman A, Levsky JM, Zalta B, Haramati LB. CT pulmonary angiography in pregnant and postpartum women: low yield, high dose. Clin Imaging. 2015;39(2): 251-253. King-Im JM, Freeman SJ, Boylan T, Cheow HK. Quality of CT pulmonary angiography for suspected pulmonary embolus in pregnancy. Eur Radiol. 2008;18(12):2709-2715. Litmanovich D, Boiselle PM, Bankier AA, Kataoka ML, Pianykh O, Raptopoulos V. Dose reduction in computed tomographic angiography of pregnant patients with suspected acute pulmonary embolism. J Comput Assist Tomogr. 2009;33(6):961-966. Moradi M, Monfared LJ. Qualitative evaluation of pulmonary CT angiography findings in pregnant and postpartum women with suspected pulmonary thromboembolism. J Res Med Sci. 2015;20(11):1088-1093. Moriarty JM, Bolster F, O'Connor C, et al. Frequency of nonthromboembolic imaging abnormalities in pregnant women referred for computed tomography pulmonary arteriography. Can Assoc Radiol J. 2015; 66(1):24-29. Ridge CA, McDermott S, Freyne BJ, Brennan DJ, Collins CD, Skehan SJ. Pulmonary embolism in pregnancy: comparison of pulmonary CT angiography and lung scintigraphy. AJR Am J Roentgenol. 2009;193(5): 1223-1227. Ridge CA, Mhuircheartaigh JN, Dodd JD, Skehan SJ. Pulmonary CT angiography protocol adapted to the hemodynamic effects of pregnancy. AJR Am J Roentgenol. 2011;197(5):1058-1063. Sellem A, Elajmi W, Mahjoub Y, Hammami H. Diagnosis pulmonary embolism in pregnancy: contribution of lung scintigraphy. Retrospective study about 116 cases. MED NUCL. 2013;37(10-11):432-438. Shahir K, McCrea JM, Lozano LA, Goodman LR. Reduced z-axis technique for CT pulmonary angiography in pregnancy-validation for practical use and dose reduction. Emerg Radiol. 2015;22(6):651-656. Abele JT, Sunner P. The clinical utility of a diagnostic imaging algorithm incorporating low-dose perfusion scans in the evaluation of pregnant patients with clinically suspected pulmonary embolism. Clin Nucl Med. 2013;38(1):29-32.

28. Astani SA, Davis LC, Harkness BA, Supanich MP, Dalal I. Detection of pulmonary embolism during pregnancy: comparing radiation doses of CTPA and pulmonary scintigraphy. Nucl Med Commun. 2014;35(7):704-711. 29. Bajc M, Olsson B, Gottsater A, Hindorf C, Jogi J. V/P SPECT as a diagnostic tool for pregnant women with suspected pulmonary embolism. Eur J Nucl Med Mol Imaging. 2015;42(8):1325-1330. 30. Gruning T, Mingo RE, Gosling MG, et al. Diagnosing venous thromboembolism in pregnancy. Br J Radiol. 2016;89(1062): 20160021. 31. Hamilton EJ, Green AQ, Cook JA, Nash H. Investigating for pulmonary embolism in pregnancy: five year retrospective review of referrals to the acute medical unit of a large teaching hospital. Acute Med. 2016;15(2): 58-62. 32. Ramsay R, Byrd L, Tower C, James J, Prescott M, Thachil J. The problem of pulmonary embolism diagnosis in pregnancy. Br J Haematol. 2015;170(5):727-728. 33. Revel MP, Cohen S, Sanchez O, et al. Pulmonary embolism during pregnancy: diagnosis with lung scintigraphy or CT angiography? Radiology. 2011;258(2):590598. 34. Richard MC, Lambert R, Rey E, Turpin S. Is perfusion scintigraphy sufficient in pregnant or post-partum women? Med Nucl. 2015;39(6):479-485. 35. Scarsbrook AF, Bradley KM, Gleeson FV. Perfusion scintigraphy: diagnostic utility in pregnant women with suspected pulmonary embolic disease. Eur Radiol. 2007;17(10): 2554-2560. 36. Scott K, Rutherford N, Fagermo N, Lust K. Use of imaging for investigation of suspected pulmonary embolism during pregnancy and the postpartum period. Obstet Med. 2011;4(1):20-23. 37. Shahir K, Goodman LR, Tali A, Thorsen KM, Hellman RS. Pulmonary embolism in pregnancy: CT pulmonary angiography versus perfusion scanning. AJR Am J Roentgenol. 2010;195(3):W214-W220. 38. Bowlen L, Hofman M, Santos P. A retrospective review of lung ventilation and perfusion (V/Q) scanning in pregnant women. Intern Med J. 2011;Conference:41. 39. Butt N, Coffey J, Hill JC. Pulmonary embolism imaging in pregnancy - a 3 year retrospective review. Eur J Nucl Med Mol Imaging. 2011;Conference: S401. 40. Edwards R. Imaging techniques for pregnant patients for suspected PE-a retrospective review. Inter Med J. 2012;Conference:18-19. 41. Hufton A, Albert P, Phitidis ME, Kwok A, Duffy N. Investigating suspected pulmonary embolism (PE) in pregnancy - review of practice at Aintree University Hospital. Eur Respir J. 2015;(1):52. 42. Hullah P, Harris B, Hibbert M. A case review study of investigating for pulmonary

187

C. Tromeur et al.

43. 44. 45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

188

embolism in pregnancy. Respirology. 2011;16(Suppl 1):37-86. Ma HY, Jia CP, Moadel R, Freeman LR. Evaluation for pulmonary emboli in pregnancy. Clin Nucl Med. 2014;Conference: 313. Ma HY, Natenzon A, Sheen J-J, et al. Evaluation of pulmonary emboli in pregnancy. J Nucl Med. 2015;Conference. Nijkeuter M, Tan M, Middeldorp S, Kroft LJM, Beenen L, Huisman MV. Safety of ruling out pulmonary embolism (PE) in pregnancy by computed tomography pulmonary angiography (CTPA). J Thromb Haemost. 2013;Conference:130. Potton EC, Marsh S, Hasford C. Incidence of non-diagnostic CT pulmonary angiography images in the pregnant population: a 4-year experience. Thorax. 2009;Conference:A88. Roseverne L, Schembri G, Roach P. Value of ventilation SPECT scintigraphy in pregnant patients with suspected PE. Intern Med J. 2011;Conference(var.pagings):16. Slim I, Ben GT, Yeddes I, et al. Lung perfusion scintigraphy: diagnostic utility of pulmonary embolism during pregnancy. Eur J Nucl Med Mol Imaging. 2012;Conference: S484. Tomas HS, Phillips J, Bebbington N, Ogunremi T, Guest P. Investigating pulmonary embolism (PE) in pregnancy: single photon emission computed tomography ventilation-perfusion scintigraphy (SPECT VQ). Eur J Nucl Med Mol Imaging. 2013;Conference:S386. Vanes NK, O'Connor H, Baillie P, Sinha A. Diagnosis and management of suspected venous thromboembolism in pregnancy: Experience in a birmingham maternity unit. Arc Dis Child Fetal Neonatal Ed. 2014;Conference:A125. Ezwawah O, Alkoteesh J, Barry JE, Ryan M. Pulmonary embolism in pregnancy: is nuclear medicine imaging still a valid option? Ir Med J. 2008;101(9):281-284. Armstrong L, Gleeson F, Mackillop L, Mutch S, Beale A. Survey of UK imaging practice for the investigation of pulmonary embolism in pregnancy. Clin Radiol. 2017;72(8):696-701. Golfam M, Pham X, Samaan S, Moher D. Optimization of lung scintigraphy in pregnant women at the Ottawa Hospital, Canada. J Nucl Med. 2017;58(Suppl1):1209. Mitchell DP, Rowan M, Loughman E, Ridge CA, MacMahon PJ. Contrast monitoring techniques in CT pulmonary angiography: An important and underappreciated contributor to breast dose. Eur J Radiol. 2017;86: 184-189.

55. Sheen JJ, Haramati LB, Natenzon A, et al. Performance of low dose perfusion scintigraphy and CT pulmonary angiography for pulmonary embolism in pregnancy. Chest. 2018;153(1):152-160. 56. Yeo JH, Zhou L, Lim R. Indeterminate CT pulmonary angiogram: why and does it matter? J Med Imaging Radiat Oncol. 2017;61(1):18-23. 57. Halpenny D, Park B, Alpert J, et al. Low dose computed tomography pulmonary angiography protocol for imaging pregnant patients: can dose reduction be achieved without reducing image quality? Clinical Imaging. 2017;44:101-105. 58. Chatterson LC, Leswick DA, Fladeland DA, Hunt MM, Webster S, Lim H. Fetal shielding combined with state of the art CT dose reduction strategies during maternal chest CT. Eur J Radiol. 2014;83(7):1199-1204. 59. Chatterson LC, Leswick DA, Fladeland DA, Hunt MM, Webster ST. Lead versus bismuth-antimony shield for fetal dose reduction at different gestational ages at CT pulmonary angiography. Radiology. 2011;260 (2):560-567. 60. Doshi SK, Negus IS, Oduko JM. Fetal radiation dose from CT pulmonary angiography in late pregnancy: a phantom study. Br J Radiol. 2008;81(968):653-658. 61. Hurwitz LM, Yoshizumi T, Reiman RE, et al. Radiation dose to the fetus from body MDCT during early gestation. AJR Am J Roentgenol. 2006;186(3):871-876. 62. Iball GR, Kennedy EV, Brettle DS. Modelling the effect of lead and other materials for shielding of the fetus in CT pulmonary angiography. Br J Radiol. 2008;81(966):499503. 63. Kennedy EV, Iball GR, Brettle DS. Investigation into the effects of lead shielding for fetal dose reduction in CT pulmonary angiography. Br J Radiol. 2007;80(956):631638. 64. Litmanovich D, Tack D, Lin P-J, Boiselle PM, Raptopoulos V, Bankier AA. Female breast, lung, and pelvic organ radiation from dosereduced 64-MDCT thoracic examination protocols: a phantom study. AJR Am J Roentgenol. 2011;197(4):929-934. 65. Perisinakis K, Seimenis I, Tzedakis A, Damilakis J. Perfusion scintigraphy versus 256-slice CT angiography in pregnant patients suspected of pulmonary embolism: comparison of radiation risks. J Nucl Med. 2014;55(8):1273-1280. 66. Winer-Muram HT, Boone JM, Brown HL, Jennings SG, Mabie WC, Lombardo GT. Pulmonary embolism in pregnant patients:

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

fetal radiation dose with helical CT. Radiology. 2002;224(2):487-492. Isidoro J, Gil P, Costa G, Pedroso de Lima J, Alves C, Ferreira NC. Radiation dose comparison between V/P-SPECT and CTangiography in the diagnosis of pulmonary embolism. Phys Med. 2017;41:93-96. Motavalli LR, Azghadi EH, Hakimabad HM, Akhlaghi P. Pulmonary embolism in pregnant patients: assessing organ dose to pregnant phantom and its fetus during lung imaging. Med Phys. 2017;44(11):6038-6046. McCollough CH, Schueler BA, Atwell TD, et al. Radiation exposure and pregnancy: when should we be concerned? Radiographics. 2007;27(4):909-917. van Mens TE, Scheres LJ, de Jong PG, Leeflang MM, Nijkeuter M, Middeldorp S. Imaging for the exclusion of pulmonary embolism in pregnancy. Cochrane Database Syst Rev. 2017;1:CD011053. Dronkers CEA, van der Hulle T, Le Gal G, et al. Towards a tailored diagnostic standard for future diagnostic studies in pulmonary embolism: communication from the SSC of the ISTH. J Thromb Haemost. 2017;15(5): 1040-1043. van der Hulle T, Cheung WY, Kooij S, et al. Simplified diagnostic management of suspected pulmonary embolism (the YEARS study): a prospective, multicentre, cohort study. Lancet. 2017;390(10091):289-297. Konstantinides SV, Torbicki A, Agnelli G, et al. 2014 ESC guidelines on the diagnosis and management of acute pulmonary embolism. Eur Heart J. 2014;35(43):3033-3080. Leung AN, Bull TM, Jaeschke R, et al. American Thoracic Society documents: an official American Thoracic Society/Society of Thoracic Radiology clinical practice guideline--evaluation of suspected pulmonary embolism in pregnancy. Radiology. 2012;262(2):635-646. Burton KR, Park AL, Fralick M, Ray JG. Risk of early onset breast cancer among women exposed to thoracic computed tomography in pregnancy or early postpartum. J Thromb Haemost. 2018;16(5):876-885. Schueller-Weidekamm C, Schaefer-Prokop CM, Weber M, Herold CJ, Prokop M. CT angiography of pulmonary arteries to detect pulmonary embolism: improvement of vascular enhancement with low kilovoltage settings. Radiology. 2006;241(3):899-907. Shahir K, McCrea JM, Lozano LA, Goodman LR. Reduced z-axis technique for CT Pulmonary angiography in pregnancy-validation for practical use and dose reduction. Emerg Radiol. 2015;22(6):651-656.

haematologica | 2019; 104(1)

ARTICLE

Stem Cell Transplantation

Machine learning reveals chronic graft-versushost disease phenotypes and stratifies survival after stem cell transplant for hematologic malignancies Jocelyn S. Gandelman,1,2,3,4 Michael T. Byrne,1 Akshitkumar M. Mistry,3,5 Hannah G. Polikowsky,3,4 Kirsten E. Diggins,2,3 Heidi Chen,6 Stephanie J. Lee,7 Mukta Arora,8 Corey Cutler,9 Mary Flowers,7 Joseph Pidala,10 Jonathan M. Irish2,3,4* and Madan H. Jagasia1,3*

Department of Medicine, Division of Hematology/Oncology, Vanderbilt University Medical Center, Nashville, TN; 2Department of Cell and Developmental Biology, Vanderbilt University, Nashville, TN; 3Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN; 4Department of Pathology, Microbiology and Immunology, Vanderbilt University Medical Center, Nashville, TN; 5Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, TN; 6Department of Biostatistics, Vanderbilt University Medical Center, Nashville, TN; 7Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA; 8Division of Hematology, Oncology and Transplantation, University of Minnesota, Minneapolis, MN; 9Stem Cell/Bone Marrow Transplantation Program, Dana-Farber Cancer Institute, Boston, MA and 10H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA 1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):189-196

ABSTRACT

he application of machine learning in medicine has been productive in multiple fields, but has not previously been applied to analyze the complexity of organ involvement by chronic graft-versushost disease. Chronic graft-versus-host disease is classified by an overall composite score as mild, moderate or severe, which may overlook clinically relevant patterns in organ involvement. Here we applied a novel computational approach to chronic graft-versus-host disease with the goal of identifying phenotypic groups based on the subcomponents of the National Institutes of Health Consensus Criteria. Computational analysis revealed seven distinct groups of patients with contrasting clinical risks. The high-risk group had an inferior overall survival compared to the low-risk group (hazard ratio 2.24; 95% confidence interval: 1.363.68), an effect that was independent of graft-versus-host disease severity as measured by the National Institutes of Health criteria. To test clinical applicability, knowledge was translated into a simplified clinical prognostic decision tree. Groups identified by the decision tree also stratified outcomes and closely matched those from the original analysis. Patients in the high- and intermediate-risk decision-tree groups had significantly shorter overall survival than those in the low-risk group (hazard ratio 2.79; 95% confidence interval: 1.58-4.91 and hazard ratio 1.78; 95% confidence interval: 1.06-3.01, respectively). Machine learning and other computational analyses may better reveal biomarkers and stratify risk than the current approach based on cumulative severity. This approach could now be explored in other disease models with complex clinical phenotypes. External validation must be completed prior to clinical application. Ultimately, this approach has the potential to reveal distinct pathophysiological mechanisms that may underlie clusters. Clinicaltrials.gov identifier: NCT00637689. haematologica | 2019; 104(1)

Correspondence: jonathan.irish@vanderbilt.edu or madan.jagasia@vanderbilt.edu Received: March 17, 2018. Accepted: August 17, 2018. Pre-published: September 20, 2018. doi:10.3324/haematol.2018.193441 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/189 ÂŠ2019 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.

189

J.S. Gandelman et al.

Introduction Stem cell transplantation is an important treatment for hematologic malignancies offering a potential cure and a treatment option for advanced disease. However, chronic graft-versus-host disease (GvHD) is a major cause of morbidity and mortality after a transplant.1 Chronic GvHD is a multisystem disease, however its current grading system categorizes disease compositely as mild, moderate or severe.2-4 The current grading system may overlook clinically relevant patterns of chronic GvHD organ scores. For example, a patient with severe skin sclerosis and a patient with highly elevated liver enzymes are both classified as having severe chronic GvHD, despite starkly different clinical manifestations of the disease.3 To date, it has not been straightforward to align the National Institutes of Health (NIH) overall severity classification system and biomarkers.5 There have been some associations between the severity of chronic GvHD, as determined by the NIH classification system (NIHSeverity) and biomarkers, but biomarkers have not been able to predict clinical outcomes as strongly in chronic GvHD as in acute GvHD.6-9 Previous analyses examined disease severity in individual organs and overall disease severity but have not combined organs for phenotypic clinical subgrouping.10 A phenotypic approach to classification has the potential to characterize the pathogenesis of chronic GvHD better. Furthermore, a computational workflow capable of analyzing patterns of chronic GvHD may also have the power to elucidate patterns in other diseases in oncology and throughout clinical medicine. Machine learning and clustering techniques have successfully exposed patterns in medicine, including identifying breast cancer metastases and genetically targeted therapy for acute myeloid leukemia.11-15 Machine learning has the potential to find patterns in clinical data that may be missed by the human observer and traditional approaches alone.16 A potential advantage of machine learning approaches compared to traditional statistical approaches is that results can go beyond a preformed hypothesis

allowing for discovery of novel associations and clusters.17 Additionally, with high-dimensional data, such as the types and grades of organ involvement in chronic GvHD, the multiple comparisons required in conventional statistics can lead to false-positives, whereas a machine learning-inspired approach allows for processing of multidimensional data.15,18,19 Furthermore, an algorithmic approach has outperformed traditional statistics in recent clinical studies.15,20 We used a computational approach to classify patients with chronic GvHD according to organ scores, identify phenotypic subgroups and stratify survival. We hypothesized that machine learning methods could identify distinct clusters of clinical phenotypes and survival patterns among patients with chronic GvHD.

Methods Study population and chronic graft-versus-host disease assessment Research was conducted with informed consent, Institutional Review Board approval and in accordance with the Declaration of Helsinki. The clinical data used were from 339 patients with incident chronic GvHD enrolled in the Chronic GvHD Consortium study, a pre-existing multicenter prospective observational clinical database.21 Incident disease was defined as new chronic GvHD within the 3 months preceding the first study visit and only adult patients (â&#x2030;Ľ18 years of age) were included. The original cohort size was 341; three patients were excluded because of missing organ scores, leaving 339 patients in the final analysis. Demographics and the patientsâ&#x20AC;&#x2122; characteristics were collected at enrollment and through abstraction from clinical charts (Online Supplementary Table S1). At enrollment, NIH 2005 consensus criteria scores from 0 (no involvement) to 3 (severely affected) were recorded for eye, liver, joint, mouth, gastrointestinal tract and lung. Symptom-based lung scores were used in the initial analysis. The percentage of the body surface area with erythema (% erythema) was measured. Skin sclerosis and fascia were assessed using Hopkins scores.22

Figure 1. A machine-learning workflow reveals clusters of patients with chronic graft-versus-host disease with shared organ involvement phenotypes. t-SNE/viSNE plots show organ scores (heat) for each patient (represented by a dot) on a scale where heat indicates organ involvement. Patients who are closer together are more similar while those who are farther apart are generally more different from each other. All organ domains shown were used to generate the viSNE plots, except National Institutes of Health-Severity which was not used as a parameter to generate the viSNE maps. FlowSOM clustering is shown (right) for the seven clusters of patients, with each cluster color overlaid as a dimension on the viSNE plot. For example, Cluster 7 is pink.

190

haematologica | 2019; 104(1)

Machine learning refines cGvHD classification

Machine-learning workflow Nine organ scores were analyzed via a computational workflow consisting of visualization of t-distributed stochastic neighbor embedding (viSNE) for dimensionality reduction,18,23 self-organizing maps (FlowSOM) for patient clustering24 and marker enrichment modeling (MEM) for feature enrichment scoring25,26 (Figure 1 and Online Supplementary Figure S1). viSNE is the visualization of an algorithm called t-distributed stochastic neighbor embedding (t-SNE). Therefore, on all viSNE maps the axes are called t-SNE1 and t-SNE2.23 The machine-learning algorithms are described in detail in the Online Supplementary Methods. NIH scores were squared prior to viSNE analysis and all scores were scaled from 01. FlowSOM clustering was done using t-SNE axes. Skin erythema and sclerosis were analyzed as separate skin features in order to capture type of skin involvement by chronic GvHD. Lung scores did not contribute to patient clustering; lung was neither enriched nor negatively enriched in MEM analysis of organ scores (Online Supplementary Figure S2). Cluster stability analyses were used to determine optimal clustering parameters (Online Supplementary Methods). Analysis with lung excluded from the workflow increased cluster stability, so lung was dropped from the analysis and eight organ scores were used (Online Supplementary Figure S3). Cluster stability with six, seven and eight clusters was tested based on the appearance of seven clusters in viSNE plots (Online Supplementary Figure S4). FlowSOM was run to identify seven clusters, based on similar but increased stability with this parameter. MEM labels are

reported as ▼or ▲ with OrganX where x represents a scale from -10 (most negatively enriched or ▼) to +10 (most enriched or ▲ ). Additional information on MEM and cluster stability validation is provided in the Online Supplementary Methods. De-identified data are available in FlowRepository (http://flowrepository.org/id/FR-FCM-ZYSU).

Risk analysis Kaplan-Meier survival and Cox proportional hazards models were used to analyze overall survival as well as time from stem cell transplantation to development of chronic GvHD. The survival curve of each cluster was fitted using a Cox proportional hazards model and was compared to the survival curve of the whole cohort (Figure 2). The risk coefficient from the hazards model was used as a cluster risk score. Risk groups were stratified into low, intermediate and high based on a coefficient of risk of 0 representing the overall coefficient of risk for the whole cohort, with coefficients < -0.25 indicating low risk and coefficients >0.25 indicating high risk. Non-relapse mortality was analyzed in a competing-risk analysis with relapse as a competing risk. Additional information on the multivariate models is provided in the Online Supplementary Methods.

Software Analyses were conducted using Cytobank, R software version 3.4.2 for Mac, and STATA Version 14. A seed of 42 was used for the FlowSOM analyses.

B Figure 2. Computational analysis of organ scores reveals phenotypic clusters of patients with chronic graft-versus-host disease who were stratified for overall survival. (A) Patients were grouped into seven clusters by the machine-learning workflow (Online Supplementary Figure S1) and described using marker enrichment modeling (MEM) labels (left), which captured features enriched (▲) or specifically lacking (▼) from each group relative to the others in the cohort. Risk coefficients (right) were then calculated for each group. Risk scores below -0.25 or above 0.25 were considered low and high risk, respectively, and 0 was the average risk for the cohort. Clusters 1-3 were lower risk, Cluster 4 was intermediate risk, and Clusters 5-7 were higher risk. (B) Overall survival probability was stratified for the patients with chronic graft-versus-host disease based on the low-, intermediate-, and high-risk clusters defined by the computational analysis.

haematologica | 2019; 104(1)

191

J.S. Gandelman et al.

Results Patients’ organ scores Three hundred and thirty-nine adult patients with chronic GvHD were analyzed, with predominantly intermediate (49.3%, n=167) and high (41.6%, n=141) overall NIH-Severity. Of these 339 patients, 338 had a malignancy as the indication for hematopoietic stem cell transplantation, with acute myeloid leukemia being the most common malignancy affecting 109 (32%) of the subjects. Additional characteristics are described in Online Supplementary Table S1. The organs involved by chronic GvHD at study entry by NIH criteria were the mouth (63%), gastrointestinal tract (37%), eye (43%), joint (24%), fascia (14%), skin by sclerosis (15%), skin by erythema (49%), and lung by symptom score (21%). Detailed organ scores are shown in Online Supplementary Table S2.

Unique chronic graft-versus-host disease phenotypes revealed by machine learning Computational analysis of % erythema, eye, liver, gastrointestinal tract, fascia, joint, mouth, and sclerosis scores revealed seven groups of patients with different clinical phenotypes and risks (Online Supplementary Figure S1). viSNE analysis reduced the dimensionality of chronic GvHD organ scores, with patients who are more similar to each other shown closer together and patients who are more different from each other shown further apart on the scatterplot (Figure 1). For example, a group of patients emerged with involvement of fascia and joints as well as skin sclerosis. In FlowSOM clustering analysis, this group of patients was labeled as Cluster 2 (Figure 1). FlowSOM clustering revealed a total of seven unique clusters of patients (Figures 1 and 2). • Cluster 1: ▲Eye+10 Liver+5 (7.1% of patients); unique in having predominantly ocular involvement, all with an NIH eye score of 3. • Cluster 2: ▲Joint+10, Fascia+5, Sclerosis+4, ▼Mouth-5, Liver-10 (12.7% of patients); a phenotype with enrichment for joint and fascia sclerosis, while specifically lacking mouth and liver GvHD. • Cluster 3: ▲Liver+5 (10.0% of patients); differentiated by moderate liver involvement, all patients with a NIH liver score of 2, while specifically lacking enrichment in other organ scores. • Cluster 4: ▲Mouth+5, ▼Liver-10 (28.9% of patients); enriched for mouth involvement, while lacking enrichment in other organ scores. • Cluster 5: ▲BSA Red+6, ▼Liver-10 (18.3% of patients); this cluster was differentiated by body surface area (BSA) involved by chronic GvHD. • Cluster 6: ▲Mouth+5, Eye+5, Liver+5, GI+1 (13.9% of patients); a phenotype enriched for mouth, eye, liver and gastrointestinal (GI) tract chronic GvHD. • Cluster 7: ▲Liver+10 (9.1% of patients); highly enriched for liver GvHD, all had NIH 3 liver scores while lacking specific involvement in other organ domains. The meaning of positive liver enrichment differed between cluster groups. Cluster 7 differed from other clusters with liver enrichment by capturing patients with a liver score of 3 while Clusters 1, 3 and 6 had patients with liver scores of 1 and 2.

Machine-learning clusters were stable In a cluster stability analysis involving four additional 192

runs of viSNE and FlowSOM using the same organ features, five of the seven clusters were highly stable (Online Supplementary Figure S5). Stability was defined as having a median f-measure ≥0.85. Stable clusters had phenotypically similar MEM labels between replications of analysis as well. Clusters 2-5 and 7 were highly stable. Clusters 1 and 6 were unstable with low reproducibility between replications of analysis.

Clusters of patients identified by machine learning had different overall survival Overall survival probability was stratified for chronic GvHD patients identified in low-risk (Clusters 1-3), intermediate-risk (Cluster 4), and high-risk groups (Cluster 5-7) defined by computational analysis (Figure 2). Time from the development of chronic GvHD to death differed between the high-risk group and the low-risk group [hazard ratio (HR)=2.24; 95% confidence interval (95% CI: 1.36-3.68); P=0.002) and between the intermediate-risk group and the low-risk group (HR=1.70; 95% CI: 0.992.94; P=0.055). Survival differences were not explained by NIHSeverity alone. When NIH-Severity was viewed on the viSNE scatter plot, clusters varied in NIH-Severity. For example, Cluster 2 patients had a combination of moderate and severe chronic GvHD (Figure 1). Additionally, when overall survival of all patients was stratified by NIHSeverity in a Kaplan-Meier analysis, NIH-Severity did not significantly stratify overall survival (log-rank for trend: P=0.08) (Online Supplementary Figure S6).

A physician-driven decision tree recapitulates machine-learning clusters To test clinical applicability, a decision tree was developed to classify patients into the seven clusters (Figure 3). The decision tree was based on expert physicians’ interpretation of the organs that were found together in the machine-learning workflow. The decision tree was constructed through observation of viSNE scatter plots and MEM labels from the clusters of patients identified by the machine learning (Figures 1 and 2A). Patients’ outcomes were not considered in developing the decision tree. This decision tree asks a series of seven questions and can phenotype patients in as few as one question for patients in Cluster 7. The decision tree successfully identified the seven clusters of patients, with highly similar phenotypes to those of the original analysis (Figure 3). Specifically, Clusters 3, 4 and 7 had identical phenotypes by MEM labels when compared with the original machine-learning analysis (Figure 2). The remaining clusters had similar MEM labels to those of the original machine-learning analysis.

The decision tree stratifies patients’ outcomes independently of NIH-Severity Decision-tree-determined risk groups stratified survival. Patients in decision-tree-derived Clusters 1 (ocular predominant phenotype), 2 (sclerotic phenotype) and 3 (liver predominant-moderate phenotype) were classified as low risk based on Cox proportional hazards risk coefficients (Figure 4). Patients in decision-tree-derived Clusters 4 (mixed-phenotype intermediate risk) and 5 (erythema predominant phenotype) were classified as intermediate risk, while patients in Clusters 6 (mixed phenotype-high risk phenotype) and 7 (liver predominant-severe phenotype) haematologica | 2019; 104(1)

Machine learning refines cGvHD classification

Figure 3. A simple, physiciandriven decision tree defines chronic graft-versus-host disease phenotypes. A decision tree designed to separate patients into groups with similar phenotypes and clinical risks as those revealed by the machine-learning approach in Figure 1 is shown. The decision tree is read from the top down and sequentially identifies and segregates patients in the most phenotypically distinct clusters (Y=Yes, N=No). Patients meeting the criteria at the decision point are assigned to that cluster and patients who do not meet the criteria are further advanced in the tree logic. Each circled number represents a cluster of patients. For cluster 2, two decision points were used to identify patients (arrows above and below the encircled 2). The length of the horizontal arrow is proportional to the risk coefficient and the width of the arrow is proportional to the percentage of patients in this cohort who were assigned to the cluster.

Figure 4. A simple, physician-driven decision tree created groups of patients with chronic graft-versushost disease that were similar to computational patient clusters and stratified for overall survival. (A) Cluster numbers, newly calculated marker enrichment modeling (MEM) labels, phenotype interpretations (italics), risk coefficients, and group frequencies (n=339) are shown for the new groups of patients defined using the decision tree in Figure 3. MEM labels and risk were calculated as before (Figure 1 and Methods). Phenotype interpretations were assigned by expert physicians based on analysis of MEM labels and risk. Decision tree groups 1-3 were lower risk, groups 4-5 were intermediate risk, and groups 6-7 were higher risk. (B) Overall survival probability was stratified for patients with chronic graft-versus-host disease identified in the low-, intermediate-, and high-risk groups defined by the physician-driven decision tree.

haematologica | 2019; 104(1)

193

J.S. Gandelman et al.

were classified as high risk. Patients in the high- and intermediate-risk groups had significantly shorter overall survival than those in the low-risk group (HR=2.79; 95% CI: 1.58-4.91; P<0.001 and HR=1.78; 95% CI: 1.06-3.01; P=0.03, respectively (Figure 4). Decision-tree-determined cluster risk groups were also significantly associated with non-relapse mortality (P=0.03). In a multivariate Cox proportional hazards model for overall survival, decision-tree-identified risk groups and platelet counts from 0-590 days were associated with survival (intermediate-risk: HR=1.83; 95% CI; P=0.03, highrisk: HR=2.65; 95% CI: 1.42-4.94; P=0.002; platelet count: HR=3.10; 95% CI: 1.77-5.42; P<0.0001). NIH-Severity was not predictive of survival (moderate: HR=1.49; 95% CI: 0.66-3.38; P=0.34; severe: HR=1.71; 95% CI: 0.75-3.90; P=0.20). A model of decision-tree risk group and NIHSeverity alone showed no statistically significant interaction between these variables. The association between platelet counts and machine-learning-defined clusters is illustrated in Online Supplementary Figure S7.

Individual decision-tree clusters had differential disease trajectories Outcomes and clinical trajectories in the decision-treeidentified clusters were compared. Patients in Cluster 2, a sclerotic phenotype with ▲Joint+7, Fascia+5, Sclerosis+4, ▼ Mouth-5, Liver-10, accounting for 10% of patients, had a significantly longer time from stem cell transplantation to chronic GvHD onset (log-rank: P<0.0001) (Figure 5). Worse overall survival was observed for patients in the decision-tree-derived Cluster 7, a liver predominantsevere phenotype, ▲Liver+10 (HR=1.72; 95% CI: 1.01-2.93; P=0.04) compared with patients in other clusters. Cluster 6, a mixed phenotype, ▲Mouth+5 Eye+2 GI+1, was a novel group with worse overall survival, found after ruling out the other phenotypes in the decision tree (HR=1.75; 95% CI: 1.02-2.98; P=0.04).

Decision-tree reliability and cluster-risk stability There was 86.1% concordance between clusters identified through machine learning and those identified through the decision tree (Figure 6). Bootstrapping indicated stability of risk coefficients in all but one cluster, with all clusters, except Cluster 3, having a standard deviation of risk coefficients <0.7 on ten runs of analysis (Figure 6).

Discussion Seven unique chronic GvHD patients’ phenotypes were revealed through a machine-learning workflow and successfully recapitulated with a clinically applicable decision-tree tool. The revealed groups of patients were stratified for overall survival and a unique sclerotic phenotype with different time from stem cell transplantation to development of chronic GvHD was found. The clusters of patients we describe may overcome the limitations of the current NIH classification system of disease severity which does not account for combinations of organ involvement and did not stratify survival in this cohort. The process of applying this computational workflow to chronic GvHD patients yielded clinically applicable insights. Training analyses revealed that symptom-based lung score did not contribute to clustering and that cluster stability was improved without the lung score (Online Supplementary Figures S2 and S3). In the NIH symptombased lung score, a score from 0-3 is assigned based on the degree of activity needed to cause dyspnea with a requirement for oxygen being scored 3.3 The fact that this symptom-based lung score did not contribute to patient clustering may be due to the subjective nature of the score and suggests that it reflects overall well-being rather than organ-specific involvement. However, it is important to note that the NIH symptom-based lung score has been associated with patients’ outcomes, including non-relapse mortality and overall survival, in an analysis that also included chronic GvHD Consortium patients.27 Clusters of patients identified by the computational workflow were associated with different clinical risk, demonstrated by differences in overall survival. Clusters of patients in the high-risk group were enriched for skin and liver involvement. A skin score of 3 and liver score of 3 have previously been shown to be associated with nonrelapse mortality in an analysis that included patients in this cohort.10 Groups identified by the decision tree continued to stratify survival, with patients in the intermediate-risk group having a 1.8-fold higher risk of mortality compared to those in the low-risk group and patients in the high-risk group having a 2.8-fold higher risk of mortality. Individual high-risk clusters, i.e., Clusters 6 and 7, also independently stratified overall survival when identified by the decision

Figure 5. Time from stem cell transplantation to chronic graft-versus-host disease in decision tree Cluster 2 versus other clusters. Patients in decision-tree-identified Cluster 2-sclerotic phenotype had a significantly longer time from stem cell transplantation to chronic graft-versus-host disease (cGvHD) when compared to patients in all other clusters.

194

haematologica | 2019; 104(1)

Machine learning refines cGvHD classification

Figure 6. The physician-driven decision tree recapitulates the machine-learning workflow and finds clusters with stable risk. (A) A scatter plot shows the same patients in groups resulting from the decision tree (y-axis) or computational analysis (x-axis). Patients within or touching the black boxes were those with the same group classification in both workflows (86% of patients, n=339). (B) Bootstrapping analysis revealed stability of cluster risk across ten decision-tree analysis runs using 130 of 339 randomly sampled patients. The coefficient of risk was calculated for each run of the analysis for each cluster. The standard deviation of the ten coefficients of risks was calculated and was <0.7 for all clusters, except Cluster 3.

tree. Importantly, the decision tree stratified risk of mortality independently of previously defined risk factors for chronic GvHD, including NIH-Severity. Notably, platelet count was a risk factor that continued to stratify risk significantly. Overall, the decision tree has the potential to be applied in the clinical setting to assess patientsâ&#x20AC;&#x2122; phenotypes, once further validation in prospective, independent cohorts has been completed. Additionally, this decision tree can be applied in the research setting to large cohorts of patients. Disease trajectory differed in the decision-tree-identified clusters, most notably for Clusters 2, 6 and 7. The time from stem cell transplantation to development of chronic GvHD was different in Cluster 2, a sclerotic phenotype. This is a clinically relevant and potentially biologically distinct cluster of patients. Longer time to chronic GvHD development is a known clinical finding in patients with sclerotic chronic GvHD.5,28 Previous work defined patients with sclerotic chronic GvHD as having at least one of the following: sclerosis, fascia or joint involvement.29,30 This literature did not comment on the sclerotic phenotype as one with â&#x20AC;&#x153;de-enrichmentâ&#x20AC;? of liver and mouth involvement or take into account the combination of multiple sclerotic features.29,30 The combination of enriched and de-enriched features we describe may enable better association with biomarkers and treatment response. Cluster 6, a mixed phenotype, high-risk cluster, was a novel high-risk cluster revealed by the decision tree. This cluster was defined by enrichment for mouth, eye, and gastrointestinal tract involvement. Notably, this cluster required the highest number of questions on the decision tree to reach, indicating that it was poorly defined and required that other clusters were ruled out to find patients in this phenotypic group. Patients in this cluster had significantly worse overall survival when compared to all those in all other clusters combined. A caveat is that, in stability analysis of the machine-learning workflow, Cluster 6 was not highly stable, but it did recur through all repetitions of analysis (Online Supplementary Figure S5). The combination of these areas of organ involvement has haematologica | 2019; 104(1)

not been previously cited as a risk factor for adverse outcomes in chronic GvHD and should be further explored through cellular analyses for biomarkers and evaluated in continued validation cohorts. Patients in Cluster 7 derived from the decision tree, a liver predominant-severe phenotype, also had a different disease trajectory when compared to patients in other clusters in that they had a significantly worse overall survival than patients in all other clusters combined. This decision-tree-derived cluster is supported by previous research showing that severe elevation of liver enzymes is a known risk factor for adverse outcomes in chronic GvHD.10 Prognostication by clustering is distinct from prognostication by individual organ scores alone. For example, in the machine-learning analysis, Cluster 5 lacked liver involvement and was a high-risk cluster, while high-risk Cluster 6 and Cluster 7 were specifically enriched for liver involvement. This supports the concept that this single organ score does not confer unidirectional low or high risk within the clusters. Furthermore, Liver+5 enrichment was seen in multiple low-risk clusters and one high-risk cluster. Clustering is unique in that it is not an individual organ score or characteristic but rather combinations of organ involvement and the specific absence of organ involvement that drive cluster formation and likely prognosis. Another example of this is that mouth enrichment was seen in both an intermediate-risk cluster (Cluster 4) and high-risk cluster (Cluster 6). Cluster 6, a high-risk cluster, comprises mouth, eye and liver enrichment; these individual enrichment types appear in low-risk clusters but it is perhaps the combination that makes this a high-risk cluster. However, we cannot rule out that gastrointestinal tract enrichment, uniquely present in Cluster 6, is not the driving force of adverse outcomes. A limitation of the machine-learning approach is that it is not possible to add new patients to this analysis without shifting the current clusters. This was overcome by the decision-tree approach. Validation with an external cohort as well as comparison with other risk stratification tools for chronic GvHD31 should further strengthen the findings 195

J.S. Gandelman et al.

of the computational and decision-tree analyses. We were unable to analyze whether clusters predicted response to therapy, as this was an observational cohort in which patients were on any systemic therapy at study entry. Thus, treatment response is an outcome of interest in assessing the utility of machine learning for chronic GvHD outcome stratification. An external validation cohort is pending for this analysis. External validation of machine-learning approaches is the gold standard, and external validation is necessary prior to clinical application of the findings. These results have the potential to be applied to stratify risk in the clinical setting, enhance the current chronic GvHD classification system, refine inclusion criteria for phase 2 trials, and guide biomarker discovery for more specific therapeutic targets. The distillation of machinelearning knowledge into a decision tree increases the feasibility of clinical application of the clusters. However, the clusters have not been externally validated, and this step should be explored before clinical application. Lastly, this a flexible machine learning-inspired work-

References 1. Arora M, Cutler CS, Jagasia MH, et al. Late Acute and chronic graft-versus-host disease after allogeneic hematopoietic Cell Transplantation. Biol Blood Marrow Transplant. 2016;22(3):449-455. 2. SociĂŠ G, Ritz J. Current issues in chronic graft-versus-host disease. Blood. 2014;124 (3):374-384. 3. Filipovich AH, Weisdorf D, Pavletic S, et al. National Institutes of Health Consensus Development Project on Criteria for Clinical Trials in Chronic Graft-versus-Host Disease: I. Diagnosis and Staging Working Group report. Biol Blood Marrow Transplant. 2005;11(12):945-956. 4. Jagasia MH, Greinix HT, Arora M, et al. National Institutes of Health Consensus Development Project on Criteria for Clinical Trials in Chronic Graft-versus-Host Disease: I. The 2014 Diagnosis and Staging Working Group report. Biol Blood Marrow Transplant. 2015;21(3):389-401.e381. 5. Cooke KR, Luznik L, Sarantopoulos S, et al. The biology of chronic graft-versus-host disease: a task force report from the National Institutes of Health Consensus Development Project on Criteria for Clinical Trials in Chronic Graft-versus-Host Disease. Biol Blood Marrow Transplant. 2017;23(2): 211-234. 6. Yu J, Storer BE, Kushekhar K, et al. Biomarker panel for chronic graft-versushost disease. J Clin Oncol. 2016;34(22):25832590. 7. Hartwell MJ, Ozbek U, Holler E, et al. An early-biomarker algorithm predicts lethal graft-versus-host disease and survival. JCI Insight. 2017;2(3):e89798. 8. Major-Monfried H, Renteria AS, Pawarode A, et al. MAGIC biomarkers predict longterm outcomes for steroid-resistant acute GVHD. Blood. 2018;131(25):2846-2855. 9. Paczesny S, Krijanovski OI, Braun TM, et al. A biomarker panel for acute graft-versushost disease. Blood. 2009;113(2):273-278. 10. Inamoto Y, Martin PJ, Storer BE, et al. Association of severity of organ involve-

196

11.

12.

13.

14.

15.

16. 17.

18. 19.

20.

21.

flow with numerous potential applications. The stability of the clusters suggests that this approach will be highly useful in revealing groups not only for this disease but for others that have complex phenotypes. Although the endpoint for this analysis was overall survival, this workflow could be applied to explore whether clusters of patients differ in treatment response or composite chronic GvHD endpoints, such as failure-free survival. Additionally, this workflow has the potential to be applied to other human diseases with complex classification systems such as myelodysplastic syndrome and brain tumors. This approach may change the classification of human disease by revealing otherwise unapparent, clinically relevant patterns. Acknowledgments This work was supported by funding from The Vanderbilt Medical Scholars Program and Vanderbilt Ingram Cancer Center. We thank Dr. Yan Guo for helpful discussions, Allison Greenplate and Benjamin Reisman for helpful insight on figure design, and Dr. Paul Martin for his critical review of the manuscript.

ment with mortality and recurrent malignancy in patients with chronic graft-versushost disease. Haematologica. 2014;99(10): 1618-1623. Esteva A, Kuprel B, Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):115-118. Ting DSW, Cheung CY, Lim G, et al. Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes. JAMA. 2017;318(22):2211-2223. Wang L, Liang R, Zhou T, et al. Identification and validation of asthma phenotypes in Chinese population using cluster analysis. Ann Allergy Asthma Immunol. 2017;119(4):324-332. Ehteshami Bejnordi B, Veta M, Johannes van Diest P, et al. Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer. JAMA. 2017;318(22):2199-2210. Lee SI, Celik S, Logsdon BA, et al. A machine learning approach to integrate big data for precision medicine in acute myeloid leukemia. Nat Commun. 2018;9(1):42. Deo RC. Machine learning in medicine. Circulation. 2015;132(20):1920-1930. Diggins KE, Ferrell Jr PB, Irish JM. Methods for discovery and characterization of cell subsets in high dimensional mass cytometry data. Methods. 2015;82:55-63. van der Maaten LHG. Visualizing highdimensional data using t-SNE. J Mach Learn Res. 2008;9:2579-2605. Saeys Y, Van Gassen S, Lambrecht BN. Computational flow cytometry: helping to make sense of high-dimensional immunology data. Nat Rev Immunol. 2016;16(7):449462. Singal AG, Mukherjee A, Elmunzer BJ, et al. Machine learning algorithms outperform conventional regression models in predicting development of hepatocellular carcinoma. Am J Gastroenterol. 2013;108(11):17231730. Chronic GVHD Consortium. Rationale and design of the chronic GVHD cohort study:

22.

23.

24.

25.

26.

27.

28.

29.

30. 31.

improving outcomes assessment in chronic GVHD. Biol Blood Marrow Transplant. 2011;17(8):1114-1120. Inamoto Y, Pidala J, Chai X, et al. Joint and fascia manifestations in chronic graft-versushost disease and their assessment. Arthritis Rheumatol. 2014;66(4):1044-1052. Amir E-aD, Davis KL, Tadmor MD, et al. viSNE enables visualization of high dimensional single-cell data and reveals phenotypic heterogeneity of leukemia. Nature Biotechnology. 2013;31(6):545-552. Van Gassen S, Callebaut B, Van Helden MJ, et al. FlowSOM: using self-organizing maps for visualization and interpretation of cytometry data. Cytometry A. 2015;87(7):636-645. Diggins KE, Greenplate AR, Leelatian N, Wogsland CE, Irish JM. Characterizing cell subsets using marker enrichment modeling. Nat Methods. 2017;14(3):275-278. Diggins KE, Gandelman JS, Roe CE, Irish JM. Generating quantitative cell identity labels with marker enrichment modeling (MEM). Curr Protoc Cytom. 2018;83: 10.21.11-10.21.28. Palmer J, Williams K, Inamoto Y, et al. Pulmonary symptoms measured by the national institutes of health lung score predict overall survival, nonrelapse mortality, and patient-reported outcomes in chronic graft-versus-host disease. Biol Blood Marrow Transplant. 2014;20(3):337-344. Kitko CL, White ES, Baird K. Fibrotic and sclerotic manifestations of chronic graft versus host disease. Biol Blood Marrow Transplant. 2012;18(1 Suppl):S46-S52. Inamoto Y, Storer BE, Petersdorf EW, et al. Incidence, risk factors, and outcomes of sclerosis in patients with chronic graft-versushost disease. Blood. 2013;121(25):50985103. Inamoto Y, Martin PJ, Flowers MED, et al. Genetic risk factors for sclerotic graft-versushost disease. Blood. 2016;128(11):1516-1524. Arora M, Klein JP, Weisdorf DJ, et al. Chronic GVHD risk score: a Center for International Blood and Marrow Transplant Research analysis. Blood. 2011;117(24): 6714-6720.

haematologica | 2019; 104(1)

ARTICLE

Cell Therapy & Immunotherapy

Department of Hematology, Leiden University Medical Center, Leiden, the Netherlands

Ferrata Storti Foundation

Haematologica 2019 Volume 104(1):197-206

ABSTRACT

tem cell grafts from 10/10 HLA-matched unrelated donors are often mismatched for HLA-DP. In some patients, donor T-cell responses targeting the mismatched HLA-DP allele(s) have been found to induce a specific graft-versus-leukemia effect without coinciding graftversus-host disease, whereas in other cases significant graft-versus-host disease occurred. Cell-lineage-specific recognition patterns within the allogeneic HLA-DP-specific donor T-cell repertoire could explain the differential clinical effects mediated by donor T cells after HLA-DP-mismatched allogeneic stem cell transplantation. To unravel the composition of the HLA-DP T-cell repertoire, donor T-cell responses were provoked by in vitro stimulation with allogeneic HLA-DP-mismatched monocyte-derived dendritic cells. A strategy including depletion of reactivity against autologous dendritic cells allowed efficient identification and enrichment of allo-reactive T cells upon stimulation with HLA-DPmismatched dendritic cells. In this study we elucidated that the allogeneic HLA-DP-restricted T-cell repertoire contained T cells with differential cell-lineage-specific recognition profiles. As expected, some of the allogeneic HLA-DP-restricted T cells showed broad recognition of a variety of hematopoietic and non-hematopoietic cell types expressing the targeted mismatched HLA-DP allele. However, a significant proportion of the allogeneic HLA-DP-restricted T cells showed restricted recognition of hematopoietic cells, including primary malignant cells, or even restricted recognition of only myeloid cells, including dendritic cells and primary acute myeloid leukemia samples, but not of other hematopoietic and non-hematopoietic cell types. These data demonstrate that the allogeneic HLA-DP-specific T-cell repertoire contains T cells that show restricted recognition of hematopoietic cells, which may contribute to the specific graft-versus-leukemia effect without coinciding graft-versushost disease.

Correspondence: a.laghmouchi@lumc.nl

Received: March 22, 2018. Accepted: August 17, 2018. Pre-published: September 20, 2018. doi:10.3324/haematol.2018.193680 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/197 ÂŠ2019 Ferrata Storti Foundation

Introduction T-cell-depleted allogeneic stem-cell transplantation (alloSCT) can be applied in the treatment of patients suffering from hematologic malignancies.1-4 The aim of this approach is to replace the hematopoietic system of the patient (including the malignant cells) with a healthy hematopoietic system reconstituted from a stem cell graft of an HLA-matched donor.5-9 The depletion of T cells from the graft reduces the occurrence of acute graft-versus-host disease (GvHD) mediated by alloreactive donor T cells targeting normal, non-hematopoietic cells of the patient.3,10-12 However, donor T cells are essential in the beneficial graft-versusleukemia (GvL) effect and T-cell immunity is also required for anti-viral protechaematologica | 2019; 104(1)

197

A. Laghmouchi et al.

tion.3,8,13 Therefore, scheduled donor lymphocyte infusions are often given to patients following T-cell-depleted alloSCT.4,14-17 The effect of the level of HLA matching on clinical outcome after alloSCT has been extensively studied and has led to a standardized donor-patient matching procedure in which HLA-DP is frequently not taken into account.1820 The 10/10 HLA-matched stem cell grafts from unrelated donors are, therefore, often mismatched for one or two HLA-DP alleles. HLA-DP mismatches are in theory acceptable, as under non-inflammatory conditions expression of HLA-class II molecules is mainly restricted to hematopoietic cells. The balance between the induction of GvL and GvHD is one of the remaining challenges in treatment with (T-cell-depleted) alloSCT or donor lymphocyte infusion.14 Whether or not donor T cells targeting the mismatched HLA-DP allele(s) contribute to GvL and/or GvHD may depend on the magnitude, specificity and diversity of the allo-HLA-DP immune response as was illustrated for HLA-class Irestricted T-cell responses,21 and the tissue-specific pattern of HLA-DP expression.3,14,22-25 The magnitude of the T-cell response mounted against allo-HLA-DP molecules has been shown to be different depending on the specific HLA-DP allele(s) expressed in the donor and the patient.25,26 Differences at the amino acid sequence level have been found to influence the immunogenicity of specific HLA-DP molecules, probably caused by a different landscape of peptides (peptidome) presented in these HLA-DP molecules.26-28 Based on these insights, an algorithm has been proposed to predict the risk on an immune reaction in the host versus graft direction (rejection) and the graft versus host direction (GvL and/or GvHD), based on the immunogenicity of specific HLADP molecules and the differences between specific HLADP alleles.29 This has led to the distinction of two groups of HLA-DP mismatches, called the more tolerable, permissive HLA-DP mismatches that are predicted to induce T-cell responses with a lower amplitude, and the nonpermissive mismatches that induce stronger T-cell responses.29-32 In addition to the specificity and magnitude of the allo-HLA-DP T-cell response, the pattern of expression of HLA-DP on patientsâ&#x20AC;&#x2122; tissues is decisive in the induction of GvL and/or GvHD. In some patients, profound CD4 T-cell responses targeting the mismatched allo-HLA-DP allele(s) have been found to be associated with the induction of different types of GvHD (e.g. skin GvHD, gut GvHD) mediated by recognition of inflamed HLA-class II-expressing non-hematopoietic tissues.23 In other patients specific GvL reactivity without coinciding GvHD mediated by allo-HLA-DP-reactive CD4 donor T cells was demonstrated. In these patients the allo-HLA-DP response appeared to be restricted to hematopoietic cells without cross-reactivity against nonhematopoietic tissues.22,24 To initiate the allo-HLA-DP-specific immune response in vivo, donor T cells most likely first have to encounter patient-derived HLA-DP-expressing antigen-presenting cells, such as dendritic cells (DC).33,34 Since DC are of hematopoietic origin, allo-HLA-DP T cells activated by DC will probably recognize antigens expressed by hematopoietic cells presented in the mismatched HLADP molecule.14 When the allo-HLA-DP-specific immune response is initiated by DC residing in inflamed HLA-DPexpressing non-hematopoietic tissue, the immune 198

response is likely to be also directed against antigens expressed by non-hematopoietic cells presented in the mismatched HLA-DP molecule of the DC.35 The level of overlap between the antigens expressed by hematopoietic versus non-hematopoietic cells, will dictate the induction of a specific GvL response, a specific GvHD response, or a combination of both.3,14 In this study we analyzed the tissue/cell-lineage-specific recognition patterns within the allo-HLA-DP-specific T-cell repertoire provoked by stimulation with allogeneic HLA-DP-mismatched monocyte-derived DC. We observed that the allo-HLA-restricted T-cell repertoire contains T cells with a diverse spectrum of cell-lineagespecific recognition profiles, including T cells that show restricted recognition of hematopoietic cells, including primary malignant cells, or even T cells with myeloid-lineage-restricted recognition, including recognition of primary acute myeloid leukemia blasts.

Methods Cell collection and preparation The collection and preparation of cells is described in the Online Supplementary Appendix.

Induction of allo-HLA-DP-specific immune responses CD14-depleted donor peripheral blood mononuclear cells were stimulated with irradiated (25 Gy) HLA-DP mismatched allogeneic DC (alloDC) in a 10:1, responder T-cell to stimulatorcell ratio. The cells were cultured in Iscove modified Dulbecco medium (IMDM) containing 10% heat-inactivated ABOS supplemented with interleukin-7 (10 ng/mL, Miltenyi Biotec), interleukin-15 (0.1 ng/mL, Miltenyi Biotec) and interleukin-2 (50 IU/mL, Novartis Sandoz Pharmaceuticals). To activate and remove auto-reactive T cells, cell cultures were stimulated after 14 days with autologous DC (autoDC), followed by depletion of these reactive T cells around 36 h later based on activation marker (CD137) expression using CD137-allophycocyanin (BD Pharmingen, San Diego, CA, USA), APC-beads (Miltenyi Biotec) and MACS LD columns (Miltenyi Biotec) according to the manufacturerâ&#x20AC;&#x2122;s instructions. To activate allo-HLA-DP-reactive T cells, the negative fraction was subsequently specifically restimulated with the allogeneic HLA-DP-mismatched DC. After approximately 36 h, HLA-DP-mismatched-DC-reactive CD4 T cells were quantified and clonally isolated by single cell flow cytometric cell sorting based on CD137 expression using a FACSAria (BD Biosciences, San Jose, CA, USA). The time-point for CD137 isolation was chosen based on our previous work, including an unpublished clinical trial [administration of leukemia-reactive donor T cells after alloSCT or donor lymphocyte infusion to patients with persistent or relapsed mature Bcell neoplasms with blood and/or bone marrow involvement (EudraCT number 2012-003691-40)] and our experience with the isolation of virus-specific CD4 and CD8 T cells.36,37 Unstimulated peripheral blood mononuclear cells and autoDCstimulated peripheral blood mononuclear cells from the same responder cells were used as controls to determine the gating strategy. The T-cell clones were expanded using allogeneic feeder mixture consisting of IMDM containing 5% heat-inactivated ABOS and 5% heat-inactivated fetal calf serum supplemented with 5x irradiated (35 Gy) allogeneic feeder cells, 0.5x irradiated (60 Gy) allogeneic Epstein-Barr virus-transformed lymphoblastoid cell line (EBV-LCL), 100 IU/mL interleukin-2, and 800 ng/mL phytohemagglutinin (PHA-HA16, Oxoid, Altrincham, UK). haematologica | 2019; 104(1)

HLA-DP-restricted cell-lineage recognition patterns

Generation and culture of stimulator cells for functional analyses The method of generating and culturing stimulator cells is described in the Online Supplementary Appendix.

Recognition assay The procedure for testing the recognition profiles of expanded T-cell clones is described in the Online Supplementary Appendix.

Flow cytometric analyses Information on the flow cytometric analyses performed is provided in the Online Supplementary Appendix.

Approval by ethics committee Donors and patients had given written informed consent to the storage of biomaterials in the LUMC Biobank and the use of these materials was approved by the institutional medical ethical committee (protocol number B 16.039).

Results Depletion of reactivity against autologous dendritic cells allows efficient identification and enrichment of allo-reactive T cells upon stimulation with HLA-DP-mismatched dendritic cells To study the composition of the allo-HLA-DP-specific T-cell repertoire, HLA-DP-mismatched alloDC were used to provoke allo-HLA-DP-reactive T-cell responses from healthy donor peripheral blood mononuclear cells. Responder/stimulator pairs with various HLA-DP mismatches were selected (Table 1): two pairs with a targeted HLA-DP*04:01 or HLA-DP*04:02 mismatch using HLADP*04:02- or HLA-DP*04:01-expressing responder cells, respectively, classified as minimal, permissive mismatches31 (pairs 1 and 2), one pair with a targeted HLA-DP mismatch classified as permissive (pair 3), one pair with a targeted HLA-DP mismatch classified as non-permissive (pair 4), and one 9/10 HLA-matched pair with an HLA-DP mismatch classified as permissive and an additional HLADQB1 mismatch (pair 5). Since we previously illustrated that stimulation with alloDC results in coinciding stimulation of reactivity against autoDC,38 we first investigated whether this would hamper the isolation of allo-HLA-DP reactive T cells. Donor peripheral blood mononuclear cells were stimulated with HLA-DP-mismatched alloDC, cultured for 2 weeks, and then restimulated with either autoDC or

alloDC. Flow cytometric analysis of the expression of the activation marker CD137 on CD4 T cells showed that upon (re)stimulation with autoDC and alloDC similar frequencies of CD4 T cells were activated (Figure 1A), indicating that a depletion step for the autoDC reactivity could allow more efficient identification and enrichment of allo-HLA-DP reactive T cells. Therefore, a strategy was developed using initial stimulation with HLA-DP-mismatched alloDC and subsequent culture for 2 weeks, followed by a stimulation step with autoDC and a depletion step using magnetic bead separation (MACS) based on the expression of CD137 on autoDC reactive T cells. The depleted fraction (CD137-) was subsequently restimulated with alloDC to specifically activate allo-reactive T cells (Online Supplementary Figure S1). For four of the five pairs the frequencies of CD137expressing CD4 T cells were consistently higher upon specific restimulation with the HLA-DP-mismatched alloDC compared to the non-restimulated and autoDC-stimulated controls, although there was biological variability in the magnitude of the immune responses (Figure 1B). These data indicate that this strategy using a depletion step for autoDC reactivity allows more specific identification of allo-reactive T cells. Next, CD4 T cells expressing CD137 upon restimulation with HLA-DP-mismatched alloDC were sorted as single cells per well, expanded and screened (n=1679) for alloreactivity by interferon-Îł enzyme-linked immunosorbent assay upon stimulation with alloDC, using autoDC as controls (response 4, as a representative example, is shown in Online Supplementary Figure S2). The clonality of expanded T-cell clones was confirmed by T-cell receptor VÎ˛ repertoire analysis (data not shown). For responses 3-5 the majority of expanded T-cell clones were found to be allo-reactive (79-94%, n=1235). For responses 1 and 2, in which minimal, permissive HLA-DP mismatches were targeted resulting in less profound immune responses, only a limited proportion of the expanded clones were found to be allo-reactive (21-29%, n=68). A small proportion of the clones showed no reactivity or showed coinciding reactivity against the autoDC (Figure 1C).

The allo-HLA-DP-specific T-cell repertoire provoked by in vitro stimulation with HLA-DP-mismatched dendritic cells contains T cells that selectively recognize dendritic cells, but not Epstein-Barr-transformed lymphoblastoid cell lines To investigate the HLA-DP restriction of the allo-reactive CD4 T-cell clones, clones (n=1303) were tested in a

Table 1. HLA-DP typing of the responder/stimulator pairs.

In vitro response

Responder

Stimulator

HLA-DP disparity

Level of HLA-matching

HLA-class II mismatch

DPB1*04:01:01

Permissive

10/10

DPB1*04:01:01

DPB1*03:01 & DPB1*04:02 DPB1*04:01:01

Permissive

10/10

DPB1*04:02

DPB1*04:01:01

Permissive

10/10

DPB1*02:01:02

DPB1*04:01:01

Non-permissive

10/10

DPB1*03:01:01

DPB1*03:01 & DPB1*09:01

DPB1*04:01 & DPB1*04:02 DPB1*02:01:01 & DPB1*04:01:01 DPB1*03:01:01 & DPB1*04:01:01 DPB1*04:01

Permissive

9/10

DPB1*04:01 & DQB1*06:03

haematologica | 2019; 104(1)

199

A. Laghmouchi et al. A

Figure 1. Reactivities within allo-reactive T-cell responses provoked by stimulation with HLA-DP-mismatched dendritic cells. (A) Frequencies (%) of CD137+ cells in the CD4+ T-cell population at 36 h after restimulation [white bars = responder cells not restimulated (NR), gray bars = responder cells stimulated with autologous CD14-derived dendritic cells (autoDC), black bars = responder cells restimulated with allogeneic CD14-derived dendritic cells (alloDC)]; representative example. (B) Frequencies (%) of CD137+ T cells in the CD4+ Tcell populations of responses 1-5 at 36 h after restimulation of the responder cells that underwent a previous depletion step of autoDC reactive T cells (white bars = NR responder cells, gray bars = autoDC-stimulated responder cells, black bars = alloDC-restimulated responder cells). (C) Reactivity of the expanded T-cell clones from the allo-reactive T-cell responses 1-5 (n=107, 158, 540, 581, and 293, respectively) (white bars = non-reactive Tcell clones, gray bars = autoDC reactive T-cell clones, black bars = alloDC reactive T-cell clones).

stimulation assay against third-party DC and EBV-LCL expressing the mismatched HLA-DP alleles (Online Supplementary Figure S3A) that were targeted in the immune responses, using autoEBV-LCL, autoDC and alloDC as controls. The majority (48-99%, median 73%, black bars in Figure 2A) of the allo-reactive T-cell clones showed HLA-DP-restricted reactivity demonstrated by recognition of alloDC and third-party EBV-LCL expressing the targeted mismatched HLA-DP allele (response 4, as a representative example, is shown in Online Supplementary Figure S3B). A representative HLA-DPB1*03:01-restricted T-cell clone is shown that only recognizes DC and EBVLCL, but no other hematopoietic cells (CD14+ and PHAblasts) (Figure 2B). However, a significant proportion (152%, median 7%, gray bars in Figure 2A) of the allo-reactive T-cell clones recognized alloDC, but showed no reactivity against the third-party EBV-LCL expressing the targeted mismatched HLA-DP allele (response 4, as a representative example, is shown in Online Supplementary Figure S3C). Selected T-cell clones from this group were tested against third-party DC and showed recognition of these stimulator cells by production of interferon-Îł (response 4, as a representative example, is shown in Online Supplementary Figure S3D). A representative HLADPB1*03:01-restricted T-cell clone that recognizes only DC and no other hematopoietic cells is shown (Figure 2C). To confirm the specific HLA-DP restriction, T-cell clones were tested against K562 cell lines transduced with specific HLA-DP alleles or with empty vector (mock) (Online Supplementary Figure S4A). The T-cell clones only produced interferon-Îł against the K562 transduced with the target HLA-DP allele (HLA-DPB1*03:01; representative T-cell clones, response 4 in Online Supplementary Figure S4B). 200

To analyze whether the differential cell-recognition patterns of the allo-HLA-DP reactive T-cell clones is caused by recognition of a polymorphic antigen (minor histocompatibility antigen), a larger panel of DC and EBV-LCL was used as stimulator cells. The allo-reactive T-cell clones that showed HLA-DP-restricted reactivity against both DC and EBV-LCL manifested HLA-DP-restricted recognition of all DC (Figure 2D; representative HLA-DPB1*03:01restricted clone) and EBV-LCL (Figure 2F; representative HLA-DPB1*03:01-restricted clone) expressing the target HLA-DP allele, indicating that most likely a monomorphic peptide is recognized. The allo-reactive T-cell clones that showed HLA-DP-restricted reactivity against DC but not against EBV-LCL showed HLA-DP-restricted recognition of the extended panel of DC (Figure 2E; representative HLA-DPB1*03:01-restricted clone) but not of the extended panel of EBV-LCL (Figure 2G; representative HLADPB1*03:01-restricted clone). Furthermore, the restricted recognition of DC but not of EBV-LCL from the same individual (e.g. DC and EBV-LCL from individuals 2, 3, 9 and 10 in Figure 2E,G) indicates that this restricted recognition profile is most likely caused by recognition of a cell-lineage-specific monomorphic peptide in the context of the allo-HLA-DPB1*03:01 molecule. In response 5, with an additional HLA-DQB1 mismatch, 38% of the allo-reactive clones did not show HLA-DPrestricted reactivity against EBV-LCL expressing the mismatched HLA-DP molecules. Testing of these clones against a panel of DC and EBV-LCL expressing the mismatched HLA-DQB1*06:03 allele revealed that 33% of the allo-reactive clones exerted HLA-DQB1-restricted reactivity against DC and EBV-LCL, and 5% against DC, but not against EBV-LCL (bar 5b in Figure 2A; reactivity of reprehaematologica | 2019; 104(1)

HLA-DP-restricted cell-lineage recognition patterns

haematologica | 2019; 104(1)

Figure 2. Frequencies of HLAclass II-restricted CD4 T-cell clones. (A) Frequencies (%) of Tcell clones showing reactivity against allogeneic CD14-derived dendritic cells (alloDC) and thirdparty Epstein-Barr virus-transformed lymphoblastoid cell lines (EBV-LCL) expressing the targeted mismatched HLA-DP allele (black bars) or against alloDC but not against third-party EBV-LCL expressing the targeted mismatched HLA-DP allele (gray bars) in responses 1-5 (n=22, 46, 429, 549, 157, respectively). Pair 5 was a 9/10 HLA match containing an additional HLA-DQ mismatch; the HLA-DP-restricted T-cell clones are shown in bar 5a and the HLA-DQ-restricted T-cell clones in bar 5b. (B) Interferongamma (IFNÎł) production by a representative HLA-DPB1*03:01restricted T-cell clone (from response 4) after overnight incubation with hematopoietic stimulator cells (DC, CD14+ cells, PHAblasts and EBV-LCL). The clone only shows reactivity against alloDC and alloEBV-LCL. (C) Reactivity of a representative HLA-DPB1*03:01-restricted T-cell clone (from response 4) against only alloDC. (D) The representative HLA-DPB1*03:01-restricted T-cell clone from Figure 2B tested against a panel of third-party DC. (E) The representative HLADPB1*03:01-restricted T-cell clone from Figure 2C was also tested against third-party DC. (F) The representative clone of Figure 2B shows reactivity against a panel of third-party EBV-LCL, but (G) the clone of Figure 2C does not show reactivity against thirdparty EBV-LCL. The number between brackets indicates the individual from whom the stimulator cells were derived.

201

A. Laghmouchi et al. Table 2. Cell-lineage recognition patterns of HLA-class II-restricted T-cell clones provoked by HLA-class II mismatched dendritic cells.

Response

Allo HLA-class II restriction

1. 2. 3. 4. 5. a

DPB1*04:01:01 DPB1*04:02 DPB1*02:01:02 DPB1*03:01:01 DPB1*04:01

5. b

DQB1*06:03

Total tested T-cell clones

Cell-lineage recognition patterns: DC & EBV-LCL

DC, EBV-LCL, HeLa and Fibro

DC, EBV-LCL, DC, EBV-LCL, and Fibro and HeLa

16 30 423 546

10 10 219 36

9 14 29

6 11 83 435

95 43

22 3

89 65

30 37

15 9

38 6

2 10

4 3

DC: CD14-derived dendritic cells, EBV-LCL: Epstein-Barr virus-transformed lymphoblastoid cell lines, HeLa: cervical cancer cell line, Fibro: skin-derived human fibroblasts. The number of T-cell clones with the respective cell-lineage-restriction pattern are indicated for each response.

sentative HLA-DQB1*06:03-restricted T-cell clones shown in Online Supplementary Figure S5). These results illustrate that the allo-HLA-DP and allo-HLA-DQ T-cell repertoire provoked by stimulation of donor T cells with HLA-class II-mismatched DC contains T-cell clones that show HLADP- or HLA-DQ-restricted recognition of DC, but not of EBV-LCL expressing the targeted mismatched HLA-DP or HLA-DQ allele.

The allo-HLA-DP-specific T-cell repertoire comprises T cells with different cell-lineage-restricted recognition patterns To further investigate the presence of T cells with a cell-lineage-specific recognition pattern within the alloHLA-DP- and allo-HLA-DQ-specific T-cell repertoires, the reactivity pattern of the majority of allo-HLA-DPand allo-HLA-DQ-restricted T-cell clones (n=1104, some T-cell clones were discarded because of contamination) recognizing alloDC, with or without recognition of third-party EBV-LCL, was analyzed using a panel of skinderived fibroblasts and HeLa cells expressing the targeted mismatched HLA-DP alleles (Online Supplementary Figure S6A). DC and EBV-LCL were used as controls. A significant proportion (12-63%, median 53%) of the alloHLA-DP-restricted T-cell clones showed restricted reactivity against DC, with (first recognition profile column in Table 2) or without (second recognition profile column in Table 2) EBV-LCL recognition, but not against fibroblasts or HeLa cells expressing the targeted mismatched HLA-DP allele, whereas the other HLA-DP-restricted clones showed a broad recognition pattern against all stimulator cells expressing the mismatched HLA-DP allele (20-80%, median 38%, third recognition profile column in Table 2), or against a selection of the tested stimulator cells (0-28%, median 7%, fourth and fifth recognition profile columns in Table 2) (the reactivity of representative clones of response 4 is shown in Online Supplementary Figure S6B-F). A selection of allo-HLA-DPrestricted T-cell clones was tested for their cytotoxic capacity and was found to be cytotoxic with variable efficiency (data not shown). Similar reactivity patterns were observed for the HLA-DQB1-restricted clones derived from the immune response mounted in response 5 (Table 2). These data illustrate that the allo-HLA-DPand allo-HLA-DQ-specific T-cell repertoire contains T cells with various cell-lineage-specific recognition pat202

terns, as well as T cells that recognize all cells expressing the targeted mismatched HLA-DP or HLA-DQ allele, irrespectively of their cell lineage.

The allo-HLA-DP specific T-cell repertoire contains T cells that are able to recognize primary hematopoietic malignant cells, without coinciding reactivity against non-hematopoietic cells To investigate whether the allo-HLA-DP-specific T-cell repertoire contains T cells that harbor the potential to recognize primary malignant cells without recognition of various non-hematopoietic cells, a selected number of T-cell clones (response 2, n=9; response 3, n=6; response 4, n=10; and response 5a, n=10) were tested against a panel of third-party primary malignant hematopoietic cell samples (acute myeloid leukemia, 13 samples; B-cell acute lymphoblastic leukemia, 9 samples; and chronic lymphocytic leukemia, 6 samples) expressing the targeted mismatched allele (HLA-DP expression shown in Online Supplementary Figure S7A,B), and against a panel of nonhematopoietic cell lines derived from different GvHD target organs (skin-derived fibroblasts, HeLa cells, colon carcinoma, and biliary epithelial cell lines) expressing the targeted mismatched HLA-DP allele. The HLA-DP-restricted T-cell clones that previously showed a broad recognition pattern recognized all target-HLA-DP-expressing malignant and non-hematopoietic cell types (clone response 4, as a representative example, is shown in Online Supplementary Figure S7C-I). The HLA-DP-restricted T-cell clones that showed recognition of DC and EBV-LCL recognized malignant hematopoietic cell subsets of different lineages (representative HLA-DPB1*03:01-restricted clone in Figure 3A), but showed no reactivity against skin-derived fibroblasts and different non-hematopoietic cells expressing the target HLA-DP molecule (representative HLA-DPB1*03:01restricted clone in Figure 3B). The HLA-DP-restricted Tcell clones that previously showed restricted recognition of DC, but not of EBV-LCL, showed only recognition of acute myeloid leukemia, but not of malignant B cells (e.g. chronic lymphocytic leukemia or B-cell acute lymphoblastic leukemia) (a representative clone from response 4 is shown in Figure 3C and from responses 2 and 5 in Online Supplementary Figure S8) and non-hematopoietic cells (e.g. fibroblasts and cell lines) (a representative clone from response 4 is shown in Figure 3D). haematologica | 2019; 104(1)

HLA-DP-restricted cell-lineage recognition patterns

C Figure 3. Differential recognition of primary malignant and non-hematopoietic cells by alloHLA-DP-specific T cells. (A) Interferon-gamma (IFNÎł) production by a representative HLA-DPB1*03:01-restricted Tcell clone (derived from response 4) when stimulated with a variety of primary malignant cells and (B) not when stimulated with non-hematopoietic tissue cell lines. This recognition pattern was seen for the tested T-cell clones of response 2 (n=3), response 3 (n=3), response 4 (n=1) and response 5a (n=5). (C) IFNÎł production by a representative HLADPB1*03:01-restricted T-cell clone (derived from response 4) when stimulated with primary acute myeloid leukemic blasts but not when stimulated with other primary malignant cells or (D) other non-hematopoietic cells. This recognition pattern was seen for the tested T-cell clones of response 2 (n=3), response 4 (n=6) and response 5a (n=3). The number between brackets indicates the individual from whom the stimulator cells were derived. AML: acute myeloid leukemia; B-ALL: B-cell acute lymphoblastic leukemia; CLL: chronic lymphocytic leukemia.

haematologica | 2019; 104(1)

203

A. Laghmouchi et al.

Since the level of recognition of the different primary acute myeloid leukemia samples by these clones was rather heterogeneous, flow cytometric analysis was performed to measure the surface expression of HLA-DP and adhesion molecule CD54 on the leukemic blasts (Online Supplementary Figures S7 and S9). For some samples (e.g. samples 37 and 40, Online Supplementary Figure S8D,F) the absence of recognition was correlated with the lack of proper HLA-DP surface expression. However, sample 35 (Online Supplementary Figure S8B) was not recognized despite high surface HLA-DP expression. Moreover, the maturation state (e.g. co-expression of maturation markers) of this specific acute myeloid leukemia sample was not found to be different from that of other samples that were properly recognized (data not shown). These data indicate that the level of recognition is likely to be determined by a combination of factors, including the level of HLA-DP expression, the expression of adhesion molecules (e.g. CD54), and the expression of the respective antigen. These data illustrate that the allo-HLA-DP-specific Tcell repertoire comprises T cells with the capacity to recognize primary malignant cells, without recognition of non-hematopoietic tissue cells.

Discussion In this study we demonstrate that the allo-HLA-DP-specific donor T-cell repertoire provoked by stimulation with HLA-DP-mismatched DC contains a variety of cell-lineage specificities. We found T cells that recognize all cells expressing the targeted mismatched HLA-DP allele, irrespective of the cell lineage, T cells with the capacity to recognize hematopoietic cells, including primary malignant cells without recognition of non-hematopoietic tissue cell lines, and also T cells that show restricted recognition of myeloid cells, including DC and primary acute myeloid leukemia samples, but not of cells of other hematopoietic and non-hematopoietic cell lineages. Similar cell-lineagespecific recognition patterns were found for allo-HLADQ-restricted T-cell clones isolated from a 9/10 HLAmatched pair, containing an additional HLA-DQ mismatch. Although the amplitude of the T-cell response is lower in the permissive HLA-DP responses compared to the non-permissive HLA-DP responses, a variety of celllineage-specific reactivity patterns was also found in the allo-HLA-DP-specific T-cell repertoire for the permissive HLA-DP responses. It has been shown in patients receiving alloSCT and/or donor lymphocyte infusion from HLA-DP-mismatched donors, that the mismatched allo-HLA-DP alleles were targeted by CD4 T cells, resulting in the induction of GvHD in different organs (e.g. skin GvHD, and gut GvHD) indicating an upregulation of HLA-class II on inflamed non-hematopoietic tissues.23 However, in other patients allo-HLA-DP-reactive CD4 donor T cells induced specific GvL reactivity without coinciding GvHD, suggesting a hematopoietic-restricted HLA-DP response without cross-reactivity against non-hematopoietic tissues and/or absence of expression of HLA-class II on nonhematopoietic tissues in these cases.22,24 These clinical observations can be explained by different factors, the first being that the tissue restricted reactivity is dictated by the specificities present in the allo-HLA-DP T-cell repertoire. The allo-HLA-DP-specific response is expected to be ini204

tiated in secondary lymphoid organs, most likely upon stimulation with patient-derived HLA-DP-expressing antigen-presenting cells of hematopoietic origin (DC).33,34 The allo-HLA-DP-restricted T cells activated by DC probably recognize antigens expressed by hematopoietic cells and presented in the mismatched HLA-DP molecule.14 This process could lead to an immune response skewed towards restricted recognition of hematopoietic cells, explaining the clinical observation in some patients of GvL reactivity without coinciding GvHD mediated by alloHLA-DP-reactive CD4 donor T cells.22,24 In this study we showed that the allo-HLA-DP-restricted T-cell repertoire provoked by in vitro stimulation of donor T cells with HLA-DP-mismatched DC contains a broad spectrum of Tcell specificities. The restricted recognition of hematopoietic cells (e.g. DC and EBV-LCL) could indicate that in vivo T cells with comparable recognition profiles could contribute to a GvL effect in patients with HLA-DP-expressing myeloid or B-cell malignancies.24,39 On the other hand, the allo-HLA-DP-specific immune response can also be initiated by DC residing in inflamed HLA-DP-expressing non-hematopoietic tissues. If the DC in inflamed tissues are cross-presenting antigens from the damaged surrounding environment, allo-HLA-DP-restricted T cells provoked by these DC are more likely to be directed against antigens also expressed by non-hematopoietic cells and presented in the mismatched HLA-DP molecule.35 Most likely, the magnitude of the allo-HLA-DP response and, thereby, the absolute number of allo-reactive T cells as well as the recognition profile of the induced T cells will determine the balance between GvL and GvHD induction. It has been shown in vivo that the magnitude of the alloHLA-DP response is affected by the specific HLA-DP allele(s) expressed in the donor and patient.27,28 In the case of permissive HLA-DP mismatches it has been demonstrated in vivo29-32 and in vitro,40 in line with the results of our current study, that the HLA-DP response is of a lower amplitude and therefore could result in a more tolerable response than in the non-permissive HLA-DP-mismatched combinations. However, in vitro HLA-DP-specific T-cell responses showed immunogenicity of HLA-DP alleles in both permissive and non-permissive mismatched pairs.39,41 If the HLA-DP alleles and the peptidomes presented in the HLA-DP alleles are similar between donor and patient, a large proportion of the allo-HLA-DP-specific T cells is likely to be deleted during negative selection of self-reactive T cells in the thymus of the donor.42 This may explain the lower magnitude of the allo-HLA-DP-specific immune responses in permissive HLA-DP-mismatched donor/patient pairs. Donor allo-HLA-DP-restricted CD4 T cells that target peptides expressed in non-self HLA-DP molecules in hematopoietic malignancies may specifically contribute to the GvL response after alloSCT. However, the occurrence of GvHD after HLA-DP-mismatched alloSCT and donor lymphocyte infusion remains a challenge, also in permissive HLA-DP-mismatched donor/patient pairs. The aim of our study was to elucidate the HLA-DP response in the donor T-cell repertoire when exposed to HLA-DP-mismatched antigen-presenting cells, similar to the clinical setting in HLA-DP-mismatched alloSCT. New strategies resulting in the enrichment of hematopoiesis-restricted allo-HLA-DP-specific T cells from the donor repertoire for adoptive T-cell therapy may increase the likelihood of inducing a GvL effect without provoking severe GvHD. haematologica | 2019; 104(1)

HLA-DP-restricted cell-lineage recognition patterns

For example, the current procedure could be optimized by including a depletion step to remove not only auto-reactive T cells but also T cells reactive against nonhematopoietic tissue. Although the use of HLA-DP-mismatched DC from patients will give a range of antigen specificities, including specificities against polymorphic antigens, the dependency on sufficient material from patients to generate CD14-derived DC restricts applicability. Therefore, the sophisticated approach of Herr and colleagues, who used autoDC transfected with allo-HLA-DPencoding RNA as stimulator cells, shows an alternative

References 1. Miller JS, Warren EH, van den Brink MR, et al. NCI First International Workshop on The Biology, Prevention, and Treatment of Relapse After Allogeneic Hematopoietic Stem Cell Transplantation: report from the Committee on the Biology Underlying Recurrence of Malignant Disease following Allogeneic HSCT: graft-versustumor/leukemia reaction. Biol Blood Marrow Transplant. 2010;16(5):565-586. 2. von dem Borne PA, Starrenburg CW, Halkes SJ, et al. Reduced-intensity conditioning allogeneic stem cell transplantation with donor T-cell depletion using alemtuzumab added to the graft ('Campath in the bag'). Curr Opin Oncol. 2009;21 (Suppl 1): S27-29. 3. Falkenburg JH, Jedema I. Allo-reactive T cells for the treatment of hematological malignancies. Mol Oncol. 2015;9(10):18941903. 4. Barge RM, Osanto S, Marijt WA, et al. Minimal GVHD following in-vitro T celldepleted allogeneic stem cell transplantation with reduced-intensity conditioning allowing subsequent infusions of donor lymphocytes in patients with hematological malignancies and solid tumors. Exp Hematol. 2003;31(10):865-872. 5. Barrett AJ. Mechanisms of the graft-versusleukemia reaction. Stem Cells. 1997;15(4): 248-258. 6. Falkenburg JH, Warren EH. Graft versus leukemia reactivity after allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2011;17(1 Suppl):S33-38. 7. Horowitz MM, Gale RP, Sondel PM, et al. Graft-versus-leukemia reactions after bone marrow transplantation. Blood. 1990;75(3): 555-562. 8. Kolb HJ. Graft-versus-leukemia effects of transplantation and donor lymphocytes. Blood. 2008;112(12):4371-4383. 9. Riddell SR, Berger C, Murata M, Randolph S, Warren EH. The graft versus leukemia response after allogeneic hematopoietic stem cell transplantation. Blood Rev. 2003;17(3):153-162. 10. Ferrara JL, Levine JE, Reddy P, Holler E. Graft-versus-host disease. Lancet. 2009;373 (9674):1550-1561. 11. Markey KA, MacDonald KP, Hill GR. The biology of graft-versus-host disease: experimental systems instructing clinical practice. Blood. 2014;124(3):354-362. 12. Nash RA, Storb R. Graft-versus-host effect after allogeneic hematopoietic stem cell transplantation: GVHD and GVL. Curr Opin Immunol. 1996;8(5):674-680.

haematologica | 2019; 104(1)

way of provoking allo-HLA-DP-restricted donor T cells with different tissue specificities.43 Another strategy to obtain a hematopoiesis-specific T-cell product is to isolate HLA-DP-restricted T-cell receptors from our T-cell clones that showed a more hematopoiesis-specific or myeloidlineage-specific recognition profile and perform T-cell receptor gene transfer. Acknowledgments The research in this manuscript was financially supported by the Dutch Cancer Society (project UL 2013-5989).

13. Barge RM, Starrenburg CW, Falkenburg JH, Fibbe WE, Marijt EW, Willemze R. Longterm follow-up of myeloablative allogeneic stem cell transplantation using Campath "in the bag" as T-cell depletion: the Leiden experience. Bone Marrow Transplant. 2006;37(12):1129-1134. 14. Falkenburg JHF, Jedema I. Graft versus tumor effects and why people relapse. Hematology Am Soc Hematol Educ Program. 2017;2017(1):693-698. 15. Eefting M, de Wreede LC, Halkes CJM, et al. Multi-state analysis illustrates treatment success after stem cell transplantation for acute myeloid leukemia followed by donor lymphocyte infusion. Haematologica. 2016;101(4):506-514. 16. Eefting M, Halkes CJ, de Wreede LC, et al. Myeloablative T cell-depleted alloSCT with early sequential prophylactic donor lymphocyte infusion is an efficient and safe postremission treatment for adult ALL. Bone Marrow Transplant. 2014;49(2):287-291. 17. Yun HD, Waller EK. Finding the sweet spot for donor lymphocyte infusions. Biol Blood Marrow Transplant. 2013;19(4):507-508. 18. Fernandez-Vina MA, Klein JP, Haagenson M, et al. Multiple mismatches at the low expression HLA loci DP, DQ, and DRB3/4/5 associate with adverse outcomes in hematopoietic stem cell transplantation. Blood. 2013;121(22):4603-4610. 19. Moreau P, Cesbron A. HLA-DP and allogeneic bone marrow transplantation. Bone Marrow Transplant. 1994;13(6):675-681. 20. Petersdorf EW, Smith AG, Mickelson EM, et al. The role of HLA-DPB1 disparity in the development of acute graft-versus-host disease following unrelated donor marrow transplantation. Blood. 1993;81(7):19231932. 21. van Bergen CA, van Luxemburg-Heijs SA, de Wreede LC, et al. Selective graft-versusleukemia depends on magnitude and diversity of the alloreactive T cell response. J Clin Invest. 2017;127(2):517-529. 22. Rutten CE, van Luxemburg-Heijs SA, Halkes CJ, et al. Patient HLA-DP-specific CD4+ T cells from HLA-DPB1-mismatched donor lymphocyte infusion can induce graft-versus-leukemia reactivity in the presence or absence of graft-versus-host disease. Biol Blood Marrow Transplant. 2013;19(1):40-48. 23. Stevanovic S, van Bergen CA, van Luxemburg-Heijs SA, et al. HLA class II upregulation during viral infection leads to HLA-DP-directed graft-versus-host disease after CD4+ donor lymphocyte infusion. Blood. 2013;122(11):1963-1973. 24. Rutten CE, van Luxemburg-Heijs SA, Griffioen M, et al. HLA-DP as specific target

25.

26.

27.

28.

29.

30.

31.

32.

33. 34.

35. 36.

for cellular immunotherapy in HLA class IIexpressing B-cell leukemia. Leukemia. 2008;22(7):1387-1394. Petersdorf EW, Malkki M, O'HUigin C, et al. High HLA-DP expression and graft-versushost disease. N Engl J Med. 2015;373 (7):599-609. Fleischhauer K, Beelen DW. HLA mismatching as a strategy to reduce relapse after alternative donor transplantation. Semin Hematol. 2016;53(2):57-64. Cesbron A, Moreau P, Cheneau ML, et al. Crucial role of the third and fourth hypervariable regions of HLA-DPB1 allelic sequences in primary mixed-lymphocyte reaction: application in allogeneic bone marrow transplantation. Transplant Proc. 1993;25(1 Pt 2):1232-1233. Nicholson I, Varney M, Kanaan C, et al. Alloresponses to HLA-DP detected in the primary MLR: correlation with a single amino acid difference. Hum Immunol. 1997;55(2):163-169. Zino E, Frumento G, Marktel S, et al. A Tcell epitope encoded by a subset of HLADPB1 alleles determines nonpermissive mismatches for hematologic stem cell transplantation. Blood. 2004;103(4):1417-1424. Fleischhauer K, Locatelli F, Zecca M, et al. Graft rejection after unrelated donor hematopoietic stem cell transplantation for thalassemia is associated with nonpermissive HLA-DPB1 disparity in host-versusgraft direction. Blood. 2006;107(7):29842992. Fleischhauer K, Shaw BE, Gooley T, et al. Effect of T-cell-epitope matching at HLADPB1 in recipients of unrelated-donor haemopoietic-cell transplantation: a retrospective study. Lancet Oncol. 2012;13(4): 366-374. Fleischhauer K, Shaw BE. HLA-DP in unrelated hematopoietic cell transplantation revisited: challenges and opportunities. Blood. 2017;130(9):1089-1096. Ni K, O'Neill HC. The role of dendritic cells in T cell activation. Immunol Cell Biol. 1997;75(3):223-230. Reddy P, Maeda Y, Liu C, Krijanovski OI, Korngold R, Ferrara JL. A crucial role for antigen-presenting cells and alloantigen expression in graft-versus-leukemia responses. Nat Med. 2005;11(11):1244-1249. Shlomchik WD. Graft-versus-host disease. Nat Rev Immunol. 2007;7(5):340-352. Jedema I, van de Meent M, Pots J, Kester MG, van der Beek MT, Falkenburg JH. Successful generation of primary virus-specific and anti-tumor T-cell responses from the naive donor T-cell repertoire is determined by the balance between antigen-spe-

205

A. Laghmouchi et al. cific precursor T cells and regulatory T cells. Haematologica. 2011;96(8):1204-1212. 37. Wehler TC, Nonn M, Brandt B, et al. Targeting the activation-induced antigen CD137 can selectively deplete alloreactive T cells from antileukemic and antitumor donor T-cell lines. Blood. 2007;109(1):365373. 38. Lam TS, van de Meent M, Falkenburg JF, Jedema I. Monocyte-derived dendritic cells can induce autoreactive CD4(+) T cells showing myeloid lineage directed reactivity in healthy individuals. Eur J Immunol. 2015;45(4):1030-1042.

206

39. Rutten CE, van Luxemburg-Heijs SA, van der Meijden ED, et al. HLA-DPB1 mismatching results in the generation of a full repertoire of HLA-DPB1-specific CD4+ T cell responses showing immunogenicity of all HLA-DPB1 alleles. Biol Blood Marrow Transplant. 2010;16(9):1282-1292. 40. Sizzano F, Zito L, Crivello P, et al. Significantly higher frequencies of alloreactive CD4+ T cells responding to nonpermissive than to permissive HLA-DPB1 T-cell epitope disparities. Blood. 2010;116(11): 1991-1992. 41. Rutten CE, van Luxemburg-Heijs SA, van

der Meijden ED, et al. Both permissive and nonpermissive HLA-DPB1 mismatches can induce polyclonal HLA-DPB1 specific immune responses in vivo and in vitro. Blood. 2010;115(1):151-153. 42. Starr TK, Jameson SC, Hogquist KA. Positive and negative selection of T cells. Annu Rev Immunol. 2003;21:139-176. 43. Herr W, Eichinger Y, Beshay J, et al. HLADPB1 mismatch alleles represent powerful leukemia rejection antigens in CD4 T-cell immunotherapy after allogeneic stem-cell transplantation. Leukemia. 2017;31(2):434445.

haematologica | 2019; 104(1)

ARTICLE

Blood transfusion

Cold storage of platelets in additive solution: the impact of residual plasma in apheresis platelet concentrates

Ferrata Storti Foundation

Irene Marini,1* Konstanze Aurich,2* Rabie Jouni,1 Stefanie Nowak-Harnau,1 Oliver Hartwich,2 Andreas Greinacher,2 Thomas Thiele2,Ŧ and Tamam Bakchoul1,2,Ŧ

Centre for Clinical Transfusion Medicine, Medical Faculty of Tübingen, University of Tübingen and 2Institute of Immunology and Transfusion Medicine, University of Greifswald, Germany 1

*Both authors shared first authorship. ŦBoth laboratories contributed equally to this work

Haematologica 2018 Volume 103(2):207-214

ABSTRACT

latelet transfusion has become essential therapy in modern medicine. Although the clinical advantage of platelet transfusion has been well established, adverse reactions upon transfusion, especially transmission of bacterial infection, still represent a major challenge. While bacterial contamination is favored by the storage of platelets at room temperature, cold storage may represent a solution for this important clinical issue. In this study, we aimed to clarify whether plasma has protective or detrimental effects on cold-stored platelets. We investigated the impact of different residual plasma contents in apheresis-derived platelet concentrates, stored at 4°C or room temperature, on platelet function and survival. We found that platelets stored at 4°C have higher expression of apoptosis marker compared to platelets stored at room temperature, leading to accelerated clearance from the circulation in a humanized animal model. While cold-induced apoptosis was independent of the residual plasma concentration, cold storage was associated with better adhesive properties and higher response to activators. Interestingly, delta (δ) granule-related functions, such as ADP-mediated aggregation and CD63 release, were better preserved at 4°C, especially in 100% plasma. An extended study to assess cold-stored platelet concentrates produced under standard care Good Manufacturing Practice conditions showed that platelet function, metabolism and integrity were better compared to those stored at room temperature. Taken together, our results show that residual plasma concentration does not have a cardinal impact on the cold storage lesions of apheresis-derived platelet concentrates and indicate that the current generation of additive solutions represent suitable substitutes for plasma to store platelets at 4°C. Introduction Transfusion of platelet concentrates (PCs) is essential to reduce blood loss after traumatic injury or to maintain a safe platelet (PLT) count during chemotherapy. PCs are currently stored at room temperature (20-24°C) with constant agitation to ensure adequate PLT recovery, survival, and sufficient therapeutic efficacy. However, storage at room temperature (RT) not only compromises functionality both in vivo and in vitro (‘PLT storage lesions’), but also increases the potential risk of microbial growth in case of contamination.1-6 For these reasons, the shelf life of PLTs is limited to 4-7 days, depending on country specific guidelines.2 However, despite limited storage time, the incidence of bacterial contamination of PCs remains high, ranging from 1 to 10 per 50,000 units, which is a major drawback of PLTs stored at RT for clinical use. Cold storage of PCs at 4°C could be an option to reduce the risk of bacterial growth.7,8 Recent studies reported that cold-stored PLTs are functionally and metabolically superior to those stored at RT.9-11 A potential limitation of cold storage is the poorer recovery and survival of PLTs after transfusion. However, this remains controversial. Some studies have shown a decrease in survival of cold-stored PLTs haematologica | 2019; 104(1)

Correspondence: tamam.bakchoul@med.uni-tuebingen.de

Received: April 16, 2018. Accepted: August 9, 2018. Pre-published: August 16, 2018. doi:10.3324/haematol.2018.195057 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/207 ©2019 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.

207

I. Marini et al. in comparison to RT-stored PLTs,12-14 while other investigators reported that PLTs stored at 4°C can survive in the circulation for several days.15,16 In this context, the residual plasma content of PLT storage media may be relevant. Early studies have investigated cold storage of PLTs in plasma and have reported poor recovery and survival.17 Based on these studies, the concept of cold storage had been abandoned in routine clinical practice. Recently, with the availability of PLT storage in additive solutions (PAS), the cold storage of PCs has seen something of a renaissance. Storage in PAS was suggested to maintain better PLT quality and provide protection from storage lesions with the possibility of prolonging PC shelf life.18-20 However, it is still unclear whether reduced plasma content improves cold storage of PCs. Furthermore, it is not known whether recovery and survival of PLTs are better after cold storage in PAS compared to storage at RT. In this study, we investigated the impact of different residual plasma concentrations in apheresis-derived platelet concentrates (APCs) stored at 4°C or at RT. We aimed to clarify whether plasma has protective or detrimental effects on cold-stored PLTs. Moreover, we assessed in vitro PLT quality and function in APCs to define the optimal balance between cold storage in plasma and additive solution. We then initiated a validation study of coldstored APCs produced under Good Manufacturing Practice (GMP) conditions with 35% residual plasma to verify the feasibility of PLT cold storage for clinical use.

Methods Preparation of apheresis platelet concentrates Apheresis-derived platelet concentrates were collected from healthy volunteers according to the German guidelines for hemotherapy. Ten individuals donated two units of APCs collected with FENWAL AMICUS (Amicus, Fresenius Kabi, Bad Homburg, Germany) and stored in plasma or in PAS (SSP+, Macopharma, Langen, Germany) at different final plasma concentrations [100% (Plasma-APC), 35% (PAS-35-APC) or 20% (PAS-20-APC)] at 4°C and RT. See the Online Supplementary Methods for further details. Finally, PAS-35-APCs were produced under GMP-conditions (12 healthy male donors) as described,21 and stored at RT and 4°C.

gen (Horm Collagen-Takeda, Linz, Austria). PLTs from APCs (1x108 PLTs/mL) were seeded on coverslips and incubated for 1 hour (h) at RT with TRAP (0.1 mM, Hart Biologicals, Hartlepool, UK). The adherent cells were fixed with 4% paraformaldehyde (PFA) for 20 min at RT. Images were captured from 5 different microscopic fields/coverslips (x100, Olympus IX73, Tokyo, Japan).

Platelet activation and granule release

Platelets (10x106/mL) were incubated with TRAP (40 mM) for 30 min at 37°C, and fixed with 4% PFA for 20 min at RT. The expression of CD62P (CLB-Thromb/6, Beckman Coulter) and CD63 (CLB-Gran/12, Beckman Coulter) was determined by flow cytometry (FC) as well as the conformational changes of glycoproteins (GPs) IIb/IIIa complex by PAC-1-antibody (BD Bioscience).

Platelet aggregation Platelet function was analyzed by FC and light transmission PLT aggregation assay (LTA) using a 4-channel-aggregometer (Labitec, Ahrensburg, Germany). See the Online Supplementary Methods for further details.

Hypotonic shock reaction Hypotonic shock reaction (HSR) was determined by LTA. Percentage of HSR was calculated as described in the Online Supplementary Appendix.

Statistical analysis Statistical analyses were performed using GraphPad Prism 7 (La Jolla, USA). A t-test was used to analyze normally distributed results. Non-parametric tests were used when data failed to follow a normal distribution as assessed by the D’Agostino and Pearson omnibus normality test. Group comparison was performed using the Wilcoxon rank-sum test and the Fisher exact test with categorical variables. In the case of a small number of experiments (<10), the group comparison was performed using the t-test. P<0.05 was considered statistically significant.

Ethics All studies involving human subjects were approved by the ethics committees of the University Hospital of Tübingen and the Universitätsmedizin Greifswald. Animal studies were approved by the state animal ethics committees of Baden-Württemberg and Mecklenburg-Vorpommern.

In vivo studies To assess the survival of PLTs derived from APCs, we used the NOD/SCID mouse model as described previously.22,23 See the Online Supplementary Methods for further details.

Measurement of glycan changes Glycan pattern was analyzed by flow cytometer (FC) (Navious, Beckman Coulter) using ricinus communis agglutinin (RCA, 0.5 mg/mL, Vector, Burlingame, CA, USA) which binds beta (β)-galactose, as described in the Online Supplementary Methods.

Apoptosis Platelets from APCs were washed, resuspended with 1 mM CaCl2, and stained with Annexin V-FITC (Beckman Coulter) for 60 minutes (min) at RT. Freshly isolated PLTs were incubated with 10 mM ionomycin (Abcam, Cambridge, UK) and used as positive control. See the Online Supplementary Methods for further details.

Platelet adhesion Coverslips (Corning, New York, USA) were coated with 100 mg/mL of fibrinogen (Sigma Aldrich, Munich, Germany) or colla208

Results Platelet survival after cold storage at different plasma concentrations At storage day 7, PLTs were injected into the NOD/SCID mice and the survival of PLTs stored at different conditions was compared in pairs. To enable statistical analysis between PLTs from different donors, the relative survival of Plasma-APC stored at RT was considered as 1.0. Fewer PLTs were found in the circulation when APCs were stored at 4°C compared to RT, regardless of whether PLTs where stored in 35% residual plasma [PAS-35-APC stored at 4°C vs. RT mean±Standard Error of mean (SEM): 1.05±0.02 vs. 0.63±0.16, respectively, P=0.04; Figure 1A] or in 20% residual plasma (PAS-20-APC 0.58±0.05 vs. 0.34±0.05, respectively, P=0.01; Figure 1B). Regarding the plasma content, similar survival curves were observed at each temperature when PLTs were stored in 35% plasma compared to 100% plasma (Online haematologica | 2019; 104(1)

The impact of plasma on cold storage of APC

Supplementary Figure S1A). In contrast, PLTs stored in PAS with only 20% residual plasma were cleared faster from the mouse circulation compared to those stored in 100% plasma, and this was even more pronounced in coldstored PLTs (Online Supplementary Figure S1B).

Cold-induced apoptosis rather than desialylation is associated with accelerated clearance of cold-stored platelets, irrespective of residual plasma Cold storage of PLTs in plasma has been reported to induce desialylation of PLTs. Therefore, we tested the

Figure 1. Relative platelet survival after cold storage at different residual plasma concentrations (PC). Apheresis platelet concentrates (APCs) were collected, split and stored at room temperature (RT) or at 4°C, either in 100% plasma (Plasma-APC, black bars) or in platelet additive solution (PAS, gray bars) at two different residual plasma concentrations: (A) 35% (PAS-35-APC) and (B) 20% (PAS-20-APC). Platelets (PLTs) were obtained from APCs at storage day 7 and administered into the mouse circulation via the lateral tail vein. Survival of human PLTs in mouse was analyzed by collecting murine blood after two and five hours. For statistical comparison between different groups, the percentage of circulating human PLTs was normalized to Plasma-APC in each corresponding experiment (Plasma-APC was considered 1.0). Data are shown as mean±Standard Error of Mean. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001. ns: not significant (n=5). Survival curves are available in Online Supplementary Figure S1.

Figure 2. Desialylation and apoptosis after cold storage. Platelet (PLT) desialylation (A and B) and apoptosis (C and D) were analyzed after seven days of storage, at room temperature (RT) or 4°C, in apheresis platelet concentrates (APCs) containing (A and C) 35% plasma (PAS-35-APC), (B and D) 20% plasma (PAS-20-APC) or 100% plasma (Plasma-APC). Desialylation was determined by measurement of FITC-labeled RCA (0.5 mg/mL) that binds to beta (β)-galactose using flow cytometetry. Fresh PLTs incubated with or without neuraminidase (Neu) were used as positive and negative control, respectively. The percentage of apoptotic cells was measured using FITC-labeled Annexin V. As positive (Pos ctr) and negative control (Neg ctr), freshly isolated PLTs incubated in the presence or in the absence of Ionomycin were used. Data are shown as mean±Standard Error of Mean of fluorescence intensity (MFI) and percentage of positive events, respectively. *P<0.05. ns: not significant (n=4).

haematologica | 2019; 104(1)

209

I. Marini et al.

exposure of desialylated GPs in PLTs stored at different plasma concentrations. In the lectin binding assay (LBA), neither storage temperature nor plasma content had a major effect on β-galactose expression (Figure 2A and B). Only PLTs from PAS-35-APCs stored for seven days at 4°C showed a slightly higher exposure of β-galactose on their surface compared to those stored at RT. In contrast, higher exposure of phosphatidylserine was

always detected on the surface of cold-stored PLTs (percentage of Annexin V positive events mean±SEM: PAS-35APC 9±1 vs. 7±1, respectively, P=0.033; Figure 2C; PAS20-APC 22±4 vs. 17±5, respectively, P=0.317; Figure 2D). The cold-induced increase in the exposure of phosphatidylserine was not correlated to the residual plasma concentration indicating that cold storage of APCs affects PLT survival by inducing cell apoptosis in a plasma-inde-

Figure 3. Platelet adhesion to collagen and fibrinogen after storage at room temperature (RT) or 4°C. (A) Representative images of adherent platelets (PLTs) from different apheresis platelet concentrates (APCs) after seven days of storage (scale bar 10 mm). PLTs from APCs containing 100% plasma (Plasma-APC), 35% plasma (PAS-35-APC) and 20% plasma (PAS-20-APC) were allowed to adhere to collagen (B and C) or fibrinogen (D and E) surfaces in the presence of TRAP. The number and types of adherent PLTs were determined from 5 different microscopic fields per coverslip (see Online Supplementary Figure S2 and Online Supplementary Table S1A and B for further details). Freshly isolated PLTs were used as positive control (100%). Data are shown as mean±Standard Error of Mean. *P<0.05. ns: not significant (n=3).

210

haematologica | 2019; 104(1)

The impact of plasma on cold storage of APC

pendent manner. While similar Annexin V binding was observed on PLTs from PAS-35-APCs and Plasma-APCs stored at both temperatures, cells stored in 20% residual plasma (PAS-20-APC) showed higher exposure of the apoptosis marker phosphatidylserine, particularly when PLTs were cold-stored (Figure 2D).

Effect of residual plasma on adhesion of cold-stored platelets The adhesion of TRAP-activated PLTs was analyzed after seven days of storage. More PLTs from APCs stored at 4°C adhered to collagen and fibrinogen compared to PLTs stored at RT (Figure 3A). This effect was statistically significant for PLTs stored in 100% and 35% plasma (Figure 3B and D). Moreover, the individual percentage of each spreading pattern (type 1, 2, 3 and 4) (Online Supplementary Figure S2) was always higher after cold storage, regardless of plasma volume, but without reaching statistical significance (Online Supplementary Table S1A and B). Cells stored in the presence of 35% plasma showed similar adhesive response to both proteins compared to 100% plasma (Figure 3B and D). In contrast, significantly fewer PLTs adhered to both proteins from cold-stored PAS-20-APCs compared to plasma-APCs (percentage of adherent PLTs to collagen, mean±SEM: 13±4 vs. 41±4, P=0.033; Figure 3C) and to fibrinogen (mean±SEM: 16±7 vs. 16±2, respectively, P=0.028; Figure 3E). This suggests that an excessive reduction in plasma volume to lower

than 35% impairs PLTs adhesive functions, especially when PLTs were stored at 4°C.

Plasma diminishes loss of δ granule secretion during cold storage

Secretion of granule content in response to PLT activation is required for an efficient hemostatic function. All stored PLTs had a reduced release of granule markers compared to freshly isolated ones. However, storage at 4°C improved CD63 release, especially when PLTs were stored in plasma: fold increase of CD63 after activation with TRAP mean±SEM: plasma-APC versus PAS-35-APC, 3.88±0.65 versus 2.71±0.27, respectively, P=0.145 (Figure 4A); and plasma-APC versus PAS-20-APC, 4.75±1.18 versus 2.75±0.75, respectively, P=0.045 (Figure 4B). Storage solution and temperature had only a minor impact on the storage lesions causing diminished CD62P expression (Figure 4C and D).

Validation of cold PLT functions in stored APCs As a step towards optimization of cold storage in routine clinical practice, we next evaluated the effect of cold storage using PAS-35-APCs produced under GMP conditions to demonstrate the quality of cold-stored PLTs prepared for clinical use throughout extended storage for ten days. Platelet aggregation in response to collagen was significantly higher when PLTs were stored at 4°C (Figure 5A). A

Figure 4. The impact of storage temperature and residual plasma concentration on granules release upon activation. After seven days of storage, CD63 (A and B) and CD62P (C and D) expressions were determined for apheresis platelet concentrates (APCs) containing 100% plasma (Plasma-APC), 35% plasma (PAS-35-APC) and 20% plasma (PAS-20-APC) after activation with TRAP. Freshly isolated platelets (PLTs) were used as control for both markers. Data are shown as mean± Standard Error of Mean of fold increase of mean florescence intensity compared to buffer as baseline. *P<0.05. ns: not significant (n=4). RT: room temperature.

haematologica | 2019; 104(1)

211

I. Marini et al.

significantly higher response to collagen was found after four days of storage at 4°C in comparison to RT (percentage of maximal aggregation after collagen, mean±SEM 67±6 vs. 22±5, respectively, P<0.0001), consistent with their better adhesion to collagen (Figure 3). In addition, the δ-granule-dependent response to ADP was reduced less during storage in PAS at 4°C compared to RT (Figure 5B). Ristocetin-mediated PLT agglutination was slightly lower at 4°C compared to RT (Figure 5C). Along the same lines as for PLT aggregation, after activation with TRAP a slightly higher expression of CD62P and CD63 was observed on cold-stored PLTs and the gradual reduction in PLT response was slower (Figure 6A and B). In contrast, no significant impact on the ability of conformational change in the integrin alpha (α)IIb/β (β)III was found during storage, regardless of temperature (Figure 6C). Furthermore, similar glucose consumption, and lactate release was found in both storage conditions (Online Supplementary Figure S3A and B, respectively). Finally, although hypotonic shock reaction decreased in a time-dependent manner, it was always higher when PLTs were stored at RT compared to 4°C (Figure 6D).

Discussion In the present study, we investigated the impact of residual plasma concentration in APCs on cold storage lesions, including in vivo survival and in vitro hemostatic functions. Despite the reduced in vivo survival of coldstored PLTs, we showed that cold storage not only preserves PLT response to aggregation agonists, but also maintains their adhesion to thrombogenic surfaces better than RT storage. Substituting 65% of plasma with PAS did

not have any important impact on cold-induced storage lesions. However, poor survival and functional results were observed when plasma concentration was further reduced to 20%. Attempts to store APCs at 4°C were impeded by a shorter survival of cold-stored PLTs. In our study, PLTs from APCs stored at 4°C always showed inferior survival curves compared to those stored at RT. This confirmed previous reports on accelerated clearance of cold-stored PLTs.13,14 Data from an animal model suggested that desialylation of GPIb is responsible for the accelerated elimination of coldHowever, we found similar stored PLTs.13,24 β-galactose exposure on RT- and cold-stored PLTs. This fits the results of clinical studies which showed that reconstitution of sialylation in human PLTs does not prevent the accelerated clearance of cold-stored APCs.25 Interestingly, we found that cold storage triggers phosphatidylserine exposure indicating higher PLT apoptosis. Although the differences were not always significant, there is an obvious trend. This may indicate that impaired in vivo survival of cold-stored PLTs is actually caused by apoptosis-related mechanisms. In fact, cold storage had previously been found to induce GPIb α-clustering which makes the receptors an initiation site for PLT apoptosis and macrophage recognition.26 The inhibition of prostacyclin has been suggested to reduce cold-induced changes in GPIb,27 which could be emphasized by our results. Interestingly, coldinduced apoptosis in our study was not accompanied by higher spontaneous expression of P-selectin or CD63, indicating that apoptosis and activation during PLT storage are mediated by two independent mechanisms. Therefore, selective targeting of cold-induced apoptosis may provide a new approach to reduce PLT storage lesions. In fact, it was recently reported that cold storage of PLTs enhances

Figure 5. Functional analyses of cold-stored apheresis platelet concentrates. Apheresis platelet concentrates (APCs) were produced under Good Manufacturing Practice. conditions and stored in platelet additive solution (PAS) at a residual plasma concentration of 35% (PAS-35-APC) at room temperature (RT) or 4°C. The maximal aggregation ability of platelets (PLTs) was determined after activation using three inductors: (A) collagen (8 mg/mL), (B) ADP (80 mM) and (C) ristocetin (1.5 mM) (agglutination). Data are shown as mean±Standard Error of Mean. *P<0.05; ****P<0.0001. ns: not significant (n=4).

212

haematologica | 2019; 104(1)

The impact of plasma on cold storage of APC

Figure 6. Flow cytometric analysis of apheresis platelet concentrates produced under Good Manufacturing Practice (GMP) conditions. Apheresis platelet concentrates (APCs) were produced under GMP conditions. Platelet (PLT) activation was measured at the indicated storage time for room temperature (RT) and 4°C stored PLTs. The expression of (A) CD62P, (B) CD63 and (C) PAC1 was analyzed after activation with TRAP. (D) The ability of apheresis platelet concentrates (APCs) to react on a hypotonic environment was determined as hypotonic shock reaction by light transmission aggregometry. Data are shown as mean±Standard Error of Mean. ***P<0.001; ****P<0.0001. ns: not significant (n=4).

the von Willebrand factor binding to GPIb-α, inducing increased intracellular calcium concentration and phosphatidylserine exposure leading to a rapid clearance of PLTs. The authors demonstrated that the inhibition of this binding and the consequent downstream cascade, using a specific peptide that recognizes GPIb-α, significantly increased recovery and life-span of cold-stored PLTs.28 Another study showed that a specific caspase-3 inhibitor significantly improves PLT functionality and viability during seven days of storage.29 This, together with our results, suggests that inhibition of apoptosis seems to be a promising approach to reduce cold-stored lesions. Another important finding of our study is that in vivo survival of cold-stored PLTs was independent of varying plasma contents. However, remarkably poor in vitro results of PLT function were observed in APCs stored at too low a plasma concentration (PAS-20-APC). This further supports in vivo studies of Slichter et al. who used radiolabeled autologous PLTs and showed that RT storage in 80% Plasmalyte (a PAS which is FDA approved) is associated with lower recovery and survival compared to PLTs stored in 100% plasma or 35% residual plasma.30 Finally, we investigated whether our research findings could be transferred to routine production and explored the possibility of extending PC shelf-life. To exclude the influence of different donors, we designed a follow-up study where each donor donated double APCs in PAS conhaematologica | 2019; 104(1)

taining 35% plasma. Each APC was stored either at RT or at 4°C for up to ten days. We found that PLTs from APCs collected and stored under routine blood bank conditions at 4°C maintained functionality better in terms of activation (granule release in response to TRAP) and aggregation compared to those stored at RT. These results correspond to those from the split design study. Taken together, a residual plasma content of approximately 35% is feasible for cold storage of PLTs in APCs for clinical use. Our results indicate the potential of prolonging shelf-life of GMP-produced APCs stored at 4°C for clinical use. In summary, our study provides additional information on the in vitro hemostatic function and in vivo survival of cold-stored PLTs, and suggests that PLTs stored in PAS at 4°C could become an alternative to the current standard of care. Funding The study was supported by a grant from the German Red Cross, Blutspendedienst Baden-Württemberg-Hessen. The authors would like to thank Ulrike Strobel, Lars Jansen, Robert Koch, Flavianna Rigoni, Kati Sevke-Masur and Inga Miksa for their excellent technical support. Acknowledgments We thank Stephen Bosher for his contribution to the manuscript as a native English speaker. 213

I. Marini et al.

References 1 Rosenfeld BA, Herfel B, Faraday N, Fuller A, Braine H. Effects of storage time on quantitative and qualitative platelet function after transfusion. Anesthesiology. 1995;83(6):1167-1172. 2 Brecher ME, Blajchman MA, Yomtovian R, Ness P, AuBuchon JP. Addressing the risk of bacterial contamination of platelets within the United States: a history to help illuminate the future. Transfusion. 2013; 53(1):221-231. 3 Dumont LJ, Gulliksson H, van der Meer PF, et al. Interruption of agitation of platelet concentrates: a multicenter in vitro study by the BEST Collaborative on the effects of shipping platelets. Transfusion. 2007;47(9): 1666-1673. 4 Seghatchian J, Krailadsiri P. The platelet storage lesion. Transfus Med Rev. 1997;11 (2):130-144. 5 Capocelli KE, Dumont LJ. Novel platelet storage conditions: additive solutions, gas, and cold. Curr Opin Hematol. 2014;21(6): 491-496. 6 Devine DV, Serrano K. The platelet storage lesion. Clin Lab Med. 2010;30(2):475-487. 7 Currie LM, Harper JR, Allan H, Connor J. Inhibition of cytokine accumulation and bacterial growth during storage of platelet concentrates at 4 degrees C with retention of in vitro functional activity. Transfusion. 1997;37(1):18-24. 8 Sandgren P, Hansson M, Gulliksson H, Shanwell A. Storage of buffy-coat-derived platelets in additive solutions at 4 degrees C and 22 degrees C: flow cytometry analysis of platelet glycoprotein expression. Vox Sang. 2007;93(1):27-36. 9 Reddoch KM, Pidcoke HF, Montgomery RK, et al. Hemostatic function of apheresis platelets stored at 4 degrees C and 22 degrees C. Shock. 2014;41 Suppl 1:54-61. 10 Pidcoke HF, Spinella PC, Ramasubramanian AK, et al. Refrigerated platelets for the treatment of acute bleeding: a review of the literature and reexamination of current standards. Shock. 2014;41 Suppl 1:51-53.

214

11 Montgomery RK, Reddoch KM, Evani SJ, Cap AP, Ramasubramanian AK. Enhanced shear-induced platelet aggregation due to low-temperature storage. Transfusion. 2013;53(7):1520-1530. 12 Lee DH, Blajchman MA. Novel treatment modalities: new platelet preparations and substitutes. Br J Haematol. 2001; 114(3):496-505. 13 Hoffmeister KM, Felbinger TW, Falet H, et al. The clearance mechanism of chilled blood platelets. Cell. 2003;112(1):87-97. 14 Murphy S, Gardner FH. Effect of storage temperature on maintenance of platelet viability--deleterious effect of refrigerated storage. N Engl J Med. 1969;280(20):1094-1098. 15 Pidcoke HF, McFaul SJ, Ramasubramanian AK, et al. Primary hemostatic capacity of whole blood: a comprehensive analysis of pathogen reduction and refrigeration effects over time. Transfusion. 2013;53 Suppl 1:137S-149S. 16 Jobes D, Wolfe Y, O'Neill D, et al. Toward a definition of "fresh" whole blood: an in vitro characterization of coagulation properties in refrigerated whole blood for transfusion. Transfusion. 2011;51(1):43-51. 17 Valeri CR. Circulation and hemostatic effectiveness of platelets stored at 4 C or 22 C: studies in aspirin-treated normal volunteers. Transfusion. 1976;16(1):20-23. 18 Sweeney J, Kouttab N, Holme S, Kurtis J, Cheves T, Nelson E. Storage of platelet-rich plasma-derived platelet concentrate pools in plasma and additive solution. Transfusion. 2006;46(5):835-840. 19 Slichter SJ, Corson J, Jones MK, et al. Exploratory studies of extended storage of apheresis platelets in a platelet additive solution (PAS). Blood. 2014;123(2):271-280. 20 Hornsey VS, McColl K, Drummond O, et al. Extended storage of platelets in SSP platelet additive solution. Vox Sang. 2006; 91(1):41-46. 21 Thiele T, Pohler P, Kohlmann T, et al. Tolerance of platelet concentrates treated with UVC-light only for pathogen reduction--a phase I clinical trial. Vox Sang. 2015; 109(1):44-51. 22 Bakchoul T, Fuhrmann J, Chong BH,

Bougie D, Aster R, Subcommittee on Platelet I. Recommendations for the use of the non-obese diabetic/severe combined immunodeficiency mouse model in autoimmune and drug-induced thrombocytopenia: communication from the SSC of the ISTH. J Thromb Haemost. 2015; 13(5):872-875. Fuhrmann J, Jouni R, Alex J, et al. Assessment of human platelet survival in the NOD/SCID mouse model: technical considerations. Transfusion. 2016; 56(6):1370-1376. Jansen AJ, Josefsson EC, Rumjantseva V, et al. Desialylation accelerates platelet clearance after refrigeration and initiates GPIbalpha metalloproteinase-mediated cleavage in mice. Blood. 2012;119(5):12631273. Wandall HH, Hoffmeister KM, Sorensen AL, et al. Galactosylation does not prevent the rapid clearance of long-term, 4 degrees C-stored platelets. Blood. 2008;111(6): 3249-3256. van der Wal DE, Du VX, Lo KS, Rasmussen JT, Verhoef S, Akkerman JW Platelet apoptosis by cold-induced glycoprotein Ibalpha clustering. J Thromb Haemost. 2010; 8(11):2554-2562. van der Wal DE, Verhoef S, Schutgens RE, Peters M, Wu Y, Akkerman JW Role of glycoprotein Ibalpha mobility in platelet function. Thromb Haemost. 2010;103(5):10331043. Chen W, Druzak SA, Wang Y, et al. Refrigeration-Induced Binding of von Willebrand Factor Facilitates Fast Clearance of Refrigerated Platelets. Arterioscler Thromb Vasc Biol. 2017;37(12):2271-2279. Shiri R, Yari F, Ahmadinejad M, Vaeli S, Tabatabaei MR The caspase-3 inhibitor (peptide Z-DEVD-FMK) affects the survival and function of platelets in platelet concentrate during storage. Blood Res. 2014; 49(1):49-53. Slichter SJ, Bolgiano D, Corson J, et al. Extended storage of buffy coat platelet concentrates in plasma or a platelet additive solution. Transfusion. 2014;54(9):22832291.

haematologica | 2019; 104(1)