Haematologica, Volume 103, Issue 10 by Haematologica

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

Ancient Greek

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

Scientific Latin

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

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2016 JCR impact factor = 7.702

Haematologica, as the journal of the European Hematology Association (EHA), aims not only to serve the scientific community, but also to promote European cultural identify.

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

Managing Director Antonio Majocchi (Pavia)

Associate Editors Omar I. Abdel-Wahab (New York), Hélène Cavé (Paris), Simon Mendez-Ferrer (Cambridge), Pavan Reddy (Ann Arbor), Andreas Rosenwald (Wuerzburg), Monika Engelhardt (Freiburg), Davide Rossi (Bellinzona), Jacob Rowe (Haifa, Jerusalem), Wyndham Wilson (Bethesda), Paul Kyrle (Vienna), Swee Lay Thein (Bethesda), Pieter Sonneveld (Rotterdam)

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

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

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

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

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

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

Institutional Euro 600

Personal Euro 150

Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org). Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature. Direttore responsabile: Prof. Edoardo Ascari; Autorizzazione del Tribunale di Pavia n. 63 del 5 marzo 1955. Printing: Press Up, zona Via Cassia Km 36, 300 Zona Ind.le Settevene - 01036 Nepi (VT)

haematologica calendar of events

Journal of the European Hematology Association Published by the Ferrata Storti Foundation ESH 4th International Conference on Multiple Myeloma European School of Haematology Chairs: KC Anderson, MV Mateos, P Moreai October 5-7, 2018 Mandelieu, France

13th Congress of the Albanian Association of Hematology Albanian Association of Hematology Chairs: A Ivanaj, M Dimopoulos, G Gaidano, X Pivot November 5-6, 2018 Tirana, Albania

EHA-SWG Scientific Meeting on Aging and Hematology Chair: D Bron October 12-14, 2018 Warsaw, Poland

EBMT Severe Aplastic Anaemia and Autoimmune Diseases Working Parties - Joint Educational Course EBMT Chairs: C Dufour, J Snowden, RP de la Tour, A Tobias November 15-17, 2018 Firenze, Italy

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

7th National Symposium "Young Hematologist" "Hemostasis and Thrombosis" Bulgarian Medical Society of Hematology Chairs: V Kaleva, M Guenova November 16-17, 2018 Saints Constantine and Helena, Bulgaria

LIX CONGRESO NACIONAL SEHH-XXXIII CONGRESO NACIONAL SETH Spanish Society Of Hematology And Hemotherapy (SEHH) Chairs: JS Gil, AIH Mazo, JAP Fernรกndez, MEM Castellano October 26-28, 2018 Malaga, Spain

Calendar of Events updated on September 6, 2018

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

Table of Contents Volume 103, Issue 10: October 2018 Cover Figure

Bone marrow smear showing atypical cells with cytoplasmic tails in a patient with blastic plasmacytoid dendritic cell neoplasm. Courtesy of Prof. Rosangela Invernizzi.

Editorials 1579

Improving consolidation therapy in acute myeloid leukemia - a tough nut to crack Richard F. Schlenk et al.

1581

Therapy-related acute lymphoblastic leukemia Josep-Maria Ribera

1583

eGVHD App: a new tool to improve graft-versus-host disease assessment Marie Therese Rubio

Review Articles 1586

Multiple faces of succinate beyond metabolism in blood Franco Grimolizzi and Lorena Arranz

1593

Normal and pathological erythropoiesis in adults: from gene regulation to targeted treatment concepts Peter Valent et al.

Articles Hematopoiesis

1604

Repopulating hematopoietic stem cells from steady-state blood before and after ex vivo culture are enriched in the CD34+CD133+CXCR4low fraction VĂŠronique Lapostolle et al.

Iron Metabolism & its Disorders

1616

Circulating iron levels influence the regulation of hepcidin following stimulated erythropoiesis Cornel S.G. Mirciov et al.

Iron Metabolism & its Disorders

1627

Iron overload impairs normal hematopoietic stem and progenitor cells through reactive oxygen species and shortens survival in myelodysplastic syndrome mice Xin Jin et al.

Phagocyte Biology and its Disorders

1635

Bone marrow histomorphological criteria can accurately diagnose hemophagocytic lymphohistiocytosis Eric Gars et al.

Acute Myeloid Leukemia

1642

Combinatorial targeting of XPO1 and FLT3 exerts synergistic anti-leukemia effects through induction of differentiation and apoptosis in FLT3-mutated acute myeloid leukemias: from concept to clinical trial Weiguo Zhang et al.

1654

Addition of the mammalian target of rapamycin inhibitor, everolimus, to consolidation therapy in acute myeloid leukemia: experience from the UK NCRI AML17 trial Alan K Burnett et al.

Acute Lymphoblastic Leukemia

1662

Haematologica 2018; vol. 103 no. 10 - October 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation Non-Hodgkin Lymphoma

1669

CDCA7 is a critical mediator of lymphomagenesis that selectively regulates anchorage-independent growth Raúl Jiménez-P. et al.

1679

Pre-diagnosis plasma immune markers and risk of non-Hodgkin lymphoma in two prospective cohort studies Mara M. Epstein et al.

Plasma Cell Disorders

1688

Immunomodulatory drugs downregulate IKZF1 leading to expansion of hematopoietic progenitors with concomitant block of megakaryocytic maturation Ailing Liu et al.

Stem Cell Transplantation

1698

The eGVHD App has the potential to improve the accuracy of graft-versus-host disease assessment: a multicenter randomized controlled trial Helene M. Schoemans et al.

1708

Cell Therapy & Immunotherapy

1720

CD16+NK-92 and anti-CD123 monoclonal antibody prolongs survival in primary human acute myeloid leukemia xenografted mice Brent A. Williams et al.

Coagulation & ita Disorders

1730

Cytoprotective and pro-angiogenic functions of thrombomodulin are preserved in the C loop of the fifth epidermal growth factor-like domain Xiangmin Wang et al.

Blood Transfusion

1741

A subset of anti-HLA antibodies induces FcγRIIa-dependent platelet activation Maaike Rijkers et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/103/10.toc

e436

A novel non-invasive method to measure splenic filtration function in humans Sara El Hoss et al. http://www.haematologica.org/content/103/10/e436

e440

Stability of fetal hemoglobin levels in patients receiving metformin therapy Mohamed-Rachid Boulassel et al. http://www.haematologica.org/content/103/10/e440

e443

Evaluation of serum markers for improved detection of autologous blood transfusions Holly D. Cox et al. http://www.haematologica.org/content/103/10/e443

e446

Genomic characterization of spleens in patients with myelofibrosis Eran Zimran et al. http://www.haematologica.org/content/103/10/e446

e450

Quantitative competitive allele-specific TaqMan duplex PCR (qCAST-Duplex PCR) assay: a refined method for highly sensitive and specific detection of JAK2V617F mutant allele burdens Chia-Chen Hsu et al. http://www.haematologica.org/content/103/10/e450

Haematologica 2018; vol. 103 no. 10 - October 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation e455

NPM1 mutation is not associated with prolonged complete remission in acute myeloid leukemia patients treated with hypomethylating agents Pedro Henrique Prata et al. http://www.haematologica.org/content/103/10/e455

e458

Autologous T-cell activation fosters ABT-199 resistance in chronic lymphocytic leukemia: rationale for a combined therapy with SYK inhibitors and anti-CD20 monoclonal antibodies Esteban Enrique Elías et al. http://www.haematologica.org/content/103/10/e458

e462

Real-world treatment and outcomes among older adults with chronic lymphocytic leukemia before the novel agents era Anthony Mato et al. http://www.haematologica.org/content/103/10/e462

e466

Impact of ibrutinib dose intensity on patient outcomes in previously treated Waldenström macroglobulinemia Jorge J. Castillo et al. http://www.haematologica.org/content/103/10/e466

e469

Phase I study of the heparanase inhibitor roneparstat: an innovative approach for multiple myeloma therapy Monica Galli et al. http://www.haematologica.org/content/103/10/e469

e473

Safety and efficacy of vorinostat, bortezomib, doxorubicin and dexamethasone in a phase I/II study for relapsed or refractory multiple myeloma (VERUMM study: vorinostat in elderly, relapsed and unfit multiple myeloma) Johannes M. Waldschmidt et al. http://www.haematologica.org/content/103/10/e473

e480

Interleukin-22 levels are increased in gastrointestinal graft-versus-host disease in children Dana T. Lounder, et al. http://www.haematologica.org/content/103/10/e480

Case Reports Case Reports are available online only at www.haematologica.org/content/103/10.toc

e483

Contribution of alternative complement pathway to delayed hemolytic transfusion reaction in sickle cell disease Satheesh Chonat et al. http://www.haematologica.org/content/103/10/e483

e486

Pitfalls in the molecular follow up of NPM1 mutant acute myeloid leukemia. Ulrike Bacher et al. http://www.haematologica.org/content/103/10/e486

e489

Daratumumab for relapsed/refractory Philadelphia-positive acute lymphoblastic leukemia Chezi Ganzel et al. http://www.haematologica.org/content/103/10/e489

Comments Comments are available online only at www.haematologica.org/content/103/10.toc

e491

Infection prevention in patients with hereditary hemorrhagic telangiectasia Juan Rodríguez-García et al. http://www.haematologica.org/content/103/10/e491

e492

Infections and vaccination in hereditary hemorrhagic telangiectasia: microbiological evidence-based considerations Hanny Al-Samkari et al. http://www.haematologica.org/content/103/10/e493

Haematologica 2018; vol. 103 no. 10 - October 2018 http://www.haematologica.org/

EDITORIALS Improving consolidation therapy in acute myeloid leukemia - a tough nut to crack Richard F. Schlenk,1,2 Sonia Jaramillo2 and Carsten MĂźller-Tidow2 1

NCT-Trial Center, National Center for Tumor Diseases, German Cancer Research Center, Heidelberg, and 2Department of Hematology, Oncology, and Rheumatology at Heidelberg University Hospital, University of Heidelberg, Germany E-mail: richard.schlenk@nct-heidelberg.de doi:10.3324/haematol.2018.200485

fter intensive induction therapy, 60% to 80% of younger (â&#x2030;¤60 years) and 40% to 60% of older (>60 years) patients with acute myeloid leukemia (AML) achieve a complete remission.1 However, despite intensive consolidation therapy including intensive chemotherapy, autologous or allogeneic hematopoietic cell transplantation approximately half of younger and 80% to 90% of older patients relapse and the majority of relapsed patients succumb to their disease.2 Based on these figures, it is expected that prevention of relapse by better consolidation therapy would immediately translate into better overall survival. This interrelationship appears to be simple but it is now 24 years ago that this could be demonstrated in a randomized clinical trial that showed a dose-response effect for cytarabine, with improved relapse-free and overall survival among the patients receiving high-dose cytarabine,3 which remains a cornerstone of consolidation chemotherapy in AML.1,4,5 In this issue of Haematologica, Burnett and colleagues report on a randomized comparison evaluating the addition of the mammalian target of rapamycin inhibitor, everolimus, to consolidation therapy in AML.6 Everolimus was given for a maximum of 84 days between chemotherapy courses in the experimental arm of the study. Despite the pre-clinical in vitro and in vivo rationale for everolimus, further supported by promising clinical data from phase I/II trials, the independent Data Monitoring Committee (DMC) advised study termination after randomization of 339 patients (2:1 ratio) due to excessive mortality in the everolimus arm. Toxicity of everolimus was primarily gastrointestinal (mucositis and diarrhea) and biochemical evidence of liver toxicity. The primary reason for increased mortality was infection-related deaths within the first 6 months of treatment mainly due to the immunosuppressive effects of everolimus, which reflects what has been seen with the use of this drug in solid tumors.7 This is a remarkable study and we would like to highlight two aspects: (i) the important role of the DMC in taking care of patientsâ&#x20AC;&#x2122; safety, (ii) the issue of whether we have the right strategies to improve results in consolidation therapy. Overall, the DMC played a very active role in the study by first recommending dose-reduction for the starting dose from 10 mg to 5 mg after randomization of 146 patients, based on the observation of increased side effects and reduced compliance; the DMC then recommended stopping the trial prematurely after randomization of 339 of the intended 600 patients. These DMC decisions were based, at those time points during the study, on incomplete datasets and were associated with some uncertainty and thus the decisions were not easy to take.8 In hindsight, with more trial data available, these decisions were clearly justified and prevented exposure of additional patients to an increased risk of death. This trial is, therefore, a good example of successful DMC work with useful recommendations at the right time points haematologica | 2018; 103(10)

during a study. This underlines the importance of anticipating risks already in the planning phase of a clinical trial, incorporating the identified risks into the statistical design and structure of a study with predefined interim analyses, selecting appropriate DMC members with engagement at the interim analyses as well as unscheduled analyses if necessary and, finally, of having an experienced and alert study team. The second interesting aspect of the study which we like to focus on is whether we are using the right strategy to improve consolidation therapy in AML. Based on favorable results with mammalian target of rapamycin inhibitors in preclinical models and clinical phase I/II studies in patients with active disease (newly diagnosed or relapsed AML), everolimus was added in the UK NCRI AML17 trial to consolidation therapy for patients who were in first complete remission. One assumption behind this approach is that effective biological mechanisms of action are similar during induction treatment and during consolidation therapy with residual leukemic cells hidden in bone marrow niches. It became apparent that this was not the case: in contrast to the encouraging results of everolimus in patients with active disease, the use of this inhibitor as an add-on to standard consolidation chemotherapy was associated with increased toxicity and a significant excess of deaths in remission. Even more disappointing was the fact that there was no evidence that everolimus is effective in preventing relapses. Interestingly, this mirrors data from clinical trials evaluating midostaurin and gemtuzumab ozogamicin (GO). Based on the pivotal large international multicenter randomized double-blinded phase III trial (CALGB 10603, RATIFY, clinicaltrials.gov: NCT00651261) in young adults (18-59 years) with FLT3-mutated AML, the US Food & Drug Administration (FDA) and the European Medicines Agency (EMA) recently approved midostaurin as an adjunct to conventional chemotherapy, including induction and post-remission therapy, without an upper age limit.9 A debate is ongoing how midostaurin affects overall survival and about the role of midostaurin in consolidation and maintenance therapy. A recent exploratory analysis of the trial revealed that midostaurin most effectively prevented relapse in patients who underwent allogeneic hematopoietic cell transplantation in first complete remission. These patients had a trend to better survival (P=0.07) and a significantly lower cumulative incidence of relapse (P=0.02).10 In contrast, patients in first complete remission who received high-dose cytarabine consolidation therapy had a comparable cumulative incidence of relapse rate whether they received midostaurin or not. Thus again, there is no clear evidence that midostaurin given as add-on therapy to high-dose cytarabine prevents relapse. In fact, the addition of midostaurin to first induction therapy seems to have the greatest impact on the observed beneficial effect on event-free and overall survival.9 Based on the results of the ALFA-0701 (NCT00927498) 1579

Editorials

study in newly diagnosed patients,11 the AML-19 study in patients with newly diagnosed AML unsuitable for intensive chemotherapy,12 and the MyloFrance-1 study in relapsed/refractory AML,13 GO was reapproved by the FDA in 2017 for the treatment of newly-diagnosed CD33+ AML in adults and treatment of relapsed or refractory CD33+ AML in adults and in pediatric patients 2 years and older. In Europe, GO was approved in 2018 for the treatment of patients aged 15 years and older with previously untreated, de novo, CD33+ AML. Both approvals (FDA, EMA) for newly-diagnosed CD33+ AML in adults included the addition of GO (3 mg/m2, day 1) to consolidation therapy with daunorubicin and cytarabine. However, in two trials assessing GO administered on a randomized basis in post-remission therapy, no significant impact on survival was observed.14,15 In the MRC AML-15 trial a total of 948 patients were randomly assigned to receive or not receive GO as an adjunct to first consolidation therapy.14 Once again, there was no evidence that relapses were prevented (P=0.20) and the overall survival rates of the patients in the two groups were nearly superimposable (hazard ratio, 1.02; 95% confidence interval: 0.82-1.27). In a study of the HOVON group, older patients achieving complete remission after intensive induction therapy were randomized to three cycles of GO (6 mg/m2 every 4 weeks) or no postremission therapy.15 The two treatment groups (113 patients receiving GO versus 119 control patients) were comparable with respect to age, performance status, and cytogenetics. There were no significant differences between the groups with regard to overall survival (P=0.52) and disease-free survival (P=0.40). These examples consistently show that new drugs that are active as an adjunctive therapy to standard induction are not necessarily active in consolidation therapy. But why is this the case and how can we improve the situation? In AML patients with active disease (newly diagnosed, relapsed or refractory) there is usually a bulk population of leukemic cells which can be characterized in depth by sophisticated methods and treated with targeted drugs if the target is present, such as activating FLT3 mutations, CD33 expression, or probably active AKT signaling. In contrast, the clinical situation of consolidation therapy is currently difficult to model in vitro or in vivo. Therefore, the preclinical evidence available before the initiation of a clinical trial is often limited. For example, the senescenceassociated reprogramming of non-stem bulk leukemia cells into self-renewing, leukemia-initiating stem cells,16 which may occur during the course of AML treatment, is currently not assessed within clinical trials. It is, therefore, of the utmost importance that methodologies of stem cell research are adapted to be fit for the purpose of use in clinical studies, particularly addressing consolidation research questions. We also need to improve the sensitivity and specificity of our methods of describing the depth of remission during the consolidation treatment phase. The term molecular remission has been introduced in current guidelines.1,5 Nevertheless, consolidation therapy still resembles flying blind in that after each cycle complete remission is documented but frequently without taking measurable residual disease (MRD)17 assessment into account. In addition, more than 50% of relapses are not predicted by MRD 1580

assessment and occur in MRD-negative groups.4 The consequences of this are low levels of test sensitivity of realtime quantitative polymerase chain reaction-based methods,18,19 whereas flow-cytometry and sequencing-based methods have been characterized by low levels of specificity.20-22 Thus, MRD assessment during the course of AML treatment is essential and may help to improve clinical research in consolidation therapy of AML. Nevertheless, its assessment is currently only informative for the evaluation of consolidation treatment strategies if MRD is positive and declining or rising. Negative results are still difficult to interpret.4

References 1. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 2. Schlenk RF, Müller-Tidow C, Benner A, Kieser M. Relapsed/refractory acute myeloid leukemia: any progress? Curr Opin Oncol. 2017;29(6):467-473. 3. Mayer RJ, Davis RB, Schiffer CA, et al. Intensive postremission chemotherapy in adults with acute myeloid leukemia. Cancer and Leukemia Group B. N Engl J Med. 1994;331(14):896-903. 4. Schlenk RF, Jaramillo S, Müller-Tidow C. What’s new in consolidation therapy in AML? Semin Hematol. 2018; In press. 5. NCCN Guidelines https://www.nccn.org/store/login/login.aspx? ReturnURL=https://www.nccn.org/professionals/physician_gls/pdf/a ml.pdf 6. Burnett AK, Gupta ED, Knapper S, et al. Addition of the mammalian target of rapamycin inhibitor, everolimus, to consolidation therapy in acute myeloid leukemia: experience from the UK NCRI AML17 trial. Haematologica 2018;103(10):1654-1661. 7. Kaymakcalan MD, Je Y, Sonpavde G, et al. Risk of infections in renal cell carcinoma (RCC) and non-RCC patients treated with mammalian target of rapamycin inhibitors. Br J Cancer. 2013;108(12):2478-2484. 8. Harrington D, Drazen JM. Learning from a trial stopped by a data and safety monitoring board. N Engl J Med. 2018;378(21):2031-2032. 9. Stone RM, Mandrekar SJ, Sanford BL, et al. Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation. N Engl J Med. 2017;377(5):454-464. 10. Stone RM, Mandrekar SJ, Sanford BL, et al. The addition of midostaurin to standard chemotherapy decreases cumulative incidence of relapse (CIR) in the international prospective randomized, placebo-controlled, double-blind trial (CALGB 10603 / RATIFY [Alliance]) for newly diagnosed acute myeloid leukemia (AML) patients with FLT3 mutations. Blood. 2017;130(Suppl 1):2580. 11. Castaigne S, Pautas C, Terré C, et al. Acute Leukemia French Association. 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. 12. Amadori S, Suciu S, Selleslag D, et al. Gemtuzumab ozogamicin versus best supportive care in older patients with newly diagnosed acute myeloid leukemia unsuitable for intensive chemotherapy: results of the randomized phase III EORTC-GIMEMA AML-19 Trial. J Clin Oncol. 2016;34(9):972-979. 13. Taksin AL, Legrand O, Raffoux E, et al. High efficacy and safety profile of fractionated doses of Mylotarg as induction therapy in patients with relapsed acute myeloblastic leukemia: a prospective study of the ALFA group. Leukemia. 2007;21(1):66-71. 14. Burnett AK, Hills RK, Milligan D, et al. Identification of patients with acute myeloblastic leukemia who benefit from the addition of gemtuzumab ozogamicin: results of the MRC AML15 trial. J Clin Oncol. 2011;29(4):369-377. 15. Löwenberg B, Beck J, Graux C, et al. Gemtuzumab ozogamicin as postremission treatment in AML at 60 years of age or more: results of a multicenter phase 3 study. Blood. 2010;115(13):2586-2591. 16. Milanovic M, Fan DNY, Belenki D, et al. Senescence-associated reprogramming promotes cancer stemness. Nature. 2018;553 (7686):96-100. 17. Schuurhuis GJ, Heuser M, Freeman S, et al. Minimal/measurable residual disease in AML: a consensus document from the European LeukemiaNet MRD Working Party. Blood. 2018;131(12):1275-1291. 18. Ivey A, Hills RK, Simpson MA, et al. Assessment of minimal residual disease in standard-risk AML. N Engl J Med. 2016;374(5):422-433. 19. Krönke J, Schlenk RF, Jensen KO, et al. Monitoring of minimal residual

haematologica | 2018; 103(9)

Editorials disease in NPM1-mutated acute myeloid leukemia: a study from the German-Austrian Acute Myeloid Leukemia Study Group. J Clin Oncol. 2011;29(19):2709-2716. 20. Terwijn M, van Putten WL, Kelder A, et al. High prognostic impact of flow cytometric minimal residual disease detection in acute myeloid leukemia: data from the HOVON/SAKK AML 42A study. J Clin Oncol. 2013;31(31):3889-3897.

21. Freeman SD, Virgo P, Couzens S, et al. Prognostic relevance of treatment response measured by flow cytometric residual disease detection in older patients with acute myeloid leukemia. J Clin Oncol. 2013; 31(32):4123-4131. 22. Jongen-Lavrencic M, Grob T, Hanekamp D, et al. Molecular minimal residual disease in acute myeloid leukemia. N Engl J Med. 2018;378(13):1189-1199.

Therapy-related acute lymphoblastic leukemia Josep-Maria Ribera Clinical Hematology Department. ICO-Hospital Germans Trias i Pujol, Josep Carreras Research Institute, Badalona, Universitat AutĂ˛noma de Barcelona, Spain E-mail: jribera@iconcologia.net doi:10.3324/haematol.2018.200311

herapy-related acute lymphoblastic leukemia (tALL) refers to ALL developed in patients who have received prior cytotoxic therapies, including chemotherapy and/or radiotherapy for solid or hematologic cancers. In this issue of Haematologica, Aldoss et al.1 report the largest retrospective study of patients from a single institution with analysis focused only on cases with prior exposure to cytotoxic therapies. The frequency of t-ALL was 10%, an important subset of patients showed cytogenetic abnormalities similar to those found in therapy-related acute myeloid leukemias (t-AML) or therapy-related myelodysplastic syndromes (t-MDS), and the outcome of t-ALL patients was poorer than that of the de novo ALL patients, especially for those who did not undergo allogeneic hematopoietic stem cell transplantation (alloHSCT). Similar to t-AML or t-MDS, the pathogenesis of t-ALL is attributed to the genotoxic effect of cytotoxic therapies on hematopoietic progenitor cells, but the precise mechanisms are less understood than those of therapy-related myeloid neoplasias. An underlying constitutional predisposition shared by the malignancies and ALL cannot be ruled out, especially for cases with s-ALL. In this sense, some studies have observed a higher prevalence of malignant neoplasms among first-degree relatives of patients with t-ALL or s-ALL. Ten to15% of therapy-related leukemias are t-ALL,2 and it is frequently confounded with the so-called secondary ALL (s-ALL), that refers to the patients with ALL with antecedent neoplasia but without exposure to cytotoxic therapy. Both t-ALL and sALL are infrequent (less than 2%-10% of all ALL cases)1,314 (Table 1), and are unfortunately often considered together in most case series or registry studies. The precise distinction of both types of ALL is important, especially for t-ALL, because prior exposure to cytotoxic therapies could have an impact on both treatment-related morbidity and mortality and on response to chemotherapy and the subsequent use of alloHSCT.15 Clinically, t-ALL arises in older patients than de novo ALL, and there is a female and white ethnicity preponderance in some series, different to what occurs in newly diagnosed ALL patients. Pediatric cases of t-ALL have also been described.14 An important question is to know haematologica | 2018; 103(10)

whether t-ALL is biologically different from de novo ALL. No significant differences have been reported in the white blood cell (WBC) count, the frequency of central nervous system or other extramedullary infiltration, or in the frequency of the different leukemia phenotypes (TALL or B-cell precursor ALL), although some studies have described a lower WBC count in t-ALL patients.1 However, important differences have been observed regarding genetic background, showing a predominance of high-risk genetic subtypes in t-ALL compared with de novo ALL. The most consistent genetic abnormality found across studies is the 11q23 (KMT2A) rearrangement, followed by monosomies of chromosomes 5, 7 and/or 17, hypodiploidy, and in some studies, by the Philadelphia (Ph) chromosome (Table 1).1,2-14,16 Except for the latter rearrangement, these genetic abnormalities are similar to those found in t-AML or in t-MDS and support the etiologic role of prior chemotherapy in the pathogenesis of tALL. Alterations of tumor suppressor genes at the 11q23 chromosomal regions may also predispose the cells to both solid and hematological cancers. The 11q23 rearrangements are frequently observed in patients who have received topoisomerase II inhibitors. As occurs in de novo Ph-positive ALL, the p190 BCR-ABL subtype is predominant, but the frequency of associated chromosomal abnormalities (ACA, especially monosomies), is higher in cases with Ph-positive t-ALL.16 Large molecular studies are lacking in t-ALL and, consequently, the frequency of specific subgroups such as the BCR-ABL-like is unknown. Large epidemiological studies have shown that any previous malignancy can lead to an increased incidence of sALL or t-ALL, which establishes this ALL as a separate entity.9,11,12 Among all cancer survivors, those with prior cancer treatment have a higher probability to develop ALL than those with no prior treatment, with this increased risk being observed at any age.11,12 Regarding the type of previous solid cancer, breast cancer constitutes the most common prior solid malignancy across the series, probably related to its high frequency, the elevated utilization of alkylator and topoisomerase II inhibitors as well as radiotherapy, and the excellent long-term survival for this disease. Lymphomas and other lymphoproliferative disorders encompass the most frequent antecedent of 1581

Editorials

Table 1. Main studies on therapy-related acute lymphoblastic leukemia (t-ALL) and secondary acute lymphoblastic leukemia (s-ALL).

Author (year)

N t-ALL/ s-ALL

Frequency

Age (median, [range]), years

Most frequent antecedent solid tumor

Most frequent antecedent hematologic cancer

2.3%

58 (33-78)

Breast

Hodgkinâ&#x20AC;&#x2122;s lymphoma

1012

Breast

Lymphoma

44 30/14

9.6%

65 (30-86)

Breast

Lymphoma

Abdulwahab 23 (2012) (all t-ALL) Ganzel 32 (2015) 23/9

6.9%

51 (17-75)

Breast

Lymphoma

55 (3.3-87) t-ALL: 52 (3.3-76) s-ALL: 75 (23-87)

Breast

Lymphoma

Giri3 (2015) Kelleher (2016)

1.9%

62 (21-90)

Breast

Lymphoma

24 16/8

5.3%

t-ALL: 55 (20-78) s-ALL: 65 (25-72)

Breast

Lymphoma

Breast

Lymphoma

6.6%

Breast Prostate

9.1%

55 (23-85)

Breast

Lymphoproliferative neoplasms Myeloid neoplasms Lymphoproliferative neoplasms

Pagano (1999) Shivakumar (2008)

Tang (2012)

Rosenberg3 371 (2017) 184/187 Swaika3 772 (2018) Aldoss (2018)

93 (all t-ALL)

Interval prior Main malignancy-ALL cytogenetic (median [ range]), findings1 months

27 (4-170)

t(9;22) 12 (57%) 11q23 <18 yr: 36 (3-240) 11q23 40/72 (56%) 18-59 yr: 26 (6-192) t(9;22) â&#x2030;Ľ60 yr: 22 (4-168) Complex karyotype t-ALL: 36 (6-216) 11q23, t-ALL: 18/30 s-ALL: 144 (7-420) t(9;22) (60%) -5, -7 s-ALL: 10/14 -17, -17p (71%) Hypodiploid 48 (5-360) 11q23 17/21 (81%) t(9;22) 64 (2-336) t(9,22) 25 (86%) t-ALL: 72 (9-336) Hypodiploid s-ALL: 48 (2-336) -7/7p11q23 60 (12-198) NR NR t-ALL: 37 (7-333) t(9;22) t-ALL: 15/16 s-ALL: 84 (29-219) Hypodiploidy (94%) 11q23 s-ALL: 6/8 (75%) 67 (2.6-277) NR NR 60 (2-473)

82 (10-608)

t(9;22) 79/93 (85%) 11q23 -5/5q-/-7/7qComplex karyotype

HSCT in CR1

OS/EFS

NR OS: median 5 months 14/40 OS: median 6-7 months

EFS: 13% (3-yr)

5/18

OS: 37% (3-yr)

6/25

OS: 25% (2-yr)

OS: 6.8% (5-yr)

3/21

OS: t-ALL: 71% (3-yr) s-ALL: 38% (3-yr)

NR4

OS: 10% (5-yr)

49/79

OS: 46% (2-yr)

1 In order of frequency; 2Seven own patients and 94 collected from the literature. 3Epidemiologic study. 4Patients with t-ALL were at significantly increased risk of death compared to de novo ALL patients. NR: not reported; CR: complete remission; HSCT: hematopoietic stem cell transplantation; CR1: first complete remission; OS: overall survival; EFS: event-free survival.

hematologic cancer, and the same reasons for breast cancer could be applied to explain this high frequency of tALL. In studies including large series, patients with a previous primary hematological malignancy had a higher risk for s-ALL or t-ALL as compared to solid organ neoplasms.12 The latency from prior diagnosis of cancer to tALL varies among the case series, but in general it tends to be shorter than in s-ALL and slightly longer than in tAML or in t-MDS. Among t-ALL cases, those with KMT2A rearrangements show a shorter time interval and those with a Ph-positive rearrangement show a longer interval between the previous cancer and the development of leukemia.1,2-14,16 Regarding the therapy of t-ALL patients, there is concern about the possible impact of previous exposure to chemotherapy and/or to radiotherapy on the toxicity of the chemotherapeutic agents given in induction and consolidation. However, the tolerability and the treatmentrelated mortality were similar to that observed in de novo ALL in most studies.1 In some studies, the CR rate was 1582

similar, but in others it was lower than in newly diagnosed ALL cases (Table 1). Considering the retrospective nature of most of the studies and the long period of patient recruitment, data regarding the measurable (minimal) residual disease (MRD) clearance are very limited. As the selection to proceed to alloHSCT was not based on MRD levels, the perception of their poor prognosis (similar to what occurs in t-AML and t-MDS) explains the higher use of transplantation in these patients in some series.1,15 Given the advanced age of most t-ALL patients, reduced-intensity regimens are more frequently used for conditioning. The transplant-related mortality and the rate of relapse after transplantation have shown to be similar to those of de novo ALL in some studies.1 However, when considering the transplanted and non-transplanted cases together, the survival of t-ALL patients is poorer than that observed in de novo ALL (Table 1), this difference being especially evident in the group of non-transplanted cases.1 Population-based studies also have shown a poorer outcome for ALL patients with antecedent neoplahaematologica | 2018; 103(10)

Editorials

sia.9,11,12 It is highly probable that the inferior outcome of t-ALL patients may be attributable to the high-risk of cytogenetic abnormalities in these patients rather than the antecedent cancer itself even after considering the possible relapse of the neoplasia after the ALL treatment. In Ph-positive t-ALL the combination of tyrosine kinase inhibitors and chemotherapy, generally followed by alloHSCT, yields similar results to those observed in de novo Ph-positive ALL.16 The presence of ACA to the Ph chromosome does not seem to have an impact on prognosis, but the low number of cases cannot drive solid conclusions. The current information of t-ALL is based on retrospective studies of patients treated with chemotherapy followed, when possible, by alloHSCT in first CR. The latter decision is based on the assumption of their poor prognosis, mirroring what occurs in t-AML or t-MDS. Deep molecular studies as well as the systematic use of MRD in newly diagnosed patients with t-ALL are required in order to increase the knowledge of the precise mechanisms of leukemogenesis and to make an adequate choice of the post-remission therapy. Given the scarce frequency of tALL, the response of the relapsed or refractory patients to the modern immunotherapeutic or targeted therapy approaches is largely unknown, and their possible use in first-line therapy has not been evaluated to date. Finally, the identification of prognostic factors, especially genetic biomarkers, predictive for t-ALL or s-ALL in patients with primary malignancies should be pursued in order to prevent or anticipate the occurrence of this disease.

References 1. Aldoss I, Stiller T, Tsai NC, et al. Therapy-related acute lymphoblastic leukemia has distinct clinical and cytogenetic features compared to de novo acute lymphoblastic leukemia, but outcomes are comparable in transplanted patients. Haematologica. 2018. doi: 10.3324/haematol.2018.193599. 2. Bagg A. Therapy-associated lymphoid proliferations. Adv Anat

Pathol. 2011;18(3):199-205. 3. Pagano L, Pulsoni A, Tosti ME, et al. Acute lymphoblastic leukaemia occurring as second malignancy: report of the GIMEMA archive of adult acute leukaemia. Gruppo Italiano Malattie Ematologiche Maligne dell'Adulto. Br J Haematol. 1999;106(4):1037-1040. 4. Shivakumar R, Tan W, Wilding GE, Wang ES, Wetzler M. Biologic features and treatment outcome of secondary acute lymphoblastic leukemia--a review of 101 cases. Ann Oncol. 2008;19(9):1634-1638. 5. Tang G, Zuo Z, Thomas DA, et al. Precursor B-acute lymphoblastic leukemia occurring in patients with a history of prior malignancies: is it therapy-related? Haematologica. 2012;97(6):919-925. 6. Abdulwahab A, Sykes J, Kamel-Reid S, et al. Therapy-related acute lymphoblastic leukemia is more frequent than previously recognized and has a poor prognosis. Cancer. 2012;118(16):3962-3967. 7. Ganzel C, Devlin S, Douer D, et al. Secondary acute lymphoblastic leukaemia is constitutional and probably not related to prior therapy. Br J Haematol. 2015;170(1):50-55. 8. Matnani R, Parekh V, Borate U, Brazelton J, Reddy V, Peker D. Therapy-related B-lymphoblastic leukemia associated with Philadelphia chromosome and MLL rearrangement: Single institution experience and the review of the literature. Pathol Int. 2015;65(10):536-540. 9. Giri S, Chi M, Johnson B, et al. Secondary acute lymphoblastic leukemia is an independent predictor of poor prognosis. Leuk Res. 2015;39(12):1342-1346. 10. Kelleher N, Gallardo D, Gonzalez-Campos J, et al. Incidence, clinical and biological characteristics and outcome of secondary acute lymphoblastic leukemia after solid organ or hematologic malignancy. Leuk Lymphoma. 2016;57(1):86-91. 11. Rosenberg AS, Brunson A, Paulus JK, et al. Secondary acute lymphoblastic leukemia is a distinct clinical entity with prognostic significance. Blood Cancer J. 2017;7(9):e605. 12. Swaika A, Frank RD, Yang D, et al. Second primary acute lymphoblastic leukemia in adults: a SEER analysis of incidence and outcomes. Cancer Med. 2018;7(2):499-507. 13. Kurt H, Zheng L, Kantarjian HM, et al. Secondary Philadelphia chromosome acquired during therapy of acute leukemia and myelodysplastic syndrome. Mod Pathol. 2018;31(7):1141-1154. 14. Imamura T, Taga T, Takagi M, et al. Leukemia/Lymphoma Committee; Japanese Society of Pediatric Hematology Oncology (JSPHO). Nationwide survey of therapy-related leukemia in childhood in Japan. Int J Hematol. 2018;108(1):91-97. 15. Aldoss I, Dagis A, Palmer J, et al. Therapy-related ALL: cytogenetic features and hematopoietic cell transplantation outcome. Bone Marrow Transplant. 2015;50(5):746-748. 16. Aldoss I, Stiller T, Song J, et al. Philadelphia chromosome as a recurrent event among therapy-related acute leukemia. Am J Hematol. 2017;92(2):E18-E19.

eGVHD App: a new tool to improve graft-versus-host disease assessment Marie Therese Rubio Service d’Hématologie, CHRU Nancy, Hôpital Brabois, and CNRS UMR 7365, Equipe 6, Biopole de l’Université de Lorraine, Vandoeuvre les Nancy, France E-mail: mt_rubio@hotmail.com doi:10.3324/haematol.2018.200303

ccurately diagnosing and scoring acute and chronic graft-versus-host disease (GvHD) remain challenging for many hematologists. Inconsistency between bone marrow transplant centers has been recognized in this field, in particular because of problems in following the latest recommended guidelines.1-3 In this regard, Schoemans et al. present a new electronic tool, the eGVHD application (eGVHD App), designed to improve and harmonize GvHD assessment. 4 haematologica | 2018; 103(10)

The eGVHD app was developed by the UZ Leuven (Belgium) in collaboration with the European Society for Blood and Marrow Transplantation (EBMT) Transplantation Complications Working Party and the National Institute of Health (NIH) (Bethesda, USA). This e-tool is a free, open-source web application, distributed as a normal website or a mobile application (accessible at: www.uzleuven.be/egvhd). The App allows the diagnosis of classic and late acute, as well as classic and overlap 1583

Editorials

Figure 1. Graft-versus-host disease (GvHD) assessment tools used in the study. Healthcare professionals (HCPs) in hematopoietic stem cell transplantation (HSCT) centers were randomized to evaluate GvHD with or without a new electronic tool: the eGVHD application (eGVHD App). No App group, n=40; eGVHD App group, n=37. aGvHD: acute GvHD; cGvHD: chronic GvHD.

chronic GvHD, using the most up-to date guidelines [Mount Sinai Acute GvHD International Consortium (MAGIC) for acute and NIH 2014 for chronic GvHD]. The study presented in this issue of the Journal4 was performed in 7 Belgian centers practising allogeneic hematopoietic stem cell transplantation (HSCT) and included 77 health care practitioners (HPCs), among whom 58 were physicians (75%), 15 were data managers (19%), and 4 held another position. They were invited to evaluate the diagnosis and severity score of 10 clinical vignettes (4 acute and 6 chronic GvHD) validated by a group of 10 separate GvHD experts (Expert Gold Standard) who implemented the latest guidelines. For their evaluation, the 77 HCPs were randomized to either use their usual GvHD assessment tools without the eGVHD App (No App group, n=40) or to use the eGVHD App (App group, n=37) (Figure 1). The most frequently reported GvHD guidelines referenced by HPCs in the No App group were the Glucksberg5 and the NIH 20146 criteria. They assessed GvHD of the 10 case vignettes mostly by their own knowledge (62%) or by using the 2014 GvHD evaluation sheet (23%), the 2005 NIH evaluation sheet (15%), or a self-designed scoring document (15%). The use of the 1584

App compared to the No App group improved the number of vignettes with a correct diagnosis [10/10 vs. 6.5/10, respectively; OR=6.14 (95%CI: 2.83-13.34; P<0.001)] as well as the number of vignettes with correct GvHD scoring [9/10 vs. 4.5/10, respectively; OR=6.29 (95%CI: 4.329.15; P<0.001)]. Assessment of GvHD was significantly better in the App group for both acute (aGvHD) (OR=17.89; 95%CI: 8.47-37.79; P<0.001) and chronic (cGvHD) (OR=4.34; 95%CI: 2.79-6.74; P<0.001) GvHD. As shown in Figure 1, agreement between the HPCs' results and the Expert Gold Standard evaluations also showed the superiority of the use of the eGVHD App. For GvHD diagnosis, the No App group more often misdiagnosed late acute and overlap chronic GvHD by considering them as classic cGvHD. Scoring of aGvHD was frequently false in the No App group, in particular for grades II and IV, confused with cGvHD and grade III aGvHD, respectively. Scoring for cGvHD tended to be over-estimated (15%) or under-estimated (9%) by the App group without misclassification, while both diagnosis and scoring were frequently erroneous in the No App group. Agreement between HPCs was superior in the App group (0.73 vs. 0.56 in the No App group) independently haematologica | 2018; 103(10)

Editorials

of the center, the degree of experience, and professional background. The use of the App was, however, time consuming for the HCPs who had not used it before the study. Despite this limiting factor, they found it useful and reported that they would be willing to use it in their daily practice. Thus, this well-performed randomized study demonstrates that the eGVHD App provides superior accuracy and reliability for GvHD assessment compared to usual care, even for experienced physicians.4 The improvement can mainly be explained by the use of the most up-todate guidelines, the limitation of physicianâ&#x20AC;&#x2122;s subjectivity during the evaluation, and the fact that the e-tool provides pictures and definitions of GvHD features that help physicians to better categorize GvHD symptoms. The need for harmonization in the diagnosis and scoring of GvHD has been recognized for many years and several other attempts have been made to improve this by the use of electronic tools7,8 but generally without success. The eGVHD App is available everywhere (www.uzleuven.be/egvhd) and could be used by any practitioner. The App could be of particular interest to HPCs with limited GvHD experience and can also be used for training. It remains to be seen whether the use of the eGVHD App, by improving the grading of GvHD in daily practice, could have an impact on clinical decisions and transplant outcomes. In the era of fast developing electronic devices and of 'big-data' analyses, the eGVHD App represents the first e-tool to be made widely available with the potential to improve the quality of GvHD data in clinical research.

haematologica | 2018; 103(10)

Such an App should be implemented in clinical trials aiming to evaluate and treat GvHD after allogeneic HSCT, as well as in large-scale transplant data bases.

References 1. Atkinson K, Horowitz MM, Biggs JC, Gale RP, Rimm AA, Bortin MM. The clinical diagnosis of acute graft-versus-host disease: a diversity of views amongst marrow transplant centers. Bone Marrow Transplant. 1988;3(1):5-10. 2. Weisdorf DJ, Hurd D, Carter S, et al. Prospective grading of graft-versus-host disease after unrelated donor marrow transplantation: a grading algorithm versus blinded expert panel review. Biol Blood Marrow Transplant. 2003;9(8):512-518. 3. Schoemans H, Goris K, Durm RV, et al. Development, preliminary usability and accuracy testing of the EBMT 'eGVHD App' to support GvHD assessment according to NIH criteria-a proof of concept. Bone Marrow Transplant. 2016;51(8):1062-1065. 4. Schoemans HM, Goris K, Van Durm R, Fieuws S, et al. The eGVHD App has the potential to improve the accuracy of graft-versus-host disease assessment: a multicenter randomized controlled trial. Haematologica 2018; 103(10):1698-1707. 5. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graftversus-host disease in human recipients of marrow from HL-Amatched sibling donors. Transplantation. 1974;18(4):295-304. 6. 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. 7. Levine JE, Hogan WJ, Harris AC, et al. Improved accuracy of acute graft-versus-host disease staging among multiple centers. Best Pract Res Clin Haematol. 2014;27(3-4):283-287. 8. Dierov D, Ciolino C, Fatmi S, et al. Establishing a standardized system to capture chronic graft-versus-host disease (GVHD) data in accordance to the national institutes (NIH) consensus criteria. Bone Marrow Transplant. 2017;52(Suppl 1):S102.

1585

REVIEW ARTICLE Ferrata Storti Foundation

Multiple faces of succinate beyond metabolism in blood Franco Grimolizzi1 and Lorena Arranz1,2,3

Stem Cell Aging and Cancer Research Group, Department of Medical Biology, Faculty of Health Sciences, UiT – The Arctic University of Norway; 2Department of Hematology, University Hospital of North Norway and 3Young Associate Investigator, Norwegian Center for Molecular Medicine (NCMM), Tromsø, Norway 1

Haematologica 2018 Volume 103(10):1586-1592

ABSTRACT

uccinate is an essential intermediate of the tricarboxylic acid cycle that exerts pleiotropic roles beyond metabolism in both physiological and pathological conditions. Recent evidence obtained in mouse models shows its essential role regulating blood cell function through various mechanisms that include pseudohypoxia responses by hypoxiainducible factor-1α activation, post-translational modifications like succinylation, and communication mediated by succinate receptor 1. Hence, succinate links metabolism to processes like gene expression and intercellular communication. Interestingly, succinate plays key dual roles during inflammatory responses, leading to net inflammation or anti-inflammation depending on factors like the cellular context. Here, we further discuss current suggestions of the possible contribution of succinate to blood stem cell function and blood formation. Further study will be required in the future to better understand succinate biology in blood cells. This promising field may open new avenues to modulate inflammatory responses and to preserve blood cell homeostasis in the clinical setting.

Correspondence: lorena.arranz@uit.no

Introduction: roles of succinate beyond metabolism Received: April 19, 2018. Accepted: June 27, 2018. Pre-published: June 28, 2018. doi:10.3324/haematol.2018.196097 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1586 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1586

The metabolite succinate or succinic acid is at the hub of the tricarboxylic acid (TCA) cycle, and it is mainly produced by succinyl coenzyme A synthetase from succinyl coenzyme A, in a reversible reaction that generally occurs under aerobic conditions (Figure 1). Nonetheless, when cells rely on anaerobic glycolysis, like cancer cells and certain innate immune cells upon activation, other metabolic pathways sustain succinate levels, including glutamine-dependent anerplerosis to α-ketoglutarate, and eventually citrate by reductive carboxylation.1 Similarly, succinate may derive from the γ-aminobutyric acid shunt pathway that correlates with levels of expression of the γ-aminobutyric acid transporters solute carrier family 6 members 12 and 13 (SLC6A12, SLC6A13).2,3 Under physiological hypoxia, low oxygen levels lead to reduced activity of succinate dehydrogenase (SDH), which metabolizes succinate, and other oxygen-dependent enzymes in the electron transport chain, causing succinate accumulation.4,5 Succinate functions as a competitive inhibitor for prolyl hydroxylase domain (PHD) proteins that are central to degradation of hypoxiainducible factor (HIF)-1α subunit.3-6 In fact, one of the first pieces of evidence for a role of succinate in cancer development was provided by the discovery of pseudohypoxia, which refers to activation of hypoxia signaling pathways under normal oxygen levels. Pseudohypoxia is a typical event in tumors with mutated SDH.7 Hence, succinate functions may be classified as metabolic or non-metabolic. In mitochondria, succinate plays a crucial role in metabolism and operates in both anabolic and catabolic pathways.2,3 Mitochondria are the physiological source for succinate, but accumulated succinate may be transported to the cytosol through the dicarboxylic acid translocator in the mitochondrial inner membrane and the voltage-dependent anion channel in the outer membrane (Figure 1).6 In the cytosol, succinate plays regulatory roles beyond primary metabolism. Elevated cytosolic succinate levels may promote protein post-translational modifications by addition of succinyl groups to lysine residues.8,9 A remarkable effect of succinylation is to haematologica | 2018; 103(10)

Succinate beyond metabolism in blood

alter the net charge of the protein by up to two charge units.8,9 Further, lysine succinylation is abundant and it induces significant structural changes in proteins,10 but its functional effects on protein and cellular functions have yet to be elucidated. Interestingly, succinate connects intracellular metabolic status and intercellular communication, as it may be released to the extracellular space through plasma membrane transporters of the SLC13 family (Figure 1).11 Nevertheless, expression of these transporters on blood cells has not been well characterized. In the extracellular environment, succinate contributes to intercellular signaling by a receptor-mediated mechanism.12 Under steadystate conditions, circulating levels of succinate vary from 2 to 20 mM, and pro-inflammatory stimuli such as lipopolysaccharide (LPS), interleukin (IL)-8 and tumor necrosis factor (TNF)-α boost its concentration.13,14 In addition, activation of succinate receptor (Figure 1) was shown to be a critical mediator of inflammatory responses acting in synergy with toll-like receptors (TLR), thereby enhancing TNF-α and IL-1b expression in myeloid cells.12 Succinate further functions as chemoattractant, driving immune cell precursors from site of generation to place of

maturation.12 In humans, circulating levels of succinate increase exponentially under certain circumstances like endurance exercise (93 mM)15 and pathological conditions such as type 2 diabetes (~47 to 125 mM),16,17 obesity (~80 mM),17 and ischemia-reperfusion injury following myocardial infarction.18 In addition, elevated levels of circulating succinate relate to development of solid tumors and poor prognosis in a variety of hematologic malignancies.6,19 Conversely, activation of succinate receptor on neural stem cells was recently shown to promote their anti-inflammatory activity in an experimental model of autoimmune encephalomyelitis.20 Here, we critically assess recent advances on the role of succinate, beyond its metabolic functions, at the intersection between inflammatory responses and blood cell activity with the aim of identifying gaps in the literature and proposing perspectives for further research.

Succinylation and its potential immunomodulatory effects For some time now, succinate has been known to pro-

Figure 1. Succinate production and mechanisms of action. Succinate is an intermediate of several metabolic pathways, i.e. tricarboxylic acid (TCA) cycle under normoxic conditions (blue lines), and glutamine-dependent anerplerosis and γ-aminobutyric acid (GABA) shunt under anaerobic conditions (red lines). Accumulation of succinate associates with succinylation, i.e. addition of succinyl group to a lysine residue of a protein. Succinate inhibits action of prolyl hydroxylases (PHD) and thereby causes stabilization of hypoxia-inducible factor-1α (HIF-1α). Succinate further inhibits several dioxygenases involved in epigenetic regulation like ten-eleven translocation methylcytosine dioxygenase (TET) and jumonji C domain-containing histone lysine demethylases (JMJD3). Dicarboxylate carriers (DIC) and voltagedependent anion channels (VDAC) control succinate release from mitochondria to cytosol. Succinate is released to the extracellular space through sodium-coupled citrate transporters (SLC13). GPR91 is a G protein–coupled cell surface receptor for extracellular succinate (Sucnr1). ACO: aconitase; IDH: isocitrate dehydrogenase; ODC: oxoglutarate dehydrogenase; SCS: α-succinyl-CoA synthetase; SDH: succinate dehydrogenase; FUM: fumarase; MDH: malate dehydrogenase; CSY: citrate synthase; GS: glutamine synthetase; GOGAT: glutamine oxoglutarate aminotransferase.

haematologica | 2018; 103(10)

1587

F. Grimolizzi and L. Arranz et al.

mote deoxyribonucleic acid methylation through inhibition of histone demethylases and the ten-eleven translocation family proteins (Figure 1), and thereby to play a role in the cellular epigenetic landscape.21,22 Recently, a novel posttranslational modification was associated to succinate accumulation that results in lysine succinylation.3,9 Succinylation implies dramatic structural changes in proteins and confers a negative charge at physiological pH.9,10 Many succinylation sites on histones have been identified, which may result in alterations in chromatin structure and, consequently, gene expression.8,9 Mechanisms that trigger lysine succinylation are still not clear and little is known about its potential role in immune modulation. In Escherichia coli, where succinylation was discovered, accumulation of succinate promotes succinyl coenzyme A synthetase activity, leading to higher production of succinyl coenzyme A that acts, in turn, as succinyl donor.23 In mice, succinylation is reverted by sirtuin 5, a member of the nicotinamide adenine dinucleotide-dependent family of deacetylases that exhibits desuccinylase activity.23 Succinate dehydrogenase complex mutations seem to relate to hematopoietic malignancies, particularly lymphoid leukemias, in addition to endocrine cancers.19 Interestingly, human neutrophils isolated from patients with heterozygous germ line mutations in the enzyme SDHB, display enhanced cell survival and increased succinylation.24 However, the functional consequences of SDHB defects for the inflammatory phenotype were not identified, and neither was any causal mechanism found linking SDHB mutations and succinylation of proteins in neutrophils. Mouse models have been used to further understand these processes, but many questions remain. Macrophages from mice lacking SDHB accumulate succinate.25 This strategy together with LPS stimulation leads to decreased pro-inflammatory gene expression along with enhanced expression of a variety of anti-inflammatory soluble mediators, including IL-1 receptor antagonist and IL10.25 Accordingly, inhibition of SDH with itaconate treatment, a metabolite structurally similar to succinate, abolishes inflammatory response of macrophages reducing myocardial infarct in both in vitro and in vivo models by limiting succinate oxidation.26 Thus, limiting the oxidation of succinate may have an anti-inflammatory effect. The prominent mechanisms mediating these effects seem to relate to mitochondrial metabolic reprogramming and inhibition of reactive oxygen species (ROS) production.25,26 Unfortunately, the potential role of succinylation in these processes has not been studied. Conversely, macrophages from mice lacking sirtuin 5 become hypersensitive to LPS and display increased expression of IL-1b.27 One important limitation of this strategy is that it does not allow us to rule out possible deacetylation from desuccinylation activity of sirtuin 5. Nevertheless, the Authors elegantly showed a mechanism that at least partially explains these results.27 Succinylation is a modification that may occur to multiple proteins in addition to histones. This is the case of pyruvate kinase M2, whose succinylation at K311 promotes its translocation into the nucleus. Once in the nucleus, pyruvate kinase M2 forms a transcriptional complex with HIF-1α that directly binds to the IL-1b promoter gene and activates its transcription.27 Furthermore, pyruvate kinase M2 hypersuccinylation in sirtuin 5-deficient mice renders the animals more susceptible to experimental colitis through boosted IL-1b production.27 1588

Thus, future study should examine the link between succinylation and inflammatory pathways, its regulation in different blood cell subsets, and its consequences in terms of gene expression and immune cell functions in both healthy and clinical conditions.

Succinate as intercellular communicator Activation of succinate receptor 1 (SUCNR1), or G protein-coupled receptor 91 (GPR91), was recently recognized as a regulatory mediator on a variety of cell subsets, including cardiomyocytes, adipocytes, renal and blood cells.28-30 Half-maximal effective response for GPR91 is obtained with succinate concentrations of 28-56 mM, indicative of GPR91 activation at succinate concentrations higher than the steady-state levels.30 Thus, succinate-GPR91 axis may function as an early sensor of homeostasis perturbations. In this context, selection of appropriate experimental conditions is essential to prevent biased results. For example, use of carbon dioxide as a method of euthanasia in rodents should be avoided since it rapidly raises blood succinate levels.14 G protein-coupled receptors are typically classified based on the G protein subfamily that they activate, e.g. Gαi, Gαq, Gs or G12/13. Given the enriched expression of GPR91 in kidney, human embryonic kidney cells 293 are often used to examine its mechanisms of signal transduction. However, attempts to characterize downstream pathways have led to controversial results. Some studies show that GPR91 activates both Gαi and Gαq signaling, leading to inhibition of cyclic adenosine monophosphate production and deployment of calcium mobilization, respectively.30 In contrast, others have shown that Gαq is not required for succinate-induced activation of GPR91 and Gαi alone is sufficient to quench cyclic adenosine monophosphate and mobilize calcium.29 Despite these controversies, in vitro studies in kidney cell lines seem to agree that GPR91 signaling is mediated by mobilization of intracellular calcium stores rather than being dependent on influx of extracellular calcium.29 However, stimulation of platelets with succinate failed to induce intracellular calcium mobilization, indicating the potential cell type-specific signaling transduction pathways of GPR91.29 Clues as to the cell-wide plasticity of GPR91 can further be seen from mesenchymal cells in the liver, where GPR91 activation neither increases cytosolic calcium nor inhibits adenosine monophosphate production.31 Furthermore, distinct mechanisms are responsible for GPR91 switch off in a cell-type dependent manner. Unlike human embryonic kidney cells 293, which undergo internalization of GPR91 in presence of succinate,32 MadinDarby canine kidney cells instead show a temporal desensitization of GPR91 following succinate binding.32 Thus, cell-type specific responses of GPR91 machinery and on/off dynamics may confer great plasticity to this regulatory pathway and may also underlie seemingly controversial results. Therefore, characterizing GPR91 signaling in cells of interest appears to be essential to assess the impact on functional activity and to identify specific drug targets upon pathogenesis.

GPR91 in inflammatory responses Early observations on blood cells show that GPR91 activation boosts pro-inflammatory responses, but recent haematologica | 2018; 103(10)

Succinate beyond metabolism in blood

investigations have shown that this is not always the case and that the impact of GPR91 depends on the cellular context. For example, GPR91 may indirectly protect mice from low-grade systemic inflammation under high-fat-diet conditions, since GPR91-deficient mice show progressive hyperglycemia and reduced insulin secretion.33 This was related to a function of GPR91 in white adipose tissue, a tissue where GPR91 is highly expressed.33 Monocytes that egress from the bone marrow (BM) seem to lack expression of GPR91, yet the ability to respond to activation and differentiation signals is influenced by extracellular succinate.12 Unlike immature monocytes, both naive dendritic cells and macrophages express high levels of GPR91, and its activation promotes inflammation.12 In vitro, succinate failed to influence TNF-α and IL-1b expression by dendritic cells. However, higher expression of these cytokines is achieved by addition of succinate in combination with TLR-3 and TLR-7 ligands compared to use of TLR agonists alone.12 Macrophages respond to extracellular inflammatory signals like LPS in similar ways (Figure 2A), with GPR91-mediated signal transduction yielding a strong pro-inflammatory phenotype that results in increased production of IL-1b.2,34 Furthermore, LPS-stimulated macrophages release succinate into the cell culture medium.34 This effect seems to be independent of GPR91, as extracellular succinate is found more abundantly in cultures of GPR91 deficient macrophages than in corresponding wild-type (WT) cells.34 The underlying mechanism for this succinate transport still needs to be revealed. In this regard, future studies examining the activity and expression of the SLC13 transporters in blood cells will be of great interest. In addition, GPR91 seems to be required for dendritic cell activation and licensing in vivo. It leads dendritic cell migration to lymph nodes, as this function is impaired in GPR91deficient mice. However, after maturation, GPR91 expression rapidly declines, making dendritic cells unresponsive to subsequent succinate stimulation after 24 hours of stimulation with LPS or the cytokine cocktail of TNF-α and IL1b.12 Conversely, activated macrophages become more responsive to succinate stimulation.12 In both human cell lines and monocyte-derived dendritic cells, succinate activates the extracellular signal-regulated kinases Erk1 and Erk2,12,30 which are downstream signaling components of TLR pathways. In macrophages, HIF-1α is one key mediator of their response to succinate.3,25 Accumulation of succinate in the cytosol results in competitive inhibition of PHD enzymes and subsequent stabilization of HIF-1α, even in the presence of normal oxygen levels (Figure 2A).3,6 In turn, HIF-1α activation contributes to IL-1b expression.3,25 Later, succinate oxidation in mitochondria proved central to macrophage pro-inflammatory phenotype, given that competitive inhibition of SDH abrogates IL-1b and HIF-1α expression.25 In accordance with the IL-1b findings previously described, HIF-1α stability is similar in macrophages derived from both GPR91-deficient and WT mice in presence of extracellular succinate alone, whereas GPR91 promotes pseudohypoxia upon LPS stimulation (Figure 2A).34 Recently, this mechanism has been related to pathogenesis in an experimental model of antigen-induced arthritis, where GPR91-deficient mice display reduced macrophage activation and IL-1b production.34 Surprisingly, the inflammatory impact of GPR91 deficiency varies in different myeloid cell subsets. Animal models of allergic contact dermatitis revealed increased haematologica | 2018; 103(10)

reactions in the absence of GPR91, and this was related to hyperactive mast cells in GPR91-deficient mice by means of increased TNF-α expression.35 This effect seems to be dependent on defective mast cell differentiation from BM precursors, as augmented mast cell responses could be recapitulated in WT mast cells differentiated in vitro in the absence of succinate.35 Unfortunately, no further efforts were made to explore the mechanisms involved. Intriguingly, GPR91 may induce anti-inflammatory effects through non-cell-autonomous effects (Figure 2B). In an in vivo model of autoimmune encephalomyelitis, succinate produced by type I mononuclear phagocytes accumulates in the cerebrospinal fluid of the chronically inflamed central nervous system.20 Transplanted neural stem cells sense this extracellular succinate through GPR91. In vitro, GPR91 activation leads neural stem cells to secretion of prostaglandin E2, and upregulation of members of the SLC13 family of transporters (i.e. SLC13A3 and SLC13A5) that uptake and scavenge extracellular succinate. In vivo, succinate scavenging seems to be the main mediator of the anti-inflammatory effects of transplanted neural stem cells. As a result, the beneficial effects of this strategy include reduction of succinate levels in the cerebrospinal fluid, shifting from type 1 inflammatory mononuclear phagocytes to anti-inflammatory cells, reduced post mortem tissue pathology, and improvement of behavioral defects.20 However, GPR91 is essential for these effects in vivo, as GPR91-deficient neural stem cells show reduced ability to protect from chronic neuroinflammation upon transplantation.20 It remains to be seen how broadly applicable this succinate-mediated anti-inflammatory mechanism will be to additional regulatory cells and inflammatory disease systems. However, it is interesting to hypothesize that succinate may play dual roles in inflammation both as an early driver, through cell-autonomous mechanisms, and as a late terminator by non-cell autonomous processes. Future studies should clarify the validity of this exciting hypothesis. In addition, given the contribution of GPR91 to tipping the inflammatory balance, studies are required to better understand how its modulation may provide a useful clinical tool for the control of inflammatory responses and maintenance of immune homeostasis. In this regard, recent development of GPR91-specific agonists, up to 20-fold more powerful compared to succinate,36 and antagonists,37 opens new avenues to explore the promising therapeutic value of GPR91 control with any accuracy.

New insights into hematopoiesis: what role does succinate play? Early investigations into hematopoiesis have provided hints of a role for succinate signaling in blood stem cell function, i.e. blood formation or hematopoiesis. GPR91 is expressed in monocyte-derived cell subsets, as previously discussed, while it is absent in immature monocytes, and T cells and B cells,38 indicative of selective roles for GPR91 on different blood cell subsets. In contrast, others have found expression of GPR91 transcript in purified populations of blood CD14+ monocytes and platelets, whereas its absence was confirmed in CD4+ T cells, CD8+ T cells, CD16+ granulocytes and CD19+ B cells.39 Unexpectedly, western blot analysis of protein expression showed presence of GPR91 in all blood cells except granulocytes.39 Interestingly, human BM CD34+ progenitor cells have been 1589

F. Grimolizzi and L. Arranz et al.

Figure 2. Dual role for succinate as pro- and anti-inflammatory signal. (A) Succinate as pro-inflammatory signal. Toll-like receptor (TLR) activation in macrophages causes intracellular succinate accumulation and release. In the cytosol, succinate functions as competitive inhibitor for prolyl hydroxylase domain (PHD) proteins, promoting stabilization of hypoxia-inducible factor-1Îą (HIF-1Îą), that in turn leads to pro-inflammatory interleukin-1b (IL-1b) production.3 Activation of interleukin-1 receptor (IL-1R) increases G protein-coupled receptor 91 (GPR91) expression, and extracellular succinate interacts with GPR91. This results in pseudohypoxia and IL-1b production, independent of HIF.34 (B) Succinate as anti-inflammatory signal. M1 inflammatory macrophages release succinate, which activates GPR91 on neural stem cells increasing the expression in vitro of prostaglandin E2 (PGE2) and members of the sodium-coupled citrate (SLC) family of transporters 13 (i.e. SLC13A3 and SLC13A5). In vivo, the most prominent anti-inflammation mechanism is succinate scavenging by SLC13 transporters. This reduces IL-1b and succinate extracellular levels, and promotes a shift in macrophage polarity from pro-inflammatory (in green) to anti-inflammatory phenotype (in blue).20 LPS: lipopolysaccharide; VHL: von Hippel-Lindau protein; Ub: ubiquitin.

1590

haematologica | 2018; 103(10)

Succinate beyond metabolism in blood

reported to express GPR91 on their cell surface,38 suggesting a potential role for succinate signaling in hematopoietic stem cells (HSC). At present, the lack of commercially available mouse GPR91-specific antibody makes it difficult to further explore the significance of succinate-GPR91 signaling pathway in the biology of HSC and progenitor cells. In this regard, generation of reporter mouse lines will be of great value in the hematopoietic field. In vitro, succinate stimulates the proliferation capacity of erythroid and megakaryocyte progenitors, and cell proliferation is significantly hampered in cells transfected with small interfering ribonucleic acid targeting GPR91.38 However, when megakaryocytes and erythroblasts are differentiated in vitro from human cord blood CD34+ cells, megakaryocytes show 7.2-fold higher GPR91 transcript expression as detected by cDNA microarrays.39 Of note, differentiation of CD34+ progenitors into megakaryocytes is performed in the presence of both thrombopoietin and IL-1b.39 As previously discussed, GPR91 signaling seems to be required for correct mast cell differentiation.35 GPR91deficient mice show hypomorphic mast cells that display increased responses, and this can be recapitulated in vitro using WT BM cells differentiated in the absence of succinate.35 Following chemotherapy, succinate treatment in vivo stimulates multilineage blood cell recovery, including red blood cells, platelets and neutrophils.38 In addition, inducible deletion of the mitochondrial SDHD in mouse models leads to remarkable hematopoietic defects, including depletion of BM progenitors and differentiated cells, and apoptosis in the HSC compartment defined as LSK, i.e. lineage-negative, c-kit-positive, and sca-1-positive.40 However, the consequences of this strategy on succinate levels in LSK cells and their link to cell functional changes has not been studied. Furthermore, an inducible system was used to conditionally delete SDHD controlled in a temporal manner, using Cre mediated recombination that activates upon estrogen analog tamoxifen administration. However, this genetic approach does not allow the cell subset where the deletion occurs to be delimited, either within or outside the hematopoietic system. Given the key role of regulatory cells within the healthy HSC niche, that tightly controls the function, fate and numbers of HSC in the BM,41 it may well be that these effects are partially mediated by the stem cell niche. Future studies will be needed to validate this exciting hypothesis. Interestingly, succinate is increased 24-fold in BM stromal cells derived from type 2 diabetes mellitus mice compared to normoglycemic mice.42 In this study, adherent cells were used after flushing out BM cells and culture of the adherent fraction for one week in vitro. Thus, levels of metabolites in this study may differ from those found in primary BM stromal cells in vivo. Furthermore, these adherent cells may consist of heterogeneous populations of stromal cells. Nevertheless, the Authors reveal an interesting mechanism that may contribute to bone dysregulation in metabolic disorders. In vitro, extracellular succinate binds to GPR91 on osteoclastic lineage cells and stimulates osteoclast differentiation; a process that may contribute to the bone resorption seen in vivo.42 Succinate may then have an indirect effect on hematopoiesis in diabetes, as osteoclasts function as negative regulators of HSC43 whereas osteoblasts support lymphoid progenitors.44,45 Future studies will be required to test these ideas. Furthermore, HIF-1Îą contributes importantly to cellhaematologica | 2018; 103(10)

cycle regulation in HSC,46 so identification of the potential link to succinate is of high relevance. Actually, HSC reside in hypoxic BM niches, which maintain their long-term selfrenewal by mechanisms like limiting their production of ROS.47 This is performed through adaptation of their metabolism to maintain a high glycolysis rate by HIF-1Îą activation. These conditions change during the processes of proliferation and differentiation, with activated cells depending more on oxidative phosphorylation to meet the energy requirements.47 Besides this, HSC niches are closely related to the vasculature in the BM and have been defined as perivascular, with mainly endothelial cells and mesenchymal stromal cells secreting factors that promote HSC maintenance.48 In this regard, GPR91 has been suggested to link capillary function to metabolic needs in other tissues, like retina, where GPR91 is essential to establish a neovascular network in response to injury through production of numerous angiogenic factors including vascular endothelial growth factor by retinal ganglion neurons.49 Considering that SDH mutations, HIF-1Îą accumulation and elevated levels of circulating succinate relate to pathology, particularly in hematologic malignancies,6,7,16-19 characterization of succinate signaling network, both cell- and non-cellautonomous, is pivotal to understand the link between metabolic alterations that affect HSC function and hematologic malignancies.

Conclusions Succinate is an essential intermediate of the TCA cycle at the intersection among metabolism, gene expression and intercellular communication. Initially characterized as an inflammatory signal, recent data show succinate-mediated anti-inflammatory mechanisms depending on the cellular subsets. The great plasticity of succinate biology is exemplified by cell-type specific responses of GPR91 machinery and on/off dynamics that may contribute to seemingly controversial results. In this scenario, it is interesting to hypothesize that succinate may play dual roles both as early driver of inflammation, through cell-autonomous processes, and as late terminator by intercellular communication with regulatory cells. Future studies are required to demonstrate this idea. Furthermore, we are only starting to understand how succinylation may influence protein activation and epigenetic landscape, and thereby gene expression and cell function. Studies on both succinylation regulation and impact will be highly valuable for a complete picture of succinate biology in blood cells. Finally, early hints on hematopoiesis seem to involve succinate in the survival and differentiation of blood stem cells, through mechanisms that remain unclear. Future research should carefully explore the promising therapeutic value of succinate targeting in inflammatory diseases, including hematologic malignancies.50 Acknowledgments Our work is supported by a joint meeting grant of the Northern Norway Regional Health Authority, the University Hospital of Northern Norway (UNN) and UiT The Arctic University of Norway (UiT) (2014/5668), Young Research Talent grants from the Research Council of Norway, (Stem Cell Program, 247596; FRIPRO Program, 250901), and grants from the Norwegian Cancer Society (6765150), the Northern Norway Regional Health Authority (HNF1338-17), and the Aakre-Stiftelsen Foundation (2016/9050) to LA. 1591

F. Grimolizzi and L. Arranz et al.

References 1. Wise DR, Ward PS, Shay JES, et al. Hypoxia promotes isocitrate dehydrogenase-dependent carboxylation of -ketoglutarate to citrate to support cell growth and viability. Proc Natl Acad Sci USA. 2011;10(49):1961119616. 2. Kelly B, O'Neill LA. Metabolic reprogramming in macrophages and dendritic cells in innate immunity. Cell Res. 2015;25(7):771784. 3. Tannahill GM, Curtis AM, Adamik J, et al. Succinate is an inflammatory signal that induces IL-1beta through HIF-1alpha. Nature. 2013;496(7444):238-242. 4. Epstein AC, Gleadle JM, McNeill LA, et al. C. elegans EGL-9 and mammalian homologs define a family of dioxygenases that regulate HIF by prolyl hydroxylation. Cell. 2001;107(1):43-54. 5. Jaakkola P, Mole DR, Tian YM, et al. Targeting of HIF-alpha to the von HippelLindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science. 2001;292(5516):468-472. 6. Selak MA, Armour SM, MacKenzie ED, et al. Succinate links TCA cycle dysfunction to oncogenesis by inhibiting HIF-alpha prolyl hydroxylase. Cancer Cell. 2005;7(1):77-85. 7. Morin A, Letouze' E, Gimenez-Roqueplo AP, Favier J. Oncometabolites-driven tumorigenesis: From genetics to targeted therapy. Int J Cancer. 2014;135(10):2237-2248. 8. Xie Z, Dai J, Dai L, et al. Lysine Succinylation and Lysine Malonylation in Histones. Mol Cell Proteomics. 2012;11(5): 100-107. 9. Park J, Chen Y, Tishkoff DX, et al. SIRT5mediated lysine desuccinylation impacts diverse metabolic pathways. Mol Cell. 2013;50(6):919-930. 10. Zhang ZH, Tan MJ, Xie ZY, Dai LZ, Chen Y, Zhao YM. Identification of lysine succinylation as a new post-translational modification. Nat Chem Biol. 2011;7(1):58-63. 11. Willmes DM, Birkenfeld AL. The Role of INDY in Metabolic Regulation. Comput Struct Biotechnol J. 2013;6:e201303020. 12. Rubic T, Lametschwandtner G, Jost S, et al. Triggering the succinate receptor GPR91 on dendritic cells enhances immunity. Nat Immunol. 2008;9(11):1261-1269. 13. Kushnir MM, Komaromy-Hiller G, Shushan B, Urry FM, Roberts WL. Analysis of dicarboxylic acids by tandem mass spectrometry. High-throughput quantitative measurement of methylmalonic acid in serum, plasma, and urine. Clin Chem. 2001;47(11):19932002. 14. Sadagopan N, Li W, Roberds SL, et al. Circulating succinate is elevated in rodent models of hypertension and metabolic disease. Am J Hypertens. 2007;20(11):12091215. 15. Hochachka PW, Dressendorfer RH. Succinate accumulation in man during exercise. Eur J Appl Physiol Occup Physiol. 1976;35(4):235-242. 16. van Diepen JA, Robben JH, Hooiveld GJ, et al. SUCNR1-mediated chemotaxis of macrophages aggravates obesity-induced inflammation and diabetes. Diabetologia. 2017;60(7):1304-1313. 17. Serena C, Ceperuelo-MallafrĂŠ V, Keiran N, et al. Elevated circulating levels of succinate in human obesity are linked to specific gut

1592

microbiota. ISME J. 2018;12(7):1642-1657. 18. Kohlhauer M, Dawkins S, Costa ASH, et al. Metabolomic Profiling in Acute STSegment-Elevation Myocardial Infarction Identifies Succinate as an Early Marker of Human Ischemia-Reperfusion Injury. J Am Heart Assoc. 2018;7(8). 19. Renella R, Carnevale J, Schneider KA, Hornick JL, Rana HQ, Janeway KA. Exploring the association of succinate dehydrogenase complex mutations with lymphoid malignancies. Fam Cancer. 2014;13 (3):507-511. 20. Peruzzotti-Jametti L, Bernstock JD, Vicario N, et al. Macrophage-Derived Extracellular Succinate Licenses Neural Stem Cells to Suppress Chronic Neuroinflammation. Cell Stem Cell. 2018;22(3):355-368. 21. Cervera AM, Bayley JP, Devilee P, McCreath KJ. Inhibition of succinate dehydrogenase dysregulates histone modification in mammalian cells. Mol Cancer. 2009;8:89. 22. Xiao M, Yang H, Xu W, et al. Inhibition of alpha-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes Dev. 2012;26(12):1326-1338. 23. Du J, Zhou Y, Su X, et al. Sirt5 is a NADdependent protein lysine demalonylase and desuccinylase. Science. 2011;334(6057):806809. 24. Jones R, McDonald KE, Willson JA, et al. Mutations in succinate dehydrogenase B (SDHB) enhance neutrophil survival independent of HIF-1alpha expression. Blood. 2016;127(21):2641-2644. 25. Mills EL, Kelly B, Logan A, et al. Succinate Dehydrogenase Supports Metabolic Repurposing of Mitochondria to Drive Inflammatory Macrophages. Cell. 2016;167 (2):457-470. 26. Lampropoulou V, Sergushichev A, Bambouskova M, et al. Itaconate Links Inhibition of Succinate Dehydrogenase with Macrophage Metabolic Remodeling and Regulation of Inflammation. Cell Metab. 2016;24(1):158-166. 27. Wang F, Wang K, Xu W, et al. SIRT5 Desuccinylates and Activates Pyruvate Kinase M2 to Block Macrophage IL-1beta Production and to Prevent DSS-Induced Colitis in Mice. Cell Rep. 2017;19(11):23312344. 28. Hogberg C, Gidlof O, Tan C, et al. Succinate independently stimulates full platelet activation via cAMP and phosphoinositide 3kinase-beta signaling. J Thromb Haemost. 2011;9(2):361-372. 29. Sundstrom L, Greasley PJ, Engberg S, Wallander M, Ryberg E. Succinate receptor GPR91, a Galpha(i) coupled receptor that increases intracellular calcium concentrations through PLCbeta. FEBS Lett. 2013;587(15):2399-2404. 30. He W, Miao FJ, Lin DC, et al. Citric acid cycle intermediates as ligands for orphan Gprotein-coupled receptors. Nature. 2004;429 (6988):188-193. 31. Correa PR, Kruglov EA, Thompson M, Leite MF, Dranoff JA, Nathanson MH. Succinate is a paracrine signal for liver damage. J Hepatol. 2007;47(2):262-269. 32. Robben JH, Fenton RA, Vargas SL, et al. Localization of the succinate receptor in the distal nephron and its signaling in polarized MDCK cells. Kidney Int. 2009;76(12):12581267. 33. McCreath KJ, Espada S, GĂĄlvez BG, et al.

34.

35.

36.

37.

38.

39.

40.

41. 42. 43.

44.

45.

46.

47.

48. 49.

50.

Targeted disruption of the SUCNR1 metabolic receptor leads to dichotomous effects on obesity. Diabetes. 2015;64(4):11541167. Littlewood-Evans A, Sarret S, Apfel V, et al. GPR91 senses extracellular succinate released from inflammatory macrophages and exacerbates rheumatoid arthritis. J Exp Med. 2016;213(9):1655-1662. Rubi -Schneider T, Carballido-Perrig N, Regairaz C, et al. GPR91 deficiency exacerbates allergic contact dermatitis while reducing arthritic disease in mice. Allergy. 2017;72(3):444-452. Geubelle P, Gilissen J, Dilly S, et al. Identification and pharmacological characterization of succinate receptor agonists. Br J Pharmacol. 2017;174(9):796-808. Bhuniya D, Umrani D, Dave B, et al. Discovery of a potent and selective small molecule hGPR91 antagonist. Bioorg Med Chem Lett. 2011;21(12):3596-3602. Hakak Y, Lehmann-Bruinsma K, Phillips S, et al. The role of the GPR91 ligand succinate in hematopoiesis. J Leukoc Biol. 2009;85 (5):837-843. Macaulay IC, Tijssen MR, ThijssenTimmer DC, et al. Comparative gene expression profiling of in vitro differentiated megakaryocytes and erythroblasts identifies novel activatory and inhibitory platelet membrane proteins. Blood. 2007;109(8):3260-3269. Bejarano-Garcia JA, Millan-Ucles A, Rosado IV, et al. Sensitivity of hematopoietic stem cells to mitochondrial dysfunction by SdhD gene deletion. Cell Death Dis. 2016;7(12) :e2516. Morrison SJ, Scadden DT. The bone marrow niche for haematopoietic stem cells. Nature. 2014;505(7483):327-334. Guo Y, Xie C, Li X, et al. Succinate and its Gprotein-coupled receptor stimulates osteoclastogenesis. Nat Commun. 2017;8:15621. Miyamoto K, Yoshida S, Kawasumi M, et al. Osteoclasts are dispensable for hematopoietic stem cell maintenance and mobilization. J Exp Med. 2011;208(11):2175-2181. Visnjic D, Kalajzic Z, Rowe DW, Katavic V, Lorenzo J, Aguila HL. Hematopoiesis is severely altered in mice with an induced osteoblast deficiency. Blood. 2004;103(9): 3258-3264. Zhu X1, Gui J, Dohkan J, Cheng L, Barnes PF, Su DM. Lymphohematopoietic progenitors do not have a synchronized defect with age-related thymic involution. Aging Cell. 2007;6(5):663-672. Takubo K, Goda N, Yamada W, et al. Regulation of the HIF-1 alpha Level Is Essential for Hematopoietic Stem Cells. Cell Stem Cell. 2010;7(3):391-402. Suda T, Takubo K, Semenza GL. Metabolic regulation of hematopoietic stem cells in the hypoxic niche. Cell Stem Cell. 2011;9(4): 298-310. Crane GM, Jeffery E, Morrison SJ. Adult haematopoietic stem cell niches. Nat Rev Immunol. 2017;17(9):573-590. Sapieha P, Sirinyan M, Hamel D, Zaniolo K, Joyal JS, Cho JH. The succinate receptor GPR91 in neurons has a major role in retinal angiogenesis. Nat Med. 2008;14(10):10671076. Arranz L, Arriero MDM, Villatoro A. Interleukin-1beta as emerging therapeutic target in hematological malignancies and potentially in their complications. Blood Rev. 2017;31(5):306-317.

haematologica | 2018; 103(10)

REVIEW ARTICLE

Normal and pathological erythropoiesis in adults: from gene regulation to targeted treatment concepts

Peter Valent,1,2 Guntram Büsche,3 Igor Theurl,4 Iris Z. Uras,5 Ulrich Germing,6 Reinhard Stauder,7 Karl Sotlar,8 Wolfgang Füreder,1 Peter Bettelheim,9 Michael Pfeilstöcker,2,10 Rainer Oberbauer,11 Wolfgang R. Sperr,1,2 Klaus Geissler,12 Jürg Schwaller,13 Richard Moriggl,14,15 Marie C. Béné,16 Ulrich Jäger,1,2 Hans-Peter Horny17 and Olivier Hermine18

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1593-1603

Department of Internal Medicine I, Division of Hematology & Hemostaseology, Medical University of Vienna, Austria; 2Ludwig Boltzmann Cluster Oncology, Medical University of Vienna, Austria; 3Institute of Pathology, Medizinische Hochschule Hannover, Germany; 4 Department of Internal Medicine II, Medical University Innsbruck, Austria; 5Institute of Pharmacology and Toxicology, University of Veterinary Medicine, Vienna, Austria; 6 Department of Hematology, Oncology and Clinical Immunology, Heinrich-Heine University, Düsseldorf, Germany; 7Department of Internal Medicine V, Medical University Innsbruck, Austria; 8Institute of Pathology, Paracelsus Medical University Salzburg, Austria; 9First Department of Internal Medicine, Elisabethinen Hospital, Linz, Austria; 10 rd 3 Medical Department, Hanusch Hospital, Vienna, Austria; 11Department of Nephrology and Dialysis, Medical University of Vienna, Austria; 125th Medical Department for Hematology and Oncology, Hospital Hietzing, Vienna, Austria; 13Department of Biomedicine, University Children's Hospital Basel, Switzerland; 14Ludwig Boltzmann Institute for Cancer Research, Vienna, Austria; 15Department of Biomedical Science, Institute of Animal Breeding and Genetics, University of Veterinary Medicine, Vienna, Austria; 16Hematology Biology, University Hospital, Nantes, France; 17Institute of Pathology, Ludwig-Maximilian University, Munich, Germany and 18Imagine Institute, INSERM U 1163, CNRS 8654, Université Paris Descartes, Sorbonne, Paris Cité, France 1

Correspondence: ABSTRACT

athological erythropoiesis with consequent anemia is a leading cause of symptomatic morbidity in internal medicine. The etiologies of anemia are complex and include reactive as well as neoplastic conditions. Clonal expansion of erythroid cells in the bone marrow may result in peripheral erythrocytosis and polycythemia but can also result in anemia when clonal cells are dysplastic and have a maturation arrest that leads to apoptosis and hinders migration, a constellation typically seen in the myelodysplastic syndromes. Rarely, clonal expansion of immature erythroid blasts results in a clinical picture resembling erythroid leukemia. Although several mechanisms underlying normal and abnormal erythropoiesis and the pathogenesis of related disorders have been deciphered in recent years, little is known about specific markers and targets through which prognosis and therapy could be improved in anemic or polycythemic patients. In order to discuss new markers, targets and novel therapeutic approaches in erythroid disorders and the related pathologies, a workshop was organized in Vienna in April 2017. The outcomes of this workshop are summarized in this review, which includes a discussion of new diagnostic and prognostic markers, the updated WHO classification, and an overview of new drugs used to stimulate or to interfere with erythropoiesis in various neoplastic and reactive conditions. The use and usefulness of established and novel erythropoiesisstimulating agents for various indications, including myelodysplastic syndromes and other neoplasms, are also discussed.

haematologica | 2018; 103(10)

peter.valent@meduniwien.ac.at or ohermine@gmail.com Received: March 6, 2018. Accepted: May 30, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.192518 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1593 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1593

P. Valent et al.

Introduction Erythropoiesis is one of the important physiological supply functions of the bone marrow. In healthy adults, about 200x109 red cells are produced per day in the bone marrow and are released into the peripheral blood.1 Depending on demand, red cell production can be adjusted and upregulated substantially. A complex network of oxygen sensors, cytokines, such as erythropoietin, and other factors, including regulators of iron metabolism, are involved in the control of steady-state and stress-induced erythropoiesis, thereby ensuring appropriate oxygen supply to the peripheral tissues.2-5 This regulatory network can adjust itself to physiological requirements such as the oxygen concentration (altitude) or pregnancy as well as to pathological conditions such as blood loss.2-5 However, in some pathological conditions this regulatory network is overwhelmed or is not functional, resulting in polycythemia or anemia. In the elderly, the bone marrow and other organs undergo aging. As a result, erythropoietin synthesis and red cell production may decline.6-9 However, even in very old individuals, red cell production and erythropoietin synthesis are usually adequate to keep hemoglobin levels within a reasonable range unless certain co-morbidities that lead to insufficient production of red cells have been acquired.6-9 Therefore, such other etiologies must be ruled out when hemoglobin levels drop in older individuals.7-9 Anemia is a major cause of symptomatic morbidity in daily medical practice. The etiologies contributing to anemic states are complex.6-14 Underlying disorders include trauma or coagulopathies with consequent bleeding, immunological and other inflammatory reactions as well as clonal (neoplastic) conditions. Clonal expansion of erythroid progenitor cells in the bone marrow may result in central and peripheral expansion of erythropoiesis, and thus in the clinical picture of polycythemia vera (PV), but it may also result in clonal anemic states such as the myelodysplastic syndromes (MDS) or even erythroid leukemia, characterized by a major or complete block of differentiation in early erythropoiesis.15-18 Anemic MDS are characterized by red cell dysplasia, ineffective erythropoiesis, apoptosis of late erythroid precursor cells, and the paradoxical combination of erythroid bone marrow hyperplasia and peripheral anemia. Although several mechanisms underlying normal and pathological red cell production in the bone marrow have been identified in recent years, little is known about disease-related markers and targets through which prognostication and therapy may be improved in anemic or polycythemic patients. In order to discuss novel markers, targets and mechanisms as well as new therapeutic approaches and strategies in various erythroid disorders, a workshop was organized in Vienna in April 2017 (April 28-29). The discussion in this meeting focused on pathological (neoplastic) erythropoiesis in adults. The outcomes of this workshop are summarized in this review.

Erythropoiesis Molecular mechanisms controlling erythropoiesis in health and disease A network of interconnected physiological communication networks and pathways are responsible for the pro1594

duction, distribution and turnover of red blood cells in healthy individuals (Figure 1).2-5,19-26 These networks keep the hemoglobin concentrations at a remarkably stable level throughout lifetime. Erythropoiesis starts in the bone marrow with lineage commitment of pluripotent myeloid progenitor cells and differentiation of these cells into immature erythroid progenitors that retain a certain proliferative capacity. Subsequently, these progenitor cells undergo further differentiation and maturation. A complex network of transcription factors and epigenetic regulators orchestrates this process.22-31 GATA-1 is the main regulator of lineage commitment, differentiation and survival of erythroid progenitors (Figure 1).20,22,27-30 In particular, GATA-1 triggers erythropoiesis by regulating the transcription of several erythroid differentiation-related genes, including genes involved in heme and/or globin synthesis, glycophorins, anti-apoptotic genes of the BH-3 family, genes involved in cell cycle regulation, and the gene for the erythropoietin receptor (EPOR).20,22,28 Major molecular players in these networks include, among others, classical hormones (thyroid hormones, androgens, corticosteroids, activin/inhibin and others), vitamins (e.g. vitamin B12 and folic acid), iron, regulators of iron metabolism such as the transferrin receptors-1 and -2, and early acting hematopoietic growth factors such as stem cell factor and interleukin-3 (Figure 1A).20-33 The main cytokine regulator of red cell production is erythropoietin, which acts at the level of late erythroid progenitors through a homodimeric receptor that triggers JAK2 kinase activity and subsequently STAT5 activation (Online Supplementary Figure S1). Erythropoietin acts mainly on myeloid precursor cells to ensure survival, thereby allowing the erythroid differentiation program, induced mainly by GATA-1, to occur (Figure 1A).2-5 It has been suggested that erythroid precursors in the bone marrow exhibit differential sensitivity against erythropoietin. The less sensitive cells undergo apoptosis upon caspase activation when the erythropoietin level is low, whereas at higher erythropoietin levels, most cells will survive and differentiate (Figure 1). In addition, in the bone marrow, within erythroid blood islands, late erythroid precursor cells express FAS ligand which may interact with early erythroid precursors that express FAS, resulting in caspase activation, which in turn triggers apoptosis and maturation arrest (Figure 1). In the case of a substantial need to produce more erythroid cells (e.g. after blood loss by bleeding or hemolysis), erythropoietin counteracts this activation allowing the cells to survive and differentiate, even if late erythroid precursors are abundant. In erythropoiesis, caspases, upon activation, cleave not only their main natural targets in the nucleus, but also GATA-1, which amplifies cell death and blocks erythroid differentiation (Figure 1).30 Erythropoietin synthesis is regulated in peritubular cells of the kidney by the transcription factor hypoxia-inducible factor-Îą, the main regulator of transcription of the gene encoding for erythropoietin (EPO). The stability of hypoxia-inducible factor-Îą depends on the prolyl-hydroxylase enzyme and the concentration of oxygen.32 To ensure a high level of proliferation and hemoglobin synthesis, iron absorption and the availability of iron for erythroid cells are tightly regulated by a number of modulators of iron metabolism and iron uptake. The latter include transferrin and its receptors (TfR-1 and TfR-2), as well as hepcidin and its target ferroportin, which allow iron export from various cell types, including haematologica | 2018; 103(10)

Normal and pathological erythropoiesis

Figure 1. Major pathways and molecules involved in the regulation of erythropoiesis. (A) Development of erythropoietic progenitor cells and erythroblasts. Early stages of erythropoietic development include primitive (p) and more mature (m) burst-forming unit erythroid cells (BFU-E) and colony-forming unit erythroid cells (CFUE). The differentiation and maturation of these cells are regulated by broadly acting hematopoietic cytokines, including stem cell factor (SCF) and interleukin-3 (IL-3) and their receptors (R), the SCF receptor KIT and IL-3-R. Later stages are primarily regulated by erythropoietin (EPO) and EPO-R and are dependent on iron-metabolism and the interaction between the death receptor FAS and its ligand (FAS-L). Transferrin (Tf) and Tf-receptor-1 (TfR-1) are additional major regulators of erythropoiesis. Moreover, growth differentiating factor 11 (GDF11) and polymeric immunoglobulin A (IgA) are considered to be involved in the regulation of certain stages of erythropoiesis. Later stages of erythropoiesis include proerythroblasts (ProEbl), basophilic (Baso) erythroblasts (type I and II), polychromatic (PolyC) erythroblasts and acidophilic (Acido) erythroblasts, also called late erythroblasts (Barbara Bain, personal communication to MCB). FAS-L and GDF11 are involved in the final maturation stage that leads to the generation of red cells (RC). (B) Erythroid blood island: macrophage surrounded by immature erythroblasts expressing FAS and more mature erythroblasts expressing FAS-L. The cell unit (island) acts in concert to promote erythroid differentiation and red cell production and maturation (rectangle). (C) Caspase activation during terminal erythroid differentiation: BID-dependent activation of caspase occurs in mitochondria. Various caspase targets are affected, including Rock-1, Lamin B and Acinus. However, GATA-1 is protected from caspase cleavage by heat shock protein 70 (HSP70). (D) Model of terminal erythroid differentiation and apoptosis regulated by the nuclear localization of HSP70. SCF and EPO trigger the proliferation and differentiation of erythroid progenitor cells (EPC). In RC precursors (CFU-E through erythroblasts) EPO induces maturation as HSP70 translocates into the nucleus to protect GATA-1 from caspase-induced degradation. In the absence of EPO, caspase-3 induces the cleavage of GATA-1 as HSP70 cannot translocate to the nucleus, and, as a result, apoptosis occurs.

macrophages and enterocytes, and thereby make iron available to erythroid progenitor cells.33 Erythroferone, which is synthesized by early and late erythroid progenitor cells, has recently been described as a major regulator of hepcidin synthesis in the liver.34 In addition to its role in iron uptake, transferrin receptor-1 is also able to trigger the MAP kinase and phosphoinositide 3 kinase/AKT pathways induced by erythropoietin and may thus participate in the regulation of erythropoiesis.35 In the model of regulation of red cell production by the modulation of erythropoietin sensitivity and rate of apoptosis of erythroid progenitor and precursor cells, caspases are considered to be key enzymes (Figure 1).27-31 To add to haematologica | 2018; 103(10)

the complexity of the system, it has been shown that caspases are also activated during erythroid differentiation and that this activation is absolutely necessary to prepare the cells for enucleation and thus formation of mature red cells (Figure 1C,D). In this process, GATA-1 is not cleaved by caspases, ensuring that erythroid maturation is preserved. To explain this apparent paradox it has been shown that the chaperone heat shock protein 70 (HSP70) plays an essential role in protecting GATA-1 from caspase cleavage (Figure 1C,D).25-31 Indeed, during erythroid differentiation HSP70 enters the nucleus at the time of caspase activation, and, at this level, interacts with GATA-1 to protect it against cleavage. By contrast, during apoptosis 1595

P. Valent et al.

induced by erythropoietin deprivation, HSP70 is exported from the nucleus and allows GATA-1 cleavage in erythroid progenitor cells (Figure 1D).25-31 This model, in which the fate of erythroid precursors is determined by the localization of HSP70 in the nucleus, has been shown to be altered in ineffective erythropoiesis in various anemic conditions such as b-thalassemia, MDS,36 and congenital erythroblastopenia,37 and may provide a new conceptual basis to improve anemia in these diseases.31 Apart from these established regulators of erythropoiesis, several novel molecular players regulating erythropoiesis were discussed in the workshop. Among these are members of the BCL-2 family such as BID which may play a critical role in the regulation of mitochondrial depolarization and caspase activation during erythroid differentiation and apoptosis (Figure 1C), various transcription factors, including STAT5A and STAT5B (Table 1), cytokines such as growth differentiating factor 11 (GDF11), and regulators of iron uptake and iron metabolism.33-48 GDF11, a ligand of activin receptor IIA (ActRIIA) that accumulates in erythroblasts in patients with b-thalassemia, has been implicated in the pathogenesis of anemia in these patients.42 It has also been described that polymeric immunoglobulin A (IgA) produced in the bone marrow may bind to the transferrin receptor-1 to sensitize erythroid cells to erythropoietin (Figure 1A).35 Consistent with this role as a regulator of erythropoiesis, the synthesis of polymeric IgA is increased during hypoxia. In pathological conditions, patients with IgA deficiency have higher levels of erythropoietin and, conversely, patients with unexplained polycythemia associated with an excess of polymeric IgA synthesis have recently been reported.35 Finally, a number of regulators of red cell membrane stability have recently been identified. One of these regulators may be CDK6, a cell cycle regulator that has recently been shown to serve as a membrane stabilizer of red cells in mice.49

Diagnostic evaluation of erythropoiesis: established and novel markers In most anemic patients the underlying etiology can be identified rapidly through the information gained from a thorough case history and detailed laboratory investigations. A number of different etiologies can underlie anemia, including iron deficiency, chronic inflammation, hemolysis, renal disorders or vitamin B12 deficiency. In each case, it is important to follow the principle diagnostic algorithms, to establish the correct diagnosis and to treat the underlying disorder (for example a gastrointestinal disease). In most of these cases, investigation of the bone marrow is not required. In other patients, however, the etiology remains uncertain after initial studies, so that detailed investigations of the bone marrow have to be performed. These investigations include a thorough cytological, histological and immuno-histochemical examination of the bone marrow, multi-parameter flow cytometry studies, a detailed examination of chromosomes (metaphases by conventional karyotyping and interphases by fluorescence in situ hybridization) and molecular studies.9-11,50-52 Depending on blood counts and other parameters, fluorescence in situ hybridization studies are applied to screen for the presence of MDS-related abnormalities. Our faculty recommends that in each case morphological and immunophenotypic studies, including immunohistochemistry and flow cytometry, should be employed to 1596

Table 1. Selected regulators of erythropoiesis.

Type of Regulator

Major Examples

Growth factors for multipotent and early erythropoietic progenitor cells Later-acting erythropoietic differentiation factors Essential transcription factors Important survival factors for erythropoietic cells Negative growth regulators of erythropoietic progenitor cells Essential vitamins and trace elements Iron and proteins involved in iron distribution and iron metabolism

SCF, G-CSF, IL-3

EPO, TGF-beta, GDF11, Activin A GATA-1, STAT5A, STAT5B MCL-1, BCL-xL, HSP70* Inhibin, TGF-beta, BID, FAS ligand, FAS, Caspases* Vitamin B12, Folic acid, Copper, others Ferritin, Transferrin, Transferrin receptor (CD71) Ferroportin, Hepcidin

SCF: stem cell factor; G-CSF: granulocyte colony-stimulating factor; EPO: erythropoietin; TGFbeta: transforming growth factor beta; GDF11: growth differentiation factor 11, also known as bone morphogenetic protein 11; HSP70: heat shock protein 70. *HSP70 prevents caspaseinduced cleavage of GATA1 in erythropoietic progenitor/precursor cells.

Table 2. Markers used to identify and quantify erythropoietic cells in the bone marrow.

Ery-PC

Ery

Specificity for erythropoietic cells

Flow cytometry CD36/ GP4 CD235a / Glycophorin A CD71 / TfR-1 E-Cadherin CD105 / endoglin

++ ++ ++ ++ ++

+ +++ ++ + +/-

++ +* -

Immunohistochemistry CD235a / Glycophorin A CD236 / Glycophorin C** Hemoglobin A CD71 / TfR-1 E-Cadherin* CD105 / endoglin

++ ++ +/++ ++ ++

+++ ++ ++ ++ + +/-

++ ++ ++ +/+* -

Marker

*Within the hematopoietic cell system, expression of E-cadherin is rather specific for red cells and their progenitors. **In immature erythroid leukemia, blast cells may stain negative for glycophorin A, but are still positive for glycophorin C and E-cadherin. Ery-PC: erythroid progenitor cells; Ery: erythrocytes; GP4: platelet plycoprotein-4; TfR-1: transferrin receptor-1.

define the percentage and stage of maturation of erythroid cells in bone marrow samples.52 For routine diagnostics, recommended immunohistochemical markers are glycophorin A, CD71 and E-cadherin (Table 2). Additional immunohistochemical markers include glycophorin C and hemoglobin A. E-cadherin is of special value for the detection of immature erythropoietic progenitor cells in patients with MDS and those with erythroleukemia (Figure 2). In contrast, hemoglobin A is preferentially expressed in more mature erythroblasts, and may therefore be helpful in distinguishing between more mature and more immature cells. In flow cytometry studies, recommended markers are haematologica | 2018; 103(10)

Normal and pathological erythropoiesis

Figure 2. Immunophenotypic visualization of leukemic erythroblasts. Bone marrow sections of a patient with erythroid leukemia (acute erythroleukemia) stained with (A) Wright-Giemsa solution and antibodies against (B) glycophorin A and (C) E-cadherin. Note that virtually all leukemic erythroblasts co-express abundant amounts of glycophorin A and E-cadherin, thereby confirming the immature stage of maturation of leukemic (erythroid) cells.

glycophorin A, CD71, and CD105 (Table 2).53-56 Among these, CD105 is of special value for the detection of immature erythropoietic progenitors in the bone marrow of patients with MDS. E-cadherin is also expressed specifically on the surface of erythropoietic progenitor cells in human bone marrow.57 However, E-cadherin is not used routinely as an erythroid marker in flow cytometry studies. Our faculty is of the opinion that CD105 and E-cadherin should be validated further as routine diagnostic markers in erythroid disorders of the bone marrow. Another interesting aspect is that several cell surface antigens appear to be downregulated on erythroid progenitor cells in MDS patients.58 Among these antigens, the coxsackie-adenovirus receptor (CAR) is a rather specific antigen: in fact, a decrease or lack of this receptor on erythroid progenitor cells is typically seen in patients with MDS and in other bone marrow neoplasms accompanied by marked dysplasia.58 Immunophenotypic studies are also of great importance to identify immature stages of erythropoiesis in patients with acute (myeloid/erythroid) leukemia in whom morphological and molecular studies alone are insufficient to establish a correct diagnosis (Figure 2 and Online Supplementary Figure S2). In these cases the use of immunohistochemical studies and multiparameter flow cytometry is essential. Another important aspect is that the physiological development and maturation of red cells in the bone marrow is regulated by the supporting microenvironment (consisting of macrophages and stromal cells) in so-called erythroid islands.59 These islands, also known as erythroblastic islands, are considered to represent functional units and are detectable by conventional stains and specific immunological stains (see above) on bone marrow biopsies. With age, the number of erythroid islands in the bone marrow decreases, while their size increases, which may suggest that the decrease in stem cell numbers and erythroid progenitors during aging might be compensated by an increased proliferation of local erythroid progenitors (Figure 3).60 In patients with MDS, impaired formation of erythroid islands as well as structural abnormalities within these islands have also been described (Figure 3).60 In lowrisk MDS, abnormal formation of erythroid islands may be the only histopathological change found in the bone marrow. Moreover, it has been described that erythroid island density correlates inversely with overall survival in patients with MDS.60 However, alterations of island conhaematologica | 2018; 103(10)

figuration and size (e.g. increased size) can also be detected when an increase in red cell production is required in pathological conditions, such as in hemolytic anemia or acute blood loss. In patients with aplastic anemia and other bone marrow failure syndromes, only a few or no erythroid islands are found in pathological examinations, which we consider to be relevant diagnostically. There are a number of established peripheral parameters through which red cell production can be assessed quantitatively. The most widely applied approach is to measure the numbers of reticulocytes in the peripheral blood. Another approach is to quantify the numbers of circulating erythroid progenitor cells, including multi-lineage colonyforming progenitors (CFU-GEMM) and burst-forming units (BFU-E). With this approach, clonal cytopenias (MDS) and aplastic anemia (in both conditions, CFUGEMM and BFU-E are markedly reduced or absent) are separable from ‘reactive’ and ‘deficiency’ anemias in which BFU-E and CFU-GEMM are usually within normal ranges.61-63 Even in MDS patients with 5q-, who may develop thrombocytosis, erythropoiesis and BFU-E growth are usually suppressed. Factor (erythropoietin)-independent and erythropoietin-hyperresponsive growth of BFU-E is indicative of (almost diagnostic for) the presence of PV or a related myeloproliferative neoplasm (MPN).64,65 However, the colony-assay is rather tricky and its success depends on the experience of the team performing the test. Thus, although this assay is of practical value (and was previously used as a criterion for PV), it is nowadays only employed routinely in a few specialized centers.

Anemia Classification of anemia: novel aspects and etiologies In general, anemia can be classified based on the regenerative capacity of the bone marrow (hypo- or hyperregenerative), the red cell volume (micro-, normo- or macrocytic), the etiology (bleeding, deficiency-induced, hemolytic, renal, inflammation-related, neoplastic, aplastic, others) and the dynamics with which anemia develops (acute, chronic).9-11 For each specific form of anemia, an extensive amount of published data has accumulated during the past few decades. A detailed description is beyond the scope of this article. There are also special variants of anemia that develop typically in the context of certain 1597

P. Valent et al.

physiological conditions such as pregnancy or advanced age.8,9,66,67 In these cases, the primary question is whether anemia is indicative of an undetected pathology and whether and when (at what thresholds) anemia can indeed be diagnosed. For example, in pregnant women, a hemoglobin level of 11.0 g/dL is still considered to be within the normal range by the World Health Organization (WHO). In elderly individuals, however, any decrease of hemoglobin below â&#x20AC;&#x2DC;normalâ&#x20AC;&#x2122; is considered an anemia.8,9 On the other hand, there is an ongoing discussion about the definition of anemia in the elderly and related hemoglobin thresholds.9,68-70 When no evidence of an underlying condition is found after a thorough investigation of all relevant organs and potential etiologies, the condition is termed cytopenia (anemia) of unknown significance (ICUS-A).50,52 Thus, the diagnosis of ICUS-A implies that a detailed investigation of the bone marrow, with histological, morphological (cytological), immunophenotypic, cytogenetic and molecular studies, was performed without conclusive evidence of the presence of MDS or any other underlying bone marrow neoplasm.50,52 In this regard it is worth noting that next-gener-

ation sequencing may sometimes reveal more or less specific mutations which may lead to re-classification of ICUS-A cases into either clonal cytopenia of undetermined significance or low-risk MDS, respectively.50-52 Other differential diagnoses to ICUS-A include anemia of chronic disease (inflammation-associated anemia), hemodilution, renal anemia, copper deficiency and vitamin B12 deficiency.9-11 In some elderly patients with ICUSA, inadequately low levels of erythropoietin are found even though the excretory function of the kidney is normal.71,72 In such cases, the aged kidney may be responsible for low erythropoietin production. Another equally important contributor to low red cell and erythropoietin production in elderly (otherwise healthy) people may be an age-related decrease in the production of hypophyseal and other essential hormones, resulting in a decreased supply of testosterone.73 However, unless an additional pathology (co-morbidity) is also present, these individuals have only mild anemia and are free of symptoms. As mentioned, it remains uncertain as to whether all these elderly individuals should indeed be diagnosed as having overt anemia.

1598

Figure 3. Structure and size of erythroid islands in the bone marrow. (A) Erythroid islands in the bone marrow of a 16-year old healthy male visualized by staining for CD71. (B) Erythroid islands in the bone marrow of an 82-year old female without bone marrow neoplasm. Erythroid islands were visualized by staining against hemoglobin A. Note the decreased number and increased size of erythroid islands in the bone marrow of the older healthy control. (C) Erythroid islands in the bone marrow of a 73-year old male patient with myelodysplastic syndrome with excess of blasts (59% of marrow cells, MDS-EB-1) visualized by staining for CD71. In the right images of 3A, 3B and 3C, erythroid islands are evidenced by pink circles. Note the decreased number of erythroid islands in this patient. (D, left panel) Bone marrow section of a 78-year old male with myelodysplastic syndrome with excess of blasts (10-19% of marrow cells, MDS-EB-2) stained for CD71. In this patient, numerous confluent, partially disrupted and poorly separable erythroblastic islands are seen. (D, right panel) Bone marrow section of an adult patient with hemolytic anemia. Note that erythroid islands are increased, but are clearly separable and have a regular shape (contrasting with MDS). Original magnifications: A, C, D: x 125; B: x 250.

haematologica | 2018; 103(10)

Normal and pathological erythropoiesis

Refractory anemias: the updated World Health Organization classification and novel molecular markers and targets used to grade and classify the disease In the current WHO proposal (2017) the term refractory anemia has been removed and replaced by the term MDS with an additional explanatory appendix to define the number of lineage(s) involved with dysplasia (example: MDS with single lineage dysplasia).74-76 It is important to note that MDS with single lineage dysplasia can be separated into anemic, neutropenic and thrombocytopenic types. Table 3 shows the current WHO proposal to classify MDS. Another important aspect is that only a few cytogenetic and molecular markers are used to define the subtype of MDS. For example, the 5q- syndrome is recognized as a separate entity by the WHO (MDS with isolated 5q-) and is associated with particular sensitivity to lenalidomide. It has also been described that the SF3B1 mutation is highly indicative of the presence of ring sideroblasts.74-76 Therefore, this mutation, when present, serves as a diagnostic feature. In particular, MDS variants with ring sideroblasts can be diagnosed when ring sideroblasts account for either ≥15% of all red cells (old definition) or ≥5% in the presence of a SF3B1 mutation (new adjunct).74-76 For the morphologist, this definition implies that more red cells have to be counted in the iron stain (at least 100 red cells) to arrive at a correct diagnosis.52 This individualization might be of importance because patients with this disorder usually do not respond to erythropoietin but may benefit from specific treatment, such as a GDF11 inhibitor. Another relevant change in the current WHO classification (update 2017) relates to the definition of erythroleukemia. Details are discussed below under the section ‘Red cell neoplasms’. Based on the changed definition, a new subset of poor-risk patients with MDS, formerly diagnosed as having erythroid leukemia or M6 acute myeloid leukemia (AML) according to the French-American-British classification, will require attention.74-76 In fact, these patients often behave like patients with AML, are usually resistant to cytoreductive agents and chemotherapy, and are characterized by predominant erythropoiesis (>50% of bone marrow cells). The poor prognosis of these patients is also reflected by their poor-risk cytogenetics, often in the form of a complex karyotype. If possible, debulking with poly-

Table 3. Updated WHO classification of myelodysplastic syndromes, 2017.

Variant

Abbreviation

MDS with single lineage dysplasia MDS-SLD MDS with ring sideroblasts MDS-RS MDS with ring sideroblasts and single lineage dysplasia MDS-RS-SLD MDS with ring sideroblasts and multilineage dysplasia MDS-RS-MLD MDS with multilineage dysplasia MDS-MLD MDS with excess blasts MDS-EB MDS with isolated del(5q) MDS unclassifiable MDS-U Provisional entity: refractory cytopenia of childhood -

chemotherapy (regimens used for AML) or hypomethylating agents should be considered, and in young and fit patients, intensive post-debulking therapy, preferably in the form of hematopoietic stem cell transplantation, (HSCT) is often recommended.77-79

Treatment of anemias: novel agents and concepts During the workshop, standard treatments and novel therapeutic approaches for various forms of anemias were discussed. With regard to erythropoietin, standard indications remain renal anemia and anemic patients with MDS in whom a relative erythropoietin deficiency has been documented and a response to erythropoietin is seen.80-84 For other similar indications, such as anemia with low endogenous erythropoietin levels accompanying myelomas, lymphomas or chronic leukemias, treatment with erythropoiesis-stimulating agents (ESA) may also be useful.85,86 Even in some form of ICUS (elderly ICUS-A patients) or anemia of chronic disease, the use of erythropoietin may be justified.87,88 However, our faculty is of the opinion that in all these indications, ‘inadequate’ (insufficient) production of erythropoietin should be documented before starting erythropoietin therapy. Moreover, we are of the opinion that erythropoietin therapy is no longer justified for patients with solid tumors or other malignant disorders after chemotherapy (in certain malignancies tumor cells may express erythropoietin receptor) and should no longer be considered for patients with elevated endogenous erythropoietin levels (especially levels >500 U/L). However, there are other potential indications for erythropoietin and other ESA, such as trauma or certain neurological disorders.89-91 Although beneficial effects of erythropoietin for these indications have been documented, the mechanisms contributing to these beneficial effects have not been elucidated so far. With regard to the type of erythropoietin preparation or erythropoietin-like ESA, no major differences in responses have been reported. In fact, several meta-analyses of patients with MDS and other indications have shown that short-acting and long-acting erythropoietin-based drugs all exert very similar effects.92-94 However, erythropoietin therapy should always be administered with caution, especially when hemoglobin levels rise. In fact, it has been described that hemoglobin levels above 12 g/dL can even be associated with complications and a poor outcome

Table 4. Overview of red cell neoplasms.

Neoplasm

Key Features / Criteria

Myelodysplastic syndrome with erythroid predominance

Bone marrow cell dysplasia >50% erythroid cells and myeloblasts <20% Proerythroblasts ≥30% and >80% of all BM cells are erythroid cells; myeloblasts <20% JAK2 V617F or CALR mutations and myeloblasts <20% and WHO criteria for PV fulfilled

Pure erythroid leukemia

Polycythemia vera

WHO: World Health Organization; BM: bone marrow: PV: polycythemia very; JAK2: janus kinase 2; CALR: calreticulin.

WHO: World Health Organization; MDS: myelodysplastic syndrome.

haematologica | 2018; 103(10)

1599

P. Valent et al.

compared to hemoglobin levels below 12 g/dL, especially in patients with chronic kidney disease.95 The target hemoglobin level should, therefore, be around 11 g/dL independently of the ESA applied, underlying disease, or age. For some indications, the use of additional drugs, together with erythropoietin/ESA, may be useful. For example, in patients with chronic kidney disease or chronic inflammation, addition of iron may lead to a more rapid correction of anemia.96-98 In MDS, the addition of granulocyte or granulocyte-macrophage colony-stimulating factor may work and increase the rate of responding patients.83,84 In other patients, low erythropoietin synthesis may be accompanied by other insufficiencies that need to be corrected, such as a lack of vitamin B12, iron deficiency or folate deficiency. There are also other (in part novel) emerging agents that may be useful in patients with anemias. These drugs include, among others, hypoxia-inducible factor stabilizing compounds, prolyl hydroxylase inhibitors, and activin/GDF11-trapping agents such as sotatercept or luspatercept.97-105 The latter may even work in bone marrow failure syndromes. In addition, GDF11 blockers are able to counteract anemia in patients with b-thalassemia.42,101 However, in advanced bone marrow neoplasms, such as MDS, or in aplastic anemia, the effectiveness of broadly acting growth factors, ESA and other drugs (mentioned above) is limited. In these patients, the use of diseasemodifying agents, such as demethylating agents and/or chemotherapy in MDS or immunosuppressive therapy in aplastic anemia may lead to an improvement or even correction of the anemia. In low-risk MDS patients with isolated 5q-, lenalidomide therapy usually works well and often leads to transfusion independence or even to a cytogenetic response. In patients with paroxysmal nocturnal hemoglobinuria the use of complement-targeting drugs has revolutionized the treatment of anemia. Finally, several new agents are currently being tested in patients with hemolytic anemia, thalassemia, or anemia of chronic inflammation. A detailed description of these agents and of new treatment approaches is beyond the scope of this article - we refer the interested reader to the available literature.

of ≥20% and erythroid predominance, the final diagnosis is AML [AML not otherwise specified (NOS), acute erythroid leukemia of erythroid/myeloid type] unless molecular or cytogenetic markers identify another AML subtype.74-76 In those with >80% immature erythroid precursors, ≥30% proerythroblasts and a total blast cell percentage <20%, the final diagnosis is acute erythroid leukemia (AML, pure erythroid type). In an updated revision, AML NOS, acute erythroid leukemia (erythroid/myeloid type) was replaced by AML NOS (erythroid subtype).106 Overall, the erythroid leukemias appear to comprise a heterogeneous group of malignancies. For example, erythroid leukemias can occur as primary (de novo) leukemia or as a secondary form of leukemia, for instance, following MDS or MPN (Online Supplementary Table S1). In rare cases, even the blast phase of Philadelphia chromosome-positive chronic myeloid leukemia may have a clinical and pathological picture indistinguishable from that of erythroleukemia (Online Supplementary Figure S2). It is also important to note that erythroleukemia can develop as an acute and rapidly progressive disease or a more chronic form of leukemia. The acute forms of erythroid leukemias represent a diagnostic challenge and may be overlooked. Rarely, acute erythroleukemia presents as a large-cell anaplastic neoplasm mimicking a histiocytic malignancy or even small-cell carcinoma; such cases can also easily be overlooked unless appropriate phenotypic studies with antibodies against glycophorin A/C or CD71 are conducted (H-PH, unpublished observation). The chronic form of erythroleukemia must be distinguished from PV and reactive erythrocytosis (polyglobulia), for example, in cases with an erythropoietin-producing tumor. The diagnostic criteria for PV have also changed. Whereas expression of the JAK2 V617F mutation is still considered a major criterion of PV the threshold levels of hemoglobin were changed: in the 2008 WHO definition, cut-offs were 16.5 g/dL for females and 18.5 g/dL for males, while in the 2017 revision, hemoglobin cut-offs were 16.0 g/dL for females and 16.5 g/dL for males.74,75 Although rather specific for MPN and often found in PV patients, CALR mutations are not yet regarded as diagnostic criteria of MPN/PV. A summary of erythroid neoplasms is provided in Table 4.

Red cell neoplasms

Molecular mechanisms regulating red cell neoplasms

Red cell neoplasms: classification and criteria Red cell neoplasms can be classified according to their etiology (de novo or secondary, i.e. following MDS or a mutagenic event), WHO criteria and the presence or absence of certain molecular markers. Erythroid neoplasms include PV (JAK2-mutated or wild-type JAK2), MDS with a prominent erythroid compartment (previously AML M6) and (pure) erythroid leukemia (Table 4). Based on the 2017 update of the WHO classification, erythroid neoplasms have been reclassified.74,75,106 In the previous definition provided by the French-American-British group and later by the WHO, a blast cell percentage of ≥20% in the non-erythroid compartment together with erythroid predominance (≥50% of nucleated bone marrow cells) was indicative of AML (AML M6). In the 2017 update of the WHO classification, these cases are reclassified as MDS (usually MDS with excess blasts) unless the total blast cell percentage (without subtracting erythroid cells) is ≥20%.74-76 In the case of a total blast cell percentage 1600

In many instances, the molecular mechanisms underlying red cell expansion in MPN or erythroid leukemias remain unknown. In the classical MPN, including PV, the JAK2 point mutation V617F and CALR mutations are considered to act as major disease drivers. One critical aspect is that these mutant forms initiate complex networks of signaling cascades that drive the affected cell and trigger growth factor independence. A detailed description of these networks is beyond the scope of this review. Some of the most important networks are shown in Online Supplementary Figure S1. The faculty also discussed novel preclinical models of red cell neoplasms. Based on recent molecular insights into the etiology of PV and erythroid leukemias, several mouse models have been established. In the field of PV/MPN these models are primarily based on the JAK2 mutation V617F and CALR mutations.107-110 Indeed, mice expressing a mutated and thus hyperactive Jak2 may develop a MPN-like condition over time.107-110 However, additional factors (mutations or signaling molecules) are required to convert the condition into a fullhaematologica | 2018; 103(10)

Normal and pathological erythropoiesis

blown malignancy. These additional anomalies are indeed found in patients with MPN/PV or secondary AML and are, therefore, of clinical significance.111-116 They include mutations in TP53 and in various driver genes.112-116 Several of these changes lead to hyperactive signaling in clinically relevant pro-oncogenic signaling networks. For example, molecular changes that lead to an increased production and accumulation of tyrosine-phosphorylated STAT5 (a critical JAK2-downstream transcription factor) can transform an indolent MPN-like condition into a highly fatal disease with major thromboembolism in mice (RM and PV, unpublished observation). In contrast to PV and other MPN, very little is known about the molecular mechanisms underlying the evolution and progression of erythroid-rich MDS and erythroid leukemia. In fact, despite a growing list of mutations associated with erythroid leukemia, their role in the initiation and/or maintenance of the erythroid phenotype remains largely unknown.115-118 Whereas earlier studies showed that particular viruses, such as the avian erythroblastosis virus and Friend spleen focus-forming virus, can induce neoplasms resembling erythroid leukemia in various animal models, no evidence of a viral etiology for the human disease has been identified so far. All in all, the faculty concluded that more research is needed to decipher molecular players and targets in these highly fatal neoplasms.

Therapy of red cell neoplasms Despite novel treatment options, early and advanced erythroid neoplasms are still regarded as incurable malignancies. Most PV patients can be managed quite well with phlebotomy. If phlebotomy alone is not sufficient to bring erythrocyte counts under control (target hematocrit: <45%) or thromboembolic events occur, additional cytoreductive therapy is recommended. Depending on age, risk factors, and expected adverse side effects, interferon-Îą or hydroxyurea can be prescribed. In highly symptomatic cases, ruxolitinib may be considered. For patients whose disease transforms into secondary AML, poly-chemotherapy or, in some cases, hypomethylating agents, plus HSCT should be considered. The same holds true for patients with advanced MDS and prominent erythropoiesis (previously AML M6) and cases with fullblown erythroid leukemia (fulfilling 2017 WHO criteria).77,78 In these patients, hypomethylating agents (5-

References 1. Palis J. Primitive and definitive erythropoiesis in mammals. Front Physiol. 2014;5:3. 2. Hattangadi SM, Wong P, Zhang L, Flygare J, Lodish HF. From stem cell to red cell: regulation of erythropoiesis at multiple levels by multiple proteins, RNAs, and chromatin modifications. Blood. 2011;118(24):62586268. 3. Kuhrt D, Wojchowski DM. Emerging EPO and EPO receptor regulators and signal transducers. Blood. 2015;125(23):3536-3541. 4. Liang R, Ghaffari S. Advances in understanding the mechanisms of erythropoiesis

haematologica | 2018; 103(10)

6. 7.

azacytidine or decitabine) or polychemotherapy (AML regimens) should be considered for debulking prior to HSCT. Unfortunately, however, HSCT can only be offered to a restricted group of young and fit patients. In all patients, the risk of HSCT-related mortality and morbidity must be weighed against the potential benefit. The probability of long-term disease-free survival after HSCT is probably in the same range as that seen in secondary AML following MDS. In some of these patients, hypomethylating agents may exert clinically meaningful anti-neoplastic effects.79 Especially in older patients or those who cannot tolerate intensive chemotherapy, treatment with hypomethylating agents is often recommended as first-line therapy. Hypomethylating agents can also be considered for patients who are refractory to or relapse after polychemotherapy. When hypomethylating agents fail and the patient is fit, polychemotherapy or experimental drugs should be considered. Hydroxyurea is the palliative drug of choice in multi-resistant patients and those who are old and refuse intensive or experimental therapy.

Concluding remarks Red cell production, survival and turnover in health and disease, and mechanisms regulating these processes, have attracted the attention of physicians and scientists in the past five decades and many different mechanisms and molecules involved in the regulation of red cell production and survival have been discovered. In addition, a number of molecular and immunological markers and targets have been identified in red cells and their progenitors. Several of these novel antigens may serve as diagnostic markers or as therapeutic targets. Resulting new diagnostic strategies and novel treatment concepts should advance the field and lead to more precise diagnosis, better therapies and improved prognosis in reactive and clonal erythroid disorders. Acknowledgments We would like to thank Heidi A. Neubauer, Julia NeusiedlerNicolas, Sophia-Marie Rammler and Emir Hadzijusufovic for their excellent technical assistance. This study was supported by the Austrian Science Fonds, SFB grants F4701-B20, F4704B20, F4705-B20 and F4706-B20.

in homeostasis and disease. Br J Haematol. 2016;174(5):661-673. Oburoglu L, Romano M, Taylor N, Kinet S. Metabolic regulation of hematopoietic stem cell commitment and erythroid differentiation. Curr Opin Hematol. 2016;23(3):198205. Carmel R. Anemia and aging: an overview of clinical, diagnostic and biological issues. Blood Rev. 2001;15(1):9-18. Steensma DP, Tefferi A. Anemia in the elderly: how should we define it, when does it matter, and what can be done? Mayo Clin Proc. 2007;82(8):958-966. Halawi R, Moukhadder H, Taher A. Anemia in the elderly: a consequence of aging? Expert Rev Hematol. 2017;10(4):327-335.

9. Stauder R, Valent P, Theurl I. Anemia at older age: etiologies, clinical implications and management. Blood. 2018;131(5):505514. 10. Guidi GC, Lechi Santonastaso C. Advancements in anemias related to chronic conditions. Clin Chem Lab Med. 2010;48(9): 1217-1226. 11. Valent P. Low blood counts: immune mediated, idiopathic, or myelodysplasia. Hematology Am Soc Hematol Educ Program. 2012;2012:485-491. 12. Guillaud C, Loustau V, Michel M. Hemolytic anemia in adults: main causes and diagnostic procedures. Expert Rev Hematol. 2012;5(2): 229-241. 13. Camaschella C, Nai A. Ineffective erythro-

1601

P. Valent et al.

14.

15.

16. 17. 18. 19.

20.

21. 22.

23. 24. 25.

26. 27.

28. 29.

30.

31.

32. 33. 34.

1602

poiesis and regulation of iron status in iron loading anaemias. Br J Haematol. 2016;172 (4):512-523. Shimamura A. Aplastic anemia and clonal evolution: germ line and somatic genetics. Hematology Am Soc Hematol Educ Program. 2016;2016(1):74-82. Hasserjian RP, Howard J, Wood A, Henry K, Bain B. Acute erythremic myelosis (true erythroleukaemia): a variant of AML FAB-M6. J Clin Pathol. 2001;54(3):205-209. Prchal JT. Polycythemia vera and other primary polycythemias. Curr Opin Hematol. 2005;12(2):112-116. Wang SA, Hasserjian RP. Erythroid proliferations in myeloid neoplasms. Hum Pathol. 2012;43(2):153-164. Wang W, Wang SA, Medeiros LJ, Khoury JD. Pure erythroid leukemia. Am J Hematol. 2017;92(3):292-296. Wu H, Liu X, Jaenisch R, Lodish HF. Generation of committed erythroid BFU-E and CFU-E progenitors does not require erythropoietin or the erythropoietin receptor. Cell. 1995;83(1):59-67. Cantor AB, Orkin SH. Transcriptional regulation of erythropoiesis: an affair involving multiple partners. Oncogene. 2002;21(21): 3368-3376. Donovan A, Andrews NC. The molecular regulation of iron metabolism. Hematol J. 2004;5(5):373-380. Kim SI, Bresnick EH. Transcriptional control of erythropoiesis: emerging mechanisms and principles. Oncogene. 2007;26(47):67776794. Semenza GL. Regulation of oxygen homeostasis by hypoxia-inducible factor 1. Physiology (Bethesda). 2009;24:97-106. Haase VH. Regulation of erythropoiesis by hypoxia-inducible factors. Blood Rev. 2013; 27(1):41-53. Hermine O, Arlet JB, Ribeil JA, Guillerm F, Vandekerkhove J, Courtois G. HSP70, an erythropoiesis regulator that determines the fate of erythroblasts between death and differentiation. Transfus Clin Biol. 2013;20(2): 144-147. Kautz L, Nemeth E. Molecular liaisons between erythropoiesis and iron metabolism. Blood. 2014;124(4):479-482. Simon MC, Pevny L, Wiles MV, Keller G, Costantini F, Orkin SH. Rescue of erythroid development in gene targeted GATA-1mouse embryonic stem cells. Nat Genet. 1992;1(2):92-98. Simon MC. Transcription factor GATA-1 and erythroid development. Proc Soc Exp Biol Med. 1993;202(2):115-121. Kaneko H, Shimizu R, Yamamoto M. GATA factor switching during erythroid differentiation. Curr Opin Hematol. 2010;17(3):163168. Ribeil JA, Zermati Y, Vandekerckhove J, et al. Hsp70 regulates erythropoiesis by preventing caspase-3-mediated cleavage of GATA-1. Nature. 2007;445(7123):102-105. Arlet JB, Ribeil JA, Guillem F, et al. HSP70 sequestration by free b-globin promotes ineffective erythropoiesis in b-thalassaemia. Nature. 2014;514(7521):242-246 Smith TG, Robbins PA, Ratcliffe PJ. The human side of hypoxia-inducible factor. Br J Haematol. 2008;141(3):325-334. Muckenthaler MU, Rivella S, Hentze MW, Galy B. A red carpet for iron metabolism. Cell. 2017;168(3):344-361. 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.

35. Coulon S, Dussiot M, Grapton D, et al. Polymeric IgA1 controls erythroblast proliferation and accelerates erythropoiesis recovery in anemia. Nat Med. 2011;17(11):14561465. 36. Frisan E, Vandekerckhove J, de Thonel A, et al. Defective nuclear localization of Hsp70 is associated with dyserythropoiesis and GATA-1 cleavage in myelodysplastic syndromes. Blood. 2012;119(6):1532-1542. 37. Gastou M, Rio S, Dussiot M, et al. The severe phenotype of Diamond-Blackfan anemia is modulated by heat shock protein 70. Blood Adv. 2017;1(22):1959-1976. 38. Maguer-Satta V, Bartholin L, Jeanpierre S, et al. Regulation of human erythropoiesis by activin A, BMP2, and BMP4, members of the TGFbeta family. Exp Cell Res. 2003;282 (2):110-120. 39. Socolovsky M, Murrell M, Liu Y, Pop R, Porpiglia E, Levchenko A. Negative autoregulation by FAS mediates robust fetal erythropoiesis. PLoS Biol. 2007;5(10):e252. 40. Liu Y, Pop R, Sadegh C, Brugnara C, Haase VH, Socolovsky M. Suppression of Fas-FasL coexpression by erythropoietin mediates erythroblast expansion during the erythropoietic stress response in vivo. Blood. 2006;108(1):123-133. 41. Wang Q, Huang Z, Xue H, et al. MicroRNA miR-24 inhibits erythropoiesis by targeting activin type I receptor ALK4. Blood. 2008;111(2):588-595. 42. Dussiot M, Maciel TT, Fricot A, et al. An activin receptor IIA ligand trap corrects ineffective erythropoiesis in b-thalassemia. Nat Med. 2014;20(4):398-407. 43. Koulnis M, Porpiglia E, Hidalgo D, Socolovsky M. Erythropoiesis: from molecular pathways to system properties. Adv Exp Med Biol. 2014;844:37-58. 44. Vlahakos DV, Marathias KP, Madias NE. The role of the renin-angiotensin system in the regulation of erythropoiesis. Am J Kidney Dis. 2010;56(3):558-565. 45. Kim YC, Mungunsukh O, Day RM. Erythropoietin regulation by angiotensin II. Vitam Horm. 2017;105:57-77. 46. Lee PL, Beutler E. Regulation of hepcidin and iron-overload disease. Annu Rev Pathol. 2009;4:489-515. 47. Silva B, Faustino P. An overview of molecular basis of iron metabolism regulation and the associated pathologies. Biochim Biophys Acta. 2015;1852(7):1347-1359. 48. Wang CY, Babitt JL. Hepcidin regulation in the anemia of inflammation. Curr Opin Hematol. 2016;23(3):189-197. 49. Uras IZ, Scheicher RM, Kollmann K, et al. Cdk6 contributes to cytoskeletal stability in erythroid cells. Haematologica. 2017;102 (6):995-1005. 50. Valent P, Horny HP, Bennett JM, et al. Definitions and standards in the diagnosis and treatment of the myelodysplastic syndromes: consensus statements and report from a working conference. Leuk Res. 2007;31(6):727-736. 51. Steensma DP, Bejar R, Jaiswal S, et al. Clonal hematopoiesis of indeterminate potential and its distinction from myelodysplastic syndromes. Blood. 2015;126(1):9-16. 52. Valent P, Orazi A, Steensma DP, et al. Proposed minimal diagnostic criteria for myelodysplastic syndromes (MDS) and potential pre-MDS conditions. Oncotarget 2017:8(43):73483-73500. 53. Malcovati L, Della Porta MG, Lunghi M, et al. Flow cytometry evaluation of erythroid and myeloid dysplasia in patients with myelodysplastic syndrome. Leukemia.

2005; 19(5):776-783. 54. Della Porta MG, Malcovati L, Invernizzi R, et al. Flow cytometry evaluation of erythroid dysplasia in patients with myelodysplastic syndrome. Leukemia. 2006;20(4):549-555. 55. Xu F, Wu L, He Q, Zhang Z, Chang C, Li X. Immunophenotypic analysis of erythroid dysplasia and its diagnostic application in myelodysplastic syndromes. Intern Med J. 2012;42(4):401-411. 56. Eidenschink Brodersen L, Menssen AJ, et al. Assessment of erythroid dysplasia by "difference from normal" in routine clinical flow cytometry workup. Cytometry B Clin Cytom. 2015;88(2):125-135. 57. Bühring HJ, Müller T, Herbst R, et al. The adhesion molecule E-cadherin and a surface antigen recognized by the antibody 9C4 are selectively expressed on erythroid cells of defined maturational stages. Leukemia. 1996;10(1):106-116. 58. Bauer K, Machherndl-Spandl S, Suessner S, et al. Identification of CAR as a novel mediator of erythroid differentiation and migration that is specifically downregulated in erythropoietic progenitor cells in patients with MDS. Blood. 2014;124(21):1570. 59. Chasis JA, Mohandas N. Erythroblastic islands: niches for erythropoiesis. Blood. 2008;112(3):470-478. 60. Buesche G, Teoman H, Giagounidis A, et al. Impaired formation of erythroblastic islands is associated with erythroid failure and poor prognosis in a significant proportion of patients with myelodysplastic syndromes. Haematologica. 2016;101(5):e177-e181. 61. Moriyama Y, Sato M, Kinoshita Y. Studies on hematopoietic stem cells: XI. Lack of erythroid burst-forming units (BFU-E) in patients with aplastic anemia. Am J Hematol. 1979;6(1):11-16. 62. Amato D, Khan NR. Erythroid burst formation in cultures of bone marrow and peripheral blood from patients with refractory anemia. Acta Haematol. 1983;70(1):1-10. 63. Geissler K, Hinterberger W, Jäger U, et al. Deficiency of pluripotent hemopoietic progenitor cells in myelodysplastic syndromes. Blut. 1988;57(1):45-49. 64. Manor D, Rachmilewitz EA, Fibach E. Improved method for diagnosis of polycythemia vera based on flow cytometric analysis of autonomous growth of erythroid precursors in liquid culture. Am J Hematol. 1997;54(1):47-52. 65. Geissler K, Ohler L, Födinger M, et al. Interleukin-10 inhibits erythropoietin-independent growth of erythroid bursts in patients with polycythemia vera. Blood. 1998;92(6):1967-1972. 66. Sifakis S, Pharmakides G. Anemia in pregnancy. Ann N Y Acad Sci. 2000;900:125-136. 67. Milman N. Postpartum anemia I: definition, prevalence, causes, and consequences. Ann Hematol. 2011;90(11):1247-1253. 68. Zakai NA, Katz R, Hirsch C, Shlipak MG, Chaves PH, Newman AB, Cushman M. A prospective study of anemia status, hemoglobin concentration, and mortality in an elderly cohort: the Cardiovascular Health Study. Arch Intern Med. 2005;165(19):22142220. 69. Chaves PH, Xue QL, Guralnik JM, Ferrucci L, Volpato S, Fried LP. What constitutes normal hemoglobin concentration in community-dwelling disabled older women? J Am Geriatr Soc. 2004;52(11):1811-1816. 70. Steensma DP, Tefferi A. Anemia in the elderly: how should we define it, when does it matter, and what can be done? Mayo Clin

haematologica | 2018; 103(10)

Normal and pathological erythropoiesis Proc. 2007;82(8):958-966. 71. Valent P, Horny HP. Minimal diagnostic criteria for myelodysplastic syndromes and separation from ICUS and IDUS: update and open questions. Eur J Clin Invest. 2009;39 (7):548-553. 72. Valent P. Anemia of the elderly (AOE): does it exist and does it matter in clinical practice? Eur J Clin Invest. 2008;38(10):782-783. 73. Bachman E, Travison TG, Basaria S, et al. Testosterone induces erythrocytosis via increased erythropoietin and suppressed hepcidin: evidence for a new erythropoietin/hemoglobin set point. J Gerontol A Biol Sci Med Sci. 2014;69(6):725-735. 74. Arber DA, Hasserjian RP. Reclassifying myelodysplastic syndromes: what's where in the new WHO and why. Hematology Am Soc Hematol Educ Program. 2015;2015:294298. 75. 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):2391-405. 76. Strupp C, Nachtkamp K, Hildebrandt B, et al. New proposals of the WHO working group (2016) for the diagnosis of myelodysplastic syndromes (MDS): characteristics of refined MDS types. Leuk Res. 2017;57:78-84. 77. Fouillard L, Labopin M, Gorin NC, et al; Acute Leukemia Working Party of the EBMT. Hematopoietic stem cell transplantation for de novo erythroleukemia: a study of the European Group for Blood and Marrow Transplantation (EBMT). Blood. 2002; 100(9):3135-3140. 78. Ishiyama K, Yamaguchi T, Eto T, et al. Acute megakaryoblastic leukemia, unlike acute erythroid leukemia, predicts an unfavorable outcome after allogeneic HSCT. Leuk Res. 2016;47:47-53. 79. Almeida AM, Prebet T, Itzykson R, et al. Clinical outcomes of 217 patients with acute erythroleukemia according to treatment type and line: a retrospective multinational study. Int J Mol Sci. 2017;18(4). 80. Schmid H, Schiffl H. Erythropoiesis stimulating agents and anaemia of end-stage renal disease. Cardiovasc Hematol Agents Med Chem. 2010;8(3):164-172. 81. Ramanath V, Gupta D, Jain J, Chaudhary K, Nistala R. Anemia and chronic kidney disease: making sense of the recent trials. Rev Recent Clin Trials. 2012;7(3):187-196. 82. Mimura I, Tanaka T, Nangaku M. How the target hemoglobin of renal anemia should be. Nephron. 2015;131(3):202-209. 83. Hellström-Lindberg E, Gulbrandsen N, Lindberg G, et al; Scandinavian MDS Group. A validated decision model for treating the anaemia of myelodysplastic syndromes with erythropoietin + granulocyte colonystimulating factor: significant effects on quality of life. Br J Haematol. 2003;120 (6):1037-1046. 84. Hellström-Lindberg E, van de Loosdrecht A. Erythropoiesis stimulating agents and other growth factors in low-risk MDS. Best Pract Res Clin Haematol. 2013;26(4):401-410. 85. Ludwig H, Fritz E, Kotzmann H, Höcker P, Gisslinger H, Barnas U. Erythropoietin treatment of anemia associated with multiple myeloma. N Engl J Med. 1990;322(24):16931699. 86. San Miguel JF, García-Sanz R. Recombinant human erythropoietin in the anaemia of

haematologica | 2018; 103(10)

multiple myeloma and non-Hodgkin's lymphoma. Med Oncol. 1998;15:S29-S34. 87. Agnihotri P, Telfer M, Butt Z, et al. Chronic anemia and fatigue in elderly patients: results of a randomized, double-blind, placebo-controlled, crossover exploratory study with epoetin alfa. J Am Geriatr Soc. 2007;55(10):1557-1565. 88. Agarwal N, Prchal JT. Erythropoietic agents and the elderly. Semin Hematol. 2008;45(4): 267-275. 89. Hasselblatt M, Ehrenreich H, Sirén AL. The brain erythropoietin system and its potential for therapeutic exploitation in brain disease. J Neurosurg Anesthesiol. 2006;18(2):132-138. 90. Westenbrink BD, Voors AA, Ruifrok WP, van Gilst WH, van Veldhuisen DJ. Therapeutic potential of erythropoietin in cardiovascular disease: erythropoiesis and beyond. Curr Heart Fail Rep. 2007;4(3):127-133. 91. Brines M. The therapeutic potential of erythropoiesis-stimulating agents for tissue protection: a tale of two receptors. Blood Purif. 2010;29(2):86-92. 92. Wilhelm-Leen ER, Winkelmayer WC. Mortality risk of darbepoetin alfa versus epoetin alfa in patients with CKD: systematic review and meta-analysis. Am J Kidney Dis. 2015;66(1):69-74. 93. Pérez-Ruixo JJ, Cucala-Ramos M, GarcíaGonzalo E, Del Val Romero B, Valveny N. Between subjects variability in haemoglobin and dose are not associated with the erythropoiesis-stimulating agent used to treat anaemia in dialysis: a meta-analysis. Br J Clin Pharmacol. 2013;75(1):15-25. 94. Park S, Fenaux P, Greenberg P, et al. Efficacy and safety of darbepoetin alpha in patients with myelodysplastic syndromes: a systematic review and meta-analysis. Br J Haematol. 2016;174(5):730-747. 95. Pfeffer MA, Burdmann EA, Chen CY, et al; TREAT Investigators. A trial of darbepoetin alfa in type 2 diabetes and chronic kidney disease. N Engl J Med. 2009;361(21):2019-2032. 96. Shirazian S, Grant C, Miller I, Fishbane S. How can erythropoeitin-stimulating agent use be reduced in chronic dialysis patients?: The use of iron supplementation to reduce ESA dosing in hemodialysis. Semin Dial. 2013;26(5):534-536. 97. Del Vecchio L, Locatelli F. New treatment approaches in chronic kidney disease-associated anaemia. Expert Opin Biol Ther. 2014;14(5):687-696. 98. Bonomini M, Del Vecchio L, Sirolli V, Locatelli F. New treatment approaches for the anemia of CKD. Am J Kidney Dis. 2016;67(1):133-142. 99. Macdougall IC, Ashenden M. Current and upcoming erythropoiesis-stimulating agents, iron products, and other novel anemia medications. Adv Chronic Kidney Dis. 2009;16(2):117-130. 100. Raje N, Vallet S. Sotatercept, a soluble activin receptor type 2A IgG-Fc fusion protein for the treatment of anemia and bone loss. Curr Opin Mol Ther. 2010;12(5):586-597 101. Arlet JB, Dussiot M, Moura IC, Hermine O, Courtois G. Novel players in b-thalassemia dyserythropoiesis and new therapeutic strategies. Curr Opin Hematol. 2016;23 (3):181-188. 102. Jelkmann W. The ESA scenario gets complex: from biosimilar epoetins to activin traps. Nephrol Dial Transplant. 2015;30(4): 553-559.

103. Liu J, Sun B, Yin H, Liu S. Hepcidin: A promising therapeutic target for iron disorders: a systematic review. Medicine (Baltimore). 2016;95(14):e3150. 104. Sugahara M, Tanaka T, Nangaku M. Prolyl hydroxylase domain inhibitors as a novel therapeutic approach against anemia in chronic kidney disease. Kidney Int. 2017;92(2):306-312. 105. Almeida A, Fenaux P, List AF, Raza A, Platzbecker U, Santini V. Recent advances in the treatment of lower-risk non-del(5q) myelodysplastic syndromes (MDS). Leuk Res. 2017;52:50-57 106. Arber DA. Revisiting erythroleukemia. Curr Opin Hematol. 2017;24(2):146-151. 107. Lacout C, Pisani DF, Tulliez M, Gachelin FM, Vainchenker W, Villeval JL. JAK2V617F expression in murine hematopoietic cells leads to MPD mimicking human PV with secondary myelofibrosis. Blood. 2006;108 (5):1652-1660. 108. Akada H, Yan D, Zou H, Fiering S, Hutchison RE, Mohi MG. Conditional expression of heterozygous or homozygous Jak2V617F from its endogenous promoter induces a polycythemia vera-like disease. Blood. 2010;115(17):3589-357. 109. Marty C, Pecquet C, Nivarthi H, et al. Calreticulin mutants in mice induce an MPLdependent thrombocytosis with frequent progression to myelofibrosis. Blood. 2016;127(10):1317-1324. 110. Shide K, Kameda T, Yamaji T, et al. Calreticulin mutant mice develop essential thrombocythemia that is ameliorated by the JAK inhibitor ruxolitinib. Leukemia. 2017;31(5):1136-1144. 111. Rumi E, Harutyunyan A, Elena C, et al. Identification of genomic aberrations associated with disease transformation by means of high-resolution SNP array analysis in patients with myeloproliferative neoplasm. Am J Hematol. 2011;86(12):974-979. 112. Cerquozzi S, Tefferi A. Blast transformation and fibrotic progression in polycythemia vera and essential thrombocythemia: a literature review of incidence and risk factors. Blood Cancer J. 2015;5:e366. 113. Alvarez-Larrán A, Senín A, FernándezRodríguez C, et al. Impact of genotype on leukaemic transformation in polycythaemia vera and essential thrombocythaemia. Br J Haematol. 2017;178(5):764-771. 114. Hidalgo López JE, Carballo-Zarate A, et al. Bone marrow findings in blast phase of polycythemia vera. Ann Hematol. 2018;97 (3):425-434. 115. Grossmann V, Bacher U, Haferlach C, et al. Acute erythroid leukemia (AEL) can be separated into distinct prognostic subsets based on cytogenetic and molecular genetic characteristics. Leukemia. 2013;27(9):1940-1943. 116. Cervera N, Carbuccia N, Garnier S, et al. Molecular characterization of acute erythroid leukemia (M6-AML) using targeted next-generation sequencing. Leukemia. 2016;30(4):966-970. 117. Ping N, Sun A, Song Y, et al. Exome sequencing identifies highly recurrent somatic GATA2 and CEBPA mutations in acute erythroid leukemia. Leukemia. 2017;31(1):195202. 118. Montalban-Bravo G, Benton CB, Wang SA, et al. More than 1 TP53 abnormality is a dominant characteristic of pure erythroid leukemia. Blood. 2017;129(18):2584-2587.

1603

ARTICLE

Hematopoiesis

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1604-1615

Repopulating hematopoietic stem cells from steady-state blood before and after ex vivo culture are enriched in the CD34+CD133+CXCR4low fraction Véronique Lapostolle,1,2 Jean Chevaleyre,1,2 Pascale Duchez,1,2 Laura Rodriguez,1,2 Marija Vlaski-Lafarge,1,2 Ioanna Sandvig,3 Philippe Brunet de la Grange1,2 and Zoran Ivanovic1,2

Etablissement Français du Sang Nouvelle Aquitaine, Bordeaux, France; 2U1035 INSERM/Bordeaux University, France; 3Department of Neuromedicine and Movement Science, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology (NTNU), Trondheim, Norway

ABSTRACT

Correspondence: zoran.ivanovic@efs.sante.fr

Received: November 3, 2017. Accepted: May 24, 2018. Pre-published: June 1, 2018.

doi:10.3324/haematol.2017.183962 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1604

he feasibility of ex vivo expansion allows us to consider the steadystate peripheral blood as an alternative source of hematopoietic stem progenitor cells for transplantation when growth factorinduced cell mobilization is contraindicated or inapplicable. Ex vivo expansion dramatically enhances the in vivo reconstituting cell population from steady-state blood. In order to investigate phenotype and the expression of homing molecules, the expression of CD34, CD133, CD90, CD45RA, CD26 and CD9 was determined on sorted CD34+ cells according to CXCR4 (“neg“, “low” “bright”) and CD133 expression before and after ex vivo expansion. Hematopoietic stem cell activity was determined in vivo on the basis of hematopoietic repopulation of primary and secondary recipients - NSG immuno-deficient mice. In vivo reconstituting cells in the steady-state blood CD34+ cell fraction before expansion belong to the CD133+ population and are CXCR4low or, to a lesser extent, CXCR4neg, while after ex vivo expansion they are contained only in the CD133+CXCR4low cells. The failure of the CXCR4bright population to engraft is probably due to the exclusive expression of CD26 by these cells. The limiting-dilution analysis showed that both repopulating cell number and individual proliferative capacity were enhanced by ex vivo expansion. Thus, steady-state peripheral blood cells exhibit a different phenotype compared to mobilized and cord blood cells, as well as to those issued from the bone marrow. These data represent the first phenotypic characterization of steady-state blood cells exhibiting short- and long-term hematopoietic reconstituting potential, which can be expanded ex vivo, a sine qua non for their subsequent use for transplantation.

©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1604

Introduction The introduction into clinical practice of “mobilization” from the bone marrow to peripheral blood, was an approach that resulted in an impressive increase of the number of hematopoietic progenitor cells (HPCs) and hematopoietic stem cells (HSCs) available for collection by cytapheresis. As such, this approach represented a revolutionary event in hematopoietic transplantation1 and, as a result, strategies involving steady-state peripheral blood (SS-PB)2 were abandoned. However, the procedure of mobilization of HPCs and HSCs, as well as their collection from the bone marrow, are not without risks.3 Such risks can also effectively pose a deterrent to the recruitment of voluntary donors. Besides, mobilization is contraindicated in some cases, leading to the exclusion of the potential donors. Thus, avoiding mobilization haematologica | 2018; 103(10)

Steady-state blood CD34+ HSCs are CXCR4lowCD133+

or bone marrow collection would be of great interest, especially in the context of allogeneic transplantation. Ex vivo expansion procedures have evolved over the last few years and it is now possible to amplify committed HPCs to a great extent without losing the long-term reconstituting HSCs.4,5 Recently, we demonstrated the presence of both short- and long-term reconstituting HSCs in human SS-PB and also observed that the activity of these cells increases dramatically after ex vivo expansion.6,7 In this manner, we can safely source substantial numbers of SS-PB HPCs and HSCs, thus overcoming major obstacles to subsequent transplantation. In the light of this, SSPB HPCs and HSCs should be reconsidered in the context of hematopoietic transplantation. Based on previous literature regarding HSC activity,6 it was not possible to specify whether the increase in activity of HSCs capable of reconstituting in vivo hematopoiesis of severe combined immune-deficient mice (SCID) repopulating cells (SRCs) after ex vivo expansion is: (i) due to amplification of these cells during ex vivo culture; or (ii) corresponds to pre-existing SRCs before ex vivo expansion (at time 0), which during expansion (until day 7), gained the ability to engraft after transplantation; or (iii) a combination of the above. In order to address this issue, we investigated both HSC functional capacity in in vivo assays and the expression of membrane markers known to be associated with cell adhesion and homing, such as CD9, CD26, CD49d, CD49e, CD49f and especially CXCR4, as well as markers enabling the enrichment of HSCs (CD133, CD90, CD45RA). The choice of the tetraspanin CD9 was based on the fact that it is regulated by the activity of stromal cell-derived factor-1 (SDF-1; the ligand of CXCR4 receptor)8 and CD26, since it is known to be an inhibitor of activity of the SDF-1/CXCR4 couple,9 which plays an essential role in HSC mobilization and homing.10-12 CD49d (VLA4), CD49e (VLA5), and CD49f (VLA6) are adhesion molecules of the integrin family associated with the anchorage and adhesion of cells in different situations and are considered essential for HSC homing.13 Furthermore CD49f, CD45RA and CD90 are used as markers of cord blood (CB) and/or bone marrow (BM) HSCs.14,15 Despite the fact that it largely overlaps with CD34, CD133 was chosen since it is not expressed on some subpopulations of committed progenitors and, hence, is more likely to include the HSCs.16-18 We found that HSC activity increases due to both amplification in their number and to enhancement of their individual proliferative capacity. Furthermore, in vivo reconstituting cells (both short- and long-term reconstituting cells i.e. ST-HSCs and LT-HSCs, respectively) in the fresh SS-PB CD34+ cell population belong to the subpopulation of CD133+ cells which are either CXCR4low or CXCR4neg, while after ex vivo expansion they are present only in the CD133+CXCR4low population.

Methods Human steady-state peripheral blood cells Leukocytes were recovered from leukodepletion filters (T2975, Fresenius Kabi, Louviers, France) by counterflow elution as described elsewhere6,19,20 with a slight modification, i.e. the cells were flushed directly into 50 mL tubes (Falcon, Dutscher, Brumath, France) (see Online Supplementary Methods). haematologica | 2018; 103(10)

Isolation and cryopreservation of CD34+ cells CD34+ cells were isolated from the mononuclear cell fraction using Miltenyi’s (Miltenyi Biotec, Paris, France) “indirect” immuno-magnetic technique19 (“LS” columns; Vario Macs Device). The CD34+ cell purity was 85-90% and the yield was 3-5x105 CD34+ cells per leukodepletion filter. For each sample, 20 to 24 leukodepletion filters were processed and CD34+ cells were pooled before cryopreservation (4% human serum albumin solution, 10% dimethylsulfoxide; Wak-Chemie, Steinbach, France).21 Samples were thawed in cold 4% human serum albumin and washed in selection buffer. After thawing, the CD34+ cell purity was 90-95%.

Ex vivo expansion of CD34+ cells recovered from leukodepletion filters All tests were performed on CD34+ cells after thawing, before expansion (day 0) and after expansion (day 7). Day-0 CD34+ cells were seeded at 2x104 cells/mL, and cultured in 75 cm2 flasks (NUNC, Roskilde, Denmark) for 7 days in liquid (clinical-grade serum-free medium Macopharma HP01) cultures supplemented with granulocyte colony-stimulating factor 100 ng/mL (Neuropen, Amgen SAS, Neuilly-sur-Seine, France), stem cell factor 100 ng/mL, thrombopoietin 20 ng/mL and interleukin-3 0.5 ng/mL (all from Peproteck, Rocky Hill, NJ, USA) (see Online Supplementary Methods).

CD34+ cell detection, immunophenotypic analysis and selection of cell subfractions The CD34+ cell concentrations/purities were determined as previously described.19,22 Fluorescent monoclonal antibodies were used to analyze/isolate CXCR4neg, CXCR4low, CXCR4bright subfractions, and CXCR4negCD133-, CXCR4negCD133+, CXCR4lowCD133-, CXCR4lowCD133+, CXCR4brightCD133-, and CXCR4brightCD133+ subfractions. The details are provided in the Online Supplementary Methods.

Detection of stem cells by their in vivo repopulating capacity The only way to evaluate the activity of HSCs properly is to test their in vivo capacity of hematopoietic reconstitution.23 Hence, we employed the most widely used assay based on the repopulation, by human cells, of hematopoietic tissues of immune-deficient mice, thereby evaluating the cells usually called SRCs (Figure 1). This approach enables the detection of two SRC populations: i) Short-term HSCs (ST-HSCs). ST-HSC activity was evaluated in vivo, following transplantation of different phenotypically defined fractions of human SS-PB CD34+ cells in immunodeficient [NOD/SCID/gamma-null (NSG)] mice. As described previously,24 the animal experiments were performed in compliance with French regulations (license n. 3306002) and with the approval of the Ethics Committee (n. 50120213-A). Either 1x105 CD34+ cells or 1x105 cells of sorted subfractions at day 0 were injected per mouse. After expansion, 2x105 of total day-7 cells or 2x105 cells of sorted subfractions were transplanted per mouse. In some experiments, the postculture (day-7) equivalent of a defined number of day-0 cells i.e. the total day-7 progeny of a defined day-0 cell number, was injected per mouse (Figure 1). For all experiments, 10- to 12-week old female NSG mice (central animal-keeping facility of Bordeaux University) were conditioned by means of intra-peritoneal injections of 25 mg/kg busulfan (Busilvex, Pierre Fabre, Boulogne, France),25,26 After 8 weeks, the animals were sacrificed and their femoral mononuclear BM cells isolated and analyzed for human CD45, CD19 and CD33 (with anti-human antibodies coupled with, respectively, fluorescein isothiocyanate, phycoerythrin and allophycocyanin; BD Biosciences, Le Pont de Claix, France) by flow-cytometry (FACS Canto II; BD Biosciences, Le Pont de Claix, France). To avoid false1605

V. Lapostolle et al.

positive results due to control isotype, we used the non-injected mice to establish the “positivity threshold” for CD45, which was 0.1%.24,27 Furthermore, to avoid inhibition of CXCR4 activity due to fixation of clone 12G5 antibody on the same external loop as SDF1α (CXCL12, CXCR4 specific ligand),28 we performed antibody elution by acid solution. For this, cells were incubated 20 min at 0°C in ACDA, pH 5 (Anticoagulant Citrate Dextrose solution formula A, Bioluz, Saint-Jean-de-Luz, France), then washed twice in RPMI medium before injection into mice. All cell suspensions were treated in an identical manner before injection. To determine the SRC frequency, limiting dilution analysis29,30 was performed for the CD34+CXCR4lowCD133+ subpopulation. The details are given in the Online Supplementary Methods. ii) Long-term HSCs (LT-HSCs). For the detection of LT-HSCs, secondary recipient mice (Figure 1) were conditioned as primary recipients. The BM from both femora of primary recipients was flushed, resuspended and injected intrafemorally6,25 into the secondary recipient NSG mice (Figure 1), as described in detail in the Online Supplementary Methods. The mice were sacrificed 7 or 8 weeks later and analyzed as described above.

Detection of colony-forming committed progenitors Thawed CD34+ cells were selected for their CXCR4 and CD133 expression and sorted as CXCR4negCD133-, CXCR4negCD133+, CXCR4lowCD133- and CXCR4lowCD133+ subpopulations. Day-0 sorted subfractions were expanded separately ex vivo for 7 days. Day-0 and day-7 subpopulations were plated in methylcellulose cytokine-supplemented kits “Stemα-1D” (Saint Clement les Places, France) (1000 cells/mL for each cell population) and cul-

tured for 14 days (37°C, 20% O2, 5% CO2) in 35 mm Petri dishes (NUNC, Roskilde, Denmark) in duplicate. The colonies (>50 cells) were scored19 as burst-forming unit - erythroid (BFU-E), colonyforming unit - granulocyte and macrophage (CFU-GM) and multilineage colony-forming unit (CFU-mix).

Statistical analysis The Mann-Whitney test for non-parametric values was applied. P values <0.05 were defined as statistically significant (*). P<0.01 (**) and P<0.001 (***) were highly significant values.

Results Hematopoietic stem cells with short-term reconstituting capacity To estimate ST-HSC activity directly before and after ex vivo expansion, the mice were injected with 2x105 day-0 SS-PB CD34+ cells or with their total day-7 progeny, hereafter referred to as “day-0 equivalent”. These results from the NSG mice confirmed our previous findings obtained with NOD/SCID mice6 demonstrating that 7 days of culture greatly enhanced SRC activity (P<0.05) while it also maintained the differentiation potential, as judged on the basis of the proportion of lympho (CD19)-myeloid (CD33) chimerism (Figure 2A). Regarding the SS-PB CD34+ population, the most prominent changes in culture were related to the expression of CXCR4 between day 0 (~16% cells expressing CXCR4)

Figure 1. Experimental design. Evaluation of human stem cells by employing a severe combined immunodeficiency repopulating cell assay before (day 0) and after (day 7) ex vivo expansion culture. (A) Human cell chimerism in primary mice recipients reflects the activity of short-term hematopoietic stem cells (ST-HSC) while (B) human cell chimerism in secondary mice recipients reflects the activity of long-term hematopoietic stem cells (LT-HSC). HPC: hematopoietic progenitor cell; PC: precursor cell; MC: mature cell.

1606

haematologica | 2018; 103(10)

Steady-state blood CD34+ HSCs are CXCR4lowCD133+

and the end stage of the ex vivo culture (day 7) (~67% cells expressing CXCR4) (Online Supplementary Table S1). In all experiments, three distinct subpopulations of cells with respect to CXCR4 expression level were evidenced: CXCR4neg, CXCR4low and CXCR4bright (Online Supplementary Figure S1). Flow cytometry analysis after sorting showed that the cells belonged to only one of the subpopulations, categorized according to CXCR4 expression (Online Supplementary Figure S2). Most cells with in vivo repopulating capacity in the day-0 population were predominantly concentrated in the CXCR4low fraction, although some minor activity was found in the CXCR4neg and CXCR4bright populations (Figure 2B); these HSCs exhibit a lower lymphoid differentiation potential compared to CXCR4neg and, especially, CXCR4low repopulating HSCs.

The engraftment of CXCR4neg cells prompted us to explore the hypothesis that at least some of the CXCR4neg cells can express CXCR4 once in an in vivo microenvironment of 37°C (i.e. after injection and transplantation). Thus, after an overnight incubation, 30% of the CD34+ cells that were initially CXCR4neg, became CXCR4low (Online Supplementary Figure S3). These data from the bulk CD34+ cultures were confirmed in the cultures initiated with the sorted CXCR4neg cells (Online Supplementary Figure S4). After ex vivo expansion, almost all cells with engraftment capacity (SRCs) were concentrated in the CXCR4low fraction (Figure 2C) and fully maintained their day-0 differentiation potential, although from these results it appears that SRC activity after ex vivo expansion (day-7) is lower than that of non-expanded cells at day 0. However,

haematologica | 2018; 103(10)

Figure 2. CXCR4 expression is related to the engraftment capacity of hematopoietic stem cells contained in the steady-state peripheral blood CD34+ cell population before and after ex vivo expansion. (A) The activity of severe combined immunodeficiency repopulating cells (SRCs) in CD34+ steady-state peripheral blood (SS-PB) cells is enhanced by 7 days of culture. Day 0: 2x105 SS-PB CD34+ cells were injected per mouse. Day 7: the total progeny of 2x105 day-0 SS-PB CD34+ cells was injected per mouse. (B-C) After culture day-0 (B) and day-7 (C) culture, three cell subpopulations were selected for injection into the recipient mice: P1, cells defined and sorted as the CD34+CXCR4neg subpopulation; P2, cells defined and sorted as the CD34+CXCR4low subpopulation; P3, cells defined and sorted as the CD34+CXCR4bright subpopulation. Day 0: 1x105 SS-PB total CD34+ cells or 1x105 cells from each sorted cell subpopulation were injected per mouse. Day 7: 2x105 of the total expanded cell population or 2x105 cells from each sorted cell subpopulation were injected per mouse. (D) Effect of ex vivo expansion on SRC activity in SS-PB CD34+CXCR4low cells: Day 0: 1x105 SS-PB CD34+CXCR4low cells were injected per mouse; day 7: the total progeny of 1x105 day-0 SS-PB CD34+CXCR4low cells were injected per mouse. (A-D) SRC activity was evaluated by short-term reconstitution (8 weeks) in NSG mice; each point represents the percentage of CD45+ human cells in one mouse bone marrow. For each condition (A-D) the “pie” graphs show the relative proportion of CD19+ and CD33+ cells of human origin within the huCD45+ population. Statistical significance: *P<0.05; **P<0.01; ***P<0.001.

1607

V. Lapostolle et al.

although injecting the same number of cells from each fraction into mice can show in which fraction the SRCs are concentrated, it cannot provide insight into changes of specific SRC activity during expansion culture. To obtain this information, we injected each mouse with the full day-0 day-7 progeny (equivalent) of 2x105 + low CD34 CXCR4 cells (Figure 2D) (the CD34+CXCR4low fraction was chosen since effectively all SRC activity is concentrated in this fraction). In this way, we obtained unequivocal proof that SRC activity was enhanced after the ex vivo expansion culture. To quantify these data, we performed a limiting dilution assay31 on the CD34+CXCRlowCD133+ population at the beginning and after culture (see further text for the CD133 issue), the results of which showed an ~4.2-fold expansion of SRCs after 7 days with respect to day 0 (Figure 3A-C). Furthermore, the mean individual SRC proliferative capacity was ~4-fold higher after 7 days of expansion culture compared to the capacity at day 0 (Figure 3D). In the same time period, a 14.2-fold expansion of the CD34+ CXCR4lowCD133+ fraction was found (Figure 3C; Online Supplementary Table S2). All (100%) SS-PB CD34+ cells, whatever their CXCR4 expression pattern, expressed all adhesion molecules analyzed (LFA-1, VLA-4, VLA-5, VLA-6) before and after ex vivo expansion (data not shown). Neither CD90 nor CD45RA was expressed by CD34+ SS-PB cells: all the sorted subpopulations were CD90- and CD45RA- (data not shown). Before and after culture expansion, expression of the tetraspanin CD9 correlated closely with the expression of

CXCR4 (Figure 4A). CD26 was not expressed on CXCR4neg or CXCR4low fractions of CD34+ cells at either day 0 or day 7, however 22% and 38% of CXCR4bright cells expressed CD26 on day 0 and day 7, respectively (Figure 4B). It is noteworthy that the expression of CD26 by CXCR4-expressing cells coincides with their loss of engraftment capacity (Figure 2). On day-0 CD34+ cells, CD133 was primarily expressed on CXCR4neg and CXCR4low CD34+ cell fractions (Figure 5A). In contrast, after expansion culture (day 7), the CD133+ cells were exclusively concentrated in the CXCR4low fraction of the cells remaining CD34+ (Figure 5B). When day-0 cells from these fractions defined on the basis of CXCR4 and CD133 expression were injected into NSG mice, SRCs were evidenced only in CD133+ fractions, i.e. CXCR4negCD133+ and CXCR4lowCD133+ (Figure 6A). Furthermore, after expansion at day 7, the main SRC activity remained in the CXCR4lowCD133+ fraction: with the cell dose employed, all mice were “positive” and with high chimerism (Figure 6B). At day 0, we observed that SRCs were much more frequent in the CXCR4low fraction than in the CXCR4neg fraction (P<0.01) (Figure 4A). With regard to differentiation potential, a predominant “lymphoid” profile characterized the repopulating HSCs of CD133+ fractions, while the rare repopulating HSCs detected in CD133neg fractions showed much higher proportion of, or predominantly exhibited, a myeloid differentiation potential (Figure 6A). At day 7, only the CD34+CXCR4lowCD133+ fraction yielded HSCs capable of

Figure 3. Frequencies and individual proliferative capacity of severe combined immunodeficiency repopulating cells within the CD34+CD133+CXCR4low cell population before and after ex vivo expansion. (A,B) Percentage ofmice “positive” for human CD45, 8 weeks after injection of CD34+CXCR4lowCD133+ cells, with respect to the cell dose, before expansion (A) and after expansion (B). (C) Absolute number of severe combined immunodeficiency repopulating cells (SRCs) estimated on the basis of the extreme limiting dilution assay (ELDA). (D) Mean chimerisms of the individual SRCs (only the doses giving less than 37% of positive mice were taken into consideration and only positive mice from these conditions were analyzed). The results presented were generated from the individual data given in Online Supplementary Table S2. Statistical significance: *P<0.05.

1608

haematologica | 2018; 103(10)

Steady-state blood CD34+ HSCs are CXCR4lowCD133+

in vivo reconstitution. In this case, they displayed a predominant lymphoid differentiation potential (Figure 6B).

Hematopoietic stem cells with long-term reconstituting ability While the results presented above concern ST-HSCs, we also employed the primary/secondary recipient transplantation approach to detect the LT-HSC subpopulation in SS-PB CD34+ cells depending on their CD133 and CXCR4 expression pattern before and after ex vivo expansion.6,24,32 In fact, we tested the presence of LT-HSCs at day 0 and after expansion culture (day 7) in the total cell population, in the CD34+CXCR4lowCD133+ population (which, as described above and shown in Figure 7A,B, contains most of the ST-HSCs), and in the fraction containing all remaining cells after removal of the CD34+CXCR4lowCD133+ population (Figure 7). With the number of cells injected in our experiments, the LT-HSCs were practically undetectable both at day 0 (before expansion) and at day 7 (after expansion) (see CD45 chimerism in secondary recipients, Figure 7C,D), indicating that their frequency in the total CD34+ cell population is extremely low. However, once concen-

trated in the CD34+CXCR4lowCD133+ population, LTHSCs become clearly detectable both before and after expansion (Figure 7C,D). Since we did not find “positive” secondary recipient mice after injection of BM from the primary recipient mice which had received the cell population composed of all other cells except CD34+CXCR4lowCD133+ ones, it can be concluded that the LT-HSCs are limited to the CD34+CXCR4lowCD133+ phenotype. In view of the fold expansion of the total cells (25.1 ± 9.9) (Online Supplementary Table S1) and the fact that the injected cell dose after expansion was only eight times higher than before expansion, it can be estimated that LT-HSCs were at minimum maintained during the culture. It is very interesting to note that the day-0 LTHSCs (Figure 7C) showed a relatively lower lymphoid differentiation capacity compared to cultured (day-7) LTHSCs (Figure 7D).

Committed hematopoietic progenitors The content of committed progenitors in CD34+ cells belonging to the fractions defined by CXCR4 and CD133 expression is presented in Figure 8. Interestingly, the

Figure 4. CD9 and CD26 cell expression among CXCR4 cell subpopulations. (A) CD9 expression at day 0 and after 7 days of culture. (B) CD26 expression at day 0 and after 7 days of culture. The same CD34+CXCR4neg, CD34+CXCR4low and CD34+CXCR4bright subpopulations were selected as those for the in vivo reconstitution experiments. Percentages of CD9+ or CD26+ cells are indicated for each subpopulation. Day 7: cell subpopulations were defined among the progeny of total day-0 CD34+ cells.

haematologica | 2018; 103(10)

1609

V. Lapostolle et al. A

Figure 5. CD133 expression by CD34+ cells selected on the basis of CXCR4 expression. Three gates were delimited: P1, CD34+CXCR4neg cell subpopulation; P2, CD34+CXCR4low cell subpopulation; P3, CD34+CXCR4bright cell subpopulation. Among these cell gates, CD133+ and CD133- fractions were defined and six cell subpopulations were sorted as CXCR4negCD133-, CXCR4negCD133+, CXCR4lowCD133-, CXCR4lowCD133+, CXCR4brightCD133-, and CXCR4brightCD133+ subfractions. For each of these subpopulations, the percent of CD133-expressing cells is indicated.

Figure 6. CD133 determines hematopoietic severe combined immunodeficiency repopulating cell capacity of CXCR4-expressing CD34+ steady-state peripheral blood cells, before and after ex vivo expansion. Severe combined immunodeficiency repopulating cell (SRC) activity was evaluated by short-term engraftment (8 weeks) in NSG mice. Each point of the graphs represents the percentage of CD45+ human cells in one mouse bone marrow. (A) Day 0: 1x105 SS-PB total CD34+ cells or 1x105 cells from each sorted cell subpopulation were injected per mouse. (B) Day 7: 2x105 of the total expanded cell population or 2x105 cells from each sorted cell subpopulation were injected per mouse. The â&#x20AC;&#x153;pieâ&#x20AC;? graphs in (A) and (B) show the relative proportions of CD19+ and CD33+ cells pf human origin within the huCD45+ population. Statistical significance: **P<0.01 and ***P<0.001.

1610

haematologica | 2018; 103(10)

Steady-state blood CD34+ HSCs are CXCR4lowCD133+

CD133- fractions before and after ex vivo expansion contained exclusively erythroid progenitors (BFU-E) irrespective of CXCR4 expression, while CD133+ cells always contained three classes of progenitors (CFU-GM, BFU-E and CFU-Mix). Furthermore, the committed progenitors were five times less concentrated in the CXCR4lowCD133fraction than in the CXCR4lowCD133+ one.

Discussion The findings presented in this article clearly show that ST-HSCs and LT-HSCs present in SS-PB have particular

phenotypic properties, which are different from those of HSCs in CB, BM or mobilized peripheral blood (M-PB). It is evident that the pattern of CXCR4 expression is related to the functional abilities of SS-PB ST-HSCs and LT-HSCs. This is not surprising since CXCR4 and its ligand SDF-1 have been demonstrated to have a major role in homing/mobilization of HPCs and HSCs.33,34 The presence of a small fraction (8%) of CXCR4expressing CD34+ cells in SS-PB was first observed by Lataillade et al.35 Here, we found 16% of CXCR4+CD34+ cells in the mononuclear SS-PB fraction issued from leukodepletion filters. Only a very small fraction of M-PB CD34+ cells express CXCR4; these cells exhibit an in vitro

Figure 7. Capacity of long-term hematopoietic reconstitution of NSG mice is restricted to the CD34+CXCR4lowCD133+ steady-state peripheral blood cell fraction. Before (day 0) (A, C) and after ex vivo expansion (day 7) (B, D), three cell populations were selected and sorted: total CD34+ steady-state peripheral blood (SS-PB) cells, the CD34+CXCR4lowCD133+ selected subpopulation, and total CD34+ cells without the CD34+CXCR4lowCD133+ subpopulation. (A) Day-0, short-term reconstitution in primary recipients, 2x105 cells of each subpopulation were injected intravenously per mouse. (B) Day-7, short-term reconstitution in primary recipients, 1.6x106 cells of each subpopulation were injected intravenously per mouse. Day-7 subpopulations were defined among the progeny of total day-0 CD34+ cells. (C) Day-0, long-term reconstitution in secondary recipients. Bone marrow cells from both femora of each primary recipient were injected into the bone marrow of the secondary recipient NSG mouse. (D) Day-7, long-term reconstitution in secondary recipients. Bone marrow cells from both femora of each primary recipient were injected into the bone marrow of the secondary recipient NSG mouse. For each condition (A-D) the â&#x20AC;&#x153;pieâ&#x20AC;? graphs show the relative proportions of CD19+ and CD33+ cells of human origin within the huCD45+ population. Statistical significance: *P<0.05; **P<0.01; ***P<0.001.

haematologica | 2018; 103(10)

1611

V. Lapostolle et al. A

Figure 8. Hematopoietic committed progenitors in the CD34+ cell populations selected on day 0 on the basis of CXCR4 and CD133 expression. Day-0 (A) sorted subpopulations were expanded separately ex vivo and day-7 (B) clonogenic capacities of these subpopulations were analyzed. CFU-GM: colony-forming unit granulocyte/monocyte; BFU-E: burst-forming unit – erythroid; CFU-mix: colony-forming unit - mixed. Statistical significance: *P<0.05; **P<0.01; ***P<0.001.

capacity to migrate towards a SDF-1α gradient and result in high levels of multilineage engraftment upon injection into NOD/SCID mice.10 In CB, both CXCR4+ and CXCR4neg subsets were shown to be capable of engrafting NOD/SCID mice with similar frequencies36 suggesting that CXCR4 is not a suitable marker for purification of human CB HSCs before transplantation. In addition, CXCR4 expression is rapidly regulated by environmental factors or induced ex vivo by cytokines such as granulocyte colony-stimulating factor (also used in our cultures) in CD34+ cells from all sources, including SS-PB CD34+ cells with primitive features.35 Furthermore, induction of CXCR4 expression on CB and M-PB CD34+ cells increased their capacity for in vivo engraftment in NOD/SCID mice.37 Apparently contradictory conclusions were reached in a study that found that the homing efficiency of CD34+ cells selected from BM or M-PB was not related to expression of either CXCR4 or adhesion molecules.38 However, blocking CXCR4 signaling on transplanted CB CD34+ cells prevented homing, whereas pretreatment of cells with cytokines led to up-regulation of CXCR4 expression and increased mice engraftment,39 which clearly highlights the crucial importance of CXCR4 expression for HSC engraftment. This point is important because it helps to avoid erroneous conclusions due to artifacts induced by the technical procedure related to CXCR4 expression-based cell sorting; a low repopulating capacity of CD34+CXCR4+ cells (the authors did not discriminate between low and bright populations) from CB and BM could result from the neutralizing activity of the anti-CXCR4 monoclonal antibody that was used for cell sorting. This antibody (clone 12G5) binds the site that serves for the binding and signaling of the CXCR4 specific ligand SDF-1α28,39 and, depending on its concentration, can have either inhibitory or stimulatory effects.11 To avoid any artifacts in assaying CXCR4 activity and grafting capacity of CD34+ cells, we used antibody elution after cell subfraction sorting and before any engraftment assays. In our hands, the engraftment capacity of the CD34+CXCR4neg subfraction in SS-PB was much lower than that of the CD34+CXCR4low population, as found for CB CD34+CXCR4neg versus CD34+CXCR4pos cells (including 1612

both “low” and “high” CXCR4 expression).12 However, a substantial number of CD34+CXCR4neg cells expresses CXCR4 (i.e. they become “low expressing”) after at least overnight ex vivo cytokine treatment, an effect which is even more pronounced after 4 days in culture (Online Supplementary Figure S3; confirmed also in the cultures initiated with the CD34+ cells sorted on the basis of CXCR4 expression: Online Supplementary Figure S4). This phenomenon could be proposed as the explanation for the ex vivo culture enhancing effect on the engrafting capacity of STHSCs and LT-HSCs. However, this is not the case, since practically all CXCR4neg cells which became CXCR4+ in culture lost their CD133 expression by day 4 of culture (only CD133+ cells exhibited engrafting capacity, see below) (Online Supplementary Figure S3). However, the CD133 expression was maintained for at least for 24 h in culture mimicking the in vivo situation after cell injection. This could explain some minor engraftment capacity of the CXCRneg cell population before expansion. In fact, these cells could become CXCR4low in vivo, during the first hours after injection. It should be emphasized that the highest engraftment capacity before (at day 0) or after (at day 7) cell culture is concentrated in the CXCR4low fraction and not in the CXCR4bright one, which seems to be surprising. Actually, the expression of CD26 (that, in our protocol, could be induced by granulocyte colony-stimulating factor, as shown for CD34+CD38- CB cells40), which is related only to CXCR4bright cells, can explain the decrease in CXCR4bright engrafting efficiency. It has been shown that CD26/dipeptidyl peptidase IV is a membrane-bound extracellular peptidase that cleaves polypeptides such as SDF-1, thus reducing CXCR4 activity. Furthermore, CD26 expression might be part of a mechanism regulating CXCR4 activity. The inhibition of CD26 expression on CB CD34+ cells enhances the in vitro migratory effect against the SDF-1 gradient9 and improves in vivo long-term engraftment in NOD/SCID mice.41,42 Furthermore, pretreatment of mice with a specific CD26 inhibitor (diprotin A) enhances engraftment of mouse BM cells in primary and secondary recipients.43 This is being considered among emerging strategies to improve homing and engraftment of HSCs in clinical transplants.44 A similar approach, allowing the CXCR4bright HSCs to engraft, might haematologica | 2018; 103(10)

Steady-state blood CD34+ HSCs are CXCR4lowCD133+

still enhance the engraftment efficiency of SS-PB after ex vivo expansion, although this remains to be confirmed. On the other hand, recent studies have shown that CD34+ cells also home to the BM in an SDF1-CXCR4 axis-independent manner and that “priming factors”45 as well as “mild heat treatment” facilitate incorporation of CXCR4 into functional lipid rafts.46 This might constitute another strategy to enhance engraftment of SS-PB cells. Concerning our observation of a close relationship between the expression of CXCR4 and CD9, CD9 has been implicated in the regulation of various physiological processes, including cell motility and adhesion. Trafficking and homing is a multistep process, as demonstrated for lymphocytes and myeloma cells, in which CD9 has been proven essential for transendothelial invasion.47 In human CB, CD9 is expressed by CD34+ cells and is regulated by SDF-1. Anti-CD9 antibody alters migratory and adhesive functions of CB CD34+ cells in vitro and CD9 neutralization impairs homing of transplanted CD34+ cells in NOD/SCID mice.8 The functional relationship between CD9 and CD26 on CDCXCR4bright cells remains to be elucidated. In our hands, all the sorted SS-PB CD34+ subpopulations were CD90- and CD45RA-. This phenotype is associated with committed progenitor cells in BM and CB CD34+ cells since HSCs seem to be CD90+48 and/or CD45RA+.14 However, our CD34+CXCR4lowCD133+CD90-CD45RASS-PB cells are enriched in true HSCs, as proven by efficient secondary recipient hematopoietic engraftment. CD49f, claimed to be a specific marker of CB repopulating HSCs,15 is expressed on all CD34+CD133+ SS-PB cells whatever their CXCR4 expression. Perhaps the most interesting information emerging from our study is the fact that all SS-PB HSCs exhibiting in vivo repopulating capacity (both ST- and LT-HSCs) are found to be exclusively a CD133+ population of CD34+ cells, highly concentrated in the CXCR4low population. This particular phenotypic determinant does not change after ex vivo expansion. With respect to the committed progenitors in the CD34+ population, our results (Figure 8) clearly show that before and after ex vivo expansion, CFU-GM and CFUMix reside exclusively in the CD133+ population, whereas BFU-E are present in both the CD133+ and CD133- populations. This is in line with recent findings obtained with CB CD34+ cells.49-51 CD133 has long been considered a marker of stemness for CB, BM and M-PB cells although also expressed by most committed progenitor cells.16,18,49 Here, we show that CD133 could also be used for the enrichment of SS-PB HSCs. In CB, BM and M-PB cytokine-activated CD34+ cells, CD133 is concentrated in the uropod of the polarized migrating cells.52 A functional relationship has been observed between CD133/prominin-1 and CXCR4 in specific membrane micro-domains of magnupodia,17 suggesting a favored cell migration towards the in vivo hematopoietic niche and, hence, engraftment. Since LTHSCs are present only in the CD133+ fraction of CD34+CXCR4low SS-PB cells before and after expansion, the loss of this particular phenotypically-defined population in the course of ex-vivo manipulation could be indicative of a loss of the long-term repopulating capacity of the graft. Clinical scale CD133+ selection is also considered

haematologica | 2018; 103(10)

among emerging strategies and alternative methods in clinical transplantation.53 BM mesenchymal stromal cell proliferation, but also fluctuation of the number of HSCs in peripheral blood are related to circadian oscillations.54 Since similar oscillations exist in humans,55 the circadian rhythm must be taken into consideration to optimize collection of SS-PB HSCs and HPCs. Large, phenotypic and HPC analysis was performed on CD34+ cells isolated from SS-PB.56 Ex vivo culture of CD34+ SS-PB cells enhanced the total number of HSCs exhibiting in vivo repopulating capacity as well as their individual proliferative capacities, as shown by our limiting-dilution experiments. The maintenance of the lymphoid differentiation potential of repopulating HSCs after ex vivo culture is an additional important argument, since a shift towards predominant myeloid potential, as we detected in the rare CXCR4low and CD133- repopulating HSCs, has been found to occur during aging.57 In fact, our results suggest that reducing HSC differentiation capacity to the myeloid lineage represents a degree of HSC commitment. In this respect, aging is characterized by a higher proportion of more committed HSCs58 in a context of general “consumption” and is the first sign of imminent exhaustion of the system. This suggests that ex vivo expansion can provide an adequate tool to produce enough hematopoietic stem and progenitor cells to constitute a single hematopoietic graft from the contents of only one or two steady-state leukapheresis collections. The efficiency of the expansion procedure could, most likely, be further improved by using new approaches, for example the TAT-protein transduction peptide fused to regulatory factors or inhibition of HOXB4 degradation,59-62 which is the object of our ongoing work. Furthermore, the CD34neg fraction containing immuno-competent cells (T and B lymphocytes) can be preserved and an appropriate dose of these cells injected either during transplantation or later, depending on the need for an allogeneic immuno-effect. Furthermore, lymphocyte efficiency can be enhanced and specified by ex vivo engineering. Taken together, the results presented here might help in the design of novel, advanced graft generation, which could simultaneously provide efficient immunohematopoietic reconstitution and a graft-versustumor/leukemia effect. Future work in our laboratory aims to explore this strategy. Acknowledgments The authors thank to Mr Santiago Gonzalez, Mrs Valérie De Luca, Mrs Anaëlle Stum and Mr Vincent Pitard from FlowCytometry Platform SFR Transbiomed, Bordeaux University, France, for their precious help with cell sorting experiments. The help in first-line English editing of Mrs Elisabeth Doutreloux is also gratefully acknowledged. This manuscript was funded by an EFS (French Blood Institute - Etablissement Français du Sang) grant (n. 2016-01-IVANOVIC-AQLI) and French Biomedical Agency (Agence de la Biomédecine) grant (AOR “Greffe” 2015). IS would like to acknowledge support from the Liaison Committee between the Central Norway Health Authority (RHA) and the Norwegian University of Science and Technology (NTNU).

1613

V. Lapostolle et al.

References 1. Richman CM, Weiner RS, Yankee RA. Increase in circulating stem cells following chemotherapy in man. Blood. 1976;47(6):1031-1039. 2. Reiffers J, Faberes C, Boiron JM, et al. Peripheral blood progenitor cell transplantation in 118 patients with hematological malignancies: analysis of factors affecting the rate of engraftment. J Hematother. 1994;3(3):185-191. 3. D'Souza A, Jaiyesimi I, Trainor L, Venuturumili P. Granulocyte colony-stimulating factor administration: adverse events. Transfus Med Rev. 2008;22(4):280-290. 4. Bari S, Seah KK, Poon Z, et al. Expansion and homing of umbilical cord blood hematopoietic stem and progenitor cells for clinical transplantation. Biol Blood Marrow Transplant. 2015;21(6):1008-1019. 5. Boiron JM, Dazey B, Cailliot C, et al. Largescale expansion and transplantation of CD34(+) hematopoietic cells: in vitro and in vivo confirmation of neutropenia abrogation related to the expansion process without impairment of the long-term engraftment capacity. Transfusion. 2006;46 (11):1934-1942. 6. Brunet de la Grange P, Vlaski M, Duchez P, et al. Long-term repopulating hematopoietic stem cells and "side population" in human steady state peripheral blood. Stem Cell Res. 2013;11(1):625-633. 7. Bourdieu A, Avalon M, Lapostolle V, et al. Steady state peripheral blood provides cells with functional and metabolic characteristics of real hematopoietic stem cells. J Cell Physiol. 2018;233(1):338-349. 8. Leung KT, Chan KY, Ng PC, et al. The tetraspanin CD9 regulates migration, adhesion, and homing of human cord blood CD34+ hematopoietic stem and progenitor cells. Blood. 2011;117(6):1840-1850. 9. Christopherson KW 2nd, Hangoc G, Broxmeyer HE. Cell surface peptidase CD26/dipeptidylpeptidase IV regulates CXCL12/stromal cell-derived factor-1 alpha-mediated chemotaxis of human cord blood CD34+ progenitor cells. J Immunol. 2002;169(12):7000-7008. 10. Peled A, Petit I, Kollet O, et al. Dependence of human stem cell engraftment and repopulation of NOD/SCID mice on CXCR4. Science. 1999;283(5403):845-848. 11. Plett PA, Frankovitz SM, Wolber FM, Abonour R, Orschell-Traycoff CM. Treatment of circulating CD34(+) cells with SDF-1alpha or anti-CXCR4 antibody enhances migration and NOD/SCID repopulating potential. Exp Hematol. 2002;30(9): 1061-1069. 12. Lapidot T, Kollet O. The essential roles of the chemokine SDF-1 and its receptor CXCR4 in human stem cell homing and repopulation of transplanted immune-deficient NOD/SCID and NOD/SCID/B2m (null) mice. Leukemia. 2002;16(10):19922003. 13. Peled A, Kollet O, Ponomaryov T, et al. The chemokine SDF-1 activates the integrins LFA-1, VLA-4, and VLA-5 on immature human CD34(+) cells: role in transendothelial/stromal migration and engraftment of NOD/SCID mice. Blood. 2000;95(11):32893296. 14. Majeti R, Park CY, Weissman IL. Identification of a hierarchy of multipotent hematopoietic progenitors in human cord blood. Cell Stem Cell. 2007;1(6):635-645.

1614

15. Notta F, Doulatov S, Laurenti E, Poeppl A, Jurisica I, Dick JE. Isolation of single human hematopoietic stem cells capable of longterm multilineage engraftment. Science. 2011;333(6039):218-221. 16. Bhatia M. AC133 expression in human stem cells. Leukemia. 2001;15(11):16851688. 17. Bauer N, Fonseca AV, Florek M, et al. New insights into the cell biology of hematopoietic progenitors by studying prominin-1 (CD133). Cells Tissues Organs. 2008;188 (1-2):127-138. 18. Drake AC, Khoury M, Leskov I, et al. Human CD34+ CD133+ hematopoietic stem cells cultured with growth factors including Angptl5 efficiently engraft adult NOD-SCID Il2rgamma-/- (NSG) mice. PloS One. 2011;6(4):e18382. 19. Ivanovic Z, Duchez P, Morgan DA, et al. Whole-blood leuko-depletion filters as a source of CD34+ progenitors potentially usable in cell therapy. Transfusion. 2006;46(1):118-125. 20. Peytour Y, Guitart A, Villacreces A, et al. Obtaining of CD34+ cells from healthy blood donors: development of a rapid and efficient procedure using leukoreduction filters. Transfusion. 2010;50(10):2152-2157. 21. Duchez P, Chevaleyre J, Brunet de la Grange P, et al. Cryopreservation of hematopoietic stem and progenitor cells amplified ex vivo from cord blood CD34+ cells. Transfusion. 2013;53(9):2012-2019. 22. Sutherland DR, Anderson L, Keeney M, Nayar R, Chin-Yee I. The ISHAGE guidelines for CD34+ cell determination by flow cytometry. International Society of Hematotherapy and Graft Engineering. J Hematother. 1996;5(3):213-226. 23. Frisch BJ, Calvi LM. Hematopoietic stem cell cultures and assays. Methods Mol Biol. 2014;1130:315-324. 24. Ivanovic Z, Duchez P, Chevaleyre J, et al. Clinical-scale cultures of cord blood CD34(+) cells to amplify committed progenitors and maintain stem cell activity. Cell Transplant. 2011;20(9):1453-1463. 25. Robert-Richard E, Ged C, Ortet J, et al. Human cell engraftment after busulfan or irradiation conditioning of NOD/SCID mice. Haematologica. 2006;91(10):1384. 26. Chevaleyre J, Duchez P, Rodriguez L, et al. Busulfan administration flexibility increases the applicability of scid repopulating cell assay in NSG mouse model. PloS One. 2013;8(9):e74361. 27. Ivanovic Z, Hermitte F, Brunet de la Grange P, et al. Simultaneous maintenance of human cord blood SCID-repopulating cells and expansion of committed progenitors at low O2 concentration (3%). Stem Cells. 2004;22(5):716-724. 28. Woodard LE, Nimmagadda S. CXCR4based imaging agents. J Nucl Med. 2011;52(11):1665-1669. 29. Wang JC, Doedens M, Dick JE. Primitive human hematopoietic cells are enriched in cord blood compared with adult bone marrow or mobilized peripheral blood as measured by the quantitative in vivo SCIDrepopulating cell assay. Blood. 1997;89(11):3919-3924. 30. Hu Y, Smyth GK. ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. J Immunol Methods. 2009;347(1-2):70-78. 31. Denning-Kendall PA, Evely R, Singha S, Chapman M, Bradley BA, Hows JM. In vitro expansion of cord blood does not pre-

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

vent engraftment of severe combined immunodeficient repopulating cells. Br J Haematol. 2002;116(1):218-228. Duchez P, Rodriguez L, Chevaleyre J, et al. Interleukin-6 enhances the activity of in vivo long-term reconstituting hematopoietic stem cells in "hypoxic-like" expansion cultures ex vivo. Transfusion. 2015;55(11):2684-2691. Jing D, Fonseca AV, Alakel N, et al. Hematopoietic stem cells in co-culture with mesenchymal stromal cells--modeling the niche compartments in vitro. Haematologica. 2010;95(4):542-550. Pantin J, Purev E, Tian X, et al. Effect of high-dose plerixafor on CD34+ cell mobilization in healthy stem cell donors: results of a randomized crossover trial. Haematologica. 2017;102(3):600-609. Lataillade JJ, Clay D, Dupuy C, et al. Chemokine SDF-1 enhances circulating CD34(+) cell proliferation in synergy with cytokines: possible role in progenitor survival. Blood. 2000;95(3):756-768. Rosu-Myles M, Gallacher L, Murdoch B, et al. The human hematopoietic stem cell compartment is heterogeneous for CXCR4 expression. Proc Natl Acad Sci USA. 2000;97(26):14626-14631. Kahn J, Byk T, Jansson-Sjostrand L, et al. Overexpression of CXCR4 on human CD34+ progenitors increases their proliferation, migration, and NOD/SCID repopulation. Blood. 2004;103(8):2942-2949. Herrera C, Sanchez J, Torres A, Pascual A, Rueda A, Alvarez MA. Pattern of expression of CXCR4 and adhesion molecules by human CD34+ cells from different sources: role in homing efficiency in NOD/SCID mice. Haematologica. 2004;89(9):10371045. Kollet O, Petit I, Kahn J, et al. Human CD34(+)CXCR4(-) sorted cells harbor intracellular CXCR4, which can be functionally expressed and provide NOD/SCID repopulation. Blood. 2002;100(8):27782786. Christopherson KW 2nd, Uralil SE, Porecha NK, Zabriskie RC, Kidd SM, Ramin SM. GCSF- and GM-CSF-induced upregulation of CD26 peptidase downregulates the functional chemotactic response of CD34+CD38human cord blood hematopoietic cells. Exp Hematol. 2006;34 (8):1060-1068. Campbell TB, Hangoc G, Liu Y, Pollok K, Broxmeyer HE. Inhibition of CD26 in human cord blood CD34+ cells enhances their engraftment of nonobese diabetic/severe combined immunodeficiency mice. Stem Cells Dev. 2007;16(3):347354. Christopherson KW 2nd, Paganessi LA, Napier S, Porecha NK. CD26 inhibition on CD34+ or lineage- human umbilical cord blood donor hematopoietic stem cells/hematopoietic progenitor cells improves long-term engraftment into NOD/SCID/Beta2null immunodeficient mice. Stem Cells Dev. 2007;16(3): 355-360. Broxmeyer HE, Hangoc G, Cooper S, Campbell T, Ito S, Mantel C. AMD3100 and CD26 modulate mobilization, engraftment, and survival of hematopoietic stem and progenitor cells mediated by the SDF1/CXCL12-CXCR4 axis. Ann N Y Acad Sci. 2007;1106:1-19. Ratajczak MZ, Suszynska M. Emerging strategies to enhance homing and engraftment of hematopoietic stem cells. Stem Cell Rev. 2016;12(1):121-128.

haematologica | 2018; 103(10)

Steady-state blood CD34+ HSCs are CXCR4lowCD133+

45. Ratajczak MZ. A novel view of the adult bone marrow stem cell hierarchy and stem cell trafficking. Leukemia. 2015;29(4):776782. 46. Ratajczak MZ, Adamiak M. Membrane lipid rafts, master regulators of hematopoietic stem cell retention in bone marrow and their trafficking. Leukemia. 2015;29(7): 1452-1457. 47. De Bruyne E, Andersen TL, De Raeve H, et al. Endothelial cell-driven regulation of CD9 or motility-related protein-1 expression in multiple myeloma cells within the murine 5T33MM model and myeloma patients. Leukemia. 2006;20(10):1870-1879. 48. Brendel C, Neubauer A. Characteristics and analysis of normal and leukemic stem cells: current concepts and future directions. Leukemia. 2000;14(10):1711-1717. 49. Radtke S, Gorgens A, Kordelas L, et al. CD133 allows elaborated discrimination and quantification of haematopoietic progenitor subsets in human haematopoietic stem cell transplants. Br J Haematol. 2015;169(6):868-878. 50. Gorgens A, Radtke S, Mollmann M, et al. Revision of the human hematopoietic tree: granulocyte subtypes derive from distinct hematopoietic lineages. Cell Rep. 2013;3 (5):1539-1552.

haematologica | 2018; 103(10)

51. Takahashi M, Matsuoka Y, Sumide K, et al. CD133 is a positive marker for a distinct class of primitive human cord bloodderived CD34-negative hematopoietic stem cells. Leukemia. 2014;28(6):13081315. 52. Giebel B, Corbeil D, Beckmann J, et al. Segregation of lipid raft markers including CD133 in polarized human hematopoietic stem and progenitor cells. Blood. 2004;104(8):2332-2338. 53. Bornhauser M, Eger L, Oelschlaegel U, et al. Rapid reconstitution of dendritic cells after allogeneic transplantation of CD133+ selected hematopoietic stem cells. Leukemia. 2005;19(1):161-165. 54. Ehninger A, Trumpp A. The bone marrow stem cell niche grows up: mesenchymal stem cells and macrophages move in. J Exp Med. 2011;208(3):421-428. 55. Lucas D, Battista M, Shi PA, Isola L, Frenette PS. Mobilized hematopoietic stem cell yield depends on species-specific circadian timing. Cell Stem Cell. 2008;3(4):364366. 56. Jobin C, Cloutier M, Simard C, Neron S. Heterogeneity of in vitro-cultured CD34+ cells isolated from peripheral blood. Cytotherapy. 2015;17(10):1472-1484. 57. Pang WW, Price EA, Sahoo D, et al. Human

58.

59.

60.

61.

62.

bone marrow hematopoietic stem cells are increased in frequency and myeloid-biased with age. Proc Natl Acad Sci USA. 2011;108(50):20012-20017. Cho RH, Sieburg HB, Muller-Sieburg CE. A new mechanism for the aging of hematopoietic stem cells: aging changes the clonal composition of the stem cell compartment but not individual stem cells. Blood. 2008;111(12):5553-5561. Antonchuk J, Sauvageau G, Humphries RK. HOXB4-induced expansion of adult hematopoietic stem cells ex vivo. Cell. 2002;109(1):39-45. Krosl J, Austin P, Beslu N, Kroon E, Humphries RK, Sauvageau G. In vitro expansion of hematopoietic stem cells by recombinant TAT-HOXB4 protein. Nat Med. 2003;9(11):1428-1432. Lee J, Shieh JH, Zhang J, et al. Improved ex vivo expansion of adult hematopoietic stem cells by overcoming CUL4-mediated degradation of HOXB4. Blood. 2013;121 (20):4082-4089. Codispoti B, Rinaldo N, Chiarella E, et al. Recombinant TAT-BMI-1 fusion protein induces ex vivo expansion of human umbilical cord blood-derived hematopoietic stem cells. Oncotarget. 2017;8(27): 43782-43798.

1615

ARTICLE

Iron Metabolism & its Disorders

Ferrata Storti Foundation

Circulating iron levels influence the regulation of hepcidin following stimulated erythropoiesis Cornel S.G. Mirciov,1,2 Sarah J. Wilkins,1 Grace C. C. Hung,1 Sheridan L. Helman,1,3 Gregory J. Anderson1,2,4 and David M. Frazer1,2

Iron Metabolism Laboratory, QIMR Berghofer Medical Research Institute, Herston; 2School of Medicine, The University of Queensland, St Lucia; 3School of Biomedical Sciences, Queensland University of Technology, Gardens Point and 4School of Chemistry and Molecular Bioscience, The University of Queensland, St Lucia, Australia. 1

Haematologica 2018 Volume 103(10):1616-1626

ABSTRACT

Correspondence: david.frazer@qimrberghofer.edu.au

Received: December 29, 2017. Accepted: June 11, 2018. Pre-published: June 14, 2018.

he stimulation of erythrocyte formation increases the demand for iron by the bone marrow and this in turn may affect the levels of circulating diferric transferrin. As this molecule influences the production of the iron regulatory hormone hepcidin, we hypothesized that erythropoiesis-driven changes in diferric transferrin levels could contribute to the decrease in hepcidin observed following the administration of erythropoietin. To examine this, we treated mice with erythropoietin and examined diferric transferrin at various time points up to 18 hours. We also investigated the effect of altering diferric transferrin levels on erythropoietin-induced inhibition of Hamp1, the gene encoding hepcidin. We detected a decrease in diferric transferrin levels 5 hours after erythropoietin injection and prior to any inhibition of the hepatic Hamp1 message. Diferric transferrin returned to control levels 12 hours after erythropoietin injection and had increased beyond control levels by 18 hours. Increasing diferric transferrin levels via intravenous iron injection prevented the inhibition of Hamp1 expression by erythropoietin without altering hepatic iron concentration or the expression of Erfe, the gene encoding erythroferrone. These results suggest that diferric transferrin likely contributes to the inhibition of hepcidin production in the period shortly after injection of erythropoietin and that, under the conditions examined, increasing diferric transferrin levels can overcome the inhibitory effect of erythroferrone on hepcidin production. They also imply that the decrease in Hamp1 expression in response to an erythropoietic stimulus is likely to be mediated by multiple signals.

doi:10.3324/haematol.2017.187245 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1616 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1616

Introduction Hepcidin is a key regulator of body iron homeostasis. This 25-amino acid peptide is produced predominantly by hepatocytes and is secreted into the circulation where it binds to the iron export protein ferroportin on the surface of body cells.1 This interaction causes the ferroportin/hepcidin complex to be internalized and degraded, inhibiting iron release and allowing hepcidin to regulate critical processes such as dietary iron absorption and the recycling of erythrocyte iron by macrophages. The production of hepcidin is tightly regulated. The main stimuli triggering changes in hepcidin synthesis are the level of iron in the body, referred to as the stores regulator, and the adequacy of iron supply to the erythroid marrow, termed the erythroid regulator.2,3 The molecular basis of the stores regulator has been relatively well characterized. As body iron stores increase, non-parenchymal cells in the liver secrete bone morphogenetic protein 6 (BMP6), which binds to the BMP receptor complex on the surface of hepatocytes and stimulates hepcidin expression by activating the SMAD signaling pathway.4-6 This increase in hepcidin production inhibits dietary iron absorption, limiting further increases in body iron stores, and tissue iron release, and resulting in the storage of excess iron within macrophages and hepatocytes. Various other molecules, such as hemojuvelin and transmemhaematologica | 2018; 103(10)

Erythropoiesis, circulating iron and hepcidin

brane protease, serine 6 modulate BMP/SMAD signaling and refine hepcidin production, allowing the precise regulation necessary to maintain iron homeostasis.4,7-9 The influence of the erythroid regulator is readily apparent in conditions such as b-thalassemia, in which the increased iron demands of the expanded erythroid marrow signal a decrease in hepcidin production and subsequent iron loading.10 However, the molecular mechanism by which developing erythrocytes signal their iron needs to hepatocytes remains poorly understood. One signaling molecule that has recently been identified is erythroferrone, a member of the tumor necrosis factor superfamily of cytokines which is encoded by the ERFE gene.11 Erythroferrone is produced by erythroblasts11 and is detectable in the circulation following stimulated erythropoiesis.12 Importantly, recombinant erythroferrone has been shown to inhibit hepcidin production in both primary hepatocytes and in mice.11 In addition, the inhibition of hepcidin by stimulated erythropoiesis is blunted in Erfe knockout mice and hepcidin expression normalizes in b-thalassemic animals lacking erythroferrone,11 providing strong evidence supporting a role for erythroferrone in hepcidin regulation. There is clear evidence that erythroferrone can influence hepcidin expression in response to stimulated erythropoiesis, but the level of iron in the circulation has also been proposed to be involved.13-15 Iron in the plasma is predominantly bound to the protein transferrin.16 As each transferrin molecule can bind two iron atoms and physiological levels of circulating iron are not high enough to saturate all transferrin binding sites, circulating transferrin can exist in four different states: apo-transferrin; two forms of monoferric transferrin; and diferric transferrin.17 Iron bound to transferrin is taken up by cells via receptor-mediated endocytosis following the binding of transferrin to cell surface transferrin receptor 1 (TFR1).16 By far the largest sink for transferrinbound iron in the circulation is the developing erythrocytes of the bone marrow10 and, as TFR1 has the greatest affinity for the diferric isoform,16 increases in the erythropoietic rate would be expected to preferentially reduce circulating diferric transferrin levels. Indeed, we have previously demonstrated such a change following treatment with the hemolytic agent phenylhydrazine in rats.18 As diferric transferrin has been implicated in the regulation of hepcidin expression,13,19 changes in diferric transferrin levels following a stimulus to increase erythropoiesis could augment the effect of erythroferrone and play a role in the inhibition of hepcidin production. Such a mechanism has been suggested by Nai et al. to explain the reduction in hepcidin expression following the administration of erythropoietin to mice.14 While Kautz et al. saw no change in serum iron levels in mice injected with erythropoietin, they did not specifically examine diferric transferrin.11 In order to determine whether diferric transferrin might contribute to the decrease in hepcidin expression that occurs following stimulated erythropoiesis, we examined its level in erythropoietin-injected mice. We also examined the effect of altering diferric transferrin levels in mice following stimulated erythropoiesis.

Methods Animals Six-week old male C57BL/6 mice obtained from the Animal Resources Centre (Perth, Australia) were used for all experiments haematologica | 2018; 103(10)

and were maintained on a standard rodent chow (120 mg/kg iron, Norco Stockfeed, Lismore, Australia). To stimulate erythropoiesis, mice were intravenously administered 10 U/g body weight of human erythropoietin (Epoetin alfa, Eprex, Janssen, Macquarie Park, Australia). The mice were euthanized 0, 5, 9, 12, 15 or 18 h later. Prior to euthanasia, mice were anesthetized (200 mg/kg ketamine, 10 mg/kg xylazine) and blood was withdrawn by cardiac puncture. Blood for diferric transferrin quantitation was collected in heparin-coated tubes, briefly centrifuged, and the plasma stored at -80Â°C. Blood for serum iron determination was allowed to clot and the serum stored at -80Â°C. Following euthanasia, the liver, spleen and bone marrow were removed and snap frozen for subsequent analysis. To study the effects of altering serum iron on erythropoietininduced inhibition of hepcidin expression, four groups of mice were used. Two groups were intravenously injected with erythropoietin (10 U/g body weight) 9 h prior to euthanasia, while the remaining two groups served as uninjected controls. Of the two groups injected with erythropoietin, one group was administered iron intravenously (2.5 mg/g body weight ferric citrate monohydrate in 5 mM citrate buffer, pH 7.0) while the other group was injected with sodium citrate (equimolar with the citrate in the ferric citrate solution) in 5 mM citrate buffer as a control. Similar injections were administered to the two groups not given erythropoietin. All ferric citrate or sodium citrate injections were given 4 h prior to euthanasia. To confirm that the injected iron led to an increase in circulating diferric transferrin, additional groups of mice were euthanized 5 min after the iron or citrate injections and blood was collected for analysis. To control for circadian variations in serum iron levels and the expression of Hamp1, the gene encoding hepcidin, all experimental animals were euthanized between 10:00 am and 12:30 pm local time. We did not observe any consistent alterations in hepatic Hamp1 expression during this period. Male mice were used in all studies as differences in the absolute levels of some iron parameters have been reported in males and females.20 However, the mechanisms regulating iron homeostasis and hepcidin expression are thought to be similar in both genders. All animal experiments were approved by the QIMR Berghofer Animal Ethics Committee.

Analysis of blood and tissue samples Details of the analysis of serum iron levels, liver iron concentration and gene and protein expression are included in the Online Supplementary Methods.

Statistics All results are expressed as mean Âą standard error of the mean (SEM). Statistical differences between groups were calculated using ANOVA followed by either Tukey post hoc testing for samples with equal variance or Games-Howell post hoc testing for samples with unequal variance. IBM SPSS Statistics version 22 software (IBM Australia, St Leonards, Australia) was used. A P value of less than 0.05 was considered statistically significant.

Results Reduced hepatic Hamp1 expression is associated with increased splenic and bone marrow Erfe production following injection of erythropoietin Previous studies in mice have implicated erythroferrone in the inhibition of hepcidin production following administration of erythropoietin.11,12 Other factors, such as the level of diferric transferrin in the circulation, might also 1617

C.S.G. Mirciov et al.

play a role. To examine this, we established a model of erythropoietin-induced hepcidin inhibition. In this model, we ensured that all mice were euthanized within a narrow window of time to ensure that known circadian variations in both hepatic Hamp1 expression and serum iron levels21,22 did not influence the results. In agreement with findings of previous studies,11,14 hepatic Hamp1 levels progressively decreased following erythropoietin administration, with significant inhibition first observed after 9 h (Figure 1A). By 18 h, Hamp1 expression was only 12% of the control value. Only minor changes in spleen weight were observed (Figure 1B), possibly because the time points examined were too soon after erythropoietin injection for significant increases in the number of erythroid precursors to have occurred. Despite this, a large increase in Erfe expression (>86-fold) was seen in both the spleen (Figure 1C, D) and bone marrow (Figure 1E, F) of injected animals at the 5 h time point. This increase in Erfe expression was observed at the 5, 9, 12 and 15 h time points, but splenic levels had reduced by 18 h after erythropoietin injection. Normalizing Erfe expression to the general housekeeper hypoxanthine guanine phosphoribosyl transferase (Hprt) or the erythroid precursor specific marker glycophorin A (Gypa) did not alter the expression pattern seen. Our results agree with those of

previous studies11,14 and confirm that our model can be used to examine the potential role of diferric transferrin in erythropoietin-induced hepcidin inhibition.

Erythropoietin administration increases Tfr1 expression in the bone marrow and spleen and causes a transient decrease in circulating diferric transferrin levels Developing erythrocytes are the major sink for circulating iron and this iron demand is increased when erythropoiesis is stimulated.10 To determine how rapidly this increase in demand occurs, we examined the expression of the Tfr1 gene, the product of which is central to the main cellular iron import pathway in developing erythrocytes.10 In the bone marrow, Tfr1 expression increased rapidly after erythropoietin administration, showing maximal expression at 5 h when normalized to either Hprt (Figure 2A), or the erythroid-specific housekeeper, Gypa (Figure 2B). The increase in splenic Tfr1 was not as pronounced, with significant increases only observed when its levels were normalized to those of Gypa (Figure 2C,D). These results indicate that there is a very rapid increase in the iron demands of developing erythroid cells following erythropoietin administration, and imply that iron is removed from the circulation more rapidly than normal,

Figure 1. Hamp1 and Erfe expression following erythropoietin injection in mice Six-week old male C57BL/6 mice were euthanized 0, 5, 9, 12, 15 or 18 h following the intravenous injection of 10 U/g body weight human erythropoietin and tissues were taken for analysis. Hepatic Hamp1 expression (A), spleen weight (B), splenic Erfe expression (C, D) and bone marrow Erfe expression (E, F) were determined for each time point. Gene expression levels were calculated relative to either the general housekeeping gene Hprt or the erythroid specific marker Gypa, and are expressed as a proportion of the values at 0 h. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Statistical significance is shown relative to the 0 h group. *P<0.05; **P<0.01; ***P<0.005.

1618

haematologica | 2018; 103(10)

Erythropoiesis, circulating iron and hepcidin

possibly leading to reduced diferric transferrin levels. To determine whether circulating iron levels were indeed reduced by the increase in erythroid Tfr1 expression following erythropoietin injection, we examined total serum iron and transferrin saturation at each time point. While there was a small decrease in both serum iron (Figure 3A) and transferrin saturation (Figure 3B) at 5 h, these changes were not significantly different from control values. However, TFR1 preferentially takes up diferric transferrin,16 and so any increase in TFR1-mediated iron uptake due to enhanced erythropoiesis would preferentially affect diferric transferrin levels. We hypothesized that the small changes in circulating iron levels seen (Figures 3A, B) might reflect larger changes in the level of circulating diferric transferrin. To test this hypothesis, we examined the various transferrin species by western blotting. Diferric transferrin is expressed as a proportion of total transferrin, so we first confirmed that there were no significant changes in total transferrin by examining total iron binding capacity (and hence total transferrin levels) in the samples analyzed (Figure 3C). We then demonstrated that there was a significant reduction in diferric transferrin levels at the 5 h time point (51% of the control value) (Figure 3D). This decrease was transient, with diferric transferrin returning to control levels after 12 h. As diferric transferrin has been implicated in the regulation of hepcidin production, these results suggest that reductions in diferric transferrin levels might contribute to the decrease in Hamp1 expression following erythropoietin injection. Indeed, the decrease in diferric transferrin levels preceded the decrease in Hamp1 expression (Online Supplementary Figure S1). The significant increases in serum iron, transferrin saturation and diferric transferrin that were observed at the latter time points likely reflect the increase in iron

haematologica | 2018; 103(10)

release that would occur following a decrease in hepcidin production.

Intravenous injection of iron raises diferric transferrin levels without increasing hepatic iron stores Having established that a transient reduction in diferric transferrin occurs following injection of erythropoietin, we sought to determine the consequences of increasing diferric transferrin levels during this period. To achieve this, mice were intravenously injected with a dose of ferric citrate estimated to be twice the amount required to fully saturate circulating transferrin, ensuring that transferrin was as close to saturated as possible, with any remaining non-transferrin bound iron at such low levels that it would not significantly alter tissue iron stores once removed from the circulation. Control mice were injected with an equimolar amount of citrate as sodium citrate. The iron was injected 5 h after erythropoietin administration as, at this time point, diferric transferrin was reduced whereas Hamp1 expression had not yet been affected (Figures 1A and 3D), allowing us to determine whether Hamp1 levels would decrease if diferric transferrin was elevated during the treatment period. In order to check that the injected iron bound to circulating transferrin, a cohort of mice was euthanized 5 min after the iron injections. As non-transferrin bound iron is rapidly cleared from the circulation with a half-life of 30 s,23 any iron remaining in the bloodstream after 5 min should be bound to transferrin. Total serum iron (Figure 4A) and transferrin saturation (Figure 4B) were significantly elevated in mice injected with iron compared to those injected with sodium citrate, regardless of whether they had received a prior injection of erythropoietin, with transferrin saturations above 80% being achieved. The

Figure 2. Tfr1 expression following erythropoietin injection in mice. Six-week old male C57BL/6 mice were euthanized 0, 5, 9, 12, 15 or 18 h following the intravenous injection of 10 U/g body weight human erythropoietin and tissues were taken for analysis. Bone marrow Tfr1 expression (A, B) and splenic Tfr1 expression (C, D) were determined for each time point. Gene expression levels were calculated relative to either the general housekeeping gene Hprt or the erythroid-specific marker Gypa, and are expressed as a proportion of the values at 0 h. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Statistical significance is shown relative to the 0 h group. *P<0.05; ***P<0.005.

1619

C.S.G. Mirciov et al. A

D Figure 3. Circulating iron parameters following erythropoietin injection in mice. Six-week old male C57BL/6 mice were euthanized 0, 5, 9, 12, 15 or 18 h following the intravenous injection of 10 U/g body weight human erythropoietin and blood was taken for analysis. Total serum iron (A), transferrin saturation (B), total iron binding capacity (C) and relative diferric transferrin levels (D) were determined for each time point. The relative diferric transferrin levels represent the percentage of transferrin in the diferric form expressed as a proportion of the values at 0 h. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Apo: apotransferrin; Mono: monoferric transferrin; Di: diferric transferrin. Statistical significance is shown relative to the 0 h group. *P<0.05.

iron injections had no effect on total iron binding capacity, a measure of total transferrin levels (Figure 4C). The analysis of transferrin species showed that transferrin was almost fully saturated 5 min after iron injection, with no apo-transferrin detectable (Figure 4D). In agreement with our earlier observations (Figure 3D), a decrease in diferric transferrin was detected in mice injected with erythropoietin and sodium citrate when compared to those injected with sodium citrate alone. A second cohort of mice was euthanized 4 h after iron injection (9 h after erythropoietin injection). Total serum iron and transferrin saturation in these iron-injected mice were lower than those observed at the 5 min time point (Figure 4A,B), with significant differences seen only in the groups injected with erythropoietin (Figure 5A,B). Once again, the injections had no influence on total iron binding capacity (i.e. total transferrin levels) (Figure 5C). Diferric transferrin levels remained significantly elevated 4 h after iron injection (Figure 5D), although the levels were reduced when compared with those 5 min after injection (Figure 4D). An analysis of liver tissue showed no change in hepatic iron levels in the groups treated with iron (Figure 5E), indicating that, although iron had been removed from the circulation in the 4 h since the iron injection, the amounts were not enough to influence storage iron levels.

Increasing serum iron levels can overcome the inhibitory effect of erythropoietin injection on Hamp1 expression without altering Erfe production The effect of increasing serum iron levels on Hamp1 expression was examined in our erythropoietin-treated 1620

mice sacrificed 4 h after iron injection. We found that increasing serum iron was able to overcome the inhibitory effect of stimulated erythropoiesis, with Hamp1 expression in mice injected with both erythropoietin and iron remaining at levels similar to those in mice not injected with erythropoietin (Figure 6A). The level of phosphorylated SMAD1/5/8 was decreased in the erythropoietininjected mice and increased with subsequent iron injection (Figure 6B); this finding supports those of previous studies indicating that the SMAD pathway is involved in the regulation of Hamp1 expression by both circulating erythroferrone and iron levels.14,15,24 Despite Hamp1 expression returning to normal following iron injection, splenic and bone marrow Erfe levels remained elevated in response to erythropoietin (Figure 7A-D). We also observed a significant increase in spleen weight in the erythropoietin-injected group receiving iron, possibly indicating an increase in erythroblast proliferation and implying that erythropoiesis is iron restricted in the hours following an erythropoietic stimulus (Figure 7E). Consistent with this, we detected an increase in the expression of Tfr1 (normalized to Hprt) (Figure 8A) in the spleen of mice treated with both erythropoietin and iron when compared to those treated with erythropoietin alone, although the increase failed to reach statistical significance (P=0.089). A similar non-significant increase in splenic Erfe/Hprt was also seen (Figure 7A). In both cases, the difference was lost when gene expression was normalized to Gypa (Figures 7B and 8B), indicating that the changes were due to an increase in erythroblast number rather than individual cell expression. In contrast, we detected a decrease in Tfr1 expression in the bone marrow haematologica | 2018; 103(10)

Erythropoiesis, circulating iron and hepcidin

D Figure 4. Circulating iron parameters immediately after iron injection of erythropoietin-treated mice. Six-week old male C57BL/6 mice were injected intravenously with 10 U/g body weight human erythropoietin. Five hours later, mice were intravenously injected with either 2.5 mg/g body weight ferric citrate or an equimolar amount of citrate as sodium citrate. Mice were euthanized 5 min after this final injection and blood taken for analysis. Total serum iron (A), transferrin saturation (B), total iron binding capacity (C) and relative diferric transferrin levels (D) were determined for each group. The relative diferric transferrin levels represent the percentage of transferrin in the diferric form expressed as a proportion of the values in mice injected with sodium citrate but not with erythropoietin. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Con: control mice that were injected with sodium citrate; Fe: mice that were injected with ferric citrate; No Epo: mice that were not injected with erythropoietin; Epo: mice that were injected with erythropoietin; Apo: apotransferrin; Mono: monoferric transferrin; Di: diferric transferrin. ***P<0.005.

of mice treated with both erythropoietin and iron when compared with those treated with erythropoietin alone (Figure 8C). Interestingly, the decrease remained when Tfr1 levels were normalized to Gypa (Figure 8D), indicating lower Tfr1 expression in individual erythroblasts and a reduction in iron demand following iron injection.

Discussion Developing erythrocytes are by far the largest sink for iron in the body,10 so it is not surprising that erythropoiesis is a major regulator of hepcidin production. Our understanding of how the erythroid mass in the bone marrow and spleen signal changes in HAMP expression in hepatocytes has been advanced considerably with the recent discovery of erythroferrone, a protein produced by developing erythrocytes and secreted into the circulation, which inhibits hepatic hepcidin production,11,12 As a result, iron absorption and storage iron release are increased, providing more iron for erythrocyte development. However, other factors are also known to regulate hepcidin expression. Of these, serum iron levels, particularly diferric transferrin, may be relevant to hepcidin expression following stimulated erythropoiesis. We have previously haematologica | 2018; 103(10)

shown that diferric transferrin correlates with inhibited Hamp expression in rats following treatment with the hemolytic agent phenylhydrazine,18 and have suggested that the level of diferric transferrin in the circulation might signal the iron demands of the erythroid marrow to the liver and influence hepcidin production.13,19 In the current study we have demonstrated for the first time that the rapid decrease in Hamp1 expression that occurs following erythropoietin injection in mice is preceded by a transient decrease in circulating diferric transferrin. The evidence in favor of serum iron and, therefore, diferric transferrin, influencing hepcidin production is strong. We have previously shown that reductions in serum iron levels correlate closely with the inhibition of Hamp expression in rats switched to an iron-deficient diet25 and following phenylhydrazine treatment.18 We and others have also demonstrated that increased serum iron levels stimulate hepcidin expression in mice.15,19 In addition, a mechanism by which hepatocytes might detect the levels of circulating diferric transferrin has been described. It has been proposed that the hemochromatosis protein HFE on the surface of hepatocytes competes with diferric transferrin for binding to TFR1, with the HFE/TFR1 complex being favored when diferric transferrin levels are low.13,26 Thus, diferric transferrin would not be expected to 1621

C.S.G. Mirciov et al. A

E Figure 5. Circulating iron parameters and hepatic iron concentration 4 hours after iron injection in erythropoietin-treated mice. Six-week old male C57BL/6 mice were injected intravenously with 10 U/g body weight human erythropoietin. Five hours later, mice were intravenously injected with either 2.5 mg/g body weight ferric citrate or an equimolar amount of citrate as sodium citrate. Mice were euthanized 4 h after this final injection and blood and liver tissue taken for analysis. Total serum iron (A), transferrin saturation (B), total iron binding capacity (C), relative diferric transferrin levels (D) and hepatic iron concentration (E) were determined for each group. The relative diferric transferrin levels represent the percentage of transferrin in the diferric form expressed as a proportion of the values in mice injected with sodium citrate but not with erythropoietin. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Con: control mice injected with sodium citrate; Fe: mice injected with ferric citrate; No Epo: mice that were not injected with erythropoietin; Epo: mice that were injected with erythropoietin; Apo: apotransferrin; Mono: monoferric transferrin; Di: diferric transferrin. *P<0.05; **P<0.01; ***P<0.005.

Figure 6. Hamp1 expression and SMAD phosphorylation 4 hours after iron injection in erythropoietintreated mice. Six-week old male C57BL/6 mice were injected intravenously with 10 U/g body weight human erythropoietin. Five hours later, mice were intravenously injected with either 2.5 mg/g body weight ferric citrate or an equimolar amount of citrate as sodium citrate. Mice were euthanized 4 h after this final injection and various tissues were taken for analysis. Hepatic Hamp1 expression (A) and the amount of phosphorylated SMAD1/5/8 (B) were determined for each group. Gene expression levels were calculated relative to the general housekeeping gene Hprt and are expressed as a proportion of the values in mice injected with sodium citrate but not with erythropoietin. The relative phosphorylated SMAD1/5/8 levels are expressed as phosphoprotein/total SMAD1/Actin and presented in the graph as a proportion of the values in mice injected with sodium citrate but not with erythropoietin. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Con: control mice injected with sodium citrate; Fe: mice injected with ferric citrate; No Epo: mice that were not injected with erythropoietin; Epo: mice that were injected with erythropoietin. *P<0.05; ***P<0.005.

1622

haematologica | 2018; 103(10)

Erythropoiesis, circulating iron and hepcidin

be able to regulate HAMP expression in an individual with HFE-related hemochromatosis, although it could still be regulated through other pathways. An additional detection mechanism involving TFR2, a TFR1 homolog, has also been described.13 Although the precise pathways by which these molecules signal changes in hepcidin production remains unclear, their involvement in detecting diferric transferrin levels is widely accepted.3,10,27,28 While showing a correlation only, when viewed in conjunction with the evidence supporting diferric transferrin as a regulator of hepcidin, the current study suggests that diferric transferrin levels are likely to contribute to the initial reduction in Hamp1 expression seen following an erythropoietic stimulus. However, whether the decrease in diferric transferrin levels is required for the inhibition of hepcidin is unclear. Erfe expression was also increased at the 5 h time point, although the lack of a reliable commercial assay for mouse erythroferrone made it difficult to determine the amount of functional erythroferrone at this time. Further research is required to determine whether both stimuli are required or whether there is a degree of redundancy in the pathways inhibiting hepcidin at these early time points. In contrast, serum iron is clearly not involved in reducing Hamp1 expression towards the end of the time course, as

serum iron parameters quickly returned to normal levels and even exceeded control values at these later time points. Our results also show that increasing serum iron levels soon after erythropoietin injection can overcome the inhibitory effect of stimulated erythropoiesis and prevent any decrease in Hamp1 expression. While our data suggest that changes in diferric transferrin levels are the likely cause, a role for non-transferrin-bound iron (NTBI) cannot be excluded, as this is the form of iron injected in our studies. However, there are several reasons why a major role for NTBI in hepcidin regulation is unlikely. Firstly, studies using the hypotransferrinemic mouse suggest that it is the transferrin-bound iron in the circulation that is important for hepcidin regulation. These mice have greatly reduced transferrin levels and, as a consequence, high levels of NTBI.29,30 With limited transferrin, iron supply to the marrow is compromised and the resulting anemia reduces hepcidin production. So hepcidin levels are low, despite very high levels of NTBI.29 Even in hypotransferrinemic mice subjected to myeloablation to remove the effect of stimulated erythropoiesis, hepcidin expression could only be increased if transferrin was administered, making it unlikely that NTBI has a major effect on hepcidin produc-

Figure 7. Erfe expression and spleen weight 4 hours after iron injection in erythropoietin-treated mice. Six-week old male C57BL/6 mice were injected intravenously with 10 U/g body weight human erythropoietin. Five hours later, mice were intravenously injected with either 2.5 mg/g body weight ferric citrate or an equimolar amount of citrate as sodium citrate. Mice were euthanized 4 h after this final injection and various tissues were taken for analysis. Splenic Erfe expression (A, B), bone marrow Erfe expression (C, D) and spleen weight (E) were determined for each group. Gene expression levels were calculated relative to either the general housekeeping gene Hprt or the erythroid-specific marker Gypa and are expressed as a proportion of the values in mice injected with sodium citrate but not with erythropoietin. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Con: control mice injected with sodium citrate; Fe: mice injected with ferric citrate; No Epo: mice that were not injected with erythropoietin; Epo: mice that were injected with erythropoietin. *P<0.05; ***P<0.005.

haematologica | 2018; 103(10)

1623

C.S.G. Mirciov et al. A

tion.31 This latter finding is consistent with that of an earlier study (carried out before hepcidin was identified) showing that injected transferrin was able to suppress iron absorption in hypotransferrinemic mice in which erythropoiesis had been normalized.32 However, as transferrin levels subsequently declined, iron absorption increased. Secondly, although NTBI in the form of ferric citrate was administered to the mice in the current study, it had previously been demonstrated that any NTBI formed in this way is removed very rapidly from the circulation with a half-life of <30 s.23 In contrast, we have shown that diferric transferrin levels remain elevated 4 h after iron injection (Figure 5D). Thirdly, as mentioned previously, a mechanism for the detection of diferric transferrin levels in the circulation and the subsequent regulation of Hamp1 expression has been proposed and is supported by in vivo evidence,33,34 whereas no such mechanism has been described for NTBI. Finally, these results are consistent with multiple cell culture studies showing that treatment of cells with iron salts does not directly stimulate hepcidin production.35,36 In fact, Hamp expression could only be stimulated when transferrin-bound iron was supplied.36 Therefore, despite not being able to definitively exclude a role for NTBI, it is far more likely that diferric transferrin is the major stimulator of Hamp1 expression in our studies. The increase in Hamp1 expression occurred despite significant increases in Erfe expression in the bone marrow or 1624

Figure 8. Tfr1 expression 4 hours after iron injection in erythropoietin-treated mice. Six-week old male C57BL/6 mice were injected intravenously with 10 U/g body weight human erythropoietin. Five hours later, mice were intravenously injected with either 2.5 mg/g body weight ferric citrate or an equimolar amount of citrate as sodium citrate. Mice were euthanized 4 h after this final injection and tissues were taken for analysis. Splenic Tfr1 expression (A, B) and bone marrow Tfr1 expression (C, D) were determined for each group. Gene expression levels were calculated relative to either the general housekeeping gene Hprt or the erythroid-specific marker Gypa and are expressed as a proportion of the values in mice injected with sodium citrate but not with erythropoietin. The data represent the mean Âą SEM with the number of mice in each group indicated in parentheses along the x-axis. Con: control mice injected with sodium citrate; Fe: mice injected with ferric citrate; No Epo: mice that were not injected with erythropoietin; Epo: mice that were injected with erythropoietin. *P<0.05; ***P<0.005.

spleen. In fact, Erfe expression in both tissues was higher in mice that received both erythropoietin and iron compared to those that received erythropoietin only, although the changes were not significant. However, as spleen size was also increased, it is highly likely that circulating erythroferrone levels are higher in erythropoietin-treated animals injected with iron, implying that increased serum iron levels can overcome the effect of erythroferrrone, at least in the early stages of erythroid stimulation. However, this does not always occur. For example, we observed a significant increase in diferric transferrin levels 18 h after erythropoietin injection despite Hamp1 expression being at its lowest level at this time. While the reason for this is unclear, it is possible that circulating erythroferrone levels increased over the time course. While no such change was observed in Erfe message levels, a previous study showed that Erfe mRNA expression does not always correspond to serum erythroferrone levels.12 Another possible explanation is that the increase in diferric transferrin seen later in our time course had not had sufficient time to influence Hamp1 expression, as changes in serum iron levels can take several hours to exert an effect on hepcidin production.15 Certain mouse models of b-thalassemia exhibit high transferrin saturation yet continue to load with iron due to low hepcidin levels.37 It is likely that, in these instances, the expansion of the erythroid marrow due to chronic haematologica | 2018; 103(10)

Erythropoiesis, circulating iron and hepcidin

anemia would result in a much larger increase in serum erythroferrone than would occur following a single injection of erythropoietin, allowing the erythroid regulator to overcome the effect of increases in diferric transferrin. In contrast, the current study involves the acute stimulation of erythropoiesis and, as such, any expansion of the erythroid marrow would be minimal, allowing the stimulating effect of increased diferric transferrin to overcome the inhibitory effect of erythroferrone and prevent any reduction in Hamp1 expression. Such competition between pathways influencing hepcidin production has been reported previously12,38,39 and these studies indicate that it is the strength of each signal, rather than its origin, that dictates HAMP expression. It is also possible that, in b-thalassemia, the preferential uptake of diferric transferrin by the greatly expanded erythroid precursor mass causes circulating diferric transferrin levels to be relatively low despite higher than normal transferrin saturation, and that this might contribute to the observed decrease in hepcidin. The increase in spleen size following iron injection also suggests that erythropoiesis is iron-restricted shortly after stimulation with erythropoietin, with the additional iron allowing further expansion of the erythroid marrow. This is supported by the observation of decreased Tfr1 expression in the bone marrow in iron-injected animals. Interestingly, a recent study has suggested that there is an erythroferrone-independent role for erythroid cell TFR1 in the regulation of hepcidin production.40 The authors propose that TFR1 on erythroid precursor cells regulates the production of an unknown soluble factor that influences hepcidin expression in the liver. We would suggest that a

References 1. Nemeth E, Tuttle MS, Powelson J, et al. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science. 2004;306(5704):20902093. 2. Finch C. Regulators of iron balance in humans. Blood. 1994;84(6):1697-1702. 3. Ganz T. Systemic iron homeostasis. Physiol Rev. 2013;93(4):1721-1741. 4. Rishi G, Subramaniam VN. The liver in regulation of iron homeostasis. Am J Physiol Gastrointest Liver Physiol. 2017;313 (3):G157-G165. 5. Kautz L, Meynard D, Monnier A, et al. Iron regulates phosphorylation of Smad1/5/8 and gene expression of Bmp6, Smad7, Id1, and Atoh8 in the mouse liver. Blood. 2008;112(4):1503-1509. 6. Enns CA, Ahmed R, Wang J, et al. Increased iron loading induces Bmp6 expression in the non-parenchymal cells of the liver independent of the BMP-signaling pathway. PLoS One. 2013;8(4):e60534. 7. Papanikolaou G, Samuels ME, Ludwig EH, et al. Mutations in HFE2 cause iron overload in chromosome 1q-linked juvenile hemochromatosis. Nat Genet. 2004;36 (1):77-82. 8. Du X, She E, Gelbart T, et al. The serine protease TMPRSS6 is required to sense iron deficiency. Science. 2008;320(5879):10881092. 9. Folgueras AR, de Lara FM, Pendas AM, et al.

haematologica | 2018; 103(10)

10. 11.

12.

13.

14.

15.

16. 17.

novel factor is unnecessary, as alterations in erythroid TFR1 production will rapidly and preferentially influence the levels of circulating diferric transferrin, which will, in turn, affect hepcidin production. Our results also indicate that such a change in diferric transferrin would not necessarily result in a significant difference in serum iron levels, and thus could remain undetected in many studies examining the regulation of hepcidin. In conclusion, we have demonstrated that diferric transferrin levels are transiently decreased following the stimulation of erythropoiesis by erythropoietin injection and suggest that this molecule may contribute to the initial inhibition of hepcidin expression that occurs. We also show that, in certain situations, diferric transferrin can overcome the inhibitory effect of erythroferrone to prevent hepcidin inhibition. However, this is likely to depend on the strength of the individual signals, with the effect of erythroferrone clearly overcoming increases in diferric transferrin levels at later time points. As the changes in diferric transferrin levels are not always detected by measuring serum iron or transferrin saturation, we suggest that studies investigating the regulation of hepcidin would benefit from direct measurements of diferric transferrin. Acknowledgments This work was supported by a Project Grant (APP1051764) (GJA and DMF) and a Senior Research Fellowship (GJA) from the National Health and Medical Research Council of Australia, an Australian Government Research Training Program Scholarship (CSGM) and a Top Up Award from the QIMR Berghofer Higher Degrees Committee (CSGM).

Membrane-bound serine protease matriptase-2 (Tmprss6) is an essential regulator of iron homeostasis. Blood. 2008;112(6):25392545. Muckenthaler MU, Rivella S, Hentze MW, Galy B. A red carpet for iron metabolism. Cell. 2017;168(3):344-361. 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. Kautz L, Jung G, Du X, et al. Erythroferrone contributes to hepcidin suppression and iron overload in a mouse model of beta-thalassemia. Blood. 2015;126(17):2031-2037. Frazer DM, Anderson GJ. The orchestration of body iron intake: how and where do enterocytes receive their cues? Blood Cells Mol Dis. 2003;30(3):288-297. Nai A, Rubio A, Campanella A, et al. Limiting hepatic Bmp-Smad signaling by matriptase-2 is required for erythropoietinmediated hepcidin suppression in mice. Blood. 2016;127(19):2327-2336. Corradini E, Meynard D, Wu Q, et al. Serum and liver iron differently regulate the bone morphogenetic protein 6 (BMP6)-SMAD signaling pathway in mice. Hepatology. 2011;54(1):273-284. Frazer DM, Anderson GJ. The regulation of iron transport. Biofactors. 2014;40(2):206214. Williams J, Moreton K. The distribution of iron between the metal-binding sites of transferrin human serum. Biochem J. 1980; 185(2):483-488.

18. Frazer DM, Inglis HR, Wilkins SJ, et al. Delayed hepcidin response explains the lag period in iron absorption following a stimulus to increase erythropoiesis. Gut. 2004;53(10):1509-1515. 19. Wilkins SJ, Frazer DM, Millard KN, McLaren GD, Anderson GJ. Iron metabolism in the hemoglobin-deficit mouse: correlation of diferric transferrin with hepcidin expression. Blood. 2006;107(4):1659-1664. 20. Courselaud B, Troadec MB, Fruchon S, et al. Strain and gender modulate hepatic hepcidin 1 and 2 mRNA expression in mice. Blood Cells Mol Dis. 2004;32(2):283289. 21. Troutt JS, Rudling M, Persson L, et al. Circulating human hepcidin-25 concentrations display a diurnal rhythm, increase with prolonged fasting, and are reduced by growth hormone administration. Clin Chem. 2012;58(8):1225-1232. 22. Schaap CC, Hendriks JC, Kortman GA, et al. Diurnal rhythm rather than dietary iron mediates daily hepcidin variations. Clin Chem. 2013;59(3):527-535. 23. Craven CM, Alexander J, Eldridge M, Kushner JP, Bernstein S, Kaplan J. Tissue distribution and clearance kinetics of nontransferrin-bound iron in the hypotransferrinemic mouse: a rodent model for hemochromatosis. Proc Natl Acad Sci USA. 1987;84(10):3457-3461. 24. Wang CY, Core AB, Canali S, et al. Smad1/5 is required for erythropoietin-mediated suppression of hepcidin in mice. Blood. 2017;130(1):73-83.

1625

C.S.G. Mirciov et al. 25. Frazer DM, Wilkins SJ, Becker EM, et al. Hepcidin expression inversely correlates with the expression of duodenal iron transporters and iron absorption in rats. Gastroenterology. 2002;123(3):835-844. 26. Townsend A, Drakesmith H. Role of HFE in iron metabolism, hereditary haemochromatosis, anaemia of chronic disease, and secondary iron overload. Lancet. 2002;359 (9308):786-790. 27. Muckenthaler MU. How mutant HFE causes hereditary hemochromatosis. Blood. 2014;124 (8):1212-1213. 28. Chen J, Enns CA. Hereditary hemochromatosis and transferrin receptor 2. Biochim Biophys Acta. 2012;1820(3):256-263. 29. Simpson RJ, Cooper CE, Raja KB, et al. Nontransferrin-bound iron species in the serum of hypotransferrinaemic mice. Biochim Biophys Acta. 1992;1156(1):19-26. 30. Bu JT, Bartnikas TB. The use of hypotransferrinemic mice in studies of iron biology. Biometals. 2015;28(3):473-480. 31. Bartnikas TB, Andrews NC, Fleming MD.

1626

32.

33.

34. 35.

36.

Transferrin is a major determinant of hepcidin expression in hypotransferrinemic mice. Blood. 2011;117(2):630-637. Raja KB, Pountney DJ, Simpson RJ, Peters TJ. Importance of anemia and transferrin levels in the regulation of intestinal iron absorption in hypotransferrinemic mice. Blood. 1999;94(9):3185-3192. Schmidt PJ, Toran PT, Giannetti AM, Bjorkman PJ, Andrews NC. The transferrin receptor modulates Hfe-dependent regulation of hepcidin expression. Cell Metab. 2008;7(3):205-214. Robb A, Wessling-Resnick M. Regulation of transferrin receptor 2 protein levels by transferrin. Blood. 2004;104(13):4294-4299. Gehrke SG, Kulaksiz H, Herrmann T, et al. Expression of hepcidin in hereditary hemochromatosis: evidence for a regulation in response to the serum transferrin saturation and to non-transferrin-bound iron. Blood. 2003;102(1):371-376. Lin L, Valore EV, Nemeth E, Goodnough JB, Gabayan V, Ganz T. Iron transferrin regu-

37.

38.

39.

40.

lates hepcidin synthesis in primary hepatocyte culture through hemojuvelin and BMP2/4. Blood. 2007;110(6):2182-2189. Gardenghi S, Ramos P, Marongiu MF, et al. Hepcidin as a therapeutic tool to limit iron overload and improve anemia in beta-thalassemic mice. J Clin Invest. 2010;120(12): 4466-4477. Darshan D, Frazer DM, Wilkins SJ, Anderson GJ. Severe iron deficiency blunts the response of the iron regulatory gene Hamp and pro-inflammatory cytokines to lipopolysaccharide. Haematologica. 2010;95 (10):1660-1667. Huang H, Constante M, Layoun A, Santos MM. Contribution of STAT3 and SMAD4 pathways to the regulation of hepcidin by opposing stimuli. Blood. 2009;113(15):35933599. Keel SB, Doty R, Liu L, et al. Evidence that the expression of transferrin receptor 1 on erythroid marrow cells mediates hepcidin suppression in the liver. Exp Hematol. 2015;43(6):469-478 e466.

haematologica | 2018; 103(10)

ARTICLE

Iron Metabolism

Iron overload impairs normal hematopoietic stem and progenitor cells through reactive oxygen species and shortens survival in myelodysplastic syndrome mice

Ferrata Storti Foundation

Xin Jin,1* Xiaoyuan He,1* Xiaoli Cao,2 Ping Xu,3 Yi Xing,2 Songnan Sui,3 Luqiao Wang,3 Juanxia Meng,3 Wenyi Lu,3 Rui Cui,3 Hongyan Ni4** and Mingfeng Zhao3,1**

Nankai University School of Medicine, Tianjin; 2Tianjin Children's Hospital; 3Department of Hematology, Tianjin First Central Hospital and 4Department of Radiology, Tianjin First Central Hospital, Tianjin, PR China 1

*XJ and XYH, and **HYN and MFZ contributed equally to this work.

Haematologica 2018 Volume 103(10):1627-1634

ABSTRACT

here is increasing clinical evidence to suggest a suppressive effect on hematopoiesis in myelodysplastic syndrome patients with iron overload. However, how iron overload influences hematopoiesis in myelodysplastic syndrome (MDS) remains unknown. Here, the RUNX1S291fs-transduced bone marrow mononuclear cells were yielded and transplanted into lethally irradiated recipient mice together with radioprotective bone marrow cells to generate MDS mice. Eight weeks post transplantation, the recipient mice received an intraperitoneal injection of 0.2 mL iron dextran at a concentration of 25 mg/mL once every other day for a total of 8 times to establish an iron overload model. In the present study, we show that iron overload impairs the frequency and colony-forming capacity of normal hematopoietic stem and progenitor cells, especially in erythroid, in MDS mice, which is due, at least in part, to growth differentiation factor 11-induced reactive oxygen species, shortening survival of MDS mice. Given that we are the first to construct an iron overload model in MDS mice, we hope this model will be helpful for further exploring the influence and mechanism of iron overload on MDS.

Introduction Myelodysplastic syndrome (MDS) is a heterogeneous group of myeloid malignancy characterized by dysplastic changes in one or more cell lineages, ineffective hematopoiesis, and high risk of leukemic transformation.1,2 Because of chronic anemia, up to 80% of MDS patients become transfusion dependent.3 Due to repeated blood transfusion and ineffective hematopoiesis, most MDS patients can eventually develop iron overload. Excessive iron can deposit in the liver, heart, spleen, pancreas, bone marrow (BM) and other tissues, resulting in tissue damage and fibrosis, and a series of complications which seriously affect the prognosis of MDS patients.4-6 Therefore, iron overload in MDS patients has received great attention from clinicians. Here the question is whether and how iron overload impacts the hematopoietic system in MDS. Indeed, iron-chelation therapy often improves the prognosis of MDS patients, implicating that iron overload has a suppressive effect on hematopoiesis in these subjects. However, how iron overload influences hematopoiesis in MDS still has to be addressed. RUNX1 (also known as AML1/CBFA2), one of the most common mutation genes in MDS patients, has an important role in hematopoiesis. RUNX1-S291fs is one type of RUNX1 mutation. Previous studies have proved that RUNX1-S291fs mutant can be used to construct an MDS mouse model.7,8 In order to investigate how iron overload affects hematopoiesis in MDS, we generated a RUNX1-S291fsinduced MDS mouse model and subsequently administered iron dextran by intraperitoneal injection to establish an iron overload model in MDS mice. In this study, we found that iron overload impairs the frequency and function of haematologica | 2018; 103(10)

Correspondence: mingfengzhao@sina.com

Received: March 10, 2018. Accepted: June 7, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2018.193128 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1627 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1627

X. Jin et al.

normal hematopoietic stem and progenitor cells (HSPCs), especially in erythroid cells, which may result from growth differentiation factor 11(GDF11)-induced reactive oxygen species (ROS) and shortens survival in MDS mice. Here we are the first to utilize RUNX1-S291fs-induced MDS mice to successfully construct an iron overload model. This can provide the experimental basis for further exploring the influence and mechanism of iron overload on MDS.

Statistical analysis Results were analyzed with the GraphPad Prism program (GraphPad Software, Inc., San Diego, CA, USA). Data obeyed normal distribution were presented as means±Standard Deviation (SD) and multiple group comparisons were performed by using one-way analysis of variance (ANOVA), whereas data with nonnormal distribution were showed by median and quartiles, and compared by Kruskal-Wallis test. The survival curves were analyzed using the Kaplan-Meier method with the log-rank test. P<0.05 was considered statistically significant.

Methods Results Ethical considerations Our experiment has been approved by the Ethics Committee of Tianjin First Central Hospital. All animal experiments were conducted in accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the Institute of Radiation Medicine, Chinese Academy of Medical Sciences (n. 1204).

Vector and retrovirus construct The Empty control vector (pMYs-Empty control-IRES-GFP) and the RUNX1-S291fs mutant vector (pMYs-RUNX1 S291fs-IRESGFP) were kindly provided by Dr Atsushi Iwam (Kumamoto University, Japan). Production of retrovirus construct has been described previously.8

Mice and treatment C57BL/6 mice were purchased from Vital River (Beijing, China) at 6-8 weeks of age. The animals were quarantined and allowed to acclimatize for one week. They were maintained in a room at 22±2°C, with a relative humidity of 50%±10%. The mice were housed 5 individuals per cage and used at a weight of approximately 20.0-22.0g. Retroviral construct transduced into bone marrow mononuclear cells (BMMNCs) was performed as previously described.9 Briefly, retrovirus-infected BMMNCs were cultured in αMEM medium containing 10 ng/mL of human IL-6, 10 ng/mL of mouse IL-3, and 100 ng/mL of mouse SCF for three days, and then these cells were intravenously injected into 9Gy lethally irradiated recipient mice together with radioprotective dose of 1×106 BM cells to generate MDS mice or control mice. Eight weeks post transplantation, the recipient mice were administered an intraperitoneal injection of 0.2 mL iron dextran at a concentration of 25 mg/mL once every other day for a total of 8 times to establish an iron overload model. At the same time, normal saline was given to the control group.

Flow cytometric analysis Flow cytometry was performed by using the following monoclonal mouse antibodies: CD45.2, CD45.1, Gr1, Mac1/CD11b, Ter119, CD71, B220, CD19, CD3, CD5, CD41, CD117/c-Kit, Sca1 and CD34. All Flow cytometric analysis of the stained cells was performed with Coulter Altra (Beckman Coulter) equipped with Cytexpert software (Beckman Coulter).

Western blot analysis To detect the expression of RUNX1S291fs protein, BM cells and spleen cells were lysed and blotted as previously described.10 Polyclonal anti-RUNX1 and anti-α-Tubulin antibodies were used to detect RUNX1S291fs protein and α-Tubulin, respectively.

Diagnosis The diagnosis was made according to the Bethesda proposals for classification of MDS in mice.11 1628

Iron overload model can be established in RUNX1S291fs-induced MDS mice In order to explore the role of iron overload in MDS, we first transduced a pMYs-RUNX1S291fs-IRES-GFP or a pMYs-Empty control-IRES-GFP retroviral construct into BMMNCs (Figure 1A). After five days culture in vitro, the transduced cells were transplanted into lethally irradiated recipient mice together with radioprotective BM cells to generate MDS mice or control mice (Figure 1A). Eight weeks after transplantation, the recipient mice were administered iron dextran or normal saline by intraperitoneal injection. Hereafter, we refer to the Empty control mice with normal saline, the Empty control mice with iron dextran, the RUNX1S291fs mice with normal saline, and the RUNX1S291fs mice with iron dextran as Empty/NS, Empty/FE, RX291/NS and RX291/FE mice, respectively. The ratio of GFP+ cells gradually increased over 24 weeks after the transplantation in RX291/NS and RX291/FE mice, but not in mice who received transplants of the Empty control-transduced cells (Figure 1B), which indicated that the RUNX1S291fs-transduced cells successfully engrafted and established chimerism in the peripheral blood (PB). To verify the establishment of an MDS model, several mice were sacrificed at 24 weeks post transplantation. The RUNX1S291fs protein was detected high expression in spleen and low expression in BM in RX291/NS and RX291/FE groups, but not in Empty/NS and Empty/FE mice (Figure 1C), suggesting that RUNX1S291fs was successfully transduced and expressed. We then examined whether RUNX1S291fs contributes to the development of MDS in recipient mice. The mice received RUNX1S291fstransduced cells showed significantly reduced white blood cell (WBC) counts (Figure 1D), hemoglobin (HG) levels (Figure 1E), platelet counts (PLT) (Figure 1F), and increased mean corpuscular volume (MCV) (Figure 1G) compared with Empty control recipient mice. The RUNX1S291fs mutant mice also showed dysplastic cells, such as dot-like and polychromatic erythrocytes, giant platelets, pseudoPelger-Huet granulocytes in the PB, together with dualnucleated granulocytes, increased immature blasts (less than 20%) in the BM, characteristic of human MDS (Figure 1H). Then we examined the pathological changes in the femur, liver and spleen. The femur showed extremely active myeloid hyperplasia, mainly in blast cells, with the increased ratio of granulocyte/red cells and abnormal localization of blast precursor cells in RUNX1S291fs mice but not Empty control mice (Figure 1I). Moreover, the RUNX1S291fs mice showed many nucleated cells around the central vein and the portal area of the liver, and exhibited enlargement and irregular shape haematologica | 2018; 103(10)

IO impairs normal HSPCs and survival in MDS mice

Figure 1. RUNX1S291fs induced mice model can phenotypically recapitulate human myelodysplastic syndrome. (A) Experimental scheme of our model mouse using RUNX1S291fs mutant. (B) Chimerism of GFP+ cells post transplantation. (C) Successful expression of RUNX1S291fs protein detected by western blotting. (D) White blood cell (WBC) count. (E) Hemoglobin (HG) count. (F) Platelet (PLT) count. (G) Mean corpuscular volume (MCV). (H) The morphological abnormality observed in the bone marrow and spleen size. (I) Pathological changes in the femur, liver and spleen. *P<0.05, **P<0.01, ***P<0.001.

of white pulp and narrow red pulp, whereas such pathological changes were not found in Empty control mice (Figure 1I). Collectively, our RUNX1S291fs-induced mice model showed neutropenia, anemia, thrombocytopenia and multilineage dysplasia, together with less than 20% blasts, matching the criteria of MDS in the Bethesda proposal and phenotypically recapitulating human MDS. Next, we examined whether iron overload model can be established in RUNX1S291fs induced MDS mice. We and others previously reported that iron overload can cause liver and spleen enlargement.12,13 There was a clear increase in liver and spleen weight in both control mice and in MDS mice administered iron dextran treatment, implying that iron can deposit in the liver and spleen in RUNX1S291fs-induced MDS mice (Figures 1H, and 2A and B). We further performed Perl’s iron staining to verify iron deposition in mice tissues and organs. Significant iron deposition could be observed in the liver, spleen and BM in RX291/FE group compared to RX291/NS mice (Figure 2C), supporting iron overload in RX291/FE mice. In addition, our data showed the level of ferritin was almost undetectable in Empty/NS, while it was significantly higher in RX291/NS mice (Figure 2D), which can be clearly explained by ineffective erythropoiesis in MDS. However, the ferritin in Empty/FE and RX291/FE group was comparable, and was more significant than in Empty/NS and RX291/NS mice, respectively (Figure 2D), indicating that iron overload can also be established in MDS mice. Taken haematologica | 2018; 103(10)

together, administration of iron dextran by intraperitoneal injection confers iron deposition in organs, liver and spleen enlargement, and can be used to facilitate the establishment of an iron overload model in RUNX1S291fsinduced MDS mice.

Iron overload impairs the frequency of normal HSPCs and affects erythroid maturation in MDS mice To understand how iron overload influences the hematopoietic system, we first evaluated WBC, PLT and HG counts. However, our experiment showed no statistical difference but a downward trend in the PB between the group treated with iron and that treated with normal saline in both control mice and MDS mice (Figure 1D-F). We further analyzed the frequency of HSCs (Lin–c-Kit+Sca1+/LSK) and HPCs (Lin–c-Kit+Sca1–/LK) in BM (Figure 3A) and we found that there were significantly more normal (GFP–) HSCs in MDS mice than in control mice, which may due to extremely active myeloid hyperplasia in MDS. In addition, iron overload can significantly decrease the number of GFP– HSCs in both control and MDS mice. However, we did not observe any significance about mutant (GFP+) HSCs in RX291/NS and RX291/FE mice (Figure 3B). A similar result can be seen regarding the change in normal and mutant HPCs in these groups (Figure 3C). Next, we analyzed erythroid differentiation through CD71 and Ter119 gating (Figure 3D). We referred to proerythroblasts, basophilic erythroblasts, polychro1629

X. Jin et al.

matophilic erythroblasts and orthochromatic erythroblasts as E1, E2, E3 and E4, respectively. Compared to Empty mice, MDS mice showed a significant increase in the number of GFP- E1 and a significant decrease in GFPE2 (Figure 3E), which may due to erythroid dysmaturity in MDS. Of note, iron overload can significantly reduce the frequency of E3 and E4 in control mice and GFP– E1 in MDS mice, while it did not exhibit any significant inhibition in GFP+ MDS erythroid cells, which may suggest that iron overload has a negative effect on normal hematopoiesis in erythroid, but not malignant clones. Collectively, our study shows that iron overload impairs the frequency of normal HSPCs and may inhibit erythroid maturation in MDS mice.

Iron overload compromises the erythroid colony-forming capacity of normal HSPCs in MDS mice We next wanted to examine if iron overload has a negative impact on HSPCs function in MDS mice. Therefore, we collected BM cells and determined the colony-forming capacity or hematopoietic cells reconstitution ability of these cells. Our data showed that normal (GFP–) HSPCs in MDS mice have a significantly lower ability to form colony-forming unit erythroid (CFU-E), burst-forming unit erythroid (BFU-E), and colony-forming unit granulocyte, erythroid, macrophage, megakarycyte (CFU-GEMM) colonies compared to control mice (Figure 4A). What is more, iron overload can markedly inhibit CFU-E and BFUE of GFP- HSPCs in both control and MDS mice. We also tested a serial replating assay about GFP– HSPCs (Figure 4B). Despite the fact that the first plating did not show sig-

nificantly reduced clonogenic numbers, but there seemed to be a declining trend between Empty/NS and Empty/FE mice, the count of clonogenic HSPCs exhibited a marked decline at the second and subsequent platings (Figure 4B), which supported the view that the colony-forming capacity of normal HSPCs was suppressed by iron overload in MDS mice. Because GFP+ HSPCs were unable to form colonies but clusters compared to GFP– HSPCs (Figure 4C), we performed competitive repopulation assay to identify the effect of iron overload on the hematopoietic reconstitution capacity of GFP+ cells (Figure 4D). To our surprise, the chimerism of GFP+ cells from RX291/FE mice was not significant compared to those of RX291/NS mice (Figure 4D), indicating that iron overload may not damage the hematopoietic reconstitution capacity of GFP+ HSPCs. Together, we conclude that iron overload can compromise the erythroid colony-forming capacity of normal HSPCs, but may not affect the hematopoietic reconstitution capacity of abnormal HSPCs in MDS.

Iron overload inhibits erythroid hematopoiesis in MDS mice through ROS These data showed that iron overload inhibits erythroid hematopoiesis in MDS mice. We next investigated the mechanisms involved in this process. Previous studies have reported that enhanced apoptosis of BM cells in MDS can cause pancytopenia.7 Indeed, our data showed that the level of apoptosis in RX291/NS and RX291/FE mice was significantly increased compared to Empty/NS and Empty/FE group (Figure 5A), respectively, consistent with previous studies.8 In addition, iron overload can

Figure 2. Iron overload model can be established in RUNX1S291fs induced myelodysplastic syndrome mice. (A) Weight of liver. (B) Weight of spleen. (C) Perl’s iron staining of liver, spleen and bone marrow (BM). (D) Ferritin level detected by ELISA assay. **P<0.01, ***P<0.001.

1630

haematologica | 2018; 103(10)

IO impairs normal HSPCs and survival in MDS mice

induce apoptosis of BM cells in normal mice (Figure 5A), which has previously been confirmed by others.14 Of note, normal (GFP–) BM cells in RX291/FE mice have a markedly higher apoptosis level than those of RX291/NS mice, but not mutant (GFP+) cells (Figure 5A), implying that iron overload can promote apoptosis of normal BM

cells in MDS. It has been reported that ROS can induce apoptosis.14 Our data showed that the ROS level was also clearly higher in RX291/FE mice than that of RX291/NS mice (Figure 5B) in BM. And erythroid cells in BM presented similar changes (Figure 5C). We also detected the mRNA levels of NOX4 related to ROS gen-

Figure 3. Iron overload impairs the frequency of normal hematopoietic stem and progenitor cells (HSPCs) and affects erythroid maturation in myelodysplastic syndrome (MDS) mice. (A) Gating hematopoietic stem cells (HSCs) and hematopoietic progenitor cells (HPCs) by flow cytometry. (B) The frequency of HSCs in different groups. (C) Frequency of HPCs in different groups. (D) Different stages of erythroid through CD71 and Ter119 gating. (E) Effect of iron overload on erythroid differentiation. *P<0.05, **P<0.01. BMMNCs: bone marrow mononuclear cells; GFP-: normal; GFP+: mutant; LSK: Lin–c-Kit+Sca1+, HSCs; LK: Lin–c-Kit+Sca1–, HPCs; E1: proerythroblasts; E2: basophilic erythroblasts; E3: polychromatophilic erythroblasts; E4: orthochromatic erythroblasts.

Figure 4. Iron overload compromises the erythroid colony-forming capacity of normal hematopoietic stem and progenitor cells (HSPCs) in myelodysplastic syndrome mice. (A) The colonies generated by GFP- cells. (B) A serial replating assay about GFP– HSPCs. (C) Comparison of the colony-forming capacity between GFP– and GFP+ cells. Left column is in bright field; right column is in fluorescence field. Red arrow indicates the location of GFP+ cells. (D) Experimental scheme of our mouse model using GFP+ cells to perform competitive repopulation assay and detecting chimerism of GFP+ cells post transplantation. *P<0.05. CFU-E: colony-forming unit erythroid; BFU-E: burst-forming unit erythroid; GM: colony-forming unit granulocyte-macrophage; GEMM: colony-forming unit granulocyte, erythroid, macrophage, megakarycyte.

haematologica | 2018; 103(10)

1631

X. Jin et al.

eration and GPX1 involved in ROS clearance in erythroid cells. Interestingly, the expression of NOX4 was significantly higher in RX291/FE mice than that of RX291/NS mice (Figure 5D), while the level of GPX1 was significantly lower in the RX291/FE group (Figure 5E), supporting the view that iron overload can induce ROS to damage erythroid hematopoiesis. Previous studies have reported that TGF-b signaling is myelosuppressive and inhibits erythroid differentiation by induction of ROS and apoptosis in erythroblasts.15-17 We then examined the mRNA levels of TGF-b superfamily including GDF8, GDF11, GDF15, Activin A, Activin B, Acvr2b, ALK4, and ALK5, and found that GDF11 mRNA levels were significantly increased in RX291/FE mice compared to RX291/NS mice (Figure 5F). We further detected the GDF11 protein levels in serum, which showed similar results in these two groups (Figure 5G). Collectively, iron overload can damage erythroid hematopoiesis in MDS mice, which may partially be due to GDF11-induced ROS, leading to enhanced apoptosis of normal BM cells and inhibition of their function in MDS.

Iron overload shortens survival MDS mice Given that iron overload results in a suppressive effect in the frequency and function of normal HSPCs in MDS mice, we postulated that iron overload can affect the median survival of these mice. The mice were observed for 360 days post transplantation and each group has more than ten mice. In our study, MDS mice have an obviously shorter median survival time than control mice (Figure 6), mostly due to AML transformation. For control mice, the survival time did not differ between Empty/NS and Empty/FE mice because the experiment was terminated at the end of day 360. However, RX291/FE mice exhibited a significantly shorter median survival than RX291/NS mice (Figure 6), suggesting that iron overload can shorten the survival time of MDS mice.

Discussion Myelodysplastic syndrome is a heterogeneous group of clonal stem cell disorders characterized by multilineage dysplasia, ineffective hematopoiesis, pancytopenia, and with a high risk of developing into acute myeloid leukemia.18 Over recent years, investigators all over the world have been trying hard to establish an MDS model to further study this disease. However, only a few mouse models are available for human MDS. RUNX1 is a frequently mutated gene in MDS7 and RUNX1S291fs is one of the most common mutations in RUNX1. It has been demonstrated that transduction of this mutant into BM cells can be used to construct MDS mice.8 Indeed, our RUNX1S291fs-induced mice showed neutropenia, anemia, thrombocytopenia, multilineage dysplasia, together with less than 20% blasts, matching the criteria of MDS in the Bethesda proposal and phenotypically recapitulating human MDS. Due to frequent blood transfusions and ineffective hematopoiesis, most MDS patients will eventually develop iron overload. Excessive iron deposits in the tissues and organs cause a series of complications, affecting the prognosis of MDS patients. In turn, iron overload has a negative role in hematopoiesis in MDS, further deteriorating the situation of these patients. To understand how iron overload influences hematopoiesis in MDS, we first established an iron overload model in RUNX1S291fsinduced MDS mice by intraperitoneal injection of iron dextran. Our data presented more iron deposition in organs, and more significant liver and spleen enlargement in RX291/FE mice, supporting the successful establishment of an iron overload model in RUNX1S291fs-induced MDS mice. In the present study, although there was no statistical significance in the levels of WBC, PLT and HG in the PB between Empty/NS and Empty/FE, we observed a declining trend in Empty/FE relative to Empty/NS mice,

Figure 5. Iron overload inhibits erythroid hematopoiesis in myelodysplastic syndrome (MDS) mice through reactive oxygen species (ROS). (A) Level of apoptosis in MDS mice. (B) ROS level in the bone marrow (BM) and erythrocytes. (C) ROS level in the erythrocytes. (D) mRNA expression of NADPH oxidase 4 (NOX4) gene. (E) mRNA expression of glutathione peroxidase 1 (GPX1) gene. (F) mRNA expression of growth differentiation factor 11 (GDF11) gene. (G) The concentration of GDF11 protein in serum detected by ELISA assay. *P<0.05, **P<0.01, ***P<0.001. BMMNC: bone marrow mononuclear cells.

1632

haematologica | 2018; 103(10)

IO impairs normal HSPCs and survival in MDS mice

Figure 6. Iron overload shortens the survival time of myelodysplastic syndrome mice.

similar to our previous study.13 We also obtained similar results in the PB between the two groups of MDS mice. In addition, we found that iron overload impairs the frequency of normal HSPCs, especially in erythroid cells, but not abnormal HSPCs in MDS mice. We further investigated the impact of iron overload on HSPCs function and found that iron overload compromises the erythroid colonyforming capacity of normal HSPCs, but may not affect the hematopoietic reconstitution capacity of abnormal HSPCs in MDS mice. In addition, RX291/FE mice have a shorter median survival than RX291/NS mice, indicating that iron overload can shorten survival time in MDS mice. There are increasing data to show that oxidative stress was increased in BM cells of patients with iron overload, and antioxidant or iron chelator therapy could partially rescue the impaired hematopoietic function of patients, which indicates the presence of ROS-induced cellular injury.19,23 Of interest, we detected the level of ROS and found it significantly increased in RX291/FE compared with RX291/NS group, suggesting that iron overload impairs normal HSPCs, at least in part, via inducing ROS in MDS, consistent with previous studies.19,24 Our data showed that iron overload decreased the number of HSPCs partially due to ROS-induced apoptosis. However, whether the reduced HSPCs was regulated by HIF-1a/ROS or NF-ÎşB pathway warrants further investigation.20,24 It has been reported that the TGF-b pathway is myelosuppressive and inhibits erythroid differentiation by induction of ROS and apoptosis in erythroblasts.15-17,25 Our data showed in GDF11, one of the TGF-b superfamilies, mRNA and protein levels significantly increased in RX291/FE mice compared to RX291/NS mice, suggesting that iron overload can damage erythroid hematopoiesis in MDS mice, which may partially be due to GDF11-induced ROS, leading to enhanced apoptosis of normal BM cells and inhibition of their function in MDS. However, additional studies are needed to clarify whether GDF11-induced ROS and apoptosis of erythroid was related to the Fas-Fas ligand path-

haematologica | 2018; 103(10)

way.16,25 Interestingly, Masayo et al. found that iron overload can activate glucose metabolism and increase DNA methylation, which is associated with MDS pathogenesis and progression, and iron chelation can reverse these effects.26 Therefore, future studies should be conducted to evaluate the impact of iron overload on metabolic pathways such as glucose and lipid involved in MDS pathogenesis in our model. In addition, our mice can be used to illustrate the association between oxidative imbalance and iron overload in MDS.18 Our previous study showed that damaged mesenchymal stromal cells (MSC) were related to iron overload induced ROS in normal mice.21 However, Zheng et al. reported that iron overload damages MSC through AMPK/MFF/Drp1 pathway in MDS.27 Thus, further research should investigate the effect of iron overload on the BM microenvironment in MDS. In conclusion, our preliminary findings suggest that iron overload impairs the frequency and function of normal HSPCs, particularly in erythroid, at least in part via GDF11-induced ROS, and shortens survival in MDS. Given that there are a few MDS models available, and we are the first to utilize RUNX1-S291fs-induced MDS mice to successfully construct an iron overload model, we hope this model will be helpful for further exploring the influence and mechanism of iron overload on MDS. Acknowledgments The authors would like to thank Dr Atsushi Iwam for the valuable plasmids. We also thank Dr Gang Huang for excellent technical assistance. Funding This work was supported by grants from the National Natural Sciences Foundation of China (81400092), Tianjin Key Natural Science Foundation (17JCZDJC35800, 15JCQNJC45500), and Tianjin Key Science and Technology Program (2015K215, 15KG134, 16KG110), as well as Tianjin First Central Hospital.

1633

X. Jin et al.

References 1. Harada H. MDS: Recent progress in molecular pathogenesis and clinical aspects. Rinsho Ketsueki. 2017;58(10):1941-1950. 2. Ganguly BB, Banerjee D, Agarwal MB. Impact of chromosome alterations, genetic mutations and clonal hematopoiesis of indeterminate potential (CHIP) on the classification and risk stratification of MDS. Blood Cells Mol Dis. 2018;69:90-100. 3. Imran F, Phatak P. Decision points in the treatment of transfusional iron overload in patients with myelodysplastic syndromes: why, when, and how to chelate. Expert Rev Hematol. 2017;10(1):53-64. 4. Zeidan AM, Pullarkat VA, Komrokji RS. Overcoming barriers to treating iron overload in patients with lower-risk myelodysplastic syndrome. Crit Rev Oncol Hematol. 2017;117:57-66. 5. Gattermann N. Iron overload in myelodysplastic syndromes (MDS). Int J Hematol. 2018;107(1):55-63. 6. Wong C, Wong S, Leitch HA. Iron overload in lower international prognostic scoring system risk patients with myelodysplastic syndrome receiving red blood cell transfusions: Relation to infections and possible benefit of iron chelation therapy. Leuk Res. 2018;67:75-81. 7. Sood R, Kamikubo Y, Liu P. Role of RUNX1 in hematological malignancies. Blood. 2017;129(15):2070-2082. 8. Sashida G, Harada H, Matsui H, et al. Ezh2 loss promotes development of myelodysplastic syndrome but attenuates its predisposition to leukaemic transformation. Nat Commun. 2014;5:4177. 9. Morita S, Kojima T, Kitamura T. Plat-E: an efficient and stable system for transient packaging of retroviruses. Gene Ther. 2000;7(12):1063-1066. 10. Unnisa Z, Clark JP, Roychoudhury J, et al.

1634

11.

12. 13.

14.

15.

16.

17. 18.

19.

20.

Meis1 preserves hematopoietic stem cells in mice by limiting oxidative stress. Blood. 2012;120(25):4973-4981. Kogan SC, Ward JM, Anver MR, et al. Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice. Blood. 2002;100(1):238-245. Ramanathan G, Olynyk JK, Ferrari P. Diagnosing and preventing iron overload. Hemodial Int. 2017;21 Suppl 1:S58-S67. Chai X, Li D, Cao X, et al. ROS-mediated iron overload injures the hematopoiesis of bone marrow by damaging hematopoietic stem/progenitor cells in mice. Sci Rep. 2015;5:10181. Muto Y, Nishiyama M, Nita A, Moroishi T, Nakayama KI. Essential role of FBXL5mediated cellular iron homeostasis in maintenance of hematopoietic stem cells. Nat Commun. 2017;8:16114. Mies A, Platzbecker U. Increasing the effectiveness of hematopoiesis in myelodysplastic syndromes: erythropoiesis-stimulating agents and transforming growth factor-beta superfamily inhibitors. Semin Hematol. 2017;54(3):141-146. Dussiot M, Maciel TT, Fricot A, et al. An activin receptor IIA ligand trap corrects ineffective erythropoiesis in beta-thalassemia. Nat Med. 2014;20(4):398-407. Kim A, Nemeth E. New insights into iron regulation and erythropoiesis. Curr Opin Hematol. 2015;22(3):199-205. Ivars D, Orero MT, Javier K, et al. Oxidative imbalance in low/intermediate1-risk myelodysplastic syndrome patients: The influence of iron overload. Clin Biochem. 2017;50(16-17):911-917. Angelucci E, Cianciulli P, Finelli C, Mecucci C, Voso MT, Tura S. Unraveling the mechanisms behind iron overload and ineffective hematopoiesis in myelodysplastic syndromes. Leuk Res. 2017;62:108-115. Meunier M, Ancelet S, Lefebvre C, et al.

21.

22.

23.

24.

25.

26.

27.

Reactive oxygen species levels control NFkappaB activation by low dose deferasirox in erythroid progenitors of low risk myelodysplastic syndromes. Oncotarget. 2017;8(62):105510-105524. Zhang Y, Zhai W, Zhao M, et al. Effects of iron overload on the bone marrow microenvironment in mice. Plos One. 2015;10(3):e120219. Lu W, Zhao M, Rajbhandary S, et al. Free iron catalyzes oxidative damage to hematopoietic cells/mesenchymal stem cells in vitro and suppresses hematopoiesis in iron overload patients. Eur J Haematol. 2013;91(3):249-261. Chen J, Lu WY, Zhao MF, et al. Reactive oxygen species mediated T lymphocyte abnormalities in an iron-overloaded mouse model and iron-overloaded patients with myelodysplastic syndromes. Ann Hematol. 2017;96(7):1085-1095. Zheng QQ, Zhao YS, Guo J, et al. Iron overload promotes erythroid apoptosis through regulating HIF-1a/ROS signaling pathway in patients with myelodysplastic syndrome. Leuk Res. 2017;58:55-62. Arlet JB, Dussiot M, Moura IC, Hermine O, Courtois G. Novel players in beta-thalassemia dyserythropoiesis and new therapeutic strategies. Curr Opin Hematol. 2016;23(3):181-188. Yamamoto M, Tanaka H, Toki Y, et al. Iron-induced epigenetic abnormalities of mouse bone marrow through aberrant activation of aconitase and isocitrate dehydrogenase. Int J Hematol. 2016;104(4):491501. Zheng Q, Zhao Y, Guo J, et al. Iron overload promotes mitochondrial fragmentation in mesenchymal stromal cells from myelodysplastic syndrome patients through activation of the AMPK/MFF/Drp1 pathway. Cell Death Dis. 2018;9(5):515.

haematologica | 2018; 103(10)

ARTICLE

Phagocyte Biology and its Disorders

Bone marrow histomorphological criteria can accurately diagnose hemophagocytic lymphohistiocytosis

Ferrata Storti Foundation

Eric Gars,1 Natasha Purington,1 Gregory Scott,1 Karen Chisholm,2 Dita Gratzinger,1 Beth A. Martin1* and Robert S. Ohgami1* Stanford University, CA and 2Seattle Children's Hospital and University of Washington, WA, USA

BAM and RSO contributed equally to this work.

Haematologica 2018 Volume 103(10):1635-1641

emophagocytic lymphohistiocytosis (HLH) is a rare multi-system inflammatory disorder with diagnostic criteria based on the HLH2004 trial. Hemophagocytosis is the only histomorphological criterion, but in isolation is neither specific nor sensitive for the diagnosis of HLH. While objective thresholds for clinical and laboratory criteria have been established, specific criteria for histomorphological evidence of hemophagocytosis in HLH have not been rigorously evaluated or established. We sought to determine if numerical and objective criteria for morphological hemophagocytosis could be identified, and if such criteria would aid in the diagnosis of HLH. We analyzed the morphological features of hemophagocytosis in 78 patients presenting with clinical features suspicious for HLH: 40 patients with and 38 patients without HLH. We demonstrate that non-nucleated erythrophagocytosis alone is a non-specific finding, while hemophagocytosis of granulocytes [1 per 1000 cells, area under the curve (AUC): 0.92, 95% Confidence Interval (CI): 0.86, 0.99], nucleated erythrocytes (4 per 1000 cells, AUC: 0.92, 95%CI: 0.87, 0.98), and at least one hemophagocyte containing multiple nucleated cells (AUC: 0.91, 95%CI: 0.85, 0.95) are strongly associated with HLH. Joint modeling of hemophagocytes containing engulfed granulocytes, nucleated erythrocytes, and lymphocytes effectively distinguished between HLH and non-HLH (cross-validated AUC: 0.90, 95%CI: 0.83, 0.97). Introduction Hemophagocytic lymphohistiocytosis (HLH) is a rare life-threatening syndrome that occurs secondary to severe systemic immune activation.1 Cytotoxic T-cell proliferation leads to increased cytokine production and activation of tissue resident macrophages. Ultimately, multi-system end organ damage caused by massive inflammation may lead to a fatal outcome without timely diagnosis and initiation of appropriate therapy.2 Hemophagocytic lymphohistiocytosis affects patients of all ages and occurs as an inherited disease, or secondarily in the setting of predisposing conditions that alter the normal immune response. The inherited form of the disease presents in early childhood and is associated with homozygous mutations in genes involved in CD8+ T-cell- and NK-cell-mediated immunity.3 These genetic forms of HLH are uniformly fatal without hematopoietic cell transplant or gene therapy. Secondary HLH may occur sporadically in healthy individuals, but is more often encountered in patients with hematologic malignancy, autoimmune disease, and iatrogenic immunosuppression. Virtually all cases are thought to require an infectious or noninfectious trigger to initiate the aberrant immune response, regardless of the underlying immune dysfunction.4-7 Hemophagocytic lymphohistiocytosis presents abruptly over a period of several days to weeks with a consistent pattern of fever, pancytopenia, and splenomegaly. Common laboratory abnormalities include hyperferritinemia, hypofibrinogenemia, hypertriglyceridemia, elevated soluble IL-2 receptor, and abnormal liver func-

haematologica | 2018; 103(10)

Correspondence: ericgars@stanford.edu or rohgami@stanford.edu Received: December 17, 2017. Accepted: June 13, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2017.186627 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1635 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1635

E. Gars et al.

Figure 1. Examples of hemophagocytosis in patients with hemophagocytic lymphohistiocytosis (HLH). (A) Histiocytes in patients with HLH often display rounded contour with cytoplasmic projections. (B-D) Hemophagocytes with a single ingested mature red blood cell (RBC), nucleated RBC progenitor, and granulocyte, respectively. Hematopoietic progenitor cells (HPCs) often contain single nucleated hematopoietic cells in addition to multiple mature RBCs (E); however, the presence of multiple nucleated cells within the cytoplasm of a single HPC (F and G) is highly predictive of the diagnosis of HLH. (H) An example of a histiocyte with degenerating nuclear debris, indistinct cytoplasmic contour, and equivocal intracytoplasmic nucleated RBCs that we do not consider to be a definite hemophagocyte.

tion tests.1 The most widely used diagnostic criteria for HLH were developed for inclusion in the HLH-2004 trial which requires genetic evidence of a mutation associated with HLH or fulfillment of 5 of 8 clinical criteria including fever, splenomegaly, bicytopenia, hypertriglyceridemia or hypofibrinogenemia, evidence of hemophagocytosis in bone marrow or other tissues, low or absent NK-cell activity, elevated ferritin, and elevated soluble IL-2 receptor.3 Although not validated for adults, these HLH-2004 criteria are broadly applied in patients of all ages. Pathologists play a critical role in the diagnostic workup of patients suspected of having HLH. Bone marrow examination is performed to evaluate for hemophagocytosis, identify underlying malignancy, and exclude benign or neoplastic mimics. The presence of hemophagocytosis in the marrow fulfills one of the HLH-2004 diagnostic criteria; however, no accepted diagnostic threshold or reporting guidelines have been established. The lack of evidence-based guidelines leads to considerable uncertainty among pathologists as to what degree of hemophagocytosis is sufficient to satisfy this criterion. Adding to the challenge is that hemophagocytosis is not specific to the diagnosis of HLH in the absence of other clinical features of the disease. Rare erythrophagocytosis is commonly seen in bone marrow aspirates and increased hemophagocytosis may be encountered in the setting of sepsis, blood transfusions, hematopoietic transplantation, chemotherapy, and myelodysplastic syndrome.8-11 Given the lack of a defined threshold to fulfill the criterion for diagnosis of HLH, we designed this retrospective study to interrogate whether quantitative or qualitative morphological features of hemophagocytosis in bone marrow aspirates are predictive of the eventual diagnosis of HLH. We identified a cohort of patients presenting with clinical characteristics that were of concern for HLH and their aspirates were examined blindly. Hemophagocytes were enumerated per 1000 nucleated 1636

cells according to the lineage of their ingested hematopoietic contents [mature red blood cells (RBCs), nucleated RBCs (nRBCs), granulocytes, and lymphocytes] (Figure 1). In addition to quantitative features, we evaluated a binary morphological feature, the presence of multiple nucleated cells within a single hemophagocyte, as a possible predictive characteristic of HLH.

Methods Patient selection We searched the pathology Laboratory Information Service database (Powerpath) using the following keywords: “hemophagocytic lymphohistiocytosis”, “hemophagocytosis”, “erythrophagocytosis”, and “HLH”. This search returned 258 results between the dates 1st January 2013 and 7th January 2017, and included text from anywhere within the diagnostic report including the provided clinical information, microscopic description, diagnostic line, and/or diagnostic comment (Figure 2). These patients’ medical records were reviewed by EG to assess whether clinical suspicion for HLH was present at the time of bone marrow aspiration, either indicated on the specimen requisition form (i.e. “rule out HLH” or “concern for HLH”) or listed in the differential diagnosis in the electronic medical record (EMR) within one week prior to biopsy. Demographic information, clinical characteristics, diagnostic impressions, pathological features, and laboratory values at the time of biopsy were collected for each patient. Patients were classified as “HLH” and “non-HLH” based on the diagnostic impression of the consulting hematologists described in the clinical notes. The final diagnosis in all cases was determined based on the HLH-2004 criteria in conjunction with the overall clinical picture. Patients were excluded from the analysis if hemophagocytosis was incidentally noted independent of clinical concern for HLH, slides were not available for review, HLH was considered but the diagnosis was equivocal after workup, or a documented history of HLH-directed treatment was noted prior to biopsy. This study was approved by Stanford University’s institutional review board. haematologica | 2018; 103(10)

Hemophagocytic lymphohistiocytosis in marrows

Table 1. Clinical and laboratory findings of patients in hemophagocytic lymphohistiocytosis (HLH) and non-HLH groups.

Patients' characteristics Age in years, mean (range) Male, % HLH-2004 Criteria† Fever ≥ 38.5°C, n/N (%)† Splenomegaly, n/N (%)† Bicytopenia, n/N (%)† Triglycerides >265 mg/dL, n/N (%)† Triglycerides, mean (range), mg/dL Fibrinogen <150 mg/dL, n/N (%)† Fibrinogen, mean (range) Ferritin >500 ng/mL, n/N (%)† Ferritin, median (range), ng/mL Elevated soluble IL-2 receptor (>2400 U/mL), n/N (%)† Soluble IL-2 receptor, median (range), U/mL Decreased NK-cell function, n/N (%)† Hemophagocytosis, n/N (%)† Malignancy, n/N (%) Autoimmune disease, n/N (%) Hemophagocytosis Criteria Sum (excluding hemophagocytosis), n/N (%) 0 1 2 3 4 5 6 7

HLH (N=40)

Non-HLH (N=38)

50 (10mo-96) 54%

41 (1mo-85) 55%

38/40 (95%) 26/37 (70%) 38/40 (95%) 15/38 (39%) 301 (49-812) 17/39 (44%) 262 (57-779) 39/40 (98%) 19,331 (425-40000) 30/32 (94%) 14,963 (1215-69,000) 2/12 (17%) 40/40 (100%) 22/40 (56%) 11/40 (28%)

29/38 (76%)* 16/32 (42%)* 22/38 (58%)* 2/31 (6%)* 174 (39-479)* 3/33 (9%)* 413 (64-1200)* 27/36 (75%)* 6,553 (30-40,000)* 7/21 (33%)* 1782 (139-4,431)* 3/8 (38%)* 12/38 (32%)* 9/38 (24%)* 15/38 (38%)

0 0 0 2 (5%) 9 (23%) 16 (40%) 12 (30%) 1 (3%)

1 (3%) 4 (11%) 7 (18%) 14 (37%) 8 (21%) 4 (11%) 0 0

n/N: number; mo: months; *P<0.05. †One of HLH-2004 criteria (one criteria is fibrinogen <150 mg/dL or triglycerides >265 mg/dL).

Evaluation of bone marrow aspirates Bone marrow aspirates (Wright-Giemsa stained) of HLH and non-HLH patients were evaluated blindly. Every aspirate slide was initially examined at low power (4x) to identify areas with hemophagocytosis and select an appropriate slide for enumerating hemophagocytes. Two hundred and fifty intact nucleated cells were counted in each quadrant on a single aspirate per case in areas with the highest density of hemophagocytes. Histiocytes were tallied by the lineage of ingested cells (mature RBCs, nRBCS, granulocytes, and lymphocytes) and the hematopoietic progenitor cell (HPC) sum was calculated as the total number of histiocytes containing ingested hematopoietic cells. The presence of multiple nucleated cells in a single hemophagocyte was also noted. The methods used for the statistical analysis are described in the Online Supplementary Appendix.

Results Patients’ characteristics Patients' characteristics of the 40 patients with HLH and 38 patients without HLH are summarized in Table 1. There were no significant differences in age or sex. HLH patients were more likely to present with underlying haematologica | 2018; 103(10)

malignancy compared to the non-HLH group (56% vs. 24%; P<0.05), with diffuse large B-cell lymphoma (DLBCL) being the most common primary diagnosis among patients with HLH (Tables 1 and 2). Epstein-Barr virus (EBV) was the most common infectious trigger identified in patients with HLH; 32.5% of the patients in the HLH group had evidence of EBV infection by peripheral blood PCR detection or immunohistochemistry compared to 10.5% in the non-HLH group (P<0.01). As expected, patients diagnosed with HLH were more likely to present with clinical and laboratory findings satisfying each of the HLH-2004 diagnostic criteria (Table 1). Significant differences were observed between the HLH and non-HLH groups in the average value of each laboratory test (triglycerides, fibrinogen, ferritin, and soluble IL2r) and the number of patients satisfying each individual HLH-2004 criterion, with the exception of natural killer (NK)-cell function. Although impaired or absent NK-cell function testing is considered a valid screening tool for patients with genetic defects in cytotoxicity, this test is rarely ordered in adults with secondary HLH and has a high failure rate, limiting its diagnostic utility. Hemophagocytosis was reported in the diagnostic bone 1637

E. Gars et al.

marrow report of all 40 patients with HLH, compared to 12 of 38 (32%) in the non-HLH group (P<0.01). Although evidence of hemophagocytosis may be reassuring for clinicians in making the diagnosis of HLH, hemophagocytosis is not required. In our cohort, the majority of patients ultimately diagnosed with HLH (73%) fulfilled 5 or more criteria independently of evidence of hemophagocytosis in the bone marrow aspirate (Table 1). However, 9 patients (23%) in the HLH group and 8 patients (21%) in the non-HLH group met 4 criteria excluding hemophagocytosis, indicating that morphological assessment of the bone marrow aspirate was critical to the diagnosis in 22% of patients presenting with clinical features of concern for HLH. Two patients in the HLH group satisfied fewer than 5 of the HLH-2004 criteria; however, both of these patients were surgically asplenic and the remainder of their clinical and laboratory findings, including the presence of hemophagocytosis on bone marrow aspirate, were compatible with the diagnosis. Four patients in the non-HLH group satisfied 5 criteria but were not diagnosed with HLH. The clinical notes of the treating physicians indicate that the chronicity of symptoms was considered to be inconsistent with HLH. None of these patients had hemophagocytosis identified in their bone marrow.

Figure 2. Flow chart for classification of patients. *Patients were excluded from the analysis if hemophagocytosis was incidentally noted independent of clinical concern for hemophagocytic lymphohistiocytosis (HLH), slides were not available for review, HLH was considered but the diagnosis was equivocal after workup, or a documented history of HLH-directed treatment was noted prior to biopsy.

1638

Developing predictive models for a morphological diagnosis of hemophagocytic lymphohistiocytosis Patients diagnosed with HLH (n=40) displayed significantly higher values of total hemophagocytes, and hemophagocytes with any of the cell lineages (RBCs, nRBCs, granulocytes, and lymphocytes) compared to non-HLH patients (n=38) (P<0.001 for all lineages) (Figures 1 and 3). Correlation analysis demonstrated that each of the clinical and laboratory criteria included in the HLH-2004 criteria (excluding NK function) has a significant positive correlation with total number of hemophagocytes and hemophagocytes with each of the individual cell lineages, indicating that the degree of hemophagocytosis correlates with the diagnosis of HLH (Online Supplementary Figure S1). Independently of one another, each lineage distinguished between HLH and non-HLH fairly well. Ingested nRBCs and ingested granulocytes had the highest area under the curve (AUC), indicating a high degree of correct classification of HLH and non-HLH subjects (AUC: 0.92), with threshold values of 2 ingested cells and 1 ingested cell, respectively. The sum of all four lineages also performed well at distinguishing between HLH and nonHLH with a threshold value of 6 (AUC: 0.92, 95%CI: 0.85, 0.98). Dichotomizing each lineage based on the threshold values described above and including all four in a decision tree, hemophagocytes ingesting granulocytes was chosen as the most important predictor of HLH, followed by nRBCs and lymphocytes (Figure 4A). Patients with an absence of hemophagocytes ingesting granulocytes had a 3% chance of having an HLH diagnosis. Patients with a presence of at least one hemophagocyte with an ingested granulocyte, two or more hemophagocytes with ingested nRBCs, and one hemophagocyte with ingested lymphocytes together were guaranteed to have an HLH diagnosis (100% chance). Cross-validated (CV) AUC of this CART was 0.90; 95%CI: 0.83-0.97. Identifying lymphocytes within HPCs is rare, even in the presence of florid HLH. The highest number of HPCs containing lymphocytes was identified in a patient with

Figure 3. Values of hematopoietic progenitor cell (HPC) lineages and sum by hemophagocytic lymphohistiocytosis (HLH) diagnosis. Patients ultimately diagnosed as HLH had significantly higher values of all variables compared to nonHLH patients (Kruskal-Wallis rank sum test, P<0.001).

haematologica | 2018; 103(10)

Hemophagocytic lymphohistiocytosis in marrows

DLBCL and based on morphology likely represented ingested tumor cells. Additionally, distinguishing between nucleated erythrocytes, lymphocytes, and hematopoietic progenitor cells within the cytoplasm of a histiocyte is challenging and subject to interpretive variability. As such, we created an additional CART excluding lymphocytes. This CART is identical to the first two levels of the previous model, with patients having a 92% chance of having an HLH diagnosis with a presence of at least one hemophagocyte with an ingested granulocyte and two or more hemophagocytes with ingested nRBCs (Figure 4B). While the CV AUC was identical to that of the first, overall accuracy was slightly higher in the CART without lymphocytes (88% CV accuracy vs. 86%).

Qualitative evaluation The initial description of virus-associated hemophagocytic syndrome described histiocytes that were â&#x20AC;&#x153;filledâ&#x20AC;? with ingested hematopoietic elements.11 We frequently observe the phenomenon of multiple nucleated cells within individual histiocytes in patients with established HLH and hypothesized that this finding may be indicative of a pathological hemophagocytic state. To evaluate the significance of the degree of hemophagocytosis within individual hemophagocytes, we analyzed the presence of multiple nucleated cells within histiocytes (Figure 1). Our analysis ultimately demonstrated that at least one hemophagocyte-containing multiple nucleated cells was identified in 37 patients with HLH compared to only 4 patients in the non-HLH group (AUC: 0.91, 0.845-0.947), indicating that this binary qualitative feature performs similarly to the quantitative metrics described above for distinguishing patients with HLH from non-HLH patients (Table 3).

Discussion Given the rarity of the diagnosis and non-specific clinical presentation, HLH is a challenging diagnosis for clinicians and pathologists. Once considered in the differential, expedient workup including evaluation of each of the HLH-2004 criteria is commonly pursued, as early therapeutic intervention improves outcomes in these often critically ill patients. Unfortunately, there is no defined threshold to satisfy the diagnostic criterion of hemophagocytosis in the bone marrow and evidence-based guidelines for reporting findings have not been established. Towards this goal, we designed this retrospective study to determine whether quantitative features of hemophagocytosis at the time of initial bone marrow assessment are predictive of the ultimate diagnosis of HLH in patients presenting with clinical features concerning for the diagnosis. Given that HLH classically presents with multilineage cytopenia that is thought to result from consumption of hematopoietic cells by activated macrophages,12 we suspected that patients with HLH would be more likely to display hemophagocytosis of nucleated erythrocytes, granulocytes and lymphocytes compared to patients without HLH. In addition to evaluating for possible quantitative differences in hemophagocytosis, we simultaneously assessed for the presence of multiple nucleated cells within a single hemophagocyte as a candidate binary morphological feature that may differentiate patients with pathological hemophagocytosis. Overall, we found that patients with HLH displayed significantly higher numbers of HPCs by each of the lineages examined (Table 3). We also identified quantitative thresh-

Table 2. Underlying medical conditions of patients in hemophagocytic lymphohistiocytosis (HLH) and non-HLH groups.

HLH Diagnosis (N. of patients) Diffuse large B-cell lymphoma (9) Peripheral T-cell lymphoma (6) NK/T-cell lymphoma/leukemia (2) EBV+ lymphoproliferative disorder (2) Classical Hodgkin Lymphoma (EBV+) (1) Langerhans cell histiocytosis (1) Primary effusion lymphoma (HIV+ and HHV-8+) (1) Juvenile xanthogranuloma (1) B-cell acute lymphoblastic leukemia (1) Rheumatoid arthritis (3) Systemic lupus erythematosus (2) Sarcoidosis (2) Dermatomyositis (1) Atypical Kawasaki disease (1) Ulcerative colitis (1) Idiopathic (5) Post ventricular assist device (3) Griscelli syndrome (homozygous Rab27a mutation) (1) Homozygous perforin mutation (1)

Non-HLH Diagnosis (N. of patients) Classical Hodgkin lymphoma (EBV+) (2) Diffuse Large B-cell lymphoma (1) Peripheral T-cell lymphoma (2) T-Cell Acute lymphoblastic leukemia (1) FIP1L1-PDGFRa myeloid leukemia (1) Blastic plasmacytoid dendritic cell neoplasm (1) Autoimmune hemolytic anemia (2) Systemic lupus erythematosus (5) Adult onset still's disease (3) Autoimmune hepatitis (1) Severe acquired neutropenia (1) Aplastic anemia (1) Autoimmune disorder NOS Dermatomyositis (1) Autoimmune lymphoproliferative Syndrome (1) Mixed connective tissue disease (1) Infection (6) White cell aplasia (1) Other (15)

N: number; NK: natural killer cell; EBV: Epstein-Barr virus; NOS: not otherwise specified.

haematologica | 2018; 103(10)

1639

E. Gars et al.

olds that can be used to accurately diagnose most cases of HLH: one granulocyte or two nucleated erythroid cells per 1000 nucleated cells. In addition, we utilized classification and regression tree analysis to identify the best combination of variables to create an even more specific and predictive model for discriminating patients with HLH. Finally, we showed that the presence of multiple nucleated cells within a single hemophagocyte was additionally predictive of a diagnosis of HLH among patients presenting with clinical findings of concern for the disease. The initial description of virus-associated hemophagocytic syndrome (VAHS) described florid hemophagocytosis in 19 cases of immunosuppressed and previously healthy patients presenting with clinical features compatible with HLH.11 In 60 consecutive bone marrow aspirates used for comparison, they found erythrophagocytosis in 29 of 60 cases; however, the degree of phagocytosis was â&#x20AC;&#x153;never of a degree to be confused with the VAHSâ&#x20AC;?. This study conducted prior to the development of the HLH-2004 criteria provided initial evidence of the association of hemophagocytosis with VAHS (what we

would now call HLH), and showed that the finding of erythrophagocytosis is frequently identified in bone marrow aspirates of patients without HLH, demonstrating a lack of specificity in isolation of clinical findings. We conducted a similar non-blinded evaluation of 87 bone marrow aspirates from patients with de novo and post-treatment myeloid and lymphoid malignancies, benign cytopenias, and negative staging marrows, and identified predominantly erythrophagocytosis of mature RBCs in 39% of these cases. Applying the quantitative threshold determined from our analysis for engulfed non-nucleated RBCs (erythrophagocytes) to this dataset, we find only 3 of the 87 patients (5%) would satisfy this morphological criterion (4 per 1000 cells). Similarly, if we apply the quantitative threshold for granulocytes (1 per 1000 cells), only 4 of the 87 patients (4%) would satisfy this morphological criterion. Therefore, although we can indeed find rare examples of erythrophagocytosis in aspirate smears of a large minority of cases, the application of quantitative thresholds reveals a low incidence of clinically significant hemophagocytosis in a patient pop-

Figure 4. Classification and regression trees (CART). Lineages were dichotomized based on the threshold values obtained from Table 3 and entered into the CART. Lineages are ordered in terms of relative importance to hemophagocytic lymphohistiocytosis (HLH), where variables on the higher levels are deemed more important. The shaded area in each box corresponds to the probability of having an HLH diagnosis based on the path that leads to it. All four lineages were entered into the CART in (A), while lymphocytes were excluded from the CART in (B).

1640

haematologica | 2018; 103(10)

Hemophagocytic lymphohistiocytosis in marrows

Table 3. Quantities of hemophagocytic cells by lineage of ingested hematopoietic cells and optimal cutoff values derived from Youden’s index.

Lineage RBC nRBC Granulocyte Lymphocyte HPC sum mHPC

Mean (range)* HLH

Non-HLH

7.7 (1-21) 5.2 (1-15) 4.3 (0-21) 1.3 (0-11) 18.4 (4-44) 37

3.1 (0-18) 0.8 (0-6) 0.5 (0-5) 0 4.3 (0-19) 4

Threshold

AUC (95% CI)

4 2 1 1 6 NA

0.83 (0.74-0.93) 0.92 (0.87-0.98) 0.92 (0.86-0.99) 0.80 (0.72-0.88) 0.92 (0.85-0.98) 0.91 (0.85-0.95)

RBC: red blood cells; nRBC: nucleated RBCs; HPC sum: total hematopoietic progenitor cell count from all four lineages; mHPC: multiple nucleated cells within a single HPC. *Per 1000 cells.

ulation that reflects those seen in routine diagnostic practice. A recent study by Ho et al. examined the specificity of hemophagocytosis for HLH by quantifying the absolute amount of hemophagocytosis identified in bone marrow aspirates of patients whose diagnostic pathology report described hemophagocytosis.10 They demonstrated that the presence of hemophagocytosis, even when present in a high amount, lacks specificity for HLH. Our institutional experience is consistent with their conclusion that significant hemophagocytosis is not predictive of the diagnosis of HLH in the absence of clinical features of concern for the disease; however, the presence of substantial hemophagocytosis is a relatively rare finding. Incidental hemophagocytosis was reported in 86 of 8097 (1.1%) in-house bone marrow biopsy reports at Stanford University Hospital from 2013-2017. The majority of reports indicate “rare” hemophagocytosis (61%) while the remainder describe “scattered”, “occasional”, or “brisk” hemophagocytosis. A subset (n=12) of these latter cases was reviewed. All 12 cases demonstrated erythrophagocytosis. Half of the cases show ingested granulocytes and 3 cases demonstrated multiple nucleated cells within individual HPCs. None of these patients were ultimately diagnosed with HLH. One major limitation of this study is its retrospective nature. Our population is limited to patients in whom the clinical diagnosis was thoroughly evaluated and a definitive determination was made concerning the diagnosis of

References 1. Jordan MB, Allen CE, Weitzman S, Filipovich AH, McClain KL. How I treat hemophagocytic lymphohistiocytosis. Blood. 2011;118(15):4041-4052. 2. George MR. Hemophagocytic lymphohistiocytosis: review of etiologies and management. J Blood Med. 2014;569-86. 3. Henter J-I, Horne A, Aricó M, et al. HLH2004: Diagnostic and therapeutic guidelines for hemophagocytic lymphohistiocytosis. Pediatr Blood Cancer. 2007;48(2):124131. 4. Buyse S, Teixeira L, Galicier L, et al. Critical care management of patients with hemophagocytic lymphohistiocytosis. Intensive Care Med. 2010;36(10):1695-1702. 5. Dhote R, Simon J, Papo T, et al. Reactive hemophagocytic syndrome in adult sys-

haematologica | 2018; 103(10)

HLH. We excluded patients in whom the ultimate diagnosis was ambiguous. Additionally, although our cohort includes all of the patients at our institution that met criteria for evaluation, including children, only a single patient in our cohort had a homozygous mutation diagnostic of primary HLH, limiting the applicability of our findings to patients with genetic forms of the disease. Finally, we note that while CD107a testing has been described as a more sensitive test for primary HLH,13 this recently developed assay was not utilized in this retrospective cohort. Ultimately, the diagnosis of HLH rests on the thorough assessment of patients in the appropriate clinical context. Similarly, microscopic examination of bone marrow aspirate smears in patients suspected of having HLH requires careful evaluation for the presence of hemophagocytosis. Our results demonstrate that quantitative thresholds of the lineage of ingested cells, either alone or in combination, accurately predict the eventual diagnosis of HLH. We additionally demonstrate that identification of a single hemophagocyte containing multiple nucleated hematopoietic cells within its cytoplasm performs similarly to our quantitative approach. With external validation and further prospective study, we hope that these data will help to provide a method for pathologists and clinicians to systematically evaluate and accurately classify patients with HLH and contribute to establishment of consensus guidelines of diagnosis and reporting.

temic disease: Report of twenty-six cases and literature review. Arthritis Care Res. 2003;49(5):633-639. Machaczka M, Vaktnäs J, Klimkowska M, Hägglund H. Malignancy-associated hemophagocytic lymphohistiocytosis in adults: a retrospective population-based analysis from a single center. Leuk Lymphoma. 2011;52(4):613-619. Ramos-Casals M, Brito-Zerón P, LópezGuillermo A, Khamashta MA, Bosch X. Adult haemophagocytic syndrome. Lancet Lond Engl. 2014;383(9927):1503-1516. Goel S, Polski JM, Imran H. Sensitivity and specificity of bone marrow hemophagocytosis in hemophagocytic lymphohistiocytosis. Ann Clin Lab Sci. 2012;42(1):21-25. Glasser L, LeGolvan M, Horwitz HM. Florid histiocytic hemophagocytosis following therapy with long acting G-CSF (pegfilgrastim). Am J Hematol. 2007;

82(8):753-757. 10. Ho C, Yao X, Tian L, Li F-Y, Podoltsev N, Xu ML. Marrow assessment for hemophagocytic lymphohistiocytosis demonstrates poor correlation with disease probability. Am J Clin Pathol. 2014;141(1):6271. 11. Risdall RJ, McKenna RW, Nesbit ME, et al. Virus-associated hemophagocytic syndrome: a benign histiocytic proliferation distinct from malignant histiocytosis. Cancer. 1979;44(3):993-1002. 12. Zoller EE, Lykens JE, Terrell CE, et al. Hemophagocytosis causes a consumptive anemia of inflammation. J Exp Med. 2011; 208(6):1203-1214. 13. Rubin TS, Zhang K, Gifford C, et al. Perforin and CD107a testing is superior to NK cell function testing for screening patients for genetic HLH. Blood. 2017; 129(22):2993-2999 .

1641

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1642-1653

Weiguo Zhang,1 Charlie Ly,1 Jo Ishizawa,1 Hong Mu,1 Vivian Ruvolo,1 Sharon Shacham,2 Naval Daver3 and Michael Andreeff1,3

Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX; 2Karyopharm Therapeutics Inc., Newton, MA and 3Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA. 1

ABSTRACT

Correspondence: mandreef@mdanderson.org

Received: November 20, 2017. Accepted: May 16, 2018. Pre-published: May 17, 2018.

doi:10.3324/haematol.2017.185082 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1642 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1642

argeted therapies against FLT3-mutated acute myeloid leukemias have shown limited clinical efficacy primarily because of the acquisition of secondary mutations in FLT3 and persistent activation of downstream pro-survival pathways such as MEK/ERK, PI3K/AKT, and STAT5. Activation of these additional kinases may also result in phosphorylation of tumor suppressor proteins promoting their nuclear export. Thus, co-targeting nuclear export proteins (e.g., XPO1) and FLT3 concomitantly may be therapeutically effective. Here we report on the combinatorial inhibition of XPO1 using selinexor and FLT3 using sorafenib. Selinexor exerted marked cell killing of human and murine FLT3-mutant acute myeloid leukemia cells, including those harboring internal tandem duplication and/or tyrosine kinase domain point mutations. Interestingly, selinexor treatment of murine FLT3-mutant acute myeloid leukemia cells activated FLT3 and its downstream MAPK or AKT signaling pathways. When combined with sorafenib, selinexor triggered marked synergistic pro-apoptotic effects. This was preceded by elevated nuclear levels of ERK, AKT, NFÎşB, and FOXO3a. Five days of in vitro combination treatment using low doses (i.e., 5 to 10 nM) of each agent promoted early myeloid differentiation of MOLM13 and MOLM14 cells without noticeable cell killing. The combinatorial therapy demonstrated profound in vivo anti-leukemia efficacy in a human FLT3mutated xenograft model. In an ongoing phase IB clinical trial the selinexor/sorafenib combination induced complete/partial remissions in six of 14 patients with refractory acute myeloid leukemia, who had received a median of three prior therapies (ClinicalTrials.gov: NCT02530476). These results provide pre-clinical and clinical evidence for an effective combinatorial treatment strategy targeting XPO1 and FLT3 in FLT3mutated acute myeloid leukemias.

Introduction Acute myeloid leukemia (AML) is a molecularly heterogeneous hematologic disease defined by the accumulation of immature myeloid cells in blood and bone marrow, which results from a dysregulation of normal proliferation, differentiation, and apoptosis in these cells.1 Mutations of the fms-like tyrosine kinase-3 (FLT3) gene, including internal tandem duplication (ITD) and the tyrosine kinase domain (TKD) mutations, are common in patients with AML with approximately one-third of newly diagnosed AML patients carrying these mutations. These gain-of-function haematologica | 2018; 103(10)

Effects of XPO1 and FLT3 inhibition on AML

mutations constitutively activate FLT3 and its downstream effectors MEK/ERK, PI3K/AKT, and STAT5 by phosphorylation. The latter effectors also activate target genes such as p21, p53, and cyclin D1.2,3 Thus, aberrant FLT3-mediated activation of effector proteins leads to uncontrolled proliferation, inhibition of differentiation and reduction of apoptosis in transformed hematopoietic blasts, and is also associated with poor prognosis in AML.4 Targeted therapies against FLT3 in FLT3-mutated AML using small molecule inhibitors such as sorafenib, quizartinib, midostaurin, crenolenib, and gilteritinib have shown clinical activity by reducing circulating leukemic blasts, and achieving temporary remission. However, these effects are apparently ineffective against leukemic stem cells in the bone marrow microenvironment and therefore the basis for temporary remission.5-10 In fact, we have reported marked upregulation of MAPK signaling following treatment with FLT3 inhibitors in AML/stroma co-culture, hypoxia, and in clinical samples ex vivo; this upregulation could be partially overcome by the novel dual FLT3ITD/MAPK inhibitor E6201.11 However, recent studies indicate that mutations of FLT3 are late events in leukemogenesis, suggesting that they are acquired rather than founder mutations in leukemia-initiating cells.12 Targeting FLT3 alone is, therefore, unlikely to be sufficient to eradicate leukemia-initiating cells. Recently, exportin 1 (XPO1), also known as the nuclear export protein (CRM1), has been identified.13 XPO1 is a nuclear receptor involved in the active transport of a large number of cargo proteins, including Foxo3A, p53, p21 and NPM1, across the nuclear membrane14 along with microRNAs.15,16 XPO1 overexpression is common in hematologic malignancies including AML, and it was reported by us to be associated with poor disease prognosis.17 Leukemic cells depend on the continuous nuclear export of one or more oncoproteins, and the removal of tumor suppressor proteins, which require nuclear localization for their functions.18 Targeting nuclear membrane proteins such as XPO1 could, therefore, restore tumor suppressor function in AML. The small molecule XPO1 inhibitor, selinexor (KPT-330) is a first–in-class, orally bioavailable selective inhibitor of nuclear export compound that has shown promising anti-leukemia activity in vitro and in vivo.19,20 Selinexor was effective in inducing apoptosis in cells from established AML cell lines that are in the G0/G1 phase of the cell cycle,21 and targeting it also abrogated hypoxia-induced drug resistance in multiple myeloma cells.22 These results suggest that targeting XPO1 with selinexor may have potent anti-proliferative effects against non-proliferating or slowly proliferating leukemiainitiating cells in primary AML unlike the limitation observed when using FLT3 inhibitors.19 In addition, the recent results of phase I/II trials using selinexor as monotherapy (e.g. NCT02091245 and NCT02088541) or in combinations with conventional chemotherapeutic drugs (e.g. NCT02249091), have shown promising antileukemia activity with a high rate of blast clearance and complete remissions.20,23-28 Initial problems with gastrointestinal toxicities and anorexia have largely been overcome by dose reduction without loss of clinical efficacy.23,24 However, the anti-leukemia activity of selinexor in AML patients with FLT3 mutations, including those with acquired secondary mutations found in relapsed/refractory disease following FLT3-targeted therapy, has not been established. haematologica | 2018; 103(10)

In this study, we report that selinexor has marked proapoptotic effects against AML cells harboring FLT3-ITD and/or TKD mutations. However, compensatory upregulation of phosphorylated FLT3 and its downstream signaling pathways was observed in most of the FLT3-mutated cell lines tested in vitro. We, therefore, combined selinexor with sorafenib. This combinatorial drug regimen achieved markedly synergistic leukemia cell killing in cells harboring ITD and/or TKD mutations, which usually show resistance to FLT3-targeted therapy.29,30 Of note, the combinatorial regimen also achieved encouraging clinical efficacy including molecular complete responses in an ongoing phase IB/II clinical trial of selinexor plus sorafenib in patients who were refractory to FLT3 inhibitor therapy. Thus, this combinatorial approach may abrogate selinexor-mediated FLT3 activation, resulting in abrogation of resistance to FLT3 inhibitors and induction of durable remissions in patients with additional acquired FLT3 mutations.

Methods Reagents and antibodies Selinexor was provided by Karyopharm Therapeutics (Newton, MA, USA). Sorafenib was purchased from Selleckchem (Houston, TX, USA). Their molecular structures are shown in Online Supplementary Figure S1. The antibodies against human phosphorylated (p)-p44/42 MAPK (ERK1/2)(Thr202/Tyr204), phosphoAKT(Ser473), phospho-FLT3(Tyr589/591), phosphoS6K(Ser240/244), AKT, S6K, Bcl-xL, C/EBPα, PU.1, STAT3, cMyc and cleaved caspase-3 were purchased from Cell Signaling Technology (Danvers, MA, USA), against Bcl-2 from Dako (Carpinteria, CA, USA), against phospho-STAT5 A/B from Upstate (Lake Placid, NY, USA), against total STAT5A/B from R&D Systems Inc. (Minneapolis, MN, USA), against ERK2, FLT3, p53, IκB alpha, phospho-Stat3, and Mcl-1 from Santa Cruz Biotechnology (Santa Cruz, CA, USA), against Bim and Puma from CalBiochem (San Diego, CA, USA), against HIF1α from BD Biosciences (San Diego, CA, USA), and against phospho-IκB alpha (ser32/36) from Novus (Littleton, CO, USA). The antiluciferase antibody was purchased from Promega (Madison, WI, USA).

Acute myeloid leukemia cell lines and patients’ samples The Baf3/FLT3, Baf3/ITD, and Baf3/D835Y cell lines were kindly provided by Dr. Donald Small (Department of Pediatric Oncology, Johns Hopkins University, Baltimore, MD, USA) and Baf3/ITD+D835Y and Baf3/ITD+D835H cells by Dr. Neil Shah (Department of Medicine, The University of California at San Francisco, San Francisco, CA, USA). The FLT3-inhibitor-resistant cells Baf3/ITD+F691 and Baf3/ITD+Y842, which harbor FLT3ITD plus F691L and Y842C mutations, respectively, were established by us as described previously.30 The human AML cell lines THP-1, Kasumi-1, and MV4-11 were obtained from the American Type Culture Collection (Manassas, VA, USA), and MOLM13 and MOLM14 from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschweig, Germany). All cell lines were validated by short tandem repeat DNA fingerprinting using the AmpFISTR Identifiler kit according to manufacturer's instructions (Applied Biosystems cat. n. 4322288). All cells were maintained in RPMI medium supplemented with 10% fetal bovine serum, and interleukin-3-dependent murine Baf3/FLT3 cells were maintained in the presence of 2 ng/mL of interleukin-3. The FLT3 status of the AML cell lines used in this study is shown in Table 1. 1643

W. Zhang et al. Table 1. The half maximal responding concentrations (EC50s and IC50 values)* of selinexor in leukemia cell lines.

Cell lines

FLT3 status

EC50 (µm)

95% confidence interval (lower/upper)

IC50 (mm)

95% confidence interval (lower/upper)

Murine cells

Baf3-FLT3 Wildtype 2.55 0.66/9.73 2.89 0.36/23.4 Baf3-ITD ITD mut** 0.48 0.30/0.77 0.03 0.01/0.16 Baf3-D835Y D835Y mut 0.65 0.40/1.04 0.06 0.02/0.17 Baf3-ITD+D835Y ITD+D835Y mut 0.32 0.07/1.50 0.09 0.06/0.12 Baf3-ITD+D835H ITD+D835H mut 0.22 0.04/1.30 0.08 0.05/0.13 Baf3-ITD+Y842 ITD+Y842C mut 0.56 0.16/2.01 0.08 0.02/0.38 Baf3-ITD+F691 ITD+F691L mut 0.61 0.17/2.23 0.06 0.02/0.25 Human cells THP-1 Wildtype 20 5.02/48.19 0.12 0.04/0.25 Kasumi-1 Wildtype 9.2 2.87/29.38 9.18 2.87/29.4 MOLM13 ITD mut 0.74 0.21/1.53 0.03 0.01/0.09 MV4-11 ITD mut 0.35 0.21/0.61 0.03 0.01/0.10 * EC50: the half maximal responding concentration to induce apoptosis; IC50: the half-maximal concentration to inhibit cell growth. ** mut: mutations

AML patients’ samples with FLT3-ITD mutations were obtained after written informed consent following institutional guidelines of the University of Texas MD Anderson Cancer Center and in accordance with the principles of the Declaration of Helsinki. The mononuclear cells in these samples were purified by Ficoll-Hypaque (Sigma-Aldrich) density-gradient centrifugation, and the cells were cultured in RPMI 1640 culture medium supplemented with 10% fetal calf serum, as described above, prior to treatment.

Cell viability and apoptosis assays The number of viable cells was determined using a Vi-CELL XR Cell Counter (Beckman Coulter Inc., Indianapolis IN, USA) with the trypan blue dye exclusion method, and apoptosis was determined via fluorescence-activated cell sorting (FACS) by annexin V positivity and propidium iodide positivity, as described previously.31 The 50% inhibitory concentration (IC50) for inhibition of cell growth and the 50% effective concentration (EC50) for induction of apoptosis were calculated using CalcuSyn software (BioSoft, Cambridge, UK).

Immunoblot analyses Protein levels in treated cells were determined by western blot analysis, as described previously.6 Briefly, the treated cells were collected for preparation of cell lysates which were then resolved by electrophoresis on 12% precast sodium dodecyl sulfate-polyacrylamide gels, and transferred to Hybond-P membranes. After immunoblotting with antibodies, signals were detected by using the Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA) and semi-quantitatively assessed using the Scion Imaging system and software (beta version 4.03; Scion, Frederick, MD, USA).

Cellular morphology and assessment of differentiation FLT3-ITD mutated AML cell lines MOLM13, MOLM14 and primary AML cell samples were plated at an initial density of 1.0×105 cells/mL in the presence of the indicated agents or combinations. Following 5 to 6 days of culture, cellular morphology was assessed after cytospinning onto slides and staining with Giemsa solution. Expression of the myeloid differentiation marker CD11b was determined by staining with anti-CD11b antibody (cell lines). The stained cells were washed twice with phosphate-buffered saline containing 2% bovine serum albumin. Morphology was evaluated by light microscopy; the percentage of CD11b cells and fluores1644

cence intensity were determined using a FACS Calibur flow cytometer (Becton Dickinson).

Animal studies The animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Texas, MD Anderson Cancer Center. NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NOG) mice (8-week old females; n=40; The Jackson Laboratory, Bar Harbor, ME, USA) were injected intravenously with 0.5x106 of MOLM13-Luci-GFP cells that were lentivirally infected with firefly luciferase.32 Mice (10 for each group) were treated with selinexor (15 mg/kg, dissolved in 0.6% Pluronic F-68 and 0.6% Plasdone K-29/32) or sorafenib [10 mg/kg, dissolved in cremophor EL/ethanol/water (12.5/12.5/75)] alone, or in combination (n=10); starting on day 4 after leukemia cell injection when an unambiguous luciferase signal was recorded. Animals injected with vehicle (only the solvents mentioned above in the same ratios without drugs) via gavage, at a once daily x 5/week schedule, served as controls. Mice were noninvasively imaged in a Xenogen-200 in vivo bioluminescence imaging system (Xenogen, Hopkinton, MA, USA) after injection with luciferin substrate (D-luciferin, GoldBoi, St Louis, MO, USA) at a concentration of 4 mg/mouse. Bioluminescence images were obtained and quantitated as described in detail previously.6 Three mice for each group were sacrificed on day 18 after tumor cell injection, and spleen, liver, lung, and bone marrow samples were collected for immunohistochemical analysis. Briefly, the collected tissues were fixed in 10% neutral buffered formalin solution at 4ºC overnight, then dehydrated, embedded in paraffin, and sectioned. After antigen retrieval, the slides were incubated with anti-luciferase antibodies.

Clinical trial We initiated a phase IB/II clinical study of selinexor in combination with sorafenib in relapsed/refractory patients with FLT3-ITD or FLT3-D835 mutations (NCT02530476). The study included a dose-escalation phase IB portion in which sorafenib 400 mg BID continuously was combined with escalating doses of selinexor at a dose of 40 mg twice/week, 60 mg twice/week, and 80 mg twice/week to identify the recommended phase II dose of the combination. The selinexor was given twice/week for 3 weeks with 1 week off per 28-day cycle. Response to therapy was defined according to the International Working Group criteria.33 Briefly, a complete remission (CR) was defined as ≤5% bone marrow blasts, a neutrophil count ≥1.0×109/L, and platelet count haematologica | 2018; 103(10)

Effects of XPO1 and FLT3 inhibition on AML ≥100×109/L. Briefly, a complete remission (CR) was defined as ≤5% bone marrow blasts, a neutrophil count ≥1.0×109/L, and a platelet count ≥100×109/L. CRi was defined as meeting all CR criteria except residual neutropenia (<1.0×109/L) and/or thrombocytopenia (<100×109/L). CRp (complete remission with incomplete platelet recovery) was defined as meeting all CR criteria except thrombocytopenia (<100 ×109/L).

Statistical analyses The Student t-test was used to analyze immunoblot and cell apoptosis data. A P-value ≤0.05 was considered statistically significant. All statistical tests were two-sided and the results are expressed as the mean of triplicate samples/experiments ± standard deviation/95% confidence intervals (error bars). The efficacy

of selinexor on survival was estimated by the Kaplan–Meier method,34 with log-rank statistics used to test for differences in survival.

Results Selinexor, alone, or in combination with sorafenib, exerts marked pro-apoptotic effects in human and murine FLT3-mutated acute myeloid leukemia cells We first investigated anti-leukemia effects of selinexor on AML cells with different FLT3 mutational status. Selinexor triggered profound induction of apoptosis and inhibition of cell growth, at sub-micromolar concentra-

Figure 1. Targeting XPO1 with selinexor induces profound apoptosis in FLT3-mutated leukemic cells. Human (A) and murine (B, C) AML cell lines were treated with selinexor at the indicated concentrations for 72 h. Apoptosis induction was assessed as the percentage of annexin V–positive cells by flow cytometry. Data are the mean of three independent determinations. Error bars correspond to 95% confidence intervals. The protein levels of correlated signaling pathways (D) and phosphorylation status (E) were determined by immunoblotting after treatment with selinexor for 24 h. GAPDH was used as a loading control.

haematologica | 2018; 103(10)

1645

W. Zhang et al.

tions, in all human and murine AML cell lines that harbor ITD, TKD, or dual mutations of FLT3. The agent was much less effective in this regard in FLT3 wildtype Baf3/FLT3, THP-1, and Kasumi-1 cell lines (FLT3 mutant cells, regardless of whether they had single or dual mutations of ITD and TKD, showed 5- to 10-fold lower EC50 values than those of FLT3-wildtype ones) (Figure 1A-C, Table 1 and Online Supplementary Figures S2 and S3). We next evaluated effects of selinexor on protein expression using immunoblot analysis after 24 h of exposure to selinexor. Selinexor inhibited expression of the anti-apoptotic protein Mcl-1 and upregulated the pro-apoptotic protein Bim. The tumor suppressor proteins p53, p21, and p27 were upregulated as well (Figure 1D). Unexpectedly, activations of FLT3 and its downstream signaling pathways were upregulated, as evidenced by increasing levels of phosphorylated FLT3, -ERK, and -AKT after exposure to

selinexor for 24 h (Figure 1E), which were observed only in the FLT3-mutated cells. In addition, total FLT3 levels were also upregulated in FLT3-ITD and -ITD plus Y842 mutated cells, but not in ITD plus D835Y cells: this was true for both protein levels (Figure 1E) and mRNA transcriptional levels (Online Supplementary Figure S4). Kinetic analysis revealed that the upregulation of phospho-FLT3 was observable at 1 h, and phospho-ERK and -AKT at 6 h, after selinexor treatment (Online Supplementary Figure S5). These findings suggest that co-targeting FLT3 signaling, to suppress its downstream signaling pathways, simultaneously with nuclear export may potentially trigger synergistic cytotoxic effects in these cells. We tested this hypothesis using combinatorial treatment with selinexor and sorafenib. The combinatorial regimen did indeed trigger synergistic pro-apoptotic effects in murine FLT3-ITD mutated, ITD plus Y842C and ITD plus

Figure 2. Combination treatment with selinexor and sorafenib triggers synergistically pro-apoptotic effects in murine cells with ITD or ITD plus TKD2 point mutations. (A) FLT3 mutated cells (Baf3/ITD, Baf3/ITD+Y842 and Baf3/ITD+D835Y) and wildtype cells (Baf3/FLT3-wt) were treated with either selinexor or sorafenib alone or the combination for 48 h, and examined for apoptosis induction (annexin V positivity) as described in Figure 1. (B) FLT3-mutated cells were treated with single agent(s) alone or the combination for 16 h, and levels of correlated phosphorylated proteins were measured by immunoblotting. GAPDH was used as a loading control. (C) MOLM13 cells were treated with the indicated concentrations of agent(s) for 16 h. Cytosolic and nuclear fractions were separated using a nuclear extract kit (Active Motif) following the manufacturerâ&#x20AC;&#x2122;s instructions. The correlated protein levels were determined with immunoblotting. Poly (ADP-ribose) polymerase (PARP) was used as a loading control of the nuclear fraction and tubulin was used as a loading control of the cytosolic fraction. Dimethylsulfoxide (DMSO) was used as a control. Seli: selinexor; Sora: sorafenib; Combi: combination. Solid line box indicates upregulation and dotted line box indicates downregulation.

1646

haematologica | 2018; 103(10)

Effects of XPO1 and FLT3 inhibition on AML

D835Y mutated cells (Figure 2A) and triggered additive pro-apoptotic effects in human FLT3-ITD-mutated MOLM13 and MV4-11 cells (Online Supplementary Figure S5). Of note, the combination regimen also demonstrated a synergistic pro-apoptotic effect in a primary AML sample with FLT3-ITD plus D839G dual mutations (Online Supplementary Figure S6). Immunoblot analysis showed that the combinatorial treatment markedly abrogated the upregulation of phospho-FLT3, -ERK and -AKT (Figure 2B) that was observed in the experiments of selinexor treatment alone. In fact, all phospho-proteins were suppressed far below basal levels by the combination treatment. Interestingly, we also observed that ERK, AKT, FOXO3a, NF-kB, p53, p27, and p21 were preferentially retained in the AML cell nuclei after 16 h of treatment with selinexor and sorafenib. This treatment also completely abrogated anti-apoptotic Mcl-1 expression and decreased c-Myc levels (Figure 2C), suggesting greater AML cell sensitivity to apoptosis.

Co-targeting XPO1 and FLT3 partially abrogates hypoxia-mediated chemoprotection The hypoxic bone marrow microenvironment is a reservoir for leukemia-initiating cells, and it is associated with resistance to AML chemotherapy.30,35 We examined apoptosis induction using the indicated concentrations of selinexor and/or sorafenib for a 68-h exposure under normoxic and hypoxic conditions. The combination partially abrogated hypoxia-mediated chemoprotection and induced synergistic apoptotic effects compared to those observed following treatment with either agent alone in

FLT3-ITD-mutated MOLM13 and MOLM14 cells (Figure 3A,B). Immunoblot analysis indicated that the combination treatment profoundly suppressed the hypoxia-mediated upregulation of CXCR4 and HIF1Îą, suppressed phospho-FLT3, -ERK and -AKT, decreased Mcl-1, and increased the cleavage of caspase-3 (Figure 3C).

Co-targeting XPO1 and FLT3 enhances myeloid differentiation of FLT3-ITD-mutated human acute myeloid leukemia cells and human primary acute myeloid leukemia samples Since selinexor and sorafenib treatments alone have been reported to induce differentiation of leukemic cells,21,36 we investigated if the combination could enhance the differentiation of FLT3-ITD-mutated leukemic cells in vitro. MOLM13 and MOLM14 cells were exposed to nanomolar concentrations of either agent alone or the combination for 5 days. At these doses only growth arrest was observed. Morphological changes were observed, including indentation and bending of the nuclei and a decrease of nuclear/cytoplasmic ratio, with single-agent treatment in both ITD-mutated AML cell lines (Giemsa staining) suggesting a metamyelocyte stage of granulocytic differentiation. The combination treatment markedly enhanced these morphological changes. In addition, the myeloid differentiation marker CD11b+ significantly increased in the cells following the combination treatment (Figure 4A,B). We used the same regimen to treat two primary, FLT3ITD-mutated, human AML samples (Online Supplementary Table S1) in vitro for 6 days. Enhanced morphological myeloid differentiation described above was observed fol-

Figure 3. Combination treatment partially abrogates hypoxia protection and triggers synergistic apoptosis induction. MOLM13 (A) and MOLM14 (B) cells were treated with selinexor and/or sorafenib for 68 h. Apoptosis induction (annexin V positivity) was determined by flow cytometry. (C) MOLM13 cells were treated with the combination for 24 h and correlated protein levels were determined by immunoblotting.

haematologica | 2018; 103(10)

1647

W. Zhang et al.

lowing the combination treatment, as was a profound increase of the CD11b+ population, which was more significant in the ITD-mutated AML sample (AML #2) than in the D835 TKD-mutated AML sample (AML #1) (Figure 4C). In addition, a decrease of the CD34+ population was observed in both tested primary AML samples (Online Supplementary Figure S8). Of note, the doses of either drug alone, or in combination, were not enough to kill the primary AML cells, but were sufficient for growth arrest of the CD34+ cells, implying that the combinatorial regimen could be beneficial by impairing the self-renewal capacity of the leukemic CD34+ compartment. Mechanistically,

marked upregulation of CCAAT/enhancer binding protein α (C/EBPα), one of the leucine zipper transcription factors that is important for normal myeloid cell differentiation, was observed in the cells following the combination treatment. This was accompanied by an increase in the C/EBPα/PU-1 ratio (Figure 4D), which has been reported to be associated with granulocytic myeloid cell differentiation.37 Interestingly, upregulation of cell proliferation-related proteins, such as phospho-ERK, -STAT5, -STAT3, and tumor suppressor proteins, including p53 and p21, was also observed after exposing cells to low doses of both drugs for 5 days (Figure 4D).

Figure 4. Combination treatment triggers myeloid differentiation of leukemic cells. FLT3-ITD mutated cell lines MOLM13 (A), MOLM14 (B) and primary AML patients’ samples (C) were treated with the indicated concentrations of selinexor and/or sorafenib for 5 or 6 days in vitro; morphological changes of the cells were checked with Giemsa staining. Expression of the myeloid differentiation marker CD11b was determined using flow cytometry. The histograms present individual assays which measured fluorescent intensity of CD11b. Each column was generated from four individual assays. (D) Differentiation-related proteins were evaluated using immunoblotting. The expression levels and ratios of C/EBPα/PU-1 were measured (first lane which was defined as 1). Dimethylsulfoxide (DMSO) was used as a control. Seli: selinexor; Sora: sorafenib; Combi: combination.

1648

haematologica | 2018; 103(10)

Effects of XPO1 and FLT3 inhibition on AML

The selinexor and sorafenib combination has marked anti-leukemia efficacy in a mouse xenograft model of human FLT3-ITD-mutated acute myeloid leukemia We next assessed in vivo efficacy of the selinexor and sorafenib combination in a murine leukemia model. NOG mice bearing xenografts of MOLM13-Luc-GFP cells were treated with either selinexor or sorafenib alone or the drug combination. The vehicle served as a control. The mice received 39 days of treatment starting from day 4 after injection of leukemia cells. The median survival in the vehicle, sorafenib, selinexor, and combination treatment groups was 16, 23, 31, and 51 days, respectively (P<0.001) (Figure 5A). In addition, the combination significantly reduced leukemia burden compared with that in the control mice after 10 days of treatment (day 14). The mean luminescence was ~1x107 photons/s in the vehicle-treated group versus ~2x105 photons/s in the group treated with the combination (Figure 5B,C). The mice tolerated the individual drugs and the combination well, without signs of anorexia, weight loss, or other signs/symptoms of distress. One week after treatment cessation (i.e., day 49), the mice in the combination group developed increased leukemia burden and succumbed to AML (Figure 5D). Further analysis revealed that the infiltration of leukemic cells was significantly reduced in peripheral blood after receiving 14 days of either single-agent treatment or combination treatment (Figure 6A). However, the

bone marrow environment protected against sorafenib treatment-mediated leukemia cell killing, while selinexor alone had anti-leukemia efficacy. Impressively, the combination treatment almost eliminated leukemic cells from the bone marrow (i.e., approximately 100-fold lower levels than following treatment with the vehicle, P<0.001) (Figure 6B), which suggests potent killing of leukemia cells by this combinatorial regimen in the bone marrow environment. Immunohistochemical analysis further confirmed the profound reduction of leukemia cells in the bone marrow and other organs (Figure 6C).

The selinexor and sorafenib combination exerted anti-leukemia effects in patients with FLT3-mutated acute myeloid leukemia in a phase IB clinical trial The preclinical synergy of selinexor and sorafenib led to the initiation of a phase IB/II study of the combination in relapsed/refractory patients harboring FLT3 mutations. Fourteen patients enrolled in this clinical trial were eligible for analysis. Remarkably, four of the 14 patients (29%) achieved a sustained CRp/CRi and two others (14%) had >50% blast reduction (Table 2). Six of the 11 (55%) patients treated previously with FLT3 inhibitors achieved responses. These patients comprised: (i) two FLT3-ITD mutated patients, previously treated with sorafenib, who achieved CRp with molecular CR with selinexor plus sorafenib; (ii) one FLT3-ITD and FLT3-D835 dual mutated

Figure 5. Combination treatment significantly improves mice survival and reduces leukemia burden in a MOLM13-engrafted murine leukemia model. (A) The median survival were assessed for each group by the Kaplanâ&#x20AC;&#x201C;Meier method, and log-rank statistics applied to test for differences in survival. (B) Serial bioluminescence images of representative mice at day 4 and day 14 after injection of leukemic cells in the groups treated with selinexor or sorafenib alone or the combination. (C) Quantitative analysis of the leukemia burden. (D) Serial bioluminescence images of representative mice at 26, 33, 40 and 49 days after leukemia cell injection.

haematologica | 2018; 103(10)

1649

W. Zhang et al.

patient, previously treated with quizartinib and idarubicin with cyatarabine plus crenolanib, who achieved CRi with molecular CR for both FLT3-ITD and FLT3-D835 with selinexor plus sorafenib; (iii) a FLT3-mutated patient previously treated with crenolanib, sorafenib and allogeneic stem cell transplantation who achieved CRi without molecular CR with selinexor plus sorafenib; and (iv) two FLT3-ITD and FLT3-D835 dual mutated patients who had received prior sorafenib therapy and achieved PR + >50% blast reduction with selinexor plus sorafenib (Daver et al., ASH annual meeting, 2017, Abstract #1344). We have no information regarding differentiation induction in these patients at this time.

Discussion A previous study using the XPO1 inhibitor KPT-185 demonstrated strong post-transcriptional downregulation of total FLT3 protein expression in AML cell lines and primary AML samples, which was associated with antileukemia efficacy.21 We confirmed the suppression of phospho-FLT3 and accompanying downregulation of total FLT3 protein in human MOLM13 and MV4-11 cells after 24 h of selinexor treatment (Online Supplementary Figure S9). Unexpectedly, however, we observed marked upregulation of phospho-FLT3 in murine AML cells, including those harboring ITD, ITD plus D835Y or Y842C dual mutations, after exposure to the same concentrations of selinexor. Furthermore, total FLT3 protein and mRNA levels were upregulated in addition to the upregulation of FLT3 downstream phospho-ERK and -AKT levels in cells with these mutations, but not in the ITD plus D835Y mutants (Figure 1E, Online Supplementary Figure S4), suggesting that a possible transcriptional mechanism was

involved in FLT3 upregulation in these murine cells. The modulation profiling suggests that selinexor-induced apoptosis is independent of the suppression of FLT3 and its downstream pathways. In fact, cells with the dual Baf3/ITD+D835 mutation showed greater sensitivity to selinexor-induced apoptosis compared with other dualmutated cells, Baf3/ITD+Y842 and Baf3/ITD+F691 (EC50 values were 0.2 and 0.3 mM versus 0.56 and 0.6 mM, respectively, Baf3/ITD+D835H and Baf3/ITD+D835Y versus Baf3/ITD+Y842 and Baf3/ITD+F691 cells), and suppressed phospho-ERK and â&#x20AC;&#x201C;AKT, but upregulated phospho-FLT3 levels. These results strongly support that selinexor induces apoptosis in an FLT3-independent manner in FLT3-mutated AML cells. In addition, we observed upregulation of XPO1 cargo proteins p53, p21 and p27, as well as downregulation of pro-survival Mcl-1 after 24 h of selinexor treatment in both human and murine FLT3mutated AML cells (Figures 1D and 2C). We, therefore, postulated that the modulation of pro- and anti-apoptotic regulators was the driving force behind selinexor-triggered apoptosis in the FLT3-mutated AML cells. These results extend our previous observations in wild-type FLT3 AML cells, which are likely also dependent on p53.17 We further postulated that the selinexor-induced upregulation of phospho-FLT3 and its downstream components may provide a rationale for the combinatorial treatment with FLT3 inhibitors. Concomitantly targeting XPO1 and FLT3 triggered profound synergistic pro-apoptotic effects in murine FLT3-mutated AML cells. However, only additive induction of apoptosis was observed in human FLT3-mutated AML cells (Online Supplementary Figure S5). Of note, we have reported that FLT3-targeted therapy upregulates FLT3 and its downstream proteins in human clinical samples and in most sorafenib-resistant AML cells in vitro.5,11,30 Thus, it is reasonable to speculate that co-targeting FLT3 and XPO1

Table 2. Characteristics of patients who responded to the combination treatment.

Patients

Age FLT3 Status (years)

Additional mutationsa

Karyotypes

Blasts (%)

Dose level

Best response

Response duration

Prior therapies

-Selinexor 80 mg twice per week -Sorafenib 400 mg BID -Selinexor 60 mg twice per week -Sorafenib 400 mg BID -Selinexor 80 mg twice per week -Sorafenib 400 mg BID

CRp with undetectable FLT3-ITD by PCR CRp, FLT3 PCR pending

180 days

Aza+sorafenib, Anti-CD25 trial

75 days

7+3, CLIA+sorafenib

Case 1

ITD mutb

TP53, IDH2, RUNX1, WT1

Diploid

Case 2

ITD mut

HRAS. WT1

t(6;9)

Case 3

ITD + D835 mut

NRAS, TET2, IDH1

Diploid

Case 4

ITD mut

RUNX1

Miscellaneous

Case 5

ITD + D835 mut DNMT3A, RUNX1

Diploid

Case 6

ITD + D835 mut

Miscellaneous

WT1

-Selinexor 60 mg twice per week -Sorafenib 400 mg BID -Selinexor 60 mg twice per week -Sorafenib 400 mg BID -Selinexor 60 mg twice per week -Sorafenib 400 mg BID

CRi with 60 days then undetectable to autologous FLT3-ITD and SCT FLT3-D835 by PCR CRi with 60 days FLT3 positive by PCR Hi-blast reduction 35 days

Hi-blast reduction

35 days

7+3, MEC, quizartinib, IA+crenolanib 7+3+sorafenib, GCLAC +sorafenib, SCT Aza, aza+sorafenib

7+3, MEC, MUD, SCT, Dac+sorafenib,

Based on screening of bone marrow samples with a 28-gene mutation panel. bmut: mutations. PCR: polymerase chain reaction; Aza: azacytidine; CLIA: cladribine idarubicin, cytarabine; IA: idarubicin, cytarabine; MEC: mitoxantrone, etoposide, cytarabine; GCLAC: granulocyte colony-stimulating factor, cladribine, cytarabine; SCT: stem cell transplantation; MUD: matched unrelated donor; Dac: decitabine. a

1650

haematologica | 2018; 103(10)

Effects of XPO1 and FLT3 inhibition on AML

could achieve synergistic efficacy by overcoming the resistance to FLT3-targeted therapy. Our data demonstrate a synergistic anti-leukemia efficacy of the drug combination in vivo. In addition, the combination regimen produced a profound reduction of CXCR4 and HIF1Îą levels in hypoxia in vitro (Figure 3C), and also marked elimination of leukemic cells in a bone marrow in vivo murine AML model (Figure 6B), suggesting a potential benefit for overcoming resist-

ance provided by the microenvironment. Remarkably, the combination regimen has achieved complete remissions, including ongoing complete molecular remissions, in 30% (3 of the first 10) relapsed/refractory AML patients in our on-going phase IB/II clinical trial of combinatorial therapy of selinexor and sorafenib. Of note, the three patients with CRp/CRi, who were previously clinically refractory to quizartinib/sorafenib monotherapy or to combinations

Figure 6. Combination treatment significantly eliminates leukemia cell engraftment in bone marrow and other organs. Leukemia cell infiltration was evaluated in (A) blood circulation and (B) bone marrow (BM) at day 18 by flow cytometry after gating GFP-positive cells, and in (C) other soft organs using immunostaining with anti-luciferase antibody in paraffin sections. The bar represents 100 mm.

haematologica | 2018; 103(10)

1651

W. Zhang et al.

with conventional chemotherapeutic drugs, achieved FLT3ITD negativity as determined by the real-time polymerase chain reaction assay. One hallmark of AML is the differentiation arrest of leukemic blasts. Promoting differentiation may therefore be beneficial for achieving and maintaining remissions in leukemias. Targeting FLT3-mutated AML cells with FLT3 inhibitors has been reported to induce cell-cycle arrest and differentiation, rather than apoptosis, which is reportedly driven by overexpression of C/EBPα and PU.1.36,38 Schepers et al. described that upregulation of C/EBPα led to growth arrest of CD34+ leukemia cells, which impaired the selfrenewal capacity of the leukemic CD34+ cells, and corresponded with enhanced myeloid differentiation as well.39 Also, KPT-185 induced cell-cycle arrest and myeloid differentiation in AML cells, including patients’ samples, which increased C/EBPα.21 In fact, C/EBPα is a p53-regulated DNA damage-inducible gene,40 and p53 induction is involved in myeloid differentiation.41 Our data indicate that selinexor deactivated the nuclear export of a number of cargo proteins including p53. Furthermore, sorafenib mediated the upregulation of p21, and its combination with selinexor markedly enhanced levels of p53, p21, and C/EBPα (Figure 4D) as noted above, the last being one of the key hematopoietic-specific transcription factors mediating myeloid differentiation of leukemia cells. However, we did not observe an increase in another transcription factor, PU.1, which is reportedly an upregulated effector of C/EBPα. The precise function of PU.1 is still unclear. Dahl et al. suggested that lower levels of PU.1 direct granulocyte differentiation, whereas higher levels are required for macrophage differentiation.37 Nevertheless, our data imply that an increase of C/EBPα levels was sufficient to induce myeloid differentiation of FLT3-ITD-mutated leukemic cells and decrease the CD34+ population, especially for the combination of sorafenib and selinexor. This treatment restores nuclear p53 level by blocking XPO1, and then upregulates C/EBPα to enhance C/EBPα/PU-1 and granulocyte differentiation as shown in Figure 4 and Online Supplementary Figure S8. Signal transducer and activator of transcription (STAT)

References 1. Estey E, Dohner H. Acute myeloid leukaemia. Lancet. 2006;368(9550):18941907. 2. Takahashi S. Downstream molecular pathways of FLT3 in the pathogenesis of acute myeloid leukemia: biology and therapeutic implications. J Hematol Oncol. 2011;4:13. 3. Girardot M, Pecquet C, Chachoua I, et al. Persistent STAT5 activation in myeloid neoplasms recruits p53 into gene regulation. Oncogene. 2015;34(10):1323-1332. 4. Kottaridis PD, Gale RE, Linch DC. Prognostic implications of the presence of FLT3 mutations in patients with acute myeloid leukemia. Leuk Lymphoma. 2003;44(6):905-913. 5. Borthakur G, Kantarjian H, Ravandi F, et al. Phase I study of sorafenib in patients with refractory or relapsed acute leukemias. Haematologica. 2011;96(1):62-68. 6. Zhang W, Konopleva M, Shi YX, et al.

1652

7. 8.

10.

family proteins are reportedly involved in regulation of myeloid progenitor cell differentiation.42 In fact, STAT5 plays an important role in early myeloid differentiation, and lacking expression of STAT5 reduced lymphomyeloid repopulating activity from adult bone marrow and fetal liver of mice.43 STAT3 activation has also been reported to be a critical step in terminal differentiation of myeloid cells.44 On the other hand, upregulation of MAPK has also shown to be critical in both monocytic and granulocytic differentiation of myeloid cell lines, which can be abrogated by using the MEK inhibitor U0126.45 All of these lines of evidence imply that high activation of these proteins may contribute to myeloid differentiation of leukemia cells. Of note, we observed profound upregulation of phosphorylated STAT3, STAT5 and ERK levels after combination treatment with low doses of sorafenib and selinexor in FLT3-mutated MOLM13 and MOLM14 cells (Figure 4D), suggesting that upregulation of STATs and/or MAPK signaling pathways may also contribute to differentiation induction of the combination regimen in FLT3 mutated AML cells. In summary, our combinatorial strategy targeting FLT3 and XPO1 showed synergistic anti-leukemia effects in FLT3 inhibitor-resistant cells in vitro and in vivo. The combination of XPO1 and FLT3 inhibitors may also be able to eliminate leukemia-initiating cells by arresting cell growth and impairing the self-renewal capacity of leukemic CD34+ cells. These results should provide a solid basis for examining these agents further in patients with FLT3mutated AML, including those who have acquired resistance to FLT3-targeted therapy. Acknowledgments The authors would like to thank Dr. Neil Shah for FLT3-ITD and TKD double mutant cells and Dr. Numsen Hail, Jr. for providing critical review and editorial assistance in the preparation of this manuscript. Funding This work was supported in part by the NIH/NCI grants CA143805, CA100632, CA016672, and CA049639 (to MA).

Mutant FLT3: a direct target of sorafenib in acute myelogenous leukemia. J Natl Cancer Inst. 2008;100(3):184-198. Levis M. Quizartinib for the treatment of FLT3/ITD acute myeloid leukemia. Future Oncol. 2014;10(9):1571-1579. Fischer T, Stone RM, Deangelo DJ, et al. Phase IIB trial of oral midostaurin (PKC412), the FMS-like tyrosine kinase 3 receptor (FLT3) and multi-targeted kinase inhibitor, in patients with acute myeloid leukemia and high-risk myelodysplastic syndrome with either wild-type or mutated FLT3. J Clin Oncol. 2010;28(28):4339-4345. Randhawa JK, Kantarjian HM, Borthakur G, et al. Results of a phase II study of crenolanib in relapsed/refractory acute myeloid leukemia patients (pts) with activating FLT3 mutations. Blood. 2014;124 (21):389a. Perl A, Altman JK, Cortes J, et al. Final results of the chrysalis trial: a first-in-human phase 1/2 dose-escalation, dose-expansion study of gilteritinib (ASP2215) in patients with relapsed/refractory acute myeloid leukemia

(R/R AML). Blood. 2016;128(21):1069a. 11. Zhang W, Borthakur G, Gao C, et al. Study of activity of E6201, a dual FLT3 and MEK inhibitor, in acute myelogenous leukemia with FLT3 or RAS mutation. Blood. 2013;122(21):2683a. 12. Corces-Zimmerman MR, Hong WJ, Weissman IL, Medeiros BC, Majeti R. Preleukemic mutations in human acute myeloid leukemia affect epigenetic regulators and persist in remission. Proc Natl Acad Sci USA. 2014;111(7):2548-2553. 13. Hing ZA, Fung HY, Ranganathan P, et al. Next-generation XPO1 inhibitor shows improved efficacy and in vivo tolerability in hematological malignancies. Leukemia. 2016;30(12):2364-2372. 14. Kanai M, Hanashiro K, Kim SH, et al. Inhibition of Crm1-p53 interaction and nuclear export of p53 by poly(ADPribosyl)ation. Nat Cell Biol. 2007;9(10):11751183. 15. Muqbil I, Bao B, Abou-Samra AB, Mohammad RM, Azmi AS. Nuclear export

haematologica | 2018; 103(10)

Effects of XPO1 and FLT3 inhibition on AML

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

mediated regulation of microRNAs: potential target for drug intervention. Curr Drug Targets. 2013;14(10):1094-1100. Azmi AS, Li Y, Muqbil I, et al. Exportin 1 (XPO1) inhibition leads to restoration of tumor suppressor miR-145 and consequent suppression of pancreatic cancer cell proliferation and migration. Oncotarget. 2017;8(47):82144-82155. Kojima K, Kornblau SM, Ruvolo V, et al. Prognostic impact and targeting of CRM1 in acute myeloid leukemia. Blood. 2013;121(20):4166-4174. Ossareh-Nazari B, Bachelerie F, Dargemont C. Evidence for a role of CRM1 in signalmediated nuclear protein export. Science. 1997;278(5335):141-144. Etchin J, Montero J, Berezovskaya A, et al. Activity of a selective inhibitor of nuclear export, selinexor (KPT-330), against AMLinitiating cells engrafted into immunosuppressed NSG mice. Leukemia. 2016;30(1): 190-199. Alexander TB, Lacayo NJ, Choi JK, Ribeiro RC, Pui CH, Rubnitz JE. Phase I study of selinexor, a selective inhibitor of nuclear export, in combination with fludarabine and cytarabine, in pediatric relapsed or refractory acute leukemia. J Clin Oncol. 2016;34(34): 4094-4101. Ranganathan P, Yu X, Na C, et al. Preclinical activity of a novel CRM1 inhibitor in acute myeloid leukemia. Blood. 2012;120(9):17651773. Muz B, Azab F, de la Puente P, Landesman Y, Azab AK. Selinexor overcomes hypoxiainduced drug resistance in multiple myeloma. Transl Oncol. 2017;10(4):632-640. Savona M, Garzon R, Brown P, et al. Phase I trial of selinexor (KPT-330), a first-in-class oral selective inhibitor of nuclear export (SINE) in patients (pts) with advanced acute myelogenous leukemia (AML). Blood. 2013;122(21):1440a. Bhatnagar B, Klisovic RB, Walker AR, et al. A phase 1 clinical trial of selinexor in combination with decitabine in patients with newly diagnosed and relapsed or refractory acute myeloid leukemia. Blood. 2016;128 (21):1651a. Sweet KL, Komrokji RS, Padron E, et al. A phase I study of selinexor in combination

haematologica | 2018; 103(10)

26.

27.

28.

29.

30.

31.

32.

33.

34. 35.

with daunorubicin and cytarabine in patients with newly diagnosed poor-risk acute myeloid leukemia. Blood. 2016;128 (22):4040a. Fiedler W, Heuser M, Chromik J, et al. Phase II results of ara-C and idarubicin in combination with the selective inhibitor of nuclear export (SINE) compound selinexor (KPT330) in patients with relapsed or refractory AML. Blood. 2016;128(22):341s. Garzon R, Savona M, Baz R, et al. A phase I clinical trial of single-agent selinexor in acute myeloid leukemia. Blood. 2017;129(24): 3165-3174. Garg M, Nagata Y, Kanojia D, et al. Profiling of somatic mutations in acute myeloid leukemia with FLT3-ITD at diagnosis and relapse. Blood. 2015;126(22):2491-2501. Smith CC, Wang Q, Chin CS, et al. Validation of ITD mutations in FLT3 as a therapeutic target in human acute myeloid leukaemia. Nature. 2012;485(7397):260-263. Zhang W, Gao C, Konopleva M, et al. Reversal of acquired drug resistance in FLT3mutated acute myeloid leukemia cells via distinct drug combination strategies. Clin Cancer Res. 2014;20(9): 2363-2374. Clodi K, Kliche KO, Zhao S, et al. Cell-surface exposure of phosphatidylserine correlates with the stage of fludarabine-induced apoptosis in chronic lymphocytic leukemia (CLL) and expression of apoptosis-regulating genes. Cytometry. 2000;40(1):19-25. Pajarinen J, Lin TH, Sato T, et al. Establishment of green fluorescent protein and firefly luciferase expressing mouse primary macrophages for in vivo bioluminescence Imaging. PLoS One. 2015;10(11): e0142736. Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. 2003;21(24):4642-4649. Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53:457-481. Tabe Y, Konopleva M. Role of microenvironment in resistance to therapy in AML. Curr

Hematol Malig Rep. 2015;10(2):96-103. 36. Zheng R, Friedman AD, Levis M, Li L, Weir EG, Small D. Internal tandem duplication mutation of FLT3 blocks myeloid differentiation through suppression of C/EBPalpha expression. Blood. 2004;103(5):1883-1890. 37. Dahl R, Walsh JC, Lancki D, et al. Regulation of macrophage and neutrophil cell fates by the PU.1:C/EBPalpha ratio and granulocyte colony-stimulating factor. Nat Immunol. 2003;4(10):1029-1036. 38. Sexauer A, Perl A, Yang X, et al. Terminal myeloid differentiation in vivo is induced by FLT3 inhibition in FLT3/ITD AML. Blood. 2012;120(20):4205-4214. 39. Schepers H, Wierenga AT, van Gosliga D, Eggen BJ, Vellenga E, Schuringa JJ. Reintroduction of C/EBPalpha in leukemic CD34+ stem/progenitor cells impairs selfrenewal and partially restores myelopoiesis. Blood. 2007;110(4):1317-1325. 40. Yoon K, Smart RC. C/EBPalpha is a DNA damage-inducible p53-regulated mediator of the G1 checkpoint in keratinocytes. Mol Cell Biol. 2004;24(24):10650-10660. 41. Meyer M, Rubsamen D, Slany R, et al. Oncogenic RAS enables DNA damage- and p53-dependent differentiation of acute myeloid leukemia cells in response to chemotherapy. PLoS One. 2009;4(11):e7768. 42. Wheadon H, Roberts PJ, Watts MJ, Linch DC. Changes in signal transduction downstream from the granulocyte-macrophage colony-stimulating factor receptor during differentiation of primary hemopoietic cells. Exp Hematol. 1999;27(6):1077-1086. 43. Bunting KD, Bradley HL, Hawley TS, Moriggl R, Sorrentino BP, Ihle JN. Reduced lymphomyeloid repopulating activity from adult bone marrow and fetal liver of mice lacking expression of STAT5. Blood. 2002;99(2):479-487. 44. Minami M, Inoue M, Wei S, et al. STAT3 activation is a critical step in gp130-mediated terminal differentiation and growth arrest of a myeloid cell line. Proc Natl Acad Sci USA. 1996;93(9):3963-3966. 45. Miranda MB, McGuire TF, Johnson DE. Importance of MEK-1/-2 signaling in monocytic and granulocytic differentiation of myeloid cell lines. Leukemia. 2002;16(4): 683-692.

1653

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1654-1661

Addition of the mammalian target of rapamycin inhibitor, everolimus, to consolidation therapy in acute myeloid leukemia: experience from the UK NCRI AML17 trial Alan K Burnett,1 Emma Das Gupta,2 Steve Knapper,3 Asim Khwaja,4 Marion Sweeney, 3Lars Kjeldsen,5 Timothy Hawkins,6 Sophie E Betteridge,7 Paul Cahalin,8 Richard E Clark,9 Robert K Hills9 and Nigel H Russell2 on behalf of the UK NCRI AML Study Group

Formerly Department of Haematology, Cardiff University School of Medicine, UK; Department of Haematology, Nottingham University Hospital NHS Trust, UK; 3Department of Haematology, University Hospital of Wales, Cardiff, UK; 4University College, London Cancer Institute, UK; 5Department of Haematology, Rigshospitalet, Copenhagen, Denmark; 6 Department of Haematology, Auckland City Hospital, New Zealand; 7Centre for Trials Research, Cardiff University School of Medicine, UK; 8Department of Haematology, Blackpool Victoria Hospital, UK and 9Department of Haematology, Royal Liverpool University Hospital, UK 1 2

ABSTRACT

Correspondence: akburnett719@gmail.com

Received: January 24, 2018. Accepted: July 4, 2018. Pre-published: July 5 2018.

doi:10.3324/haematol.2018.189514 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1654 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1654

s part of the UK NCRI AML17 trial, adult patients with acute myeloid leukemia in remission could be randomized to receive the mammalian target of rapamycin inhibitor everolimus, sequentially with post-induction chemotherapy. Three hundred and thirty-nine patients were randomised (2:1) to receive everolimus or not for a maximum of 84 days between chemotherapy courses. The primary endpoint was relapse-free survival. At 5 years there was no difference in relapsefree survival [29% versus 40%; odds ratio 1.19 (0.9-1.59) P=0.2], cumulative incidence of relapse [60% versus 54%: odds ratio 1.12 (0.82-1.52): P=0.5] or overall survival [45% versus 58%: odds ratio 1.3 (0.94-1.81): P=0.11]. The independent Data Monitoring Committee advised study termination after randomization of 339 of the intended 600 patients because of excess mortality in the everolimus arm without any evidence of beneficial disease control. The delivery of the everolimus dose was variable, but there was no evidence of clinical benefit in patients with adequate dose delivery compared with no treatment. This study suggests that the addition of mammalian target of rapamycin inhibition to chemotherapy provides no benefit.

Introduction The majority (70-85%) of younger patients with acute myeloid leukemia (AML) will enter complete morphological remission with any one of a variety of induction treatments. However, nearly half will relapse. It is being increasingly recognized that a substantial proportion of those subjects in morphological remission do actually have residual disease, as determined by techniques of minimal/measurable residual disease assessment (flow cytometry or quantitative polymerase chain reaction1,2). In our previous studies we endeavored to define the optimum post-remission chemotherapy. To date we have concluded that, apart from transplantation, following two induction courses of anthracycline-containing therapy, two consolidation courses of cytarabine (Ara-C) is adequate.3 One of the aims of the UK NCRI AML17 trial was to explore the effects of a further reduction in the total number of chemotherapy courses from four to three, as well as the addition of molecularly targeted treatments to consolidation therapy. Among these was the incorporation of the inhibitor of the mammalian target of rapamycin (mTOR), everolimus. haematologica | 2018; 103(10)

Everolimus added to consolidation therapy for AML

There is plausible pre-clinical evidence both in vitro and in vivo that mTOR inhibition could be beneficial in AML. mTOR is a serine/threonine protein kinase that is predominantly modulated by PI3K-AKT-dependent mechanisms and acts as a central regulator of cellular metabolism, growth and survival.4 Dysregulation of the mTOR pathway is closely associated with cancers including AML,5,6 and other human diseases. Part of the rationale is the evidence of constitutive activation of the PI3K-AKT pathway in 90% of AML samples and the demonstration that this activation is central to the survival of AML blasts but not to that of normal CD34+ cells.7 The concept that everolimus may have the potential to eliminate leukemiainitiating stem cells while sparing normal hematopoietic stem cells is also appealing. In vivo evidence in NOD/SCID mice has suggested that mTOR regulates a critical cell survival pathway in AML stem cells.8,9 In preliminary unrandomized clinical trial, the mTOR inhibitor sirolimus was administered as a single agent to nine relapsed, refractory or poor-risk AML patients for 28 days resulting in partial responses in four, and stable disease in a fifth patient.10 Dephosphorylation of downstream effectors of mTOR was demonstrated. In an ongoing UK trial, 11 elderly patients with primary, relapsed AML have been treated with the combination of low-dose Ara-C and sirolimus. Following a single 28-day course of treatment, of the seven patients eligible for analysis, one had achieved a complete remission, four had obtained a partial remission, one had profoundly hypocellular bone marrow and one patient was a non-responder (Das Gupta, unpublished data). Patients in this trial reliably maintained trough sirolimus levels of 8-16 ng/mL, which are consistent with the published concentrations required to inhibit AML cell growth in vitro. The feasibility of combining mTOR inhibition (sirolimus) with intensive chemotherapy had also been assessed in AML patients in conjunction with the more intensive MEC (mitoxantrone, etoposide and cytarabine) chemotherapy regimen in a phase I dose escalation study and reported in abstract form. In this study standard renal transplant doses of sirolimus were well tolerated and did not increase the non-hematologic toxicity of MEC chemotherapy with a median time to neutrophil recovery of 27 days.11 Based on this background information, the NCRI AML17 trial included the option for eligible patients to be randomized to receive, or not, the mTOR inhibitor everolimus daily between consolidation chemotherapy courses.

Methods The UK NCRI AML 17 trial (ISRNCTN 55675535) was a large, prospective, phase 3, multicenter trial for patients with newlydiagnosed AML or high-risk myelodysplastic syndrome (>10% marrow blasts), generally under the age of 60 years, open from April 2009 to December 2014 in more than 130 centers in the United Kingdom, Denmark and New Zealand. Through randomization of the participants, it addressed several issues (Online Supplementary Figure S1). Between October 2009 and October 2012, 499 adult patients who did not have acute promyelocytic leukemia had received a first induction course of treatment: those who did not have core-binding factor leukemia, high-risk disease (defined using a multifactorial score12) and were not in the lestaurtinib randomization for patients with FLT3 mutations, could be randomized to receive everolimus, or not, in a 2:1 ratio, between haematologica | 2018; 103(10)

subsequent consolidation chemotherapy courses. The treatment schedules have been set out elsewhere.13 Allogeneic stem cell transplantation was permitted for patients with intermediate- or poor-risk disease with a recommendation of myelo-ablative conditioning for patients aged <35 years and reduced intensity conditioning for intermediate-risk patients aged >45 years, with investigators able to choose an ablative or reduced intensity approach for patients between 35 and 45 years. Of all adult patients entering the AML17 trial while the everolimus randomization was available, 34% were eligible for such randomization. These patients were randomized to receive, or not, oral everolimus (10 mg daily from 2 days after each chemotherapy course for up to 28 days or until 2 days before the start of the subsequent course, whichever was shorter) between each course of consolidation chemotherapy. In patients allocated three courses of treatment, a final 28-day course of everolimus was given after a 1-week break. In patients with side effects thought to be due to everolimus, subsequent daily doses could be reduced by 50%. If this did not improve tolerability, dosing could be further reduced to alternate days; if these reduced doses were not tolerated, subsequent doses were to be omitted. After 65% (n=146) of the patients randomized to everolimus had been assessed, the independent data monitoring committee recommended, because of increased side effects and reduced compliance, that the starting daily dose of everolimus be reduced to 5 mg with the option to increase to 10 mg if well-tolerated. Extensive Sanger sequencing (111 genes) was undertaken in 123 patients; NPM1 status was available in 302 patients. Patients were requested to provide a trough blood sample taken immediately prior to everolimus dosing on day 14 of each treatment course to measure the level of mTOR inhibitory activity in their plasma. The methods for measuring this activity are summarized in Online Supplementary Figure S2 and will be reported more fully elsewhere.

Statistical considerations All analyses are on an intention-to-treat basis. Categorical endpoints were compared using Mantel-Haenszel tests, giving Peto odds ratios and confidence intervals. Continuous/scale variables were analyzed by Wilcoxon rank sum tests and time-to-event outcomes using the log-rank test, with Kaplan-Meier survival curves. Odds/hazard ratios (OR/HR) <1 indicate benefit for everolimus. All survival percentages refer to 5 years unless otherwise stated. Stratified analyses were performed with suitable tests for interaction14 and interpreted cautiously. It was planned to recruit 600 patients to the everolimus randomization, which would have given 85% power to detect a 12.5% difference in the primary endpoint of relapse-free survival, from 50% to 62.5% (HR 0.68). Follow-up is complete until March 1st, 2016 [median follow-up from diagnosis 53.5 months (range, 4.3 â&#x20AC;&#x201C; 76.8 months)]. The trial was conducted in accordance with the Declaration of Helsinki, sponsored by Cardiff University and approved by Wales REC3 on behalf of all UK investigators, by the Danish Medicines Agency for sites in Denmark, and by MEDSAFE for sites in New Zealand.

Results Patientsâ&#x20AC;&#x2122; characteristics The randomization opened in October 2009. In 2012, the independent data monitoring committee recommended closure of the randomization because of an excess of early mortality in remission with everolimus and no asso1655

A.K. Burnett et al.

Figure 1. CONSORT diagram. APL: acute promyelocytic leukemia; CEP-701: lestaurtinib; CBF: core-binding factor; MTOR: mammalian target of rapamycin;

ciated evidence of relapse reduction. Between October 2009 and October 2012, 332 of 482 eligible patients were randomized (Figure 1). Their characteristics are shown in Table 1. There were no significant differences in survival outcomes between eligible patients who entered the randomization and those who did not (P=0.8), although patients with higher white blood cell counts, worse performance status and secondary disease were marginally less likely to enter the randomization. The median age was 47 years (range, 16-69). The majority presented with de novo AML and had a WHO performance score of <2. The other protocol treatments given to patients in the everolimus randomization are shown in Table 1. In addition to standard daunorubicin/Ara-C induction, etoposide and gemtuzumab ozogamicin was given to 43% and 45% of patients, respectively, in induction with no difference between the arms. Overall, 132/332 (40%) of patients received a transplant (everolimus 39%, control 42%, P=0.6), with a minority of these (34/132) being allografts in first remission (20 versus 14, P=0.3). There was no evidence of differences in transplantation rates or types of transplants between the arms (any stem cell transplant 39% versus 42%, P=0.6; allograft 31% versus 34%, P=0.6; allograft in first complete remission 9% versus 13%, P=0.3) (Table 1). Extensive Sanger sequencing (111 genes) was undertaken in 123 patients: the gene panel and distribution are shown in Online Supplementary Figure S3. In addition NPM1 status, determined using previously published methods, was known for 302 patients.

Treatment compliance Of the 220 patients allocated to receive everolimus, 16 never started therapy. Approximately 25% of patients did not receive 14 days of everolimus; about half completed 1656

the first 28-day course. At the time of the second course of everolimus (course 3 of chemotherapy), 35% of patients for whom information on the second everolimus course was available did not receive the drug (Figure 2). Reasons were given for about two-thirds of patients (39/61): 11 patients had not completed the previous course; 11 patients chose to discontinue the therapy (often because of toxicity in the previous course); in three cases the data monitoring committee had recommended closure of the study with cessation of everolimus treatment; in five cases patients did not reach the starting point for everolimus therapy on protocol; in two cases the clinician decided, and in four other cases, everolimus was not given due to a variety of toxicities.

Toxicity The recorded toxicities are shown in Figure 3. There were more hematologic toxicities in the everolimus arm and these were most obvious after the first everolimus course, with a median time to platelet count recovery to >100x109/L being 9 days longer (39 versus 29 days; P=0.006); this was reflected by a significantly greater requirement for platelet support (Table 3). The kinetics of neutrophil recovery was unaffected by everolimus, but there was significantly more use of antibiotics and a longer stay in hospital with the first course of everolimus, as well as increased oral toxicity (course 1) and higher alanine transaminase levels (course 2).

Cumulative risk of relapse and death in remission The overall outcomes are shown in Table 2. The cumulative incidence of relapse at 5 years (Figure 4A) did not differ significantly between arms [60% versus 54%, HR 1.12 (0.82-1.52), P=0.5]. There was a significant excess of deaths in remission in the everolimus arm in the first 6 haematologica | 2018; 103(10)

Everolimus added to consolidation therapy for AML

Table 1. Patientsâ&#x20AC;&#x2122; characteristics.

Characteristic

Everolimus (n=220)

Age, years 16-29 33 (15%) 30-39 36 (16%) 40-49 58 (26%) 50-59 73 (33%) 60+ 20 (9%) Median 48 Range 16-69 Sex Female 117 (53%) Male 103 (47%) Diagnosis De novo 203 (92%) Secondary 5 (2%) MDS 12 (5%) WHO performance status 0 1 178 2 37 3 4 4 1 9 White blood cell count, x10 /L 0-9.9 138 (63%) 10-49.9 61 (28%) 50-99.9 15 (7%) 100+ 6 (3%) Median 5.8 Range 0.4-177.7 Cytogenetics Intermediate 194 (88%) Unknown 26 (12%) FLT3 ITD Wild/type 199 (96%) Mutant 8 (4%) Unknown 13

Control (n=112) 16 (14%) 17 (15%) 31 (28%) 37 (33%) 11 (10%) 46

70 (63%) 42 (37%) 103 (92%) 3 (3%) 6 (5%)

88 19 3 2 65 (58%) 34 (30%) 9 (8%) 4 (4%) 5.5 0.5-249.0 106 (95%) 6 (5%) 101 (99%) 1 (1%) 10

months following randomization [8% versus 1%, HR 3.57 (1.36-9.42), P=0.009], with no significant differences thereafter, leading to a non-significant excess of overall mortality with everolimus [11% versus 6%, HR 1.75 (0.833.70), P=0.14] (Figure 4B). In the first 6 months there were 17 deaths in remission in the everolimus arm versus 1 death in the control arm: the causes of these deaths were infection (9 versus 1), infection + hemorrhage (3 versus 0), hemorrhage/cardiovascular accident (3 versus 0), cardiac (1 versus 0) and multiple (1 versus 0). Beyond 6 months, there were six deaths in each of the two arms, with the causes of these deaths in remission being infection (1 versus 1), cardiac (1 versus 0), hepatic (1 versus 0), second cancer (1 versus 0), graft-versus-host disease (0 versus 1), multiple (0 versus 2) and unknown/other (2 versus 2).

Relapse-free and overall survival Both relapse-free and overall survival rates were nonsignificantly inferior in the everolimus arm (Figure 4C,D), reflecting the adverse hazard ratios for both relapse and death in remission, with no evidence of differences in salvage between arms after relapse [relapse-free survival: 29% versus 40%, HR 1.19 (0.90-1.59), P=0.2; overall survival: 45% versus 58%, HR 1.30 (0.94-1.81), P=0.11]. A sensitivity analysis censoring patients at the time of stem haematologica | 2018; 103(10)

Characteristic

Everolimus (n=220)

NPM1c Wild type 132 (65%) Mutant 70 (35%) Unknown 18 FLT3 TKD Wild type 204 (99%) Mutant 3 (1%) Unknown 13 Induction chemotherapy ADE (not randomized) 13 (6%) ADE ADE+GO3 ADE+GO6 DA+GO3 DA+GO6

29 (13%) 26 (12%) 26 (12%) 22 (10%) 26 (12%)

DA 90 mg 37 (17%) DA 60 mg 41 (19%) Risk score after course 1 Good risk 27 (13%) Standard risk 193 (87%) MRD status after course 1 (CR only) CR, MRD -ve 43 (20%) CR, MRD +ve 63 (29%) No MRD data/no CR 114 (52%) Transplanted 85 (39%) Any allograft 69 (31%) Any transplant in CR1 24 (11%) Allograft in CR1 20 (9%)

Control (n=112) 61 (61%) 39 (39%) 12 100 (98%) 2 (2%) 10 7 (6%) 14 (13%) 14 (13%) 14 (13%) 11 (10%) 12 (11%) 19 (17%) 21 (19%) 11 (10%) 101 (90%) 24 (21%) 24 (21%) 64 (57%) 47 (42%) 38 (34%) 16 (14%) 14 (13%)

MDS: myelodysplastic syndrome; WHO: World Health Organization; ITD: internal tandem duplication; TKD: tyrosine kinase domain; ADE: Ara-C, daunorubicin, etoposide; GO: gemtuzumab ozogamicin; DA: daunorubicin; CR: complete remission; MRD: minimal residual disease;

cell transplantation showed results which were consistent with the overall analysis (Table 2).

Exploratory analyses Correlations with everolimus plasma inhibitory activity, determined by the assay used in this study, did not show convincing patterns. Even patients whose samples showed deep and sustained inhibition did not have an associated reduction in relapse (Online Supplementary Figure S2). There was no relationship between the level of inhibition and toxicity or excess mortality. Prior induction chemotherapy, age, gender, white blood cell count, and minimal residual disease status after course one all had no impact on outcomes (Online Supplementary Figure S4A). In addition no relationship was found between other treatment modalities given and response, and no gene mutation, including the 111 genes assayed by Sanger sequencing in 123 patients, was shown to be associated with a differential response (Online Supplementary Figure S5). Because of concerns about compliance with everolimus treatment, relapse-free survival was compared between patients with satisfactory drug delivery (defined as at least 14 days of treatment per course), those with inadequate drug delivery (less than 14 days treatment per course) and those allocated to no 1657

A.K. Burnett et al.

Figure 2. Compliance with treatment. AE: adverse event; LFT: liver fuction tests.

Figure 3. Toxicities associated with treatment in courses 2 and 3. (A) Course 2 (B) Course 3.

1658

haematologica | 2018; 103(10)

Everolimus added to consolidation therapy for AML

treatment. Although patients in whom drug delivery was inadequate (n=85) had a worse relapse-free survival (29%) there was no difference in relapse-free survival between patients with satisfactory drug delivery (n=63) at 41% and no everolimus treatment (n=99) at 40% (Online Supplementary Figure S4).

CR/CRi MRD positivity after course 2 (CR only) 30-day mortality 60-day mortality 5-year OS 5-year RFS 5-year cumulative incidence of relapse 6-month death in CR 5-year cumulative incidence of death in CR 5-year survival after relapse 5-year OS censored at SCT

In this trial there was no benefit of the addition of everolimus to post-induction chemotherapy, despite the pre-clinical in vitro and in vivo rationale for the use of this mTOR inhibitor. The main explanations appear to be the Table 3. Recovery and supportive care in everolimus randomization.

Table 2. Clinical outcomes by treatment arm.

Everolimus Control

Discussion

OR/HR & CI

P-value

99% 63%

99% 65%

1.02 (0.09-11.2) 1.07 (0.50-2.33)

1.0 0.9

1% 4% 45% 29% 60%

1% 1% 58% 40% 54%

1.48 (0.19-11.7) 2.77 (0.70-11.0) 1.30 (0.94-1.81) 1.19 (0.90-1.59) 1.12 (0.82-1.52)

0.7 0.15 0.11 0.2 0.5

8% 11%

1% 6%

3.57 (1.36-9.42) 1.75 (0.83-3.70)

0.009 0.14

19% 57%

30% 66%

1.17 (0.81-1.70) 1.34 (0.87-2.06)

0.4 0.18

CR: complete remission; CRi: CR with incomplete blood count recovery; OS: overall survival; RE: relapse-free survival; SCT: stem cell transplantation.

Type of Care Neutrophil recovery (median days from start of course) Platelet recovery (median days from start of course) Blood (mean n. of units)

Course

Randomization Everolimus Control

P-value

0.4†

3 2 3 2 3 Platelets (mean n. of units) 2 3 Antibiotics (mean days) 2 3 Hospitalization (mean days) 2 3 Hospitalization (median days) 2 3

29 38 42 4.6 6.3 5.1 6.4 10.2 12.5 25.2 24.8 25.5 25

27 29 36 5.0 6.1 3.7 5.5 7.7 10.8 22.3 23.5 23 24.5

0.08† 0.006† 0.10† 0.08* 0.5* 0.009* 0.4* 0.002* 0.14* 0.02* 0.3*

† Logrank test. * Wilcoxon test.

Figure 4. Relapse, death in remission, relapse-free survival and overall survival within the everolimus randomisation: (A) Cumulative incidence of relapse; (B) cumulative incidence of death in remission; (C) relapse-free survival; (D) overall survival.

haematologica | 2018; 103(10)

1659

A.K. Burnett et al.

observed excess toxicity, which was primarily gastrointestinal (mucositis and diarrhea), and biochemical evidence of liver toxicity at the dose chosen. Infection was a major issue in the first 6 months of treatment with 12 versus 1 deaths attributed to infection in the everolimus and control arms, respectively. This did not appear to be the result of prolonged neutropenia but may be attributable to the immunosuppressive effects of everolimus when given with chemotherapy, which reflects what has been seen with the use of the mTOR inhibitor in solid tumors.15 This in turn contributed to sub-optimal drug delivery for many patients. The chosen schedule of 10 mg daily was not feasible in this setting, but drug delivery improved when a 5 mg daily dose was introduced. Other studies in leukemia have used equivalent schedules16,17 or a loading dose (12 mg) followed by 4 mg/day for 7 days per cycle11 or lower doses in combination with low-dose Ara-C.18 However even when the subgroup of good compliers was compared separately, there was no evidence of improved disease control. We had hoped that the development of an assay to quantitate plasma inhibitory activity would provide insight into response or toxicity, but unlike the experience of plasma inhibitory activity in the setting of an FLT3 inhibitor,19,20 consistent correlations were not found. In a phase 2 study of patients with relapsed AML treated with clofarabine and temsirolomus, correlation of response to dephosphorylation of pS6RP (S6 ribosomal protein) was demonstrated.21 However the target cells were the patients’ own blasts, which were not available in the current study and it was unclear whether the clinical outcome was superior to that which clofarabine alone could achieve. Finally the mTOR inhibitors tested to date have been inhibitors of the TORC1 pathway. This may be by-passed by the TORC2 pathway which is insensitive to this class of mTOR inhibitors, but may be sensitive to agents which produce dual inhibition. Acknowledgments The authors are grateful to Novartis for the provision of everolimus, to Cancer Research UK for research funding of the trial and to the investigators, research staff and patients in the participating sites: Aalborg Hospital: Maria Kallenbach; Aarhus University Hospital: Hans Beier Ommen, Jan Maxwell Norgaard; Aberdeen Royal Infirmary: Dominic Culligan; Addenbrookes Hospital: George Follows, Jenny Craig; Auckland City Hospital: Lucy Pemberton, Richard Doocey, Sophie Lee, Timothy Hawkins; Barnet General Hospital: Andres Virchis; Barts and the London NHS Trust: Jamie Cavenagh, Matthew Smith; Basingstoke and North Hampshire Foundation NHS Trust: Sylwia Simpson; Beatson West of Scotland Cancer Centre: Mark Drummond; Belfast City Hospital: Claire Arnold, Mary Francis McMullin, Robert Cuthbert; Birmingham Heartlands Hospital: Donald Milligan, Guy Pratt, Matthew Lumley, Shankara Paneesha; Blackpool Victoria Hospital NHS Foundation Trust: Paul Cahalin; Borders General Hospital: John Tucker; Bradford Royal infirmary: Adrian Wiliams, Lisa Newton, Sam Ackroyd; Bristol Haematology and Oncology Centre: Priyanka Mehta; Chesterfield Royal Hospital: Mark Wodzinski, Robert Cutting; Christchurch Hospital: Ruth Spearing, Steve Gibbons; Christie Hospital NHS Trust: Mike Dennis; Countess of Chester Hospital: Salah Tuegar; Crosshouse Hospital: Julie Gillies; Derby Hospitals NHS 1660

Foundation Trust: Juanah Addada; Derriford Hospital: Hannah Hunter, Tim Nokes; Doncaster Royal Infirmary: Stuti Kaul; Dorset County Hospital NHS Foundation Trust: Akeel Moosa; East Kent Hospitals University NHS Foundation Trust: Jindriska Lindsay, Vijay Ratnayake; East Sussex Hospitals NHS Trust: Richard Grace; Falkirk and District Royal Infirmary: Christopher Brammer, Marie Hughes; Glan Clwyd Hospital: Christine Hoyle, Earnest Heartin, Margaret Goodrick; Gloucestershire Royal Hospital: Adam Rye, Sally Chown; Great Western Hospital: Alex Sternberg, Atherton Gray, Norrbert Blessing; Guys and St Thomas' Foundation Trust: Kavita Raj, Robert Carr; Hairmyres Hospital: Iain Singer; Heatherwood and Wexham Park NHS Foundation Trust: Nicola Bienz, Simon Moule; Hereford County Hospital: Sara Willoughby; Herlev Hospital: Morten Krogh Jensen, Peter Moller; Hillingdon Hospital: Riaz Janmohamed, Richard Kaczmarski; Hull Royal Infirmary: Sahra Ali; James Cook University Hospital: Ray Dang; James Paget University Hospital: Cesar Gomez, Shala Sadullah; John Radcliffe Hospital: Paresh Vyas; Kettering General Hospital: Isaac Wilson-Morkeh, Matthew Lyttelten; Leicester Royal Infirmary: Ann Hunter, Murray Martin; Lincoln County Hospital: Kandeepan Saravanamuttu; Maidstone Hospital: Evangelia Dimitriadou; Manchester Royal Infirmary: Eleni Tholouli, Guy Lucas; Milton Keynes Hospital NHS Foundation Trust: Moez Dungarwalla; Monklands Hospital: John Murphy, Lindsey Mitchell, Pamela Paterson; New Cross Hospital: Sunil Hada, Supratik Basu; Ninewells Hospital and Medical Centre: Keith Gelly; Norfolk and Norwich University Hospital NHS Foundation Trust: Matthew Lawes; Northampton General Hospital: Angela Bowen, Sajan Mittal, Suchitra Krishnamurthy@Ngh.Nhs.Uk; Nottingham University Hospitals NHS Trust: Emma Dasgupta, Jenny Byrne, Kate Forman, Nigel Russell; Odense University Hospital: Claus Marcher, Lone Friis, Poul Gram Hansen; Peterborough District Hospital: Kanchan Rege; Pinderfields General Hospital: David Wright, Mary Chapple, Paul Moreton; Poole General Hospital: Fergus Jack; Queen Alexandra Hospital: Mary Ganczakowski, Tanya Cranfield; Queen Elizabeth Hospital, Birmingham: Charles Craddock, Jim Murray; Queen Elizabeth Hospital, Norfolk: Jane Keidan; Queens Hospital, Romford: Claire Hemmaway; Raigmore Hospital NHS Highland: Chris Lush, Peter Forsyth; Rigshospitalet: Carsten Niemann, Lars Kjeldsen, Ole Wei Bjerrum, Ove Juul Nielsen, Peter Kampmann; Rotherham General Hospital: Arun Alfred; Royal Berkshire Hospital: Henri Groch, Stuart Mucklow; Royal Cornwall Hospital: Bryson Pottinger, Richard Noble; Royal Devon and Exeter Hospital: Claudius Rudin, Malcolm Hamilton, Paul Kerr; Royal Free Hospital: Panos Kottaridis; Royal Hallamshire Hospital: Chris Dalley, John Snowden; Royal Liverpool University Hospital: Rahuman Salim, Richard Clark; Royal Marsden Hospital: Mark Ethell; Royal Oldham Hospital: Allameddine Allameddine, David Osborne, Hayley Greenfield, Sumaya Elhanash, Vivek Sen; Royal Surrey County Hospital: Johannes Devos, Louise Hendry; Royal Sussex County Hospital: Timothy Corbett; Russell’s Hall Hospital: Jeff Neilson; Salford Royal Hospital: John Houghton, Simon Jowitt, Sonya Zaman; Salisbury Hospital NHS Foundation Trust: Jonathan Cullis, Tamara Everington; Sandwell Hospital: Farooq Wandoo, Yasmin Hasan; Singleton Hospital: Saad Ismail; South Devon Healthcare NHS Foundation Trust: Deborah Turner, Nicholas Rymes; Southampton University Hospital NHS Trust: Deborah Richardson, Kim Orchard, Matthew Jenner; St Helens and Knowsley NHS Trust: Toby Nicholson; St James University Hospital: David Bowen; St Richard’s Hospital: Sarah Janes; Stafford Hospital: Andrew Amos; Stoke haematologica | 2018; 103(10)

Everolimus added to consolidation therapy for AML

Mandeville Hospital: Helen Eagleton; Sunderland Royal Hospital: Annette Nicolle, Scott Marshall; Taunton and Somerset Foundation Trust: Sarah Allford; The Newcastle upon Tyne NHS Foundation Trust: Gail Jones, Graham Jackson; University College London Hospitals: Anthony Goldstone, Asim Khwaja, Kirit Ardeshna, Nishal Patel; University Hospital Aintree: Barbara Hammer, Walid Sadik; University Hospital Coventry and Warwickshire NHS Trust: Mekkali Narayanan, Nicholas Jackson, Peter Rose, Syed Bokhari; University

References 9. 1. Buckley SA, Wood BL, Othus M, et al. Minimal residual disease prior to allogeneic hematopoietic cell transplantation in acute myeloid leukemia: a meta-analysis. Haematologica. 2017;102(5):865-873. 2. Ivey A, Hills RK, Simpson MA, et al. Assessment of minimal residual disease in standard-risk AML N Engl J Med. 2016; 374(5):422-433. 3. Burnett AK, Hills RK, Milligan DW, et al. Attempts to optimize induction and consolidation treatment in acute myeloid leukemia: results of the MRC AML12 trial. J Clin Oncol. 2010;28(4):586-595. 4. Dobashi Y, Watanabe Y, Miwa C, Suzuki S, Koyama S. Mammalian target of rapamycin: a central node of complex signaling cascades. Int J Clin Exp Pathol. 2011;4(5):476495. 5. Zoncu R, Efeyan A, Sabatini DM. mTOR: from growth signal integration to cancer, diabetes and ageing. Nat Rev Mol Cell Biol. 2011;12(1):21-35. 6. Park S, Chapuis N, Tamburini J, et al. Role of the PI3K/AKT and mTOR signaling pathways in acute myeloid leukemia. Haematologica. 2010;95(5):819-828. 7. Recher C, Beyne-Rauzy O, Demur C, et al. Antileukemic activity of rapamycin in acute myeloid leukemia. Blood. 2005;105(6):25272534. 8. Xu Q, Simpson SE, Scialla TJ, Bagg A, Carroll M. Survival of acute myeloid

haematologica | 2018; 103(10)

10.

11.

12.

13.

14.

15.

Hospital of North Staffordshire NHS Trust: Andrew Stewart, Kamaraj Karunanithi, Neil Phillips, Srinivas Pillai; University Hospital of North Tees and Hartlepool: Zor Maung; University Hospital of Wales: Jonathan Kell, Steve Knapper; Victoria Hospital NHS Fife: Stephen Rogers; Waikato Hospital: Hugh Goodman, Humphrey Pullon; Wellington Hospital: John Carter; Western General Hospital: Peter Johnson, Ph Roddie, Annielle Hung; Worcestershire Royal Hospital: Juliet Mills; Worthing Hospital: Santosh Narat.

leukemia cells requires PI3 kinase activation. Blood. 2003;102(3):972-980. Xu Q, Thompson JE, Carroll M. mTOR regulates cell survival after etoposide treatment in primary AML cells. Blood. 2005;106(13):4261-4268. Luger S, Perl A, Kemner A, et al. A phase I dose escalation study of the mTOR inhibitor sirolimus and MEC chemotherapy targetting signal transduction in leukemic stem cells for acute myeloid leukemia. Blood. 2006;108(11):161. Perl AE, Kasner MT, Tsai DE, et al. A phase I study of the mammalian target of rapamycin inhibitor sirolimus and MEC chemotherapy in relapsed and refractory acute myelogenous leukemia. Clin Cancer Res. 2009;(15):6732-6739. Burnett AK, Hills RK, Wheatley K, et al. A sensitive risk score for directing treatment in younger patients with AML Blood. 2006; 108(11):18. Knapper S, Russell N, Gilkes A, et al. A randomized assessment of adding the kinase inhibitor lestaurtinib to first-line chemotherapy for FLT3-mutated AML. Blood. 2017; 129(9):1143-1154. Early Breat Cancer Trialists Collaborative Group (EBCTCG). Treatment of early breast cancer. 1. Worldwide evidence 1985-1990, Oxford University Press, USA, 1990 Kaymakcalan,M D., Je Y. Sonpavde G. et al. Risk of infections in renal cell carcinoma (RCC) and non-RCC patients treated with mammalian target of rapamycin inhibitors. Br J Cancer. 2013;108(12):2478-2484.

16. Daver N, Boumber Y, Kantarjiian H, et al. A Phase l/ll study of the mTOR inhibitor everolimus in combination with hyperCVAD chemotherapy in patients with relapsed/refractory acute lymphoblastic leukemia. Clin Cancer Res. 2015;21(12): 2704-2714. 17. Yee AL, Zeng Z, Konople. Phase I/II study of the mammalian target or rapamycin inhibitor everolimus (RAD001) inn patients with relapsed or refractory hematologic malignancies. Clin Cancer Res. 2006;12(17): 5165-5173. 18. Wei AH, Sadawarte S, Catalono J. Phase Ib study combining mTOR inhibitor everolimus (RAD001) with low-dose cytarabine in untreated elderly AML. Blood. 2010;116(21):3299. 19. Levis M, Brown P, Smith BD, et al. Plasma inhibitory activity (PIA): a pharmacodynamic assay reveals insights into the basis for cytotoxic response to FLT3 inhibitors. Blood. 2006;108(10):3477-3483. 20. Knapper S, Burnett AK, Littlewood T, et al. A Phase 2 trial of the FLT3 inhibitor lestaurtinib (CEP701) as first-line treatment for older patients with acute myeloid leukemia not considered fit for intensive chemotherapy. Blood. 2006;108(10):3262-3270. 21. Amadori S, Stasi R, Martelli AM. Temsirolimus, an mTOR inhibitor, in combination with lower-dose clofarabine as salvage therapy for older patients with acute myeloid leukaemia: results of a phase II GIMEMA study (AML1107) Br J Haematol. 2012;156(2):205-212.

1661

ARTICLE

Acute Lymphoblastic Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1662-1668

Therapy-related acute lymphoblastic leukemia has distinct clinical and cytogenetic features compared to de novo acute lymphoblastic leukemia, but outcomes are comparable in transplanted patients Ibrahim Aldoss,1 Tracey Stiller,2 Ni-Chun Tsai,2 Joo Y. Song,3 Thai Cao,1,4 N. Achini Bandara,1 Amandeep Salhotra,1 Samer Khaled,1 Ahmed Aribi,1 Monzr M. Al Malki,1 Matthew Mei,1 Haris Ali,1 Ricardo Spielberger,1,4 Margaret Oâ&#x20AC;&#x2122;Donnell,1 David Snyder,1 Thomas Slavin,5 Ryotaro Nakamura,1 Anthony S. Stein,1 Stephen J. Forman,1 Guido Marcucci1 and Vinod Pullarkat1

Department of Hematology and Hematopoietic Cell Transplantation, Gehr Family Center for Leukemia Research, City of Hope, Duarte; 2Department of Information Sciences, Division of Biostatistics, City of Hope, Duarte; 3Department of Pathology, City of Hope, Duarte; 4Kaiser Permanente, Department of BMT, Southern California Medical Group, Los Angeles and 5Department of Medical Oncology, Division of Clinical Genetics, City of Hope, Duarte, CA, USA

ABSTRACT

Correspondence: ialdoss@coh.org

Received: March 19, 2018. Accepted: June 8, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2018.193599 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1662 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1662

herapy-related acute lymphoblastic leukemia remains poorly defined due to a lack of large data sets recognizing the defining characteristics of this entity. We reviewed all consecutive cases of adult acute lymphoblastic leukemia treated at our institution between 2000 and 2017 and identified therapy-related cases - defined as acute lymphoblastic leukemia preceded by prior exposure to cytotoxic chemotherapy and/or radiation. Of 1022 patients with acute lymphoblastic leukemia, 93 (9.1%) were classified as therapy-related. The median latency for therapy-related acute lymphoblastic leukemia onset was 6.8 years from original diagnosis, and this was shorter for patients carrying the MLL gene rearrangement compared to those with other cytogenetics. When compared to de novo acute lymphoblastic leukemia, therapy-related patients were older (P<0.01), more often female (P<0.01), and had more MLL gene rearrangement (P<0.0001) and chromosomes 5/7 aberrations (P=0.02). Although therapy-related acute lymphoblastic leukemia was associated with inferior 2-year overall survival compared to de novo cases (46.0% vs. 68.1%, P=0.001), prior exposure to cytotoxic therapy (therapy-related) did not independently impact survival in multivariate analysis (HR=1.32; 95% CI: 0.97-1.80, P=0.08). There was no survival difference (2-year = 53.4% vs. 58.9%, P=0.68) between the two groups in patients who received allogenic hematopoietic cell transplantation. In conclusion, therapy-related acute lymphoblastic leukemia represents a significant proportion of adult acute lymphoblastic leukemia diagnoses, and a subset of cases carry clinical and cytogenetic abnormalities similar to therapy-related myeloid neoplasms. Although survival of therapy-related acute lymphoblastic leukemia was inferior to de novo cases, allogeneic hematopoietic cell transplantation outcomes were comparable for the two entities.

Introduction Therapy-related leukemia has increasingly emerged as a long-term complication of cytotoxic therapy (i.e., chemotherapy and radiation) for patients who have undergone treatment for preceding malignancies.1 Therapy-related myeloid neoplasms (t-MNs) are widely recognized and comprise an established category in the WHO classification of MNs, which include therapy-related acute myeloid haematologica | 2018; 103(10)

Therapy-related ALL

leukemia (t-AML) and therapy-related myelodysplastic syndrome (t-MDS).2 T-MNs, in general, carry poor cytogenetic and molecular features at the time of diagnosis compared to de novo MNs, and are characterized by poor responsiveness to conventional treatment and inferior rates of overall outcome, such as complete remission (CR), death in remission, relapse and survival.1,3,4 Similar to t-MN, acute lymphoblastic leukemia (ALL) may also develop after prior exposure to cytotoxic therapies and is often referred to as therapy-related ALL (tALL).5-11 Similar to t-MNs, the pathogenesis of t-ALL is likely attributed to the genotoxic effect of cytotoxic therapies on hematopoietic progenitor cells. However, to date, this entity has not been fully recognized and only a few relatively small series have been reported.5-11 Unfortunately, several of these studies also include “secondary ALL,” i.e., cases with a history of prior malignancies (including non-lymphoid cancers), but no cytotoxic therapy exposure, making the accurate identification of tALL- specific clinical and genetic features somewhat challenging. Few large registry series of secondary ALL have been reported and have highlighted the inferior survival of this entity.12,13 However, these registry studies have not drawn a distinction between cases with prior malignancies that did not receive cytotoxic therapies and those that did. Additionally, these studies lack specific details on ALL genetics as well as details on prior cancer-specific therapies due to limitations of registry data.12,13 Furthermore, the optimal therapy for t-ALL as well as tALL patients’ ability to tolerate intensive treatment remain poorly defined. This becomes particularly important when a t-ALL patient has high risk features and is being considered for allogeneic hematopoietic cell transplantation (HCT). A particular concern in this regard is higher treatment-related morbidity and mortality given the prior exposure to cytotoxic therapies in t-ALL patients. Therefore, studies that clearly distinguish t-ALL are necessary in order to fill the scientific and clinical knowledge gap in this field. We report here a large, single institutional, t-ALL cohort defined using strict inclusion criteria that restrict analysis only to cases with documented exposure to cytotoxic therapy prior to developing ALL. In contrast to registry data, we were able to gather details regarding prior malignancies and therapies, clinical and genetic characteristics of the ALL, treatment, and outcomes of t-ALL from our institutional database. Our study aims to estimate the frequency of t-ALL among adult patients, to evaluate unique clinical and genetic features associated with t-ALL that are distinctive from de novo ALL, and to evaluate the prognostic impact of prior exposure to cytotoxic therapy (tALL) on clinical outcomes, including response to induction therapy, utilization and outcomes of allogeneic HCT and survival.

Methods Patients We reviewed all consecutive cases of adult ALL seen at City of Hope between 2000 and 2017 in order to identify cases of t-ALL. For the purposes of this study, t-ALL was defined as ALL occurring after prior exposure to chemotherapy and/or radiation. Any cases of ALL preceded by a malignancy but without exposure to cytohaematologica | 2018; 103(10)

toxic therapy were classified as de novo ALL. t-ALL and de novo ALL cases were then compared for distinctive demographic, clinical, and cytogenetic features and for outcomes. The study was approved by the City of Hope Institutional Review Board.

Endpoints Overall survival (OS) for all patients was defined as the time interval from ALL diagnosis to date of death from any cause or date of last contact. When analyzing patients who underwent allogeneic HCT, OS was defined as the time from transplant to date of death from any cause or date of last contact. Non-relapse mortality (NRM) was measured from time of transplant to death from any cause other than relapse/progression. Relapse/progression was treated as a competing event for NRM.

Statistical analyses Demographic, disease, and treatment characteristics were summarized using descriptive statistics. Two sample t-test, chisquared test, and fisher’s exact test were used to determine differences in demographics and disease characteristics of interest. Survival estimates were calculated using the Kaplan-Meier product-limit method and differences between Kaplan-Meier curves were assessed using the log-rank test.14 The cumulative incidence of NRM was calculated using competing risk analysis and differences between cumulative incidence curves were tested using the Gray method.15 Prognostic variables analyzed include age and white blood cell (WBC) count as continuous variables, cytogenetics (NK, Ph+, MLL, complex [≥5 abnormalities], or other/unknown), prior therapy (chemotherapy, radiation, or chemotherapy plus radiation), prior disease (solid tumor vs. blood cancer), allogeneic HCT treated as a time dependent variable, race/ethnicity (white, Hispanic, other), phenotype (T vs. B), sex (female vs. male), and use of topoisomerase II inhibitor (no vs. yes). The significance of demographic, disease, and treatment features was assessed using logistic regression to determine effect on type of ALL diagnosis and Cox proportional hazards regression analysis to determine effect on survival. All analyses performed using SAS version 9.4 (SAS Institute, Cary, NC). Data were locked for analysis January 17, 2018.

Results Comparison of clinical and pathologic characteristics t-ALL and de novo ALL Between 2000 and 2017, 1022 cases of adult ALL were evaluated and/or treated at City of Hope; 93 (9.1%) had t-ALL. When compared to de novo ALL, t-ALL patients were older (55 years vs. 37 years, P<0.01), and were more often female (57% vs. 42%, P<0.01). There was no difference in proportions of leukemia phenotypes (precursor B-cell versus T-cell) between t-ALL and de novo ALL. t-ALL patients were more often whites (52% vs. 34%) and less often Hispanics (29% vs. 48%) compared to de novo ALL (P<0.01). t-ALL cases were associated with different cytogenetic profiles (P<0.01) compared to de novo ALL. t-ALL cases were enriched with MLL gene rearrangement (KMT2A) (17% vs. 4%, P<0.01) and have less normal karyotype (18% vs. 30%, P=0.017) when compared to de novo ALL. Among patients with available conventional cytogenetics, monosomy and/or long arm deletion of chromosomes 5 and/or 7 were more common in t-ALL compared to de novo ALL (16% vs. 8%, P=0.02) (Table 1). In multivariate analysis, t-ALL was associated with older age (OR= 1.06; 95% CI:1.04-1.07, P<0.0001), female 1663

I. Aldoss et al.

sex (OR=1.64; 95% CI:1.02-2.65, P=0.04), lower WBC at presentation (OR=0.996; 95%CI:0.99-1.00, P=0.038), and MLL gene rearrangement (OR=6.52; 95%CI:2.66-15.96, P<0.0001) (Table 2).

Characteristics of the t-ALL cohort The original diagnosis prior to t-ALL onset was solid cancer in 52 (56%) patients, hematological cancer in 33 (35%) patients, combined solid and hematological cancers in 2 (2%) patients, and 6 (6%) patients had non-malignant diseases treated with cytotoxic therapies. Breast cancer was the most common prior diagnosis (n=23, 25%) followed by lymphoproliferative neoplasms (non-Hodgkin lymphoma, chronic lymphocytic leukemia, Hodgkinâ&#x20AC;&#x2122;s lymphoma) (n=21, 23%), and multiple myeloma (MM)

Table 1. Overall comparison between t-ALL and de novo ALL.

All patients De novo ALL Number Age Sex Female Male Phenotype B T ETP# B/T WBC Cytogenetic Ph+ MLL NK Complex Others UK Ch 5/7q deletion/ monosomy YES NO Ph+ & available cytogenetics ACA Isolated Ph Race White Hispanic Asian AA Others UK

t-ALL

P < 0.01

1022 39 (6-85)

929 37 (6-85)

93 55 (23-85)

440 (43) 582 (57)

387 (42) 542 (58)

53 (57) 40 (43)

870 (85) 785 (85) 85 (91) 150 (15) 142 (15) 8 (9) 20(19) 19 (19) 1 (17) 2 (<1) 2 (<1) 0 (0) 16.3 (0.2-778) 17 (0.2-778) 10 (0.9-330) 270 (26) 57 (6) 296 (29) 46 (5) 250 (24) 103 (10)

241 (26) 41 (4) 279 (30) 4 (4) 229 (25) 98 (11)

29 (31) 16 (17) 17 (18) 5 (5) 21 (23) 5 (5)

<0.01

0.09

0.10 <0.01 0.27 <0.01 0.017 0.60

De novo Therapy-related

0.02 71 (8) 779 (92) 204

58 (8) 711 (92) 182 (76)

13 (16) 68 (84) 22 (76)

107 97

91 (50) 91 (50)

16 (73) 6 (27)

368 (36) 472 (46) 86 (9) 31 (3) 22 2) 43 (4)

320 (34) 445 (48) 76 (8) 29 (3) 19 (2) 40 (4)

48 (52) 27 (29) 10 (11) 2 (2) 3 (3) 3 (3)

0.07

<0.01

ETP: early thymic T cell; WBC: white blood cell count; Ph+: Philadelphia-chromosome positive; MLL: Mixed lineage leukemia; NK: normal karyotype; UK: unknown, Ch: chromosome; ACA: additional cytogenetic abnormalities; AA: African American. # There were 108 cases of T-cell ALL (de novo = 102, t-ALL = 6) with available adequate markers upon review to make the diagnosis of ETP.

1664

(n=11, 12%). Thirty-five (38%) patients had chemotherapy alone as prior therapy for the original diagnosis, 26 (28%) had only radiotherapy, 32 (34%) had a combination of chemotherapy and radiation, 17 (18%) received an autologous hematopoietic cell transplant (HCT) as part of prior therapy, and 13 (14%) had immunomodulatory agents in combination with chemotherapy. Interestingly, 2 cases had antecedent MDS before presenting with t-ALL (Table 3). Eighty-three percent of t-ALL patients with available conventional cytogenetic and/or FISH studies had cytogenetic abnormalities. Philadelphia (Ph) chromosome was the most common finding on cytogenetics for the t-ALL cohort, and followed by normal karyotype and mixed lineage leukemia (MLL) gene rearrangement. Among 78

De novo Therapy-related

Figure 1. Survival for t-ALL and de novo ALL. A. Survival curves for all t-ALL (dashed line) and de novo ALL (solid line) and B. Survival curves for t-ALL (dashed line) and de novo (solid line) ALL in patients who underwent allogeneic HCT during ALL therapy.

Therapy-related ALL

cases with available conventional cytogenetics, 14 (18%) patients met the definition of monosomal karyotype (two or more distinct autosomal chromosome monosomies or one single autosomal monosomy in the presence of structural abnormalities).16 The median latency for developing ALL was 6.8 years (0.8-50.7) from the time of original malignancy/disease diagnosis, and it was shorter in patients carrying the MLL gene rearrangement compared to patients carrying the Ph chromosome or other cytogenetic subgroups (2.8 years vs. 7.0 years vs. 8.0 years, P<0.01), respectively. Only cytogenetics was independently associated with the interval duration for developing ALL (P=0.02) (Table 4). Topoisomerase II inhibitors were administered as part of prior therapy in 41% (n=38) of the t-ALL cohort, and were given in combination with alkylators and radiation in the majority of patients. Prior topoisomerase II inhibitor exposure did not influence the latency period between the original disease diagnosis and ALL onset (P=0.45) or the cytogenetic profile (P=0.69). All t-ALL patients except for one received induction therapy for ALL. HyperCVAD with or without tyrosine kinase inhibitors was the most commonly used regimen (n=48, 52%) to induce t-ALL patients. Median follow up for all patients and surviving t-ALL patients was 14.4 months (range: 0.2-181.7) and 17 months (range: 0.8181.7), respectively. Neither age (P=0.43), prior therapy (P=0.44), cytogenetic subgroup (P=0.51), prior diagnosis (P=0.51) nor the use allogeneic HCT (P=0.07) influenced OS for t-ALL patients in multivariate analysis (Online Supplementary Table S1). Nine (9.7%) patients had their original malignancies relapse after ALL diagnosis. However, 2-year OS was not different between patients who had recurrence of their original disease and those who did not (45% vs. 50%, P=0.91).

Comparison of outcomes of t-ALL and de novo ALL The median follow up for all patients and for surviving patients were 26 months (range: 0.2-255.5) and 43.6 months (range: 0.3-255.5), respectively. The 2-year OS for all patients was 66.2% (95% CI 63.0-69.2). CR rate was similar for both t-ALL and de novo ALL patients (85%, P=0.88) as was the percentage of patients who underwent allogeneic HCT consolidation (53% vs. 61%, P=0.15). However, more patients with t-ALL were transplanted in CR1 compared to de novo ALL (76% vs. 60%, P=0.05). The 2-year OS was inferior for t-ALL compared to de novo ALL (46.0% vs. 68.1%, P=0.001) (Figure 1A). In multivariate analysis, age at ALL diagnosis (P<0.0001), WBC at diagnosis (P=0.003), cytogenetics (P<0.0001), sex (P=0.005), HCT (P=0.02) and leukemia phenotype (P=0.02) influenced OS for all patients. Interestingly, prior exposure to cytotoxic therapy before ALL onset (t-ALL) was not an independent predictor of OS [HR=1.32; 95% CI: 0.97-1.80, P=0.08] (Table 5). When analysis was restricted to the 613 patients who underwent allogeneic HCT as part of their ALL therapy (t-ALL=49, de novo ALL=564), the median follow up was 25.5 months (range: 0.03-198.3) and 2-year OS was 58.5% (95% CI: 54.4-62.4) for all patients. The 2-year OS was similar for both t-ALL and de novo ALL (53.4% vs. 58.9%, P=0.68) despite more frequent use of reduced-intensity conditioning for t-ALL compared to de novo ALL (P<0.01) (Figure 1B). No difference was observed in non-relapse mortality (NRM) between t-ALL and de novo ALL (28.5% vs. 22.7%, P=0.38), respectively (Online Supplemental Figure S1). For the 409 patients who did not undergo allogeneic HCT (t-ALL=44, de novo ALL=365), the 2-year OS was inferior for t-ALL compared to de novo ALL (27.1% vs. 52.9%, P=0.0004) (Online Supplemental Figure S2). Again, prior cytotoxic therapy before ALL onset (t-ALL) was not

Table 2. Multivariable model for factors associated with t-ALL or de novo ALL.

Age at ALL diagnosis WBC Cytogenetic Group NK Ph+ MLL Complex Other/Unknown Race/Ethnicity White Hispanic Other Phenotype T B Sex Male Female

Therapy-related N=88

De Novo N=823

Odds Ratio

95% CI

54.5 (23-85) 9.95 (0.9-330)

38 (18-85) 17 (0.2-778)

1.06 0.996

1.04-1.07 0.99-1.00

16 (18) 28 (32) 15 (17) 5 (6) 24 (27)

257 (31) 223 (27) 39 (5) 34 (4) 270 (33)

1.46 6.52 2.16 1.41

0.74 -2.90 2.66 -15.96 0.70-6.64 0.71-2.80

<0.0001 0.038 0.0009

47 (53) 25 (29) 16 (18)

280 (34) 399 (48) 144 (18)

0.60 0.79

0.34-1.04 0.41-1.50

0.28 <0.0001 0.18 0.33 0.19 0.07 0.07 0.46

8 (9) 80 (91)

107 (13) 716 (87)

1.40

0.61-3.20

0.42

38 (43) 50 (57)

469 (57) 354 (43)

1.64

1.02-2.65

0.04

WBC: white blood cell count; NK: normal karyotype; Ph+: Philadelphia chromosome positive; MLL: mixed lineage leukemia.

haematologica | 2018; 103(10)

1665

I. Aldoss et al.

an independent predictor of survival per se when included in multivariate analysis in this cohort (P=0.11).

Discussion We present here the largest retrospective study of t-ALL with analysis solely restricted to cases with prior exposure to cytotoxic therapies. Unlike some previously published reports, we excluded cases of ALL that were preceded by other malignancies but did not receive cytotoxic chemotherapy or radiation in an attempt to more narrowly define the entity of t-ALL.7,8 Although t-ALL does not have unique defining pathologic features, we show that certain recurrent cytogenetic abnormalities are more common in t-ALL compared to de novo ALL. The cytogenetic features of t-ALL bear some resemblance to t-AML and may help define t-ALL. Therapyrelated leukemia with balanced translocations has been observed in t-AML, especially in patients with prior exposure to topoisomerase II inhibitors.3,17,18 MLL (11q23) is the prototypic cytogenetic finding among t-AML patients exposed to topoisomerase II inhibitors, and here we have shown that the incidence of MLL is also more common among t-ALL compared to de novo cases. However, we could not demonstrate association between prior topoisomerase II exposure and MLL findings, and this is likely due to the frequent administration of radiation and alkylator therapy along with topoisomerase II inhibitors. Consistent with t-AML data, we show that the latency for t-ALL onset was shorter among patients carrying the MLL gene rearrangement compared to other cytogenetic findings. Furthermore, similarly to t-MN, our t-ALL cases were associated with a higher occurrence of long arm deletions or monosomy 5 and 7.3 These cytogenetic findings support the etiologic role of prior chemotherapy in pathogenesis of attribution of t-ALL in a manner similar to t-MN. Philadelphia (Ph) chromosome is another balanced translocation and was more commonly noted among t-ALL cases, but this was not statistically significant in this cohort. Ph chromosome is rarely observed among T-cell phenotype ALL and AML cases, and prior reports have shown that some of those cases were potentially therapy-related and developed after cytotoxic exposure.6,19 Nonetheless, we have observed a trend toward higher rates of additional cytogenetic abnormalities among Ph+ t-ALL compared to Ph+ de novo ALL (73% vs. 50%, P=0.07), and this likely reflects various levels of genomic instability as a result of prior cytotoxic therapy. The incidence of Ph-like ALL would have been an interesting comparison to make between de novo and t-ALL, but unfortunately, we did not have the necessary data available in our cohort. The latency for ALL development from time of prior diagnosis was 6.8 years in our series, which is slightly longer than what is observed in t-MN (4-4.5 years).1,3 Both B and T-cell ALL phenotypes were observed in a similar proportion compared to de novo ALL. Breast cancer was the most common prior malignancy, likely related to the elevated utilization of alkylator and topoisomerase II inhibitor chemotherapy as well as radiation in early stage disease, and excellent long-term survival for breast cancer patients, allowing time for hematopoietic clonal evolution to acute leukemia. The patient demographics of our cohort also support the existence of t-ALL as a distinct entity. Interestingly, 1666

although the overall majority of ALL patients in our series were Hispanics, t-ALL was twice as common in whites compared to Hispanics. In the United States, ALL is more common in Hispanics in general20,21 and is characterized by unique genetic profiles such as the Ph-like signature,22 which in turn is associated with inherited genetic polymorphisms in the GATA3 gene.23 Although we do not have data on Ph-like ALL in our cohort, it would likely have been higher in our de novo ALL cohort given the demographics of our patient population. In contrast, the higher proportion of whites in our t-ALL cohort may be reflective of the ethnic distribution of the antecedent malignancies (e.g., breast cancer) in the t-ALL population. Given the prior exposure to chemotherapy, side effects

Table 3. Prior diagnoses and characteristics associated with t-ALL.

Number

Median latency in years for all patients (range) 6.8 (0.8-50.7) Prior diagnosis Solid cancer 52 (56) Hematological cancer 33 (35) Benign 6 (6) Both solid and hematological cancers 2 (2) Prior diagnoses More than one prior diagnosis 4 (4) Breast cancer 23 (25) Lymphoproliferative neoplasms# 21 (23) Multiple myeloma 11 (12) Thyroid cancer/disease 8 (8) Sarcoma 8 (8) Testicular 4 (4) Prostate cancer 3 (3) Gastrointestinal malignancies 3 (3) Gynecological malignancies 3 (3) Rheumatological disease 2 (2) Head and neck malignancies 2 (2) Others 8 (8) The type of prior therapy Chemotherapy 35 (38) Radiation 26 (28) Combination of chemo/radiation 32 (34) Topoisomerase II inhibitors Yes 38 (41) No 55 (59) Preceded or concurrent MDS 2 (2) Original disease relapse during or after ALL diagnosis 9 (10) Induction regimen +/- TKI HyperCVAD 48 (52) Linker 8 (9) BFM 8 (9) CALGB-9511 7 (7) DVP 6 (6) Others 15 (16) No treatment 1 (1) MDS: myelodysplastic syndrome; TKI: tyrosine kinase inhibitor #includes nonHodgkinâ&#x20AC;&#x2122;s lymphoma, chronic lymphocytic leukemia, and Hodgkinâ&#x20AC;&#x2122;s lymphoma.

haematologica | 2018; 103(10)

Therapy-related ALL

of subsequent ALL therapy is a concern in t-ALL patients. t-ALL patients achieved a high CR rate and had low induction mortality similar to de novo ALL, despite prior exposure to cytotoxic therapy. Although OS of t-ALL patients was inferior to de novo ALL patients, this was not independent in multivariate analysis. This is likely because t-ALL cases were enriched with poor prognostic factors that have driven the inferior outcomes of t-ALL cohort. Nonetheless, t-ALL patients who were able to receive allo-

geneic HCT fared better and had comparable OS to those with de novo ALL despite the more frequent use of reduced-intensity regimens. There was no increased risk of TRM among t-ALL patients who underwent allogeneic HCT despite prior cytotoxic exposure but this also could be related to earlier use of HCT (CR1) and more frequent use of RIC in this population. The limitations of our study include the retrospective nature of the data collection and the inclusion of patients

Table 4. Factors associated with latency among t-ALL patients.

Number of patients

Hazard Ratio

95% CI

0.75 0.91

0.36-1.59 0.49-1.67

0.75

31 23 30 15 27 15 5 22

1.65 3.06 0.78 1.17

0.82 -3.31 1.45-6.45 0.27-2.30 0.58-2.36

0.16 0.003 0.65 0.67

51 33

1.01

0.55-1.86

0.97

46 38

0.99

0.56-1.73

0.97

Prior Therapy Chemo Radiation Chemo/Radiation Cytogenetic Group NK Ph+ MLL Complex Other/Unknown Prior disease Solid Cancer Blood Cancer Topoisomerase II inhibitor No Yes

0.45 0.75 0.02

NK: normal karyotype; Ph+: Philadelphia chromosome positive; MLL: mixed lineage leukemia.

Table 5. Predictors of overall survival from time of diagnosisâ&#x20AC;&#x201C;multivariable model.

Number of patients

Number of events

Hazard Ratio

95% CI

911

462

1.02

1.01-1.03

<0.0001

823 88 911

412 50 462

1.32 1.001

0.97-1.80 1.001-1.002

273 251 54 39 294

138 115 28 25 156

0.63 0.83 1.22 1.19

0.48-0.82 0.54 -1.29 0.79-1.87 0.94 -1.50

404 507

188 274

1.31

1.09-1.59

0.005

115 796

50 412

1.46

1.08-1.97

0.02

359 552

169 293

1.29

1.04-1.60

0.02

Age at ALL diagnosis ALL Disease Type De novo Therapy-related WBC Cytogenetic Group NK Ph+ MLL Complex Other/Unknown Sex Female Male Phenotype T B HCT (time dependent) No Yes

0.08 0.003 <0.0001 0.0006 0.41 0.37 0.14

WBC: white blood cell count; NK: normal karyotype; Ph+: Philadelphia chromosome positive; MLL: mixed lineage leukemia; HCT: hematopoietic cell transplantation

haematologica | 2018; 103(10)

1667

I. Aldoss et al.

diagnosed over a 15 years period which introduces bias both with regard to changing treatments for the primary malignancies as well as ALL therapy. Examples include decreasing use of anthracyclines for breast cancer therapy as well as improved outcome of HCT over the time period of the study. It is also possible that some cases of t-ALL may be a coincidental occurrence of ALL after the patient has had a previous malignancy particularly in cases with long latency and lacking MLL rearrangement or monosomy karyotype. Moreover, the referral bias to our center may have introduced overestimation of t-ALL frequency. This is likely because t-ALL cases may have been perceived as being high risk, leading to earlier referral as well as earlier application of more intense therapy including allogeneic HCT. Our data suggest a good outcome for t-ALL when allogeneic HCT is used in CR1 and these patients should be considered candidates for HCT if they are in sustained remission from their primary malignancy. What remains unclear is the outcome of these cases, particularly ones treated with an intensive pediatric type ALL regimen in younger patients. The use of such regimens could be problematic in some of these patients due to

References 8. 1. Granfeldt Ostgard LS, Medeiros BC, SengelĂ¸v H, et al. Epidemiology and clinical significance of secondary and therapyrelated acute myeloid leukemia: A National Population-Based Cohort Study. J Clin Oncol. 2015;33(31):3641-3649 2. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016; 127(20):2391-2405. 3. Kayser S, Dohner K, Krauter J, et al. The impact of therapy-related acute myeloid leukemia (AML) on outcome in 2853 adult patients with newly diagnosed AML. Blood. 2011;117(7):2137-2145. 4. Smith SM, Le Beau MM, Huo D, et al. Clinical-cytogenetic associations in 306 patients with therapy-related myelodysplasia and myeloid leukemia: the University of Chicago series. Blood. 2003;102(1):43-52. 5. Aldoss I, Dagis A, Palmer J, et al. Therapyrelated ALL: cytogenetic features and hematopoietic cell transplantation outcome. Bone Marrow Transplant. 2015; 50(5):746-748. 6. Aldoss I, Stiller T, Song J, et al. Philadelphia chromosome as a recurrent event among therapy-related acute leukemia. Am J Hematol. 2017;92(2):E18-E19. 7. Pagano L, Pulsoni A, Tosti ME, et al. Acute lymphoblastic leukaemia occurring as second malignancy: report of the GIMEMA archive of adult acute leukaemia. Gruppo Italiano Malattie Ematologiche Maligne

1668

10.

11.

12.

13.

14. 15.

cumulative toxicity from treatment of their previous malignancy. This high rate of allogeneic HCT use for both de novo and t-ALL in our cohort may have minimized the survival difference between the two groups and underestimate the poor prognosis of t-ALL. In conclusion, we have attempted to define t-ALL more narrowly using stricter criteria than those used by previous reports and show that these cases have cytogenetic abnormalities that confirm a causative role for their prior cytotoxic therapy in many cases. Large molecular studies using next generation sequencing methodology and accurate correlation with clinical data regarding prior cytotoxic therapy will be required to further characterize this entity. Funding Research reported in this publication included work performed in the Biostatistics Core supported by the National Cancer Institute of the National Institutes of Health under award number P30CA033572. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

dell'Adulto. Br J Haematol. 1999; 106(4): 1037-1040. Ganzel C, Devlin S, Douer D, et al. Secondary acute lymphoblastic leukaemia is constitutional and probably not related to prior therapy. Br J Haematol. 2015; 170(1):50-55. Tang G, Zuo Z, Thomas DA, et al. Precursor B-acute lymphoblastic leukemia occurring in patients with a history of prior malignancies: is it therapy-related? Haematologica. 2012;97(6):919-925. Abdulwahab A, Sykes J, Kamel-Reid S, et al. Therapy-related acute lymphoblastic leukemia is more frequent than previously recognized and has a poor prognosis. Cancer. 2012;118(16):3962-3967. Kelleher N, Gallardo D, Gonzalez-Campos J, et al. Incidence, clinical and biological characteristics and outcome of secondary acute lymphoblastic leukemia after solid organ or hematologic malignancy. Leuk Lymphoma. 2016;57(1):86-91. Swaika A, Frank RD, Yang D, et al. Second primary acute lymphoblastic leukemia in adults: a SEER analysis of incidence and outcomes. Cancer Med. 2018;7(2):499-507. Giri S, Chi M, Johnson B, et al. Secondary acute lymphoblastic leukemia is an independent predictor of poor prognosis. Leuk Res. 2015;39(12):1342-1346. Kaplan G, Meier P. Non-parametric estimations from incomplete observations. J Am Stat Assoc. 1958;53:457-481. Gooley TA, Leisenring W, Crowley J, et al. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med. 1999;

18(6):695-706. 16. Breems DA, Van Putten WL, De Greef GE, et al. Monosomal karyotype in acute myeloid leukemia: a better indicator of poor prognosis than a complex karyotype. J Clin Oncol. 2008;26(29):4791-4797. 17. Leone G, Mele L, Pulsoni A, Equitani F, Pagano L. The incidence of secondary leukemias. Haematologica. 1999;84(10): 937-945. 18. Aldoss I, Pullarkat V. Therapy-related acute myeloid leukemia with favorable cytogenetics: still favorable? Leuk Res. 2012; 36(12):1547-1551. 19. Auer RL, Oates J, Reid S, Fegan CD, Milligan DW. Philadelphia-positive T-ALL in a patient with follicular lymphoma. Bone Marrow Transplant. 2000;26(10):1113-1115. 20. Barrington-Trimis JL, Cockburn M, Metayer C, Gauderman WJ, Wiemels J, McKean-Cowdin R. Rising rates of acute lymphoblastic leukemia in Hispanic children: trends in incidence from 1992 to 2011. Blood. 2015;125(19):3033-3034. 21. Pullarkat ST, Danley K, Bernstein L, Brynes RK, Cozen W. High lifetime incidence of adult acute lymphoblastic leukemia among Hispanics in California. Cancer Epidemiol Biomarkers Prev. 2009;18(2):611-615. 22. Jain N, Roberts KG, Jabbour E, et al. Ph-like acute lymphoblastic leukemia: a high-risk subtype in adults. Blood. 2017;129(5):572581. 23. 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.

haematologica | 2018; 103(10)

ARTICLE

Non-Hodgkin Lymphoma

CDCA7 is a critical mediator of lymphomagenesis that selectively regulates anchorage-independent growth

Ferrata Storti Foundation

Raúl Jiménez-P.,1 Carla Martín-Cortázar,1 Omar Kourani,1 Yuri Chiodo,1 Raul Cordoba,2,† María Purificación Domínguez-Franjo,3,‡ Juan Miguel Redondo,4,5 Teresa Iglesias6,7 and Miguel R. Campanero1,5

Department of Cancer Biology, Instituto de Investigaciones Biomédicas Alberto Sols, CSIC-UAM; 2Department of Hematology, University Hospital Infanta Sofía, San Sebastián de los Reyes; 3Department of Pathology, University Hospital Infanta Sofía, San Sebastián de los Reyes; 4Department of Vascular Biology and Inflammation, Centro Nacional de Investigaciones Cardiovasculares, (CNIC); 5CIBERCV, Spain; 6Department of Endocrine and Nervous Systems Pathophysiology, Instituto de Investigaciones Biomédicas Alberto Sols, CSIC-UAM, Madrid and 7Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED), Madrid, Spain 1

Present address: Lymphoma Unit, Department of Hematology, Fundacion Jimenez Diaz University Hospital, Health Research Institute IIS-FJD, Madrid, Spain

†

Present address: Department of Pathology, Hospital Rey Juan Carlos, Mostoles, Madrid, Spain

‡

Haematologica 2018 Volume 103(10):1669-1678

ABSTRACT

umor formation involves the acquisition of numerous capacities along the progression from a normal cell into a malignant cell, including limitless proliferation (immortalization) and anchorageindependent growth, a capacity that correlates extremely well with tumorigenesis. Great efforts have been made to uncover genes involved in tumor formation, but most genes identified participate in processes related to cell proliferation. Accordingly, therapies targeting these genes also affect the proliferation of normal cells. To identify potential targets for therapeutic intervention more specific to tumor cells, we looked for genes implicated in the acquisition of anchorage-independent growth and in vivo tumorigenesis capacity. A transcriptomic analysis identified CDCA7 as a candidate gene. Indeed, CDCA7 protein was upregulated in Burkitt’s lymphoma cell lines and human tumor biopsy specimens relative to control cell lines and tissues, respectively. CDCA7 levels were also markedly elevated in numerous T and B-lymphoid tumor cell lines. While CDCA7 was not required for anchorage-dependent growth of normal fibroblasts or non-malignant lymphocytes, it was essential but not sufficient for anchorage-independent growth of lymphoid tumor cells and for lymphomagenesis. These data suggest that therapies aimed at inhibiting CDCA7 expression or function might significantly decrease the growth of lymphoid tumors.

Introduction Most side effects of current therapies for cancer treatment are derived from their toxicity on actively proliferating normal cells, such as hematopoietic progenitors. These toxic effects likely occur because the targets for these therapies are also crucial for the proliferation of normal cells. The development of therapies more selective for tumor cells might be facilitated by the identification of genes involved in properties specific of these cells. Along the transformation of a normal cell into a highly malignant derivative, cells acquire numerous traits, including the ability to sustain chronic proliferation.1,2 Although immortalization is a fundamental trait of cancer cells, it is insufficient to promote malignant growth. NIH-3T3 fibroblasts, for instance, display replicative immortality but are not tumorigenic and display in vitro growth characteristics of haematologica | 2018; 103(10)

Correspondence: mcampanero@iib.uam.es

Received: January 18, 2018. Accepted: June 4, 2018. Pre-published: June 7, 2018. doi:10.3324/haematol.2018.188961 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1669 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1669

R. Jimenez-P. et al. non-transformed cells.3 Epstein-Barr virus (EBV) infection of normal lymphocytes generates immortalized lymphoblastoid B-cell lines (LCLs) unable to form tumors in immunodeficient mice but capable to replicate indefinitely in liquid culture.4 In contrast, cell lines derived from Burkittâ&#x20AC;&#x2122;s lymphoma (BL), a B-lymphocyte tumor strongly associated with EBV in some regions of Africa,5 not only display replicative immortality, but are also tumorigenic in immunodeficient mice.4 Another trait of tumor cells is their capacity to replicate and grow independently of their attachment to a rigid surface. Growth of normal tissue cells requires the signals transmitted by plasma membrane receptors that bind extracellular matrix components and transmembrane proteins from neighboring cells of the proper microenvironment. Most normal tissue cells are not viable when suspended in liquid or soft medium, and require adhesion to the surface of a culture vessel. Similarly, immortal, but non-tumoral cells, including NIH-3T3 fibroblasts and LCLs, cannot grow in semi-solid media such as soft agar,4,6,7 and are considered as anchorage-dependent. On the contrary, tumor cells do not need to adhere to a rigid surface for growth and are said to be anchorage-independent.6 Numerous genes that mediate tumorigenesis have been identified, but very limited information is available regarding genes that specifically mediate anchorage-independent growth. While anchorage-dependence has been extensively studied in fibroblasts and epithelial cells, it is unknown whether normal lymphoid cells require anchorage for proliferation. Soft agar not only limits cell binding to the culture vessel surface but also intercellular interactions. The incapacity of LCLs to grow in soft agar could therefore be attributed to lack of anchorage to a rigid substrate or to neighboring cells. It should be noted that normal lymphoid cells proliferate only in lymphoid organs in vivo, in close contact with the extracellular matrix and with other cells; with appropriate stimuli, these cells can also proliferate in vitro under conditions that permit their attachment to the culture vessel surface and to other cells. MYC deregulation is one of the most common aberrations in human tumors. The characteristic genetic marker of BL cells is a reciprocal translocation involving the MYC gene and one of three immunoglobulin gene loci that leads to deregulated MYC expression.8 MYC encodes a transcription factor and chromatin remodeler that regulates the expression of numerous genes involved in various cellular processes, including cell differentiation, proliferation, and apoptosis.9-14 Tumorigenesis by MYC (also known as C-MYC) can take place as a consequence of its overexpression, even in the absence of mutations in its coding region.15 Em-Myc transgenic mice, where Myc overexpression is targeted to B lymphocytes, give rise to lymphomas, but only after a mean latency period of about 6 months, these lymphomas being monoclonal.16 In addition, MYC overexpression in normal cells either arrests them in the G2 phase of the cell cycle17 or induces apoptosis.18 Together, these results suggest that MYC alone cannot elicit tumoral transformation of normal cells and that additional factors might cooperate with MYC in tumorigenesis. MYC regulates about 15 percent of the genes in the human genome, and it is expected that some of them participate in tumor formation. However, it remains unknown which of these genes are critical for MYCinduced transformation. CDCA7, also named JPO1, is a 1670

MYC-induced gene19 whose mRNA is deregulated in several tumor types relative to the corresponding non-proliferative control tissues.20 Little is known about the molecular function of CDCA7. Phosphorylation by AKT regulates CDCA7 subcellular localization and its association with MYC.21 There are two alternatively spliced CDCA7 isoforms which contain a zinc finger domain at the C-terminus22,23 and associate with the Helicase, Lymphoid-specific (HELLS) SNF2 family member.24 In fact, CDCA7 is required for nucleosome remodeling by HELLS and for DNA methylation maintenance.23,24 However, no functional differences between CDCA7 isoforms have been reported. CDCA7 overexpression in tumors could potentially be just a consequence of the presence of more cycling cells in the tumor tissues. Indeed, its expression is induced in the S-phase of the cell cycle25 and it is a transcriptional target not only of MYC,19 but also of E2F1-E2F4,22 factors that are active exclusively in proliferating cells. Alternatively, CDCA7 could potentially play a causative role in tumorigenesis. In this regard, CDCA7 has been assumed to participate in neoplastic transformation of B cells (http://www.uniprot.org/uniprot/Q9BWT1; https://www. ncbi.nlm.nih.gov/gene/83879), in spite of insufficient evidence supporting this assumption. Indeed, previous observations showed that forced expression of CDCA7 barely increased colony formation in a B-cell line already capable of growing in soft agar.19 In addition, transgenic mice overexpressing CDCA7 in the B-cell compartment generated lymphoid malignancies as frequently as control mice (4 out of 45 transgenic mice vs. 3 out of 28 control mice).20 Moreover, CDCA7 forced expression inhibited the induction of anchorage-independent growth induced by MYC in immortal fibroblasts.21 Therefore, a tumorigenic role for CDCA7 has not been experimentally demonstrated and it is, at minimum, controversial. The design of anti-tumor therapies that do not affect normal cells has proved to be a very difficult task. To identify potential therapeutic targets specific to tumor cells, we used a transcriptomic approach looking for genes specifically involved in anchorage-independent growth, a capacity that correlates extremely well with tumorigenesis. Here we show that CDCA7 was one of the most significantly up-regulated genes, that its encoded protein is overexpressed in lymphoid tumors, and that its silencing greatly impairs lymphomagenesis without inhibiting the proliferation of normal cells.

Methods Detailed Methods can be found in the Online Supplementary Appendix.

Patients Cases consisted of existing de-identified anonymous biopsy specimens obtained from Hospital Infanta Sofia and from the Spanish Tumor Bank Network in Centro Nacional de Investigaciones OncolĂłgicas (Madrid, Spain). The study was approved by the CSIC Ethics Committee and by the Ethics Committee of Hospital Universitario La Paz (ref. HULP: PI-1658).

Microarray Experiment Data Analysis RNA from BL and LCL cell lines was extracted and labelled as described.26 The RNA from X50-7, IB-4, and Dana LCL cell lines haematologica | 2018; 103(10)

A selective role for CDCA7 in lymphomagenesis

Figure 1. The expression profiles between BL and LCL are different. A. Experimental design. The expression profile of 5 BL cell lines (BL2, Ramos, DG-75, Akata, and Mutu-I) was compared with that of a pool of LCL lines (X50-7, IB4 and Dana). Each BL cell line was considered as a biological replica for data analysis. B. Venn diagram showing overlapping of genes significantly regulated relative to the pool of LCL. The total number of overlapping up- and down-regulated genes is shown. C. Genes up- (59 genes) or down-regulated (425 genes) >3-fold were selected and the most statistically significant genes (TOP 100) were further selected. The only genes among the TOP 100 that were up-regulated in BL cell lines are shown indicating their relative expression (Fold), adjusted P value, Ensemble gene identification number, and the biological process in which these genes might be involved.

was pooled and used as a universal control for the whole experiment. The RNA from DG-75, Ramos, BL2, Mutu-I, and Akata BL cell lines was mixed with an equal amount of universal control pool and processed for microarray analysis using 22K-oligo microarrays (CapitalBio Corporation). Genes with similar expression within BL cell lines were ranked according to their P-value for differential expression (adj. P value <0.05). Gene expression data are available at http://www.ncbi.nlm.nih.gov/projects/geo/ (accession number GSE41865).

Antibodies

Northern blotting and quantitative PCR analysis

Transformation assays

Northern blotting was performed as described27 using 32P-labeled 5â&#x20AC;&#x2122; CDCA7-2 and ACTG DNA fragments. Real-time quantitative RT-PCR (q-PCR) was performed using TaqMan Gene Expression Assays (ThermoFischer Scientific) specific for both isoforms of human CDCA7 (Hs00230589_m1), CDCA7-1 (Hs00912235_m1), CDCA7-2 (Hs00914361_m1), or TBP (Hs00427621_m1).

Transformation assays were performed as described.30 All mice were inoculated with control cells in one flank and CDCA7silenced cells in the opposite flank.

Cell Transfection, lentivirus production and cell transduction HEK-293T cells were transfected using the calcium phosphate method.28 Lentiviral particles were produced as previously described.29 MISSION pLKO.1-puro-based vectors encoded either a non-targeting shRNA (SHC002) or the following CDCA7 targeting shRNAs: sh-25 (TRCN0000140725), sh-40 (TRCN0000139240), sh-56 (TRCN0000139556) sh-83 (TRCN0000145183). DG-75, Ramos, BL2, Toledo, Molt-4 and S1F cell lines were transduced as described30 and selected in 1mg/ml puromycin >96h. haematologica | 2018; 103(10)

The anti-CDCA7 S99 polyclonal rabbit serum was raised against the CRGRHPLPGSDSQSRRPR KLH-conjugated peptide as described.31

Cell proliferation and cell cycle analysis Cell proliferation was assessed by EdU incorporation as described.32 Cell cycle analysis was performed as previously described.30

Results Transcriptomic analysis of immortalized and tumoral cells To identify genes participating in stages of malignant transformation beyond or independent of replicative immortalization, we compared the transcriptome of immortalized B cells (LCLs) and BL cell lines. Both cell types display similar growth, replication and viability rates when cultured in liquid media on plastic.30 LCLs cultured in liquid media grow mostly in clumps (Online Supplementary Figure S1). Single cells can also be found in these cultures, and most of them are anchored to the culture vessel surface, as indicated by their marked spreading 1671

R. Jimenez-P. et al. (Online Supplementary Figure S1). As previously described,4 lymphoma cells form colonies in soft agar, whereas LCLs do not grow in this medium (Online Supplementary Figure S2). Together, these results suggest that anchorage to the vessel surface or to other cells might provide LCLs the signals required for cell proliferation and that soft agar might inhibit their growth by blocking anchorage to the surface and to neighboring cells.

Figure 2. CDCA7 is overexpressed in BL cell lines. A. qPCR analysis of CDCA7 mRNA expression in the indicated BL and LCL cell lines. Expression is shown relative to that found in the LCL cell line IB4 and normalized by ACTB as mean+s.e.m. (n=3). B. Northern blot analysis of CDCA7 isoforms and ACTG1 mRNA expression in the indicated BL and LCL cell lines. Staining of the membrane with methylene blue is also shown. C. Immunoblot analysis of CDCA7 expression in the indicated BL and LCL cell lines. Tubulin is shown as loading control.

1672

Figure 3. CDCA7 expression is deregulated in biopsies from BL patients. A. qPCR analysis of CDCA7 mRNA expression in a pool of 3 reactive tonsils and in 20 BL patient samples. Data were normalized to the expression of ACTB and are shown relative to the pool of reactive tonsils as mean+s.e.m (n=3). B. Sections from the germinal center of a reactive tonsil and from 9 BL cases were analyzed by immunohistochemistry staining with anti-CDCA7 S99 antibody. A section of a BL case was also stained with the preimmune serum as negative control. Representative images are shown. C. Quantification of preimmune-positive area in immunohistochemistry-stained sections of 4 BL cases (PI) and CDCA7-positive area in immunohistochemistry-stained sections of 4 reactive tonsils (RT) and 12 BL cases. ***P<0.001; one-way ANOVA with Bonferroni post-test.

haematologica | 2018; 103(10)

A selective role for CDCA7 in lymphomagenesis

The comparison of the transcriptome of 5 BL cell lines (DG-75, Ramos, BL2, Mutu-I and Akata) with a pool of 3 LCL cell lines (X50-7, IB-4 and Dana) using whole-genome microarrays showed >1,600 genes similarly expressed in BL cells that were up- (630 genes) or down-regulated (1,033 genes) significantly (P<0.05) in BL cell lines (Figures 1A-1B) (GEO accession number GSE41865). Among the genes that were regulated >3-fold (484 genes), we selected the 100 genes showing the highest statistical significance. Only 7 of these genes were up-regulated in BL cells (Figure 1C). The information known about the proteins encoded by these genes (http://www.uniprot.org/) suggested that two of them (ID3 and CDCA7) might function as transcriptional regulators (Figure 1C) and, as such, they may have the potential to regulate gene expression programs related with malignant transformation. Indeed, ID3 is a critical mediator of BL formation.33,34 Since the role of CDCA7 in tumorigenesis was controversial, we investigated whether the elevated expression of CDCA7 might specifically play a critical role in the malignant transformation of lymphoid cells.

enriched in B cells. Quantification of the CDCA7-positive area in immunohistochemistry-stained sections confirmed a marked expression increase in BL samples (Figure 3C). Together, our results indicate that CDCA7 expression is

CDCA7 expression is deregulated in Burkittâ&#x20AC;&#x2122;s Lymphoma There are two alternatively spliced CDCA7 isoforms (CDCA7-1 and CDCA7-2) that differ only in the presence of an additional internal exon in CDCA7-1 (Online Supplementary Figure S3A). CDCA7 mRNA levels, measured by qPCR using primers that amplify both isoforms, were markedly higher in BL than in LCL cells (Figure 2A). qPCR analysis with gene expression assays specific for each isoform and Northern blotting showed that CDCA72 was more abundant in both cell types and that levels of both isoforms were markedly higher in BL cells (Figure 2B and Online Supplementary Figure S3B). CDCA7 protein levels were also compared between lymphoma and immortalized cells. Since we encountered reproducibility difficulties with commercially available antibodies (Abs), we generated a rabbit polyclonal Ab (S99) using a CDCA7 peptide present in both isoforms (Online Supplementary Figure S4A). To determine the specificity of this Ab, we confirmed that it recognized CDCA71 and CDCA7-2 ectopically expressed in HEK-293T cells (Online Supplementary Figure S4B). Immunoblot analysis of BL and LCL cells using anti-CDCA7 S99 showed that CDCA7-2 levels were markedly higher in BL than in LCLs, whereas those of CDCA7-1 were barely detectable in both cell types (Figure 2C). CDCA7-2 was not detected in BL cell lysates if the Ab was previously neutralized with the peptide used during immunization (Online Supplementary Figure 3C), further demonstrating the specificity of the S99 anti-CDCA7 Ab. To ascertain that the high expression of CDCA7 in BL cell lines was not just a consequence of their in vitro growth, we used qPCR to compare CDCA7 expression in biopsy specimens from BL patients with that in control tissues derived from reactive tonsils, which are enriched in germinal centre B cells. All BL samples expressed 5 to 20 times higher CDCA7 mRNA levels than a pool of 3 reactive tonsils (Figure 3A). Although reactive tonsils do not only contain B cells, the absence of additional cells is not expected to account for a >5-fold increase in CDCA7 mRNA expression in the tumor. In fact, CDCA7 protein immunostaining was also higher in BL samples than in the germinal centre of reactive tonsils (Figure 3B), a region haematologica | 2018; 103(10)

Figure 4. CDCA7 mediates anchorage-independent growth of BL cells. DG-75, BL2 and Ramos cells were transduced with lentivirus encoding sh-Ctl, sh-25 or sh-83 and selected in the presence of puromycin >5 days. A. Representative CDCA7 immunoblot analysis of DG-75 (n=4), Ramos (n=5) and BL2 (n=4) cells expressing the indicated shRNAs. Tubulin is shown as loading control. B. Representative images of wells containing these cells seeded in soft agar and C. colony formation quantification relative to cells expressing sh-Ctl shown as meanÂąs.e.m (n=4 independent experiments, DG-75 and BL2; n=5 independent experiments, Ramos). **P<0.01, ***P<0.001; one-way ANOVA with Bonferroni post-test.

1673

R. Jimenez-P. et al.

elevated in primary tumor samples derived from BL patients.

CDCA7 mediates anchorage-independent growth of lymphoma cells The overexpression of CDCA7 in BL cells could be causally involved in oncogenesis or be only the result of a random event unrelated to tumor formation. To ascertain whether CDCA7 played a causative role in lymphomagenesis, we knocked-down its expression in BL tumor cells by using lentivirus encoding CDCA7-specific shRNA. By screening candidate shRNAs specific for CDCA7 in lymphoma cells, we identified sh-25 and sh-83 as having high knockdown capacity relative to a non-targeting shRNA (shCtl) or to non-transduced cells (Online Supplementary Figure S5). Lentiviral transduction of DG-75 BL cells with sh-25 or sh-83 markedly inhibited CDCA7 expression relative to cells transduced with sh-Ctl (Figure 4A, left panel). CDCA7 silencing in these cells sharply decreased their colony formation capacity in soft agar (Figure 4B, left column; and Figure 4C, top panel). To determine whether CDCA7 plays a similar role in other BL cells, we transduced BL2 and Ramos BL cells with lentivirus encoding sh-25 or sh-83. These shRNAs efficiently silenced CDCA7 expression in these cells (Figure 4A, middle and right panels) and sharply decreased their growth in soft agar (Figure 4B, middle and right panels; and Figure 4C, middle and bottom panels).

The inability of CDCA7-silenced cells to grow in soft agar may potentially be caused by inhibition of their capacity to grow under liquid culture conditions (anchorage-dependent growth). However, CDCA7 silencing in BL cells did not substantially affect their cell cycle distribution [Figure 5 (top panels) and Online Supplementary Figure S6] or inhibit their proliferation in liquid culture [Figure 5 (bottom panels) and Online Supplementary Figures S7A-S7B]. To investigate the potential role of CDCA7 in the regulation of cell proliferation by cell-cell contacts, we seeded single cells in 96-well plates and assessed their colony formation capacity. It should be noted that cells were able to attach to the surface of the wells under these conditions. As shown in Online Supplementary Figure S7C, CDCA7 knockdown did not decrease the number of colonies formed by DG-75 cells. Since BL is a very rapid growing tumor, it can be subjected to poor nutrient supply. To mimic nutrient shortage, we serum-starved DG75 cells and found that CDCA7 silencing did not affect the viability of cells cultured in the presence of low serum concentrations (Online Supplementary Figure S8A). CDCA7 knockdown also had no effect on the viability of cells treated with Cisplatin or Bleomycin (Online Supplementary Figure S8B). Together, our data strongly support the notion that while CDCA7 is dispensable for anchorage-dependent growth, it is required for anchorage-independent growth

Figure 5. CDCA7 does not mediate anchorage-dependent growth. DG-75 and BL2 cells were transduced with lentivirus encoding sh-Ctl, sh-25 or sh-83 and selected in the presence of puromycin >5 days. Cell cycle and Edu incorporation analysis of (A) DG-75 and (B) BL2 cells transduced with lentivirus encoding the indicated shRNAs. (Top panels) Columns show the percentage of each of these cells in the indicated cell cycle phases as meanÂąs.e.m (n=3). (Bottom panels) Columns show normalized percentage of Edu incorporation in each of these cells as meanÂąs.e.m (n=3).

1674

haematologica | 2018; 103(10)

A selective role for CDCA7 in lymphomagenesis

of lymphoma cells. However, high CDCA7 levels in immortal, but non-malignant B cells, are not sufficient to promote anchorage-independent growth because forced expression of CDCA7-1 or CDCA7-2 in non-transformed JY cells or in JY cells transduced with lentivirus encoding constitutively active H-RAS (R12,T59) did not induce their growth in soft agar (Online Supplementary Figure S9). These results support the notion that malignant transformation requires the acquisition of multiple genetic lesions beyond those necessary to proliferate indefinitely.

CDCA7 mediates BL lymphoma growth in vivo We have shown that DG-75, BL2, and Ramos BL cells are tumorigenic in immunodeficient mice, while EBV-immortalized B-cells are not.30,32 To investigate the contribution of CDCA7 to the in vivo tumorigenicity of BL cells, we transduced DG-75, BL2, and Ramos BL cells with sh-Ctl, sh-25 or sh-83 and subcutaneously inoculated them in immunodeficient NOD-SCID mice. As expected, control-transduced DG-75, Ramos and BL cells readily elicited growth of large tumors in these mice (Figure 6). Importantly,

Figure 6. CDCA7 mediates BL tumor formation. DG-75, BL2 and Ramos cells were transduced with lentivirus encoding sh-Ctl, sh-25 or sh-83 and selected in the presence of puromycin >5 days. A. DG75, B. Ramos, and C. BL2 cells transduced with lentiviruses encoding the indicated shRNAs were inoculated subcutaneously in immunodeficient NOD-SCID mice. Tumors were extracted after 3 weeks. Circles, squares and triangles indicate the weight of individual tumors and horizontal bars indicate the mean (long bar) and s.e.m (short bar). *P<0.05, **P<0.01 and ***P<0.001 vs. sh-Ctl; one-way ANOVA with Bonferroni post-test.

haematologica | 2018; 103(10)

Figure 7. CDCA7 is overexpressed in various types of lymphoid malignancies. Representative immunoblot analysis of CDCA7 expression in the indicated BL, LCL, Diffuse Large B-Cell lymphoma (DLBCLs), Follicular Lymphoma (FL), Mantle Cell Lymphoma (MCL), uncharacterized non-Hodgkin Lymphoma (N-HL), Acute Lymphocytic B-cell leukemia (B-ALL) and T-cell leukemia (T-ALL), and Myeloid leukemia (ML) cell lines. Tubulin is shown as loading control.

1675

R. Jimenez-P. et al.

CDCA7 silencing in these cells blocked or greatly impaired their tumor formation capacity (Figure 6) without substantially affecting the expression of the Ki67 proliferation marker (Online Supplementary Figure S10). These results strongly suggest that CDCA7 is critical for BL formation and that therapies aimed at inhibiting its expression or its activity might be of interest for BL patients. Since current therapy for BL patients affects not only tumor cells but also actively proliferating normal cells from these patients, we investigated whether CDCA7 silencing affected the proliferation of primary human diploid fibroblasts IMR-90. Transduction of these cells with lentivirus encoding sh-25 or sh-83 decreased CDCA7 mRNA and protein levels relative to sh-Ctl-transduced cells (Online Supplementary Figures S11A-S11B) without affecting their proliferation rate (Online Supplementary Figure S11C).

CDCA7 was initially identified as a MYC-responsive gene19 whose expression can be transcriptionally induced also by E2F transcription factors22 and Notch.35 Although CDCA7 mRNA levels were shown to be upregulated in tumor samples relative to normal tissues, protein levels were not determined in these tumors.20 This is an important issue because the increase in mRNA expression does not necessarily imply an increase in protein content. Our results suggest that CDCA7 protein upregulation might be

CDCA7 expression is deregulated in lymphoid neoplasias and mediates their growth in vivo Next, we assessed whether CDCA7 is deregulated not only in BL, but also in other types of cancer by comparing its expression in protein extracts of several lymphoid tumors. Cell lines of Diffuse Large B-cell Lymphoma (DLBCL), Follicular Lymphoma, Mantle Cell Lymphoma, uncharacterized non-Hodgkin Lymphoma, B-cell leukemia, and T-cell leukemia expressed markedly higher levels of CDCA7-2 than LCLs (Figure 7). In contrast, CDCA7-2 expression in myeloid leukemia cell lines was similar to that of LCLs (Figure 7, bottom panels). To investigate whether CDCA7 is an essential mediator of lymphomagenesis not only in BL, but also in other lymphoid neoplasias, we silenced its expression in T-cell leukemia and DLBCL cells and analyzed their tumorigenicity in immunodeficient mice. Molt-4 T-cell leukemia and Toledo DLBCL cells expressed detectable levels of both CDCA7-1 and CDCA7-2 (Figure 8A) and their lentiviral transduction with sh-25 markedly inhibited the expression of both isoforms relative to cells transduced with sh-Ctl (Figure 8A). It should be noted that CDCA7-1 was not always detected in protein extracts from these cells. The capacity of these cells to form tumors was subsequently determined through their subcutaneous inoculation into immunodeficient mice. Molt-4 and Toledo cells transduced with sh-25 elicited formation of tumors significantly smaller than those produced by cells transduced with sh-Ctl (Figures 8B-8C). Similar to BL cells, CDCA7 silencing in Toledo cells did not substantially affect cell cycle distribution (Online Supplementary Figure S12). Together, our results strongly suggest that CDCA7 is a key mediator of lymphoid malignant transformation.

Discussion In this study, through the identification of a gene involved in anchorage-independent growth, we found a new target for therapeutic intervention in lymphoid tumors less prone to cause side effects. By comparing gene expression profiles of immortal, but non-malignant cells, with those of tumor cells from the same lineage, we have uncovered a gene, CDCA7, whose elevated protein levels in lymphoid tumor cells mediate their anchorage-independent growth and their tumorigenesis without participating in their growth under normal tissue culture conditions (liquid culture on a rigid surface). 1676

Figure 8. CDCA7 mediates tumorigenesis in T-cell leukemia and DLBCL cells. Toledo and Molt-4 cells were transduced with lentivirus encoding sh-Ctl or sh-25 and selected in the presence of puromycin >5 days. A. Representative CDCA7 immunoblot analysis of Toledo (n=3), and Molt-4 (n=3) cells expressing the indicated shRNAs. Tubulin is shown as loading control. B. Toledo and C. Molt-4 cells transduced with lentiviruses encoding the indicated shRNAs were inoculated subcutaneously in immunodeficient NOD-SCID mice. Tumors were extracted after 3 weeks. Circles and squares indicate the weight of individual tumors and horizontal bars indicate the mean (long bar) and s.e.m (short bar). **P<0.01 vs. sh-Ctl; paired samples t-test.

haematologica | 2018; 103(10)

A selective role for CDCA7 in lymphomagenesis

a common event in lymphoid tumors. CDCA7 protein levels were not higher in myeloid tumors cell lines than in LCLs, but its expression in these tumors should be compared with that in control cells of the same lineage before drawing definitive conclusions. In addition, we cannot rule out that CDCA7 expression may be enhanced in other cell lines or in primary tumors. MYC and activator E2F transcription factors are expressed only in proliferating cells, and there is a correlation between expression of their target genes and the growth rate of cells in culture.36 The increase of CDCA7 mRNA levels in tumor samples relative to control tissues could therefore have been merely the consequence of the presence of more proliferating cells in the tumor tissue. We have shown however that CDCA7 mRNA and protein levels are sharply elevated in lymphoid tumor cell lines relative to immortal cell lines (LCLs) and we had previously demonstrated that LCL and BL cell lines display indistinguishable proliferation rates and cell cycle profiles.30 These data suggest that the elevated expression of CDCA7 in lymphoid tumors is not simply the consequence of the presence of more proliferating cells relative to control tissues. Instead, our data support the notion that CDCA7 upregulation in these tumors is critical for anchorage-independent growth and tumorigenesis. Indeed, we show that its silencing in lymphoid tumor cells markedly inhibits their anchorage-independent growth and their tumor formation capacity in immunodeficient mice. These data therefore point to CDCA7 as a potential target for therapeutic intervention in lymphoid tumors. Previous reports provided contradictive evidence for the role of CDCA7 in malignant transformation. While one report showed that CDCA7 overexpression in an immortal fibroblast cell line impaired MYC-induced colony formation,21 other reports proposed that forced CDCA7 expression promoted transformation in vitro and in vivo.19,20 However, the experimental evidence supporting a transforming activity for CDCA7 was very weak. On the one hand, CDCA7 overexpression in a B-cell line or in a fibroblast cell line increased only modestly their already existing capacity to form colonies in soft agar.19 Since both cell lines had a basal capacity to grow in this semisolid medium, it seems that these cells, instead of being merely immortal, were already transformed or at least partially transformed. On the other hand, only 4 out of 45 (8.88%) transgenic mice overexpressing CDCA7 in the B-cell compartment presented lymphoid malignancies at 1 year, whereas 3 out of 28 (10.71%) control littermates presented lymphomas.20 Together, these data did not support the notion that CDCA7 overexpression induces malignant transformation. In fact, we have shown herein that CDCA7 overexpression in non-transformed B cells fails to induce their growth in soft agar even in combination with constitutively active HRAS. Our data therefore suggest that CDCA7 is necessary but not sufficient for lymphoid malignant transformation.

References 1. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell. 2000;100(1):57-70. 2. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011; 144(5):646-674. 3. Jainchill JL, Aaronson SA, Todaro GJ.

haematologica | 2018; 103(10)

Numerous genes that mediate tumor transformation and anchorage-independent growth are also essential for cell proliferation over rigid surfaces. This is the case, for instance, of E2F1, whose knock-down in BL cells inhibits tumor growth in vivo, colony formation in soft agar, and cell proliferation in liquid culture.30 The absence of anchorage of non-tumor cells to a rigid surface does not permit their proliferation likely because they miss either the growth- or the survival-promoting signals provided by anchorage,37 whereas tumor cells might autonomously generate these signals.37 While normal epithelial cells and fibroblasts require attachment to a rigid surface for survival, lymphocytes are viable in the blood and lymph. However, normal lymphocytes do not proliferate in the blood, but only in lymphoid organs where they might potentially interact with the extracellular matrix and with other cells within these organs. Thus, it seems likely that deprivation of growth-promoting signals provided by anchorage accounts for the inability of non-tumor lymphoid cells to proliferate in soft agar. Aberrantly elevated CDCA7 levels in lymphoid tumor cells might correspondingly contribute to generate the growth-promoting signals provided by anchorage. It will be therefore of great interest to investigate the molecular mechanisms involved in CDCA7-promoted anchorage-independent growth. Together, our findings identify CDCA7 as an important regulator of lymphoid tumor transformation. The inhibition of tumor formation capacities of T-cell leukemia, DLBCL, and BL cells upon CDCA7 silencing point to CDCA7 as a candidate for therapeutic intervention in lymphoid tumors. Since CDCA7 knock-down in primary diploid fibroblasts does not inhibit their proliferation, it seems plausible that therapies aimed at inhibiting CDCA7 expression or function might significantly inhibit the growth of lymphoid tumors while likely not affecting the proliferation of normal cells. Acknowledgments The authors would like to thank D. Trono for plasmids; the CNIO Tumor Bank for providing some of the cases included in this study; and the Bioinformatics facility of Centro Nacional de Investigaciones Cardiovasculares (Madrid, Spain) for analysis of microarray data. Funding This work was supported by the Spanish Ministerio de EconomĂa, Industria y Competitividad (MEIC) grants to MRC. (SAF2013-45258P and SAF2017-88881-R) and TIV. (SAF2014-52737P); and by Instituto de Salud Carlos III (CIBERNED) to TIV. The cost of this publication has been paid in part with FEDER funds. OK holds an FPI fellowship from MEIC (BES-2014-069236). The CNIC is supported by the Ministerio de Ciencia, InvestigaciĂłn y Universidades and the Pro CNIC Foundation, and is a Severo Ochoa Center of Excellence (SEV-2015-0505).

Murine sarcoma and leukemia viruses: assay using clonal lines of contact-inhibited mouse cells. J Virol. 1969;4(5):549-553. 4. Nilsson K, Giovanella BC, Stehlin JS, Klein G. Tumorigenicity of human hematopoietic cell lines in athymic nude mice. Int J Cancer. 1977;19(3):337-344. 5. Epstein MA, Achong, B. G. The relationship of the virus to Burkittâ&#x20AC;&#x2122;s lymphoma. In:

Epstein M.A. ABGe, ed. The Epstein-Barr Virus. Berlin, Heidelberg: Springer, 1979. 6. Discher DE, Janmey P, Wang YL. Tissue cells feel and respond to the stiffness of their substrate. Science. 2005; 310(5751): 1139-1143. 7. Otsuka H, Moskowitz M. Arrest of 3T3 cells in G1 phase in suspension culture. J Cell Physiol. 1975;87(2):213-219.

1677

R. Jimenez-P. et al.

8. Hecht JL, Aster JC. Molecular biology of Burkitt's lymphoma. J Clin Oncol. 2000; 18(21):3707-3721. 9. Boxer LM, Dang CV. Translocations involving c-myc and c-myc function. Oncogene. 2001;20(40):5595-5610. 10. Cole MD, McMahon SB. The Myc oncoprotein: a critical evaluation of transactivation and target gene regulation. Oncogene. 1999;18(19):2916-2924. 11. Dang CV. MYC on the path to cancer. Cell. 2012;149(1):22-35. 12. Grandori C, Cowley SM, James LP, Eisenman RN. The Myc/Max/Mad network and the transcriptional control of cell behavior. Annu Rev Cell Dev Biol. 2000; 16:653-699. 13. Nesbit CE, Tersak JM, Prochownik EV. MYC oncogenes and human neoplastic disease. Oncogene. 1999;18(19):3004-3016. 14. Pelengaris S, Khan M, Evan G. c-MYC: more than just a matter of life and death. Nat Rev Cancer. 2002;2(10):764-776. 15. Chung HJ, Levens D. c-myc expression: keep the noise down! Mol Cells. 2005;20(2):157-166. 16. Adams JM, Harris AW, Pinkert CA, et al. The c-myc oncogene driven by immunoglobulin enhancers induces lymphoid malignancy in transgenic mice. Nature. 1985;318(6046):533-538. 17. 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. 18. Packham G, Cleveland JL. c-Myc and apoptosis. Biochim Biophys Acta. 1995; 1242(1):11-28. 19. Prescott JE, Osthus RC, Lee LA, et al. A novel c-Myc-responsive gene, JPO1, participates in neoplastic transformation. J Biol Chem. 2001;276(51):48276-48284.

1678

20. Osthus RC, Karim B, Prescott JE, et al. The Myc target gene JPO1/CDCA7 is frequently overexpressed in human tumors and has limited transforming activity in vivo. Cancer Res. 2005;65(13):5620-5627. 21. Gill RM, Gabor TV, Couzens AL, Scheid MP. The MYC-associated protein CDCA7 is phosphorylated by AKT to regulate MYC-dependent apoptosis and transformation. Mol Cell Biol. 2013;33(3):498-513. 22. Goto Y, Hayashi R, Muramatsu T, et al. JPO1/CDCA7, a novel transcription factor E2F1-induced protein, possesses intrinsic transcriptional regulator activity. Biochim Biophys Acta. 2006;1759(1-2):60-68. 23. Thijssen PE, Ito Y, Grillo G, et al. Mutations in CDCA7 and HELLS cause immunodeficiency-centromeric instability-facial anomalies syndrome. Nat Commun. 2015; 6:7870. 24. Jenness C, Giunta S, Muller MM, Kimura H, Muir TW, Funabiki H. HELLS and CDCA7 comprise a bipartite nucleosome remodeling complex defective in ICF syndrome. Proc Natl Acad Sci USA. 2018; 115(5):E876-E885. 25. Whitfield ML, Sherlock G, Saldanha AJ, et al. Identification of genes periodically expressed in the human cell cycle and their expression in tumors. Mol Biol Cell. 2002;13(6):1977-2000. 26. Esteban V, Mendez-Barbero N, JimenezBorreguero LJ, et al. Regulator of calcineurin 1 mediates pathological vascular wall remodeling. J Exp Med. 2011; 208(10):2125-2139. 27. Campanero MR, Herrero A, Calvo V. The histone deacetylase inhibitor trichostatin A induces GADD45 gamma expression via Oct and NF-Y binding sites. Oncogene. 2008;27(9):1263-1272. 28. Campanero MR, Armstrong M, Flemington E. Distinct cellular factors regulate the c-

29.

30.

31.

32.

33.

34. 35.

36.

37.

myb promoter through its E2F element. Mol Cell Biol. 1999;19(12):8442-8450. Alvaro-Blanco J, Urso K, Chiodo Y, et al. MAZ induces MYB expression during the exit from quiescence via the E2F site in the MYB promoter. Nucleic Acids Res. 2017;45(17):9960-9975. Molina-Privado I, Rodriguez-Martinez M, Rebollo P, et al. E2F1 expression is deregulated and plays an oncogenic role in sporadic Burkitt's lymphoma. Cancer Res. 2009;69(9):4052-4058. Alvaro-Blanco J, Martinez-Gac L, Calonge E, et al. A novel factor distinct from E2F mediates C-MYC promoter activation through its E2F element during exit from quiescence. Carcinogenesis. 2009; 30(3): 440-448. Molina-Privado I, Jimenez PR, MontesMoreno S, et al. E2F4 plays a key role in Burkitt lymphoma tumorigenesis. Leukemia. 2012;28(10):2277-2285. Richter J, Schlesner M, Hoffmann S, et al. Recurrent mutation of the ID3 gene in Burkitt lymphoma identified by integrated genome, exome and transcriptome sequencing. Nat Genet. 2012;44(12):13161320. Love C, Sun Z, Jima D, et al. The genetic landscape of mutations in Burkitt lymphoma. Nat Genet. 2012;44(12):1321-1325. Guiu J, Bergen DJ, De Pater E, et al. Identification of Cdca7 as a novel Notch transcriptional target involved in hematopoietic stem cell emergence. J Exp Med. 2014;211(12):2411-2423. Ross DT, Scherf U, Eisen MB, et al. Systematic variation in gene expression patterns in human cancer cell lines. Nat Genet. 2000;24(3):227-235. Wolfenson H, Bershadsky A, Henis YI, Geiger B. Actomyosin-generated tension controls the molecular kinetics of focal adhesions. J Cell Sci. 2011;124(Pt 9):1425-1432.

haematologica | 2018; 103(10)

ARTICLE

Non-Hodgkin Lymphoma

Pre-diagnosis plasma immune markers and risk of non-Hodgkin lymphoma in two prospective cohort studies

Ferrata Storti Foundation

Mara M. Epstein,1 Bernard Rosner,2,3 Elizabeth C. Breen,4,5 Julie L. Batista,3 Edward L. Giovannucci,3,6,7 Larry Magpantay,4,8 Jon C. Aster,9 Scott J. Rodig,9 Kimberly A. Bertrand,10 Francine Laden,3,7 Otoniel Martínez-Maza4,8,11,12,13 and Brenda M. Birmann3

Department of Medicine and the Meyers Primary Care Institute, University of Massachusetts Medical School, Worcester, MA; 2Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA; 3Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA; 4UCLA AIDS Institute, Los Angeles, CA; 5Department of Psychiatry & Biobehavioral Sciences, David Geffen School of Medicine at UCLA, Los Angeles, CA; 6 Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA; 7 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA; 8 Department of Obstetrics & Gynecology, David Geffen School of Medicine at UCLA, Los Angeles, CA; 9Department of Pathology, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA; 10Slone Epidemiology Center at Boston University, Boston, MA; 11Department of Epidemiology, UCLA Fielding School of Public Health, Los Angeles, CA; 12Department of Microbiology, Immunology & Molecular Genetics, David Geffen School of Medicine at UCLA, Los Angeles, CA and 13Jonsson Comprehensive Cancer Center, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA 1

Haematologica 2018 Volume 103(10):1679-1687

ABSTRACT

nflammation and B-cell hyperactivation have been associated with non-Hodgkin lymphoma development. This prospective analysis aimed to further elucidate pre-diagnosis plasma immune marker profiles associated with non-Hodgkin lymphoma risk. We identified 598 incident lymphoma cases and 601 matched controls in Nurses’ Health Study and Health Professionals Follow-up Study participants with archived pre-diagnosis plasma samples and measured 13 immune marker levels with multiplexed immunoassays. Using multivariable logistic regression we calculated Odds Ratios (OR) and 95% Confidence Intervals (CI) per standard deviation unit increase in biomarker concentration for risk of non-Hodgkin lymphoma and major histological subtype, stratifying additional models by years (<5, 5 to <10, ≥10) after blood draw. Soluble interleukin-2 receptor-α, CXC chemokine ligand 13, soluble CD30, and soluble tumor necrosis factor receptor-2 were individually positively associated, and B-cell activating factor of the tumor necrosis factor family inversely associated, with all non-Hodgkin lymphoma and one or more subtypes. The biomarker combinations associated independently with lymphoma varied somewhat by subtype and years after blood draw. Of note, the unexpected inverse association between B-cell activating factor and chronic lymphocytic leukemia/small lymphocytic lymphoma risk (OR: 95%CI: 0.51, 0.43-0.62) persisted more than ten years after blood draw (OR: 0.70; 95%CI: 0.52-0.93). In conclusion, immune activation precedes non-Hodgkin lymphoma diagnosis by several years. Decreased B-cell activating factor levels may denote nascent chronic lymphocytic leukemia many years pre-diagnosis. Introduction Severe immune compromise is a strong risk factor for non-Hodgkin lymphoma (NHL), and B-cell activation and inflammation have been associated with an increased risk of AIDS-related NHL. Elevated pre-diagnosis plasma levels of markers of B-cell stimulation including CXC chemokine ligand 13 (CXCL13; a B-cell attracting chemokine),1 interleukin (IL)-6 (a B-cell stimulatory cytokine), and soluble (s) CD30 (sCD30; a soluble receptor indicative of B- and T-cell activation) prehaematologica | 2018; 103(10)

Correspondence: brenda.birmann@channing.harvard.edu

Received: November 7, 2017. Accepted: June 15, 2018. Pre-published: June 21, 2018. doi:10.3324/haematol.2017.183236 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1679 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1679

M.M. Epstein et al.

dicted risk of an AIDS-NHL diagnosis in HIV-positive persons,2-4 in some instances as early as five years pre-diagnosis.5 Several of these markers have also demonstrated an association with NHL risk in immunocompetent people in prospective studies.6-13 Of interest, plasma sCD30 levels were positively associated with NHL risk at 6-10 years9 and even 15-23 years pre-diagnosis.11 Another small nested case-control study reported a significant 2.5-fold increase in NHL risk in women with elevated soluble IL-2 receptor-α levels (sIL-2Rα; a marker of T-cell activation and IL-2 upregulation), and marginally significant increases in NHL risk in women with higher pre-diagnosis tumor necrosis factor (TNF)-α and soluble TNFreceptor-2 (sTNF-R2) levels.14 These findings collectively suggest that chronic B-cell stimulation has a role in lymphomagenesis in immunocompetent persons. Our study aimed to further characterize pre-diagnosis plasma immune marker profiles associated with risk of HIV-unrelated NHL and its major histological subtypes in two large US cohorts. This study represents one of the largest populations with prospectively collected pre-diagnosis blood samples to investigate the association between numerous immune markers and NHL risk, including those with specific NHL subtypes that are often precluded due to small sample size, and to assess the independence of biomarker-NHL associations for multiple immune markers.11,12 The long-term follow up of the study population also allowed for examination of the influence of time since blood draw on observed immune markerNHL associations, including an assessment of potential early markers of lymphomagenesis present ten years or more prior to diagnosis. The choice of immune markers was guided in part by the immune deregulation we sought to characterize and by reported findings in AIDS- or HIVunrelated NHL. We hypothesized that pre-diagnosis levels of immune markers indicative of B-cell activation or inflammation would be positively associated with risk of developing NHL and major NHL subtypes, and that the use of multi-marker models will enhance characterization of the immune milieu associated with NHL risk and suggest subtle differences by histological subtype.

Methods Study population The study population comprised Nurses’ Health Study (NHS, all female) and Health Professionals Follow-up Study (HPFS, all male) participants with archived plasma (Online Supplementary Methods).15,16 Cancer diagnoses were identified via routine questionnaires or follow up after death17,18 and confirmed by medical record review or tumor registry linkage. Participants provided written informed consent at blood collection. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health.

Case and control selection We included all participants with confirmed incident NHL diagnosed three months or more after blood draw through 31st December 2010 with no other cancer history. Study pathologists (JCA, SJR) classified NHL histological subtype19 according to World Health Organization20,21 and International Lymphoma Epidemiology (InterLymph) Consortium guidelines.22,23 We ana1680

lyzed common B-cell (B-)NHL subtypes individually [diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), and chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL)], combined less common B-NHLs (“other B-NHL”) and defined additional categories by cell type (T-NHL, B-NHL). We matched one control per case by sex (cohort), age, race, and blood draw details (Online Supplementary Methods).

Biomarker assessment Assays were performed at the University of California, Los Angeles (LM, OMM), using multiplexed kits (Fluorokine® MAP, R & D Systems, Minneapolis, MN, USA), a Bio-Plex 200 Luminex instrument and Bio-Plex analysis software (Bio-Rad, Hercules, CA, USA). Blinded laboratory personnel measured sCD30, sIL-2Rα, Bcell activating factor of the TNF family (BAFF, a B-cell stimulatory cytokine), CXCL13, sIL-6Rα, sGP130, sCD14, sTNF-R2, C-reactive protein (CRP), IL-6, IL-8, IL-10, and TNF-α concentration according to the manufacturer's directions (Online Supplementary Methods). We set TNF-α, IL-8 and CXCL13 values to missing for samples with >24-hour processing delays (NHS: n=35; HPFS: n=23). Analyte concentrations were natural log-transformed for all analyses. We observed similar measured biomarker concentrations for the NHS and HPFS (Online Supplementary Table S1) and pooled the data.

Statistical analysis We conducted batch calibration to diminish the potential influence of laboratory batch-related variability on biomarker-NHL associations.24 Outlying biomarker values were identified using the Rosner extreme Studentized deviate method25 and omitted from analyses of the marker. The primary analysis assessed batch effect-corrected, log-transformed biomarker values continuously per Standard Deviation (SD) increase in concentration, with SD units calculated for logtransformed values in the pooled controls. We calculated Odds Ratios (OR) and 95% Confidence Intervals (CI) for the association of each biomarker with NHL risk (overall and for DLBCL, FL, CLL/SLL, other B-NHL, all B-NHL and all T-NHL) using unconditional logistic regression. Models adjusted for all matching factors unless small cell counts precluded adjustment for race. We evaluated but did not observe confounding by body mass index (BMI) and autoimmune disease history. We intended a priori to identify multi-marker profiles associated with NHL risk via mutual adjustment of models for biomarkers that were individually associated. We also examined models stratified by follow-up interval (0 to <5, 5 to <10, ≥10 years) and assessed heterogeneity by time period using the contrast test.26 The Online Supplementary Methods describe additional analyses designed post hoc.

Results In total, 601 cases of NHL (345 NHS and 256 HPFS) were identified and individually matched to controls. Three cases were later excluded due to unconfirmed lymphoma status. The final analysis thus included 598 cases, including 114 DLBCL, 92 FL, 165 CLL/SLL, 132 other BNHL (4 Burkitt lymphoma, 19 lymphoplasmacytic lymphoma, 20 mantle cell lymphoma, 44 marginal zone lymphoma, 20 other B-NHL, and 25 unclassified B-NHL) and 30 T-NHL, and 601 controls. The study population was 96% Caucasian and 58% female. Cases and controls had similar covariable distributions, due in part to the matched design (Table 1). haematologica | 2018; 103(10)

Pre-diagnosis plasma immune markers and NHL risk

We omitted 109 individual biomarker measurements (<1% of all measurements) with implausible outlying values (NHS: 72; HPFS: 37), the majority (90%) of which were implausibly high for the particular marker. Omitted values ranged from one measure of IL-10 to 17 measures of IL-8. Spearman correlation coefficients ranged from -0.03 (IL-10 and CXCL13) to 0.58 (sIL-2Rα and sCD30) (Online Supplementary Table S2).

Individual immune marker models Multivariable analyses of individual log-transformed immune markers revealed significant associations for all NHL per SD increment of log-transformed sTNF-R2, sIL2Rα, CXCL13, sCD30 (all positive) and BAFF (inverse; Table 2). In subtype-specific analyses, sTNF-R2 levels were also positively associated with risk of all B-NHL, FL and CLL/SLL, while CXCL13 was positively associated

with risk of all B-NHL, DLBCL and FL (Table 2). Levels of sIL-2Rα and sCD30 were positively associated with every NHL subtype, including T-NHL. Of interest, the association of BAFF with a 17% decreased risk of all NHL appeared to be driven by CLL/SLL, for which risk decreased by 49% per SD increase in log-transformed BAFF levels (OR: 0.51; 95%CI: 0.43, 0.62; P<0.001); BAFF was not associated with other NHL subtypes in singlemarker models. We did not observe significant or consistent associations for the remaining immune markers with risk of any NHL end point. Results from cohort-specific models did not suggest marked differences by sex for these associations (Online Supplementary Table S3).

Multi-marker profiles In the model that mutually adjusted for the five logtransformed immune markers that had significant individ-

Table 1. Characteristics of non-Hodgkin lymphoma cases and matched controls from two prospective cohort studies.

Variable Cohort NHS HPFS Age, years, mean ± SD Race/ethnicity Caucasian Other BMI at blood draw, kg/m² < 22.5 22.5-24.9 25-29.9 ≥ 30 Missing BMI in young adulthood, kg/m² < 18.5 18.5-22.4 22.5-24.9 ≥ 25 Missing Autoimmune disease† Yes No Years from blood draw to index date, Mean ± SD Cell type/histological subtype of NHL‡ B-NHL DLBCL Follicular lymphoma CLL/SLL Other B-cell subtypes§ T-NHL

Cases

Controls

344 (58%) 254 (42%) 60.8 ± 8.1

345 (57%) 256 (43%) 60.8 ± 8.1

0.97

573 (96%) 25 (4%)

578 (96%) 23 (4%)

0.75

138 (23%) 139 (23%) 217 (36%) 74 (12%) 30 (5%)

118 (20%) 166 (28%) 219 (36%) 74 (12%) 24 (4%)

0.58

44 (7%) 298 (50%) 126 (21%) 106 (18%) 24 (4%)

54 (9%) 299 (50%) 112 (19%) 101 (17%) 35 (6%)

0.31

97 (16%) 501 (84%)

104 (17%) 497 (83%)

0.62

9.6 ± 5.6

0.99

0.87

503 (84%) 114 (19%) 92 (15%) 165 (28%) 132 (22%) 30 (5%)

SD: Standard Deviation; NHS: Nurses' Health Study; HPFS: Health Professionals Follow-up Study; BMI: Body Mass Index; NHL: non-Hodgkin lymphoma; B-NHL: B-cell NHL; DLBCL: diffuse large B-cell lymphoma; CLL/SLL: chronic lymphocytic leukemia/small lymphocytic lymphoma; T-NHL: T-cell NHL. *P-values from χ2 test or ANOVA. Tests for BMI at blood draw and BMI in young adulthood did not include individuals’ missing data for those variables. †Defined as any self-reported diagnosis of rheumatoid arthritis, ulcerative colitis, multiple sclerosis, psoriasis, or Sjögren syndrome. ‡Information on cell type was not available for 11% of NHL cases. §The other B-NHL subtypes include Burkitt lymphoma (n=4), lymphoplasmacytic lymphoma (n=19), mantle cell lymphoma (n=20), marginal zone lymphoma (n=44), other B-NHL (n=20), and unclassified B-NHL (n=25).

haematologica | 2018; 103(10)

1681

M.M. Epstein et al. Table 2. Associations between pre-diagnosis concentrations of 13 individual immune markers and risk of non-Hodgkin lymphoma (NHL), overall and by major histological subtype.

All NHL* DLBCL Marker N cases/ controls IL-6 IL-8 IL-10 TNF-α CRP sCD14 sGP130 sTNF-R2 sIL-6Rα BAFF sIL-2Rα CXCL13 sCD30

597/600 558/566 596/597 566/571 596/599 592/596 592/596 592/601 592/599 592/601 585/600 554/571 590/600

OR (95% CI)‡

N cases

OR (95% CI)‡

0.97(0.87,1.08) 1.00(0.88,1.13) 1.00(0.89,1.11) 1.02(0.91,1.14) 1.06(0.94,1.19) 1.01(0.90,1.15) 1.03(0.89,1.18) 1.25(1.12,1.40) 1.10(0.98,1.23) 0.83(0.75,0.92) 1.37(1.23,1.53) 1.31(1.17,1.46) 1.37(1.23,1.52)

114 106 113 108 114 114 114 114 114 114 114 107 114

1.12(0.91,1.37) 0.96(0.77,1.22) 1.14(0.93,1.40) 0.98(0.80,1.21) 1.12(0.92,1.38) 0.90(0.72,1.13) 0.87(0.66,1.15) 1.02(0.83,1.26) 0.89(0.72,1.10) 0.99(0.81,1.21) 1.26(1.04,1.54) 1.25(1.03,1.52) 1.29(1.06,1.56)

B-NHL Subtypes CLL/SLL N OR N OR cases (95% CI)‡ cases (95% CI)‡ FL

92 84 91 87 92 90 91 90 91 92 91 86 90

0.90(0.73,1.12) 1.07(0.84,1.36) 0.98(0.78,1.22) 1.16(0.92,1.47) 1.12(0.89,1.40) 0.97(0.76,1.25) 1.15(0.90,1.47) 1.37(1.10,1.70) 1.16(0.93,1.44) 0.93(0.73,1.17) 1.55(1.25,1.94) 1.58(1.28,1.95) 1.76(1.44,2.15)

165 156 165 158 165 164 163 164 165 163 162 156 163

0.99(0.84,1.17) 0.98(0.81,1.20) 0.85(0.72,1.01) 1.06(0.89,1.26) 0.93(0.77,1.12) 0.91(0.75,1.11) 1.00(0.80,1.25) 1.28(1.08,1.51) 1.15(0.97,1.36) 0.51(0.43,0.62) 1.49(1.27,1.76) 1.10(0.93,1.31) 1.33(1.13,1.57)

All T-NHL Other B-NHL† N OR cases (95% CI)‡ 131 120 132 120 130 130 130 129 128 128 126 116 131

0.89(0.74,1.08) 1.11(0.90,1.37) 1.04(0.86,1.25) 0.82(0.68,1.00) 1.15(0.94,1.40) 1.14(0.93,1.40) 0.99(0.79,1.26) 1.35(1.13,1.62) 1.13(0.93,1.37) 0.87(0.73,1.05) 1.40(1.16,1.68) 1.48(1.24,1.76) 1.29(1.09,1.54)

N cases 30 29 30 29 30 30 30 30 30 30 30 28 29

OR (95% CI)‡

P§

1.13(0.78,1.63) 0.46 0.82(0.54,1.26) 0.70 1.21(0.83,1.76) 0.18 1.14(0.78,1.67) 0.17 0.97(0.66,1.42) 0.50 0.94(0.62,1.43) 0.55 0.81(0.48,1.36) 0.57 1.03(0.70,1.50) 0.20 1.01(0.69,1.46) 0.32 1.26(0.87,1.82) <0.0001 1.97(1.37,2.85) 0.26 1.23(0.86,1.76) 0.06 1.46(1.05,2.04) 0.15

N: number; NHL: non-Hodgkin lymphoma; B-NHL: B-cell non-Hodgkin lymphoma; DLBCL: diffuse large B-cell lymphoma; FL: follicular lymphoma; CLL/SLL: chronic lymphocytic leukemia/small lymphocytic lymphoma; T-NHL: T-cell NHL; OR: Odds Ratio; CI: Confidence Interval; IL: interleukin; TNF: tumor necrosis factor; CRP: C-reactive protein; sCD14: soluble CD14; sGP130: soluble GP130; sTNF-R2: soluble tumor necrosis factor receptor-2; sIL-6Rα: soluble interleukin-6 receptor-α; BAFF: B-cell activating factor of the TNF family; sIL-2Rα: soluble interleukin-2 receptor-α; CXCL13: CXC chemokine ligand 13; sCD30: soluble CD30. *The all B-NHL (n=503 cases) results were similar to the all NHL results. †Other B-NHL subtypes include Burkitt lymphoma (n=4), lymphoplasmacytic lymphoma (n=19), mantle cell lymphoma (n=20), marginal zone lymphoma (n=44), other B-NHL (n=20), and unclassified B-NHL (n=25). ‡ Odds Ratios and 95% Confidence Intervals were calculated per 1 standard deviation increase in log biomarker concentration, based on batch-corrected values with outliers removed, for NHS and HPFS cohorts combined. All models except those for T-NHL were adjusted for age at blood draw (continuous), cohort, time of blood draw (continuous), race (Caucasian/other); the models for T-NHL were adjusted for age and cohort only. §P-values for heterogeneity by subtype from contrast tests comparing immune marker-specific estimates between DLBCL, FL, CLL/SLL, other B-NHL, and T-NHL.

ual associations with NHL end points (sTNF-R2, sIL-2Rα, CXCL13, sCD30, BAFF), sIL-2Rα, CXCL13 and sCD30 remained significantly associated with a 17-24% increased risk, and BAFF with a 26% decreased risk of all NHL per SD increase in log concentration, while sTNF-R2 was no longer significantly associated (Table 3). Results for all B-NHL risk were similar to those for all NHL, whereas mutual adjustment attenuated all the immune marker associations with DLBCL. In the multi-marker model of FL risk, sCD30 and BAFF remained independently associated, with a borderline association noted for CXCL13 (Table 4). In the multi-marker model of CLL/SLL risk, sIL-2Rα was significantly associated with a 50% increase (95%CI: 1.18-1.90), and BAFF with a significant 53% reduction (95%CI: 0.38, 0.58), per SD increase in log concentration. Lastly, only sIL-2Rα was independently associated with T-NHL risk (OR per SD increase in log concentration: 1.96; 95%CI: 1.22, 3.13) in mutually adjusted models. The 5-marker models using the polytomous logistic regression (PLR) approach yielded essentially the same effect estimates as described above for biomarker associations with the NHL end points for the full follow-up period (Online Supplementary Tables S4 and S5). sTNF-R2 had significantly different associations with B-NHL and TNHL (P-value for heterogeneity by subtype=0.04) (Online Supplementary Table S4); the associations of CXCL13 and BAFF with individual B-NHL subtypes also showed evidence of significant heterogeneity (P-values for heterogeneity by subtype =0.007 and <0.0001, respectively) (Online Supplementary Table S5). In the covariable-adjusted multi-marker models containing restricted cubic splines, there was evidence of non-lin1682

earity for two biomarkers, CXCL13 and BAFF, in their associations with risk of aggregated end points (all NHL, B-NHL and other B-NHL; P-value tests for significance of the curve <0.05), but not for biomarker associations with individual B-NHL subtypes or T-NHL. In alternative models using semi-automatic stepwise selection, the final models for all NHL and all B-NHL included sIL-2Rα, CXCL13 and sCD30, which were positively associated with risk, as well as BAFF, which was inversely associated (Online Supplementary Table S6). In comparison, for DLBCL and FL, the stepwise procedure selected only sCD30 (P=0.004 and <0.0001, respectively), and for T-NHL the procedure selected only sIL-2Rα (P=0.002) as independently (positively) associated with risk. Of interest, the stepwise procedure identified four immune markers independently associated with risk of CLL/SLL, including BAFF and IL-10 with significant inverse associations and sIL-2Rα with a significant positive association. The stepwise procedure identified three immune markers associated with the combined category of other B-NHL subtypes, including significant positive associations for CXCL13 and sIL-2Rα, and a significant inverse association for BAFF. In the model that included all 13 immune markers, sIL2Rα , CXCL13, and sCD30 again had strong positive associations with risk of all NHL and all B-NHL (Table 5). In the CLL/SLL-specific model, we observed a significant inverse association with risk for BAFF and also for sCD14 and IL-10, and a positive association with sIL2-Rα. BAFF was also significantly inversely associated with FL risk, while sCD30 was significantly positively associated with FL risk. Only sIL-2Rα was significantly associated with an increased risk of T-NHL. We observed suggestive positive haematologica | 2018; 103(10)

Pre-diagnosis plasma immune markers and NHL risk Table 3. Independent associations of multiple pre-diagnosis plasma immune markers with risk of non-Hodgkin lymphoma (NHL), overall and by B- or T-cell type of origin, for the complete follow-up period and stratified by years of follow up.

Marker

All NHL sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF All B-NHL sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF All T-NHL sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF

Complete follow-up period N cases/ OR controls (95% CI) per 1-SD *,†

Years from blood draw to diagnosis/index date 0 to less than 5 5 to less than 10 10 or more N cases/ OR N cases/ OR N cases/ OR controls (95% CI) controls (95% CI) controls (95% CI) per 1-SD *,† per 1-SD *,† per 1-SD *,†

P‡

542/571 542/571 542/571 542/571 542/571

1.05 (0.91, 1.21) 1.20 (1.03, 1.39) 1.17 (1.03, 1.32) 1.24 (1.06, 1.45) 0.74 (0.66, 0.83)

133/140 133/140 133/140 133/140 133/140

0.83 (0.60, 1.14) 1.52 (1.09, 2.11) 1.00 (0.78, 1.29) 1.52 (1.09, 2.13) 0.73 (0.59, 0.91)

149/162 149/162 149/162 149/162 149/162

1.02 (0.77, 1.35) 1.16 (0.88, 1.53) 1.30 (1.03, 1.62) 1.43 (1.07, 1.90) 0.61 (0.48, 0.78)

260/267 260/267 260/267 260/267 260/267

1.18 (0.95, 1.46) 1.11 (0.88, 1.39) 1.21 (1.01, 1.46) 0.98 (0.78, 1.23) 0.83 (0.69, 1.00)

0.20 0.28 0.32 0.02 0.15

454/570 454/570 454/570 454/570 454/570

1.07 (0.92, 1.25) 1.20 (1.03, 1.41) 1.13 (1.00, 1.29) 1.24 (1.05, 1.46) 0.73 (0.64, 0.83)

110/140 110/140 110/140 110/140 110/140

0.88 (0.63, 1.23) 1.51 (1.06, 2.14) 0.96 (0.74, 1.25) 1.59 (1.10, 2.28) 0.67 (0.53, 0.84)

118/161 118/161 118/161 118/161 118/161

1.15 (0.86, 1.54) 1.12 (0.84, 1.49) 1.17 (0.92, 1.49) 1.58 (1.14, 2.20) 0.64 (0.50, 0.81)

226/267 226/267 226/267 226/267 226/267

1.13 (0.90, 1.42) 1.15 (0.91, 1.46) 1.24 (1.02, 1.50) 0.96 (0.75, 1.22) 0.84 (0.69, 1.02)

0.40 0.38 0.31 0.02 0.14

28/569 28/569 28/569 28/569 28/569

0.62 (0.37, 1.03) 1.96 (1.22, 3.13) 1.11 (0.75, 1.65) 1.33 (0.84, 2.10) 0.88 (0.58, 1.32)

11/140 11/140 11/140 11/140 11/140

0.44 (0.17, 1.19) 2.10 (0.95, 4.68) 1.03 (0.48, 2.22) 1.68 (0.69, 4.08) 0.93 (0.52, 1.67)

10/160 10/160 10/160 10/160 10/160

0.65 (0.27, 1.58) 2.20 (0.93, 5.20) 1.37 (0.78, 2.42) 1.34 (0.56, 3.21) 0.55 (0.24, 1.28)

7/267 7/267 7/267 7/267 7/267

0.73 (0.24, 2.21) 1.04 (0.36, 3.00) 0.57 (0.24, 1.37) 1.77 (0.70, 4.43) 1.68 (0.64, 4.44)

0.78 0.50 0.26 0.90 0.23

N: number; NHL: non-Hodgkin lymphoma; B-NHL: B-cell NHL; T-NHL: T-cell NHL; OR: Odds Ratio; CI: Confidence Interval; SD: Standard Deviation; sTNF-R2: soluble tumor necrosis factor receptor-2; sIL-2Rα: soluble interleukin-2 receptor-α; CXCL13: CXC chemokine ligand 13; sCD30: soluble CD30; BAFF: B-cell activating factor of the TNF family. *Models were adjusted for age at blood draw (continuous), cohort (sex), time of blood draw (continuous) and race/ethnicity (Caucasian, non-Caucasian) and were mutually adjusted for all markers listed, except that models for all T-NHL were not adjusted for race. †Odds Ratios and 95% Confidence Intervals were calculated per 1-standard deviation increase in batch effect-corrected, log-transformed values (with cohort-specific outliers excluded) from the Nurses’ Health Study and Health Professionals Follow-up Study combined. ‡P-values from tests for heterogeneity comparing immune marker-specific estimates across time strata.

associations of DLBCL risk with IL-6, CXCL13 and sCD30 in this 13-marker model.

Time-stratified analyses The analyses stratified by time between blood draw and diagnosis/index date suggested that the individual biomarker associations with all NHL (Online Supplementary Table S7) and with NHL subtypes (Online Supplementary Table S8) varied somewhat by length of time after blood draw but did not strongly implicate any additional immune marker-NHL associations. The time-stratified 5marker models (Table 3) also suggested variability by follow-up interval in the independent associations of those immune markers with future NHL risk. For example, the association of sIL-2Rα with risk of all NHL appeared to be restricted to a shorter-term interval, specifically within five years of blood draw (OR: 1.52, 95%CI: 1.09, 2.11) (Table 3), whereas significant associations of CXCL13 with risk of all NHL were evident only five or more years after blood collection (5-<10 years; OR: 1.23, 95%CI: 1.00, 1.52; and ≥10 years; OR: 1.21, 95%CI: 1.01, 1.45). sCD30 was most strongly associated with all NHL risk within ten years of blood draw, while BAFF was consistently inversely associated with all NHL across time periods. Of note, in subtype-specific time-stratified analyses, sCD30 levels were strongly positively associated with risk of FL within five years of blood draw (OR: 4.85, 95%CI: 2.02, 11.61), haematologica | 2018; 103(10)

and the association decreased in magnitude with increasing follow-up time. In CLL/SLL-specific models, elevated sIL-2Rα was associated with a nearly 4-fold increased risk within five years of blood draw (OR: 3.71, 95%CI: 1.77, 7.76) but had no clear association with longer-term CLL/SLL risk. In contrast, BAFF had significant inverse associations with risk of CLL/SLL in all pre-diagnosis time periods, albeit with particularly strong associations with risk of CLL/SLL within five or ten years of blood draw (Table 4). When modeled using PLR, the effect estimates were virtually the same for time period-specific biomarker associations, both for the aggregated and the individual NHL end points (Online Supplementary Tables S4 and S5). The most prominent differences between the two approaches for assessing heterogeneity by time period (PLR with interaction terms vs. time-stratified unconditional logistic regression) pertained to the statistical significance of apparent heterogeneity by follow-up period for the associations of sTNF-R2 with all B-NHL and FL. For example, for the association of sTNF-R2 with all B-NHL, the P-value for heterogeneity by follow-up time was 0.04 for the cross-product term in PLR (Online Supplementary Table S4) and 0.40 for the main model contrast test (Table 3). For the association of sTNF-R2 with FL, the P-value for heterogeneity by time period was 0.0007 for the crossproduct term in PLR (Online Supplementary Table S5) and 0.11 for the main model contrast test (Table 4). Time-strat1683

M.M. Epstein et al. Table 4. Independent associations of multiple pre-diagnosis plasma immune markers with risk of non-Hodgkin lymphoma (NHL) by major histological subtype of B-cell NHL, for the complete follow-up period and stratified by years of follow up.

Marker

Years from blood draw to diagnosis/index date Complete follow-up period 0 to less than 5 5 to less than 10 10 or more N cases/ OR N cases/ OR N cases/ OR N cases/ OR controls (95% CI) controls (95% CI) controls (95% CI) controls (95% CI) per 1-SD*,† per 1-SD*,† per 1-SD*,† per 1-SD*,†

DLBCL sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF FL sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF CLL/SLL sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF Other B-NHL§ sTNF-R2 sIL-2Rα CXCL13 sCD30 BAFF

P‡

107/570 107/570 107/570 107/570 107/570

0.81 (0.62, 1.07) 1.18 (0.91, 1.53) 1.17 (0.95, 1.45) 1.23 (0.95, 1.59) 0.95 (0.76, 1.18)

25/140 25/140 25/140 25/140 25/140

0.61 (0.34, 1.10) 1.83 (1.00, 3.37) 0.71 (0.43, 1.19) 0.90 (0.48, 1.67) 0.96 (0.59, 1.55)

25/161 25/161 25/161 25/161 25/161

1.05 (0.60, 1.85) 1.19 (0.67, 2.11) 1.42 (0.95, 2.12) 1.76 (1.07, 2.89) 0.69 (0.44, 1.08)

57/267 57/267 57/267 57/267 57/267

0.83 (0.56, 1.24) 1.09 (0.75, 1.58) 1.30 (0.95, 1.79) 1.09 (0.75, 1.59) 1.08 (0.78, 1.50)

0.42 0.35 0.09 0.19 0.27

83/569 83/569 83/569 83/569 83/569

1.03 (0.77, 1.38) 1.06 (0.78, 1.46) 1.24 (0.98, 1.58) 1.69 (1.26, 2.26) 0.76 (0.59, 0.98)

18/140 18/140 18/140 18/140 18/140

0.45 (0.16, 1.25) 0.93 (0.35, 2.44) 1.10 (0.58, 2.06) 4.85 (2.02, 11.61) 0.70 (0.38, 1.30)

22/160 22/160 22/160 22/160 22/160

0.95 (0.53, 1.69) 1.09 (0.63, 1.89) 1.12 (0.72, 1.75) 1.88 (1.04, 3.40) 0.72 (0.41, 1.24)

43/267 43/267 43/267 43/267 43/267

1.35 (0.93, 1.96) 1.09 (0.70, 1.68) 1.48 (1.03, 2.13) 1.06 (0.68, 1.64) 0.78 (0.55, 1.12)

0.11 0.95 0.56 0.007 0.94

153/569 153/569 153/569 153/569 153/569

1.21 (0.96, 1.52) 1.50 (1.18, 1.90) 0.90 (0.74, 1.10) 1.15 (0.89, 1.48) 0.47 (0.38, 0.58)

36/140 36/140 36/140 36/140 36/140

0.98 (0.49, 1.93) 3.71 (1.77, 7.76) 0.78 (0.48, 1.27) 1.43 (0.71, 2.87) 0.32 (0.19, 0.53)

44/160 44/160 44/160 44/160 44/160

1.27 (0.81, 1.98) 1.39 (0.90, 2.15) 0.80 (0.54, 1.20) 1.54 (0.96, 2.46) 0.39 (0.25, 0.61)

73/267 73/267 73/267 73/267 73/267

1.16 (0.85, 1.58) 1.26 (0.87, 1.83) 1.04 (0.77, 1.41) 0.90 (0.62, 1.30) 0.63 (0.46, 0.86)

0.82 0.04 0.48 0.17 0.05

111/569 111/569 111/569 111/569 111/569

1.17 (0.92, 1.49) 1.15 (0.89, 1.48) 1.45 (1.19, 1.77) 1.08 (0.82, 1.41) 0.78 (0.64, 0.95)

31/140 31/140 31/140 31/140 31/140

1.28 (0.78, 2.12) 1.07 (0.62, 1.83) 1.50 (0.98, 2.30) 1.28 (0.73, 2.26) 0.74 (0.53, 1.04)

27/160 27/160 27/160 27/160 27/160

1.17 (0.73, 1.90) 1.05 (0.63, 1.74) 1.52 (1.07, 2.17) 1.37 (0.81, 2.32) 0.72 (0.48, 1.07)

53/267 53/267 53/267 53/267 53/267

1.11 (0.77, 1.61) 1.29 (0.87, 1.91) 1.36 (0.98, 1.88) 0.79 (0.52, 1.21) 0.89 (0.64, 1.24)

0.90 0.77 0.88 0.20 0.66

N: number; DLBCL: diffuse large B-cell lymphoma; FL: follicular lymphoma; CLL/SLL: chronic lymphocytic leukemia/small lymphocytic lymphoma; B-NHL: B-cell NHL; OR: Odds Ratio; CI: Confidence Interval; SD: Standard Deviation; sTNF-R2: soluble tumor necrosis factor receptor-2; sIL-2Rα: soluble interleukin-2 receptor-α; CXCL13: CXC chemokine ligand 13; sCD30: soluble CD30; BAFF: B-cell activating factor of the TNF family. *All models were adjusted for age at blood draw (continuous), cohort (sex), time of blood draw (continuous) and race/ethnicity (Caucasian, non-Caucasian), and were mutually adjusted for all markers listed, except that models for other B-NHL were not adjusted for race. † Odds Ratios and 95% Confidence Intervals were calculated per 1-standard deviation increase in batch effect-corrected, log-transformed values (with cohort-specific outliers excluded) from the Nurses’ Health Study and Health Professionals Follow-up Study combined. ‡P-values from tests for heterogeneity comparing immune marker-specific estimates across time strata. §Other B-NHL subtypes included Burkitt lymphoma (n=4), lymphoplasmacytic lymphoma (n=19), mantle cell lymphoma (n=20), marginal zone lymphoma (n=39), other B-NHL (n=20), and unclassified B-NHL (n=25).

ified results for the multi-marker models identified with stepwise selection largely agreed with the main results described above (Online Supplementary Table S6).

Discussion In this pooled analysis within the NHS and HPFS cohorts, we observed significant associations between NHL risk and pre-diagnosis levels of specific plasma immune markers, including a novel, inverse association between levels of BAFF and risk of CLL/SLL. Positive associations between levels of sIL-2Rα, CXCL13, and sCD30 and risk of all NHL and all B-NHL, as well as the inverse association of BAFF with risk of all NHL and CLL/SLL, were consistent and independent across several analytical approaches to constructing a multi-marker profile associated with risk. In contrast, the individual positive associations noted for sTNF-R2 with risk of all NHL and some B1684

NHL end points were attenuated upon adjustment for other immune markers, suggesting a lack of independence in the association between sTNF-R2 levels and NHL risk. Manual selection and automated stepwise selection of multi-marker profiles yielded fairly consistent results for all NHL, but also some differences for individual histological subtypes, particularly for CLL/SLL. We also observed some variation in the associations between NHL risk and immune markers by time between blood draw and diagnosis. Our findings are in agreement with previous studies reporting associations between elevated CXCL13 and/or sCD30 levels and increased NHL risk in HIV-positive and immunocompetent populations, including several reports analyzing blood samples taken many years prior to NHL diagnosis.2,3,8-13 In the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study, Purdue et al.11 prospectively investigated multi-marker models similar to those in our analysis and observed independent positive associations haematologica | 2018; 103(10)

Pre-diagnosis plasma immune markers and NHL risk Table 5. Associations of pre-diagnosis plasma immune markers with risk of non-Hodgkin lymphoma (NHL), with mutual adjustment for all thirteen immune markers, for all NHL and by major histological subtype.

All NHL DLBCL Marker N Cases/ Controls IL-6 IL-8 IL-10 TNF-α CRP sCD14 sGP130 sTNF-R2 sIL-6Rα BAFF sIL-2Rα CXCL13 sCD30

523/550 523/550 523/550 523/550 523/550 523/550 523/550 523/550 523/550 523/550 523/550 523/550 523/550

OR (95% CI) per 1-SD *,† 1.06 (0.93, 1.21) 0.96 (0.83, 1.10) 0.95 (0.84, 1.07) 1.00 (0.87, 1.15) 1.00 (0.87, 1.15) 0.84 (0.71, 1.00) 1.06 (0.85, 1.32) 1.06 (0.86, 1.31) 1.03 (0.89, 1.20) 0.73 (0.64, 0.82) 1.19 (1.02, 1.40) 1.18 (1.04, 1.34) 1.26 (1.06, 1.48)

N cases

104 104 104 104 104 104 104 104 104 104 104 104 104

OR (95% CI) per 1-SD*,† 1.22 (0.96, 1.55) 0.92 (0.72, 1.18) 1.14 (0.91, 1.43) 0.84 (0.66, 1.08) 1.07 (0.84, 1.37) 0.82 (0.61, 1.11) 1.05 (0.71, 1.55) 0.96 (0.67, 1.37) 0.86 (0.66, 1.12) 0.98 (0.78, 1.23) 1.09 (0.82, 1.44) 1.20 (0.96, 1.49) 1.27 (0.96, 1.66)

B-NHL Subtypes FL CLL/SLL N OR N OR cases (95% CI) cases (95% CI) per 1-SD*,† per 1-SD*,† 78 78 78 78 78 78 78 78 78 78 78 78 78

0.86 (0.66, 1.12) 0.94 (0.71, 1.24) 0.93 (0.72, 1.22) 1.09 (0.81, 1.46) 1.29 (0.96, 1.72) 0.84 (0.61, 1.18) 1.20 (0.78, 1.86) 0.88 (0.57, 1.36) 1.02 (0.74, 1.40) 0.74 (0.56, 0.97) 1.09 (0.77, 1.54) 1.22 (0.94, 1.59) 1.80 (1.30, 2.50)

150 150 150 150 150 150 150 150 150 150 150 150 150

1.10 (0.89, 1.37) 1.00 (0.79, 1.26) 0.80 (0.65, 0.99) 1.09 (0.87, 1.37) 0.82 (0.65, 1.04) 0.68 (0.51, 0.89) 1.11 (0.78, 1.58) 1.33 (0.94, 1.87) 1.14 (0.90, 1.45) 0.46 (0.37, 0.57) 1.59 (1.23, 2.06) 0.92 (0.74, 1.13) 1.10 (0.84, 1.45)

All T-NHL Other B-NHL‡ N OR cases (95% CI) per 1-SD*,† 106 106 106 106 106 106 106 106 106 106 106 106 106

1.03 (0.81, 1.31) 1.11 (0.88,1.38) 0.99 (0.79, 1.23) 0.86 (0.67, 1.10) 1.00 (0.77, 1.29) 0.92 (0.68, 1.24) 0.84 (0.57, 1.22) 1.16 (0.81, 1.65) 1.19 (0.91, 1.56) 0.76 (0.62, 0.94) 1.10 (0.83, 1.46) 1.45 (1.18, 1.78) 1.13 (0.85, 1.52)

N cases

OR (95% CI) per 1-SD*,†

28 28 28 28 28 28 28 28 28 28 28 28 28

1.10 (0.71, 1.73) 0.65 (0.40, 1.07) 1.19 (0.77, 1.82) 1.18 (0.74, 1.89) 0.85 (0.54, 1.34) 0.90 (0.51, 1.56) 0.76 (0.36, 1.60) 0.75 (0.38, 1.46) 1.11 (0.67, 1.84) 0.88 (0.58, 1.34) 1.95 (1.22, 3.10) 1.20 (0.80, 1.82) 1.25 (0.76, 2.05)

N: number; B-NHL: B-cell NHL; SD: Standard Deviation; OR: Odds Ratio; CI: Confidence Interval; DLBCL: diffuse large B-cell lymphoma; FL: follicular lymphoma; CLL/SLL: chronic lymphocytic leukemia/small lymphocytic lymphoma; T-NHL: T-cell NHL; IL: interleukin; TNF: tumor necrosis factor; CRP: C-reactive protein; sCD14: soluble CD14; sGP130: soluble GP130; sTNF-R2: soluble tumor necrosis factor receptor-2; sIL-6Rα: soluble interleukin-6 receptor-α; BAFF: B-cell activating factor of the TNF family; sIL-2Rα: soluble interleukin-2 receptor-α; CXCL13: CXC chemokine ligand 13; sCD30: soluble CD30. *From multivariable logistic regression models that include all 13 immune markers in each model, adjusted for age at blood draw, time of day of blood draw, race and cohort. T-NHL models were not adjusted for race due to sparse cell counts. †Odds Ratios and 95% Confidence Intervals were calculated per 1 standard deviation increase in biomarker concentration, based on batch effect-corrected log-transformed values with outliers removed, for Nurses’ Health Study and Health Professionals Follow-up Study cohorts combined. ‡Other B-NHL subtypes included Burkitt lymphoma (n=4), lymphoplasmacytic lymphoma (n=19), mantle cell lymphoma (n=20), marginal zone lymphoma (n=44), other B-NHL (n=20), and unclassified B-NHL (n=25).

for sCD30 with risk of NHL and DLBCL when adjusted for other biomarkers. Those observations were detectable more than 15 years prior to diagnosis. Also similar to our findings, positive associations observed for sTNF-R2 with NHL did not persist upon adjustment for other immune markers.11 In a prospective analysis in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, individual associations of CXCL13 and sTNF-R2 with NHL both remained significant with mutual adjustment, with correction for multiple comparisons and with restriction to samples collected 8-13 years prior to diagnosis.10 We observed an unexpected yet consistently strong inverse association between BAFF levels and CLL/SLL risk. BAFF is a member of the TNF family involved with B-cell survival and maturation.27 Pre-diagnosis serum BAFF concentrations were positively associated with AIDS-NHL, and BAFF overproduction has been associated with systemic autoimmune diseases, including systemic lupus erythematosus and Sjögren syndrome,28,29 which are associated with an increased risk of NHL in HIV-negative persons.30,31 However, systemic autoimmune disorders in HIV-negative individuals appear to be preferentially associated with NHL subtypes with a different natural history than CLL/SLL.30,32 Nonetheless, CLL cells are known to express multiple BAFF receptors (including TNFRSF13B, TNFRS13C and TNFRSF17),33 and the inverse association that we observed is biologically plausible if considered indicative of rapid uptake of circulating BAFF by nascent CLL/SLL clones,34 reflecting subclinical progression of an indolent tumor whose natural course may extend multiple decades. Consistent with this interpretation, several clinical studies have observed lower levels of BAFF in sera from CLL/SLL patients than in healthy controls.35-37 The haematologica | 2018; 103(10)

mechanism for the latter findings is unknown; our observation suggests that those underlying physiological processes may commence early in CLL pathogenesis, even ten or more years pre-diagnosis. Concurrent measurement of soluble BAFF receptors and study of cell surface expression of those molecules and classification of cases into prognostic subgroups were not feasible for the present study. Confirmation of the present findings is warranted in larger populations with specimens suitable for determining cell surface marker or gene/protein expression. Additionally, prospective studies in patients with monoclonal B-lymphocytosis would be informative to evaluate whether circulating BAFF levels can enhance risk stratification for progression to malignancy.38 We also observed significant associations of elevated sIL-2Rα levels with increased risks of all NHL, B-NHL, DLBCL, CLL/SLL and T-NHL, primarily within five years of blood draw. One other study reported a positive association between sIL-2Rα levels and NHL risk in an HIVnegative population with prospective blood collection that persisted after incorporating lag-times greater than two years.14 Of note, comparatively high sIL-2Rα levels at diagnosis were also associated with poor prognosis in patients with NHL.39-41 Biologically, sIL-2Rα and sCD30 are highly correlated (r=0.58 in this study), and both can indicate Band T-cell activation;42 in the present analysis, both markers remained independently associated with a significant increased risk of all NHL and all B-NHL after mutual adjustment. In contrast, only sIL-2Rα was significantly associated with an increased risk of T-NHL in the multimarker models, although small sample size (n=30 cases) limited statistical power to detect significant independent associations for more strongly correlated biomarkers. Of 1685

M.M. Epstein et al.

interest, we observed the strongest positive associations of sIL-2RÎą with T-NHL risk within ten years of blood draw, a novel observation that requires confirmation in other populations. We observed significant positive associations between CXCL13 and risk of all NHL, B-NHL, and FL, as well as borderline associations with DLBCL and other B-NHL, more than ten years after blood draw, suggesting an early role for an immune environment characterized by B-cell stimulation and aberrant B-cell trafficking. Consistent with this interpretation, a recent, large-scale genome-wide association study of FL identified CXCR5, which is the receptor for CXCL13, as a potential FL susceptibility locus.43 Further, genetic variation in CXCR5 and CXCL13 was associated with serum CXCL13 levels in a study of AIDS-NHL, and elevated serum CXCL13 levels were observed in AIDS-NHL cases more than three years prior to diagnosis.2 In contrast, elevated sCD30 levels were more strongly associated with increased risk of all NHL, BNHL and FL within ten years of blood draw, with a particularly strong association with FL within five years of blood draw. These findings suggest sCD30 may be capturing a more proximal pre-diagnosis increase in immune activation. When assessed with multivariable PLR models rather than the main unconditional logistic regression analysis, the associations between immune markers and NHL end points did not change substantially, whether for aggregated end points or the individual B-NHL end points. The minor discrepancies suggested somewhat improved precision in the PLR models, which yielded slightly narrower confidence intervals and slightly stronger P-values for heterogeneity by follow-up period for a few of the comparisons. None of the discrepant findings would suggest a different interpretation of the time- or subtype-specific model findings, however, and thus we retained the unconditional logistic regression models as our primary analysis for methodological consistency across the full series of analyses we conducted. In the analyses with restricted cubic splines, we observed evidence of significant non-linearity for associations of CXCL13 and BAFF with aggregated NHL end points. Of note, those end points comprise small numbers of diverse histological subtypes of NHL which may have different etiologies. Thus, we believe that the observed non-linear associations more likely reflect sampling variability and/or an artifact of potentially heterogeneous subtype-specific associations for the subtypes in the end point groups than a true biological effect. Together, our findings add new insight to previous publications on both AIDS-NHL and HIV-unrelated NHL risk,

References 1. Takagi R, Higashi T, Hashimoto K, et al. B cell chemoattractant CXCL13 is preferentially expressed by human Th17 cell clones. J Immunol. 2008;181(1):186-189. 2. Hussain SK, Zhu W, Chang SC, et al. Serum levels of the chemokine CXCL13, genetic variation in CXCL13 and its receptor CXCR5, and HIV-associated non-hodgkin B-cell lymphoma risk. Cancer Epidemiol Biomarkers Prev. 2013;22(2):295-307. 3. Breen EC, Fatahi S, Epeldegui M, Boscardin

1686

collectively suggesting that higher levels of immune activation, and in particular heightened B-cell stimulation, may affect B-cell lymphomagenesis. Interestingly, several markers of immune activation appear to be elevated many years prior to NHL diagnosis and thus could help identify populations at higher risk for developing NHL. It is important to note that some reported associations between immune markers and all NHL risk were not replicated in analyses of individual histological subtypes; this may be due in part to subtype-specific sample sizes that limited statistical power. Significant associations between immune markers and risk of all NHL may reflect commonalities in subtype etiologies; however, these findings may also conceal a more specific association with one or more of the less common subtypes, as illustrated by the present findings for BAFF and CLL/SLL. This analysis of immune markers measured from prospectively collected blood specimens from two large US cohorts with lengthy follow up identified several statistically significant associations with the risk of developing NHL, including associations that remained statistically significant for blood samples collected five or more years prior to diagnosis. Although our main results were fairly consistent across analytical approaches, slight variations in markers chosen by a priori and secondary analyses emphasize the importance of utilizing diverse panels of immune markers in future studies seeking to characterize conditions conducive to NHL development. Furthermore, our findings suggest that even though an activated immune milieu may contribute to the development of multiple types of NHL, there is evidence of subtle differences in the pathogenesis of individual NHL subtypes, some of which had not been previously reported. Larger pooled studies will be important to more accurately identify homogeneous and heterogeneous biomarkers of risk or early disease by NHL subtype and to elucidate which are more indicative of earlier or later pathogenic changes to the immune environment. Acknowledgments The authors would like to thank the participants in the Nursesâ&#x20AC;&#x2122; Health Study and Health Professionals Follow-up Study for their ongoing participation in the cohort studies. We thank Laura Burns for assistance with manuscript preparation, and wish to recognize the technical contributions of the Dana Farber/Harvard Cancer Center Specialized Histopathology Core Laboratory. We thank the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, and WY. The authors assume full responsibility for analyses and interpretation of these data.

WJ, Detels R, Martinez-Maza O. Elevated serum soluble CD30 precedes the development of AIDS-associated non-Hodgkin's B cell lymphoma. Tumour Biol. 2006;27(4):187-194. 4. Breen EC, van der Meijden M, Cumberland W, Kishimoto T, Detels R, Martinez-Maza O. The development of AIDS-associated Burkitt's/small noncleaved cell lymphoma is preceded by elevated serum levels of interleukin 6. Clin Immunol. 1999;92(3): 293-299. 5. Vendrame E, Hussain SK, Breen EC, et al.

Serum levels of cytokines and biomarkers for inflammation and immune activation, and HIV-associated non-Hodgkin B-cell lymphoma risk. Cancer Epidemiol Biomarkers Prev. 2014;23(2):343-349. 6. Edlefsen KL, Martinez-Maza O, Madeleine MM, et al. Cytokines in serum in relation to future non-Hodgkin lymphoma risk: evidence for associations by histologic subtype. Int J Cancer. 2014;135(4):913-922. 7. Saberi Hosnijeh F, Krop EJ, Scoccianti C, et al. Plasma cytokines and future risk of nonHodgkin lymphoma (NHL): a case-control

haematologica | 2018; 103(10)

Pre-diagnosis plasma immune markers and NHL risk

10.

11.

12.

13.

14.

15.

16.

17.

18. 19.

20.

study nested in the Italian European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev. 2010;19(6):1577-1584. De Roos AJ, Mirick DK, Edlefsen KL, et al. Markers of B-cell activation in relation to risk of non-Hodgkin lymphoma. Cancer Res. 2012;72(18):4733-4743. Purdue MP, Lan Q, Bagni R, et al. Prediagnostic serum levels of cytokines and other immune markers and risk of nonhodgkin lymphoma. Cancer Res. 2011;71(14):4898-4907. Purdue MP, Hofmann JN, Kemp TJ, et al. A prospective study of 67 serum immune and inflammation markers and risk of nonHodgkin lymphoma. Blood. 2013; 122(6):951-957. Purdue MP, Lan Q, Kemp TJ, et al. Elevated serum sCD23 and sCD30 up to two decades prior to diagnosis associated with increased risk of non-Hodgkin lymphoma. Leukemia. 2015;29(6):1429-1431. Purdue MP, Lan Q, Martinez-Maza O, et al. A prospective study of serum soluble CD30 concentration and risk of non-Hodgkin lymphoma. Blood. 2009;114(13):27302732. Vermeulen R, Hosnijeh FS, Portengen L, et al. Circulating soluble CD30 and future risk of lymphoma; evidence from two prospective studies in the general population. Cancer Epidemiol Biomarkers Prev. 2011;20(9):1925-1927. Gu Y, Shore RE, Arslan AA, et al. Circulating cytokines and risk of B-cell nonHodgkin lymphoma: a prospective study. Cancer Causes Control. 2010;21(8):13231333. Colditz GA, Hankinson SE. The Nurses' Health Study: lifestyle and health among women. Nat Rev Cancer. 2005;5(5):388396. Hankinson SE, Willett WC, Manson JE, et al. Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women. J Natl Cancer Inst. 1995;87(17):1297-1302. Rich-Edwards JW, Corsano KA, Stampfer MJ. Test of the National Death Index and Equifax Nationwide Death Search. Am J Epidemiol. 1994;140(11):1016-1019. Stampfer MJ, Willett WC, Speizer FE, et al. Test of the National Death Index. Am J Epidemiol. 1984;119(5):837-839. Bertrand KA, Giovannucci E, Zhang SM, Laden F, Rosner B, Birmann BM. A prospective analysis of body size during childhood, adolescence, and adulthood and risk of non-Hodgkin lymphoma. Cancer Prev Res (Phila). 2013;6(8):864-873. Swerdlow SH, Campo E, Harris NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues.

haematologica | 2018; 103(10)

Lyon, France: IARC Press, 2008. 21. Jaffe ES, Harris NL, Stein H, Vardiman JW. Pathology and Genetics of Tumours of Haematopoietic and Lymphoid Tissues. WHO/IARC Classification of Tumours, 3rd Edition, Volume 3. Lyon: International Agency for Research on Cancer, 2001. 22. Morton LM, Turner JJ, Cerhan JR, et al. Proposed classification of lymphoid neoplasms for epidemiologic research from the Pathology Working Group of the International Lymphoma Epidemiology Consortium (InterLymph). Blood. 2007;110(2):695-708. 23. Turner JJ, Morton LM, Linet MS, et al. InterLymph hierarchical classification of lymphoid neoplasms for epidemiologic research based on the WHO classification (2008): update and future directions. Blood. 2010;116(20):e90-98. 24. Rosner B, Cook N, Portman R, Daniels S, Falkner B. Determination of blood pressure percentiles in normal-weight children: some methodological issues. Am J Epidemiol. 2008;167(6):653-666. 25. Rosner B. Percentage Points for a Generalized ESD Many-Outlier Procedure. Technometrics. 1983;25(2):165-172. 26. Wang M, Spiegelman D, Kuchiba A, et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016;35(5):782-800. 27. Ng LG, Sutherland AP, Newton R, et al. B cell-activating factor belonging to the TNF family (BAFF)-R is the principal BAFF receptor facilitating BAFF costimulation of circulating T and B cells. J Immunol. 2004;173(2):807-817. 28. Rihacek M, Bienertova-Vasku J, Valik D, Sterba J, Pilatova K, Zdrazilova-Dubska L. B-Cell Activating Factor as a Cancer Biomarker and Its Implications in CancerRelated Cachexia. Biomed Res Int. 2015;2015:792187. 29. Thompson N, Isenberg DA, Jury EC, Ciurtin C. Exploring BAFF: its expression, receptors and contribution to the immunopathogenesis of Sjogren's syndrome. Rheumatology (Oxford). 2016;55(9):1548-1555. 30. Smedby KE, Hjalgrim H, Askling J, et al. Autoimmune and chronic inflammatory disorders and risk of non-Hodgkin lymphoma by subtype. J Natl Cancer Inst. 2006;98(1):51-60. 31. Morton LM, Slager SL, Cerhan JR, et al. Etiologic heterogeneity among nonHodgkin lymphoma subtypes: the InterLymph Non-Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr. 2014;2014(48):130-144. 32. Ekstrom Smedby K, Vajdic CM, Falster M, et al. Autoimmune disorders and risk of non-Hodgkin lymphoma subtypes: a

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

pooled analysis within the InterLymph Consortium. Blood. 2008;111(8):40294038. Herishanu Y, Perez-Galan P, Liu D, et al. The lymph node microenvironment promotes B-cell receptor signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011;117(2):563-574. Endo T, Nishio M, Enzler T, et al. BAFF and APRIL support chronic lymphocytic leukemia B-cell survival through activation of the canonical NF-kappaB pathway. Blood. 2007;109(2):703-710. Planelles L, Castillo-Gutierrez S, Medema JP, Morales-Luque A, Merle-Beral H, Hahne M. APRIL but not BLyS serum levels are increased in chronic lymphocytic leukemia: prognostic relevance of APRIL for survival. Haematologica. 2007;92(9):1284-1285. Bojarska-Junak A, Hus I, Chocholska S, et al. BAFF and APRIL expression in B-cell chronic lymphocytic leukemia: correlation with biological and clinical features. Leuk Res. 2009;33(10):1319-1327. Haiat S, Billard C, Quiney C, AjchenbaumCymbalista F, Kolb JP. Role of BAFF and APRIL in human B-cell chronic lymphocytic leukaemia. Immunology. 2006; 118(3):281-292. Strati P, Shanafelt TD. Monoclonal B-cell lymphocytosis and early-stage chronic lymphocytic leukemia: diagnosis, natural history, and risk stratification. Blood. 2015;126(4):454-462. Goto H, Tsurumi H, Takemura M, et al. Serum-soluble interleukin-2 receptor (sIL2R) level determines clinical outcome in patients with aggressive non-Hodgkin's lymphoma: in combination with the International Prognostic Index. J Cancer Res Clin Oncol. 2005;131(2):73-79. Mir MA, Maurer MJ, Ziesmer SC, et al. Elevated serum levels of IL-2R, IL-1RA, and CXCL9 are associated with a poor prognosis in follicular lymphoma. Blood. 2015;125(6):992-998. Goto N, Tsurumi H, Goto H, et al. Serum soluble interleukin-2 receptor (sIL-2R) level is associated with the outcome of patients with diffuse large B cell lymphoma treated with R-CHOP regimens. Ann Hematol. 2012;91(5):705-714. Birmann BM, Breen EC, Stuver S, et al. Population differences in immune marker profiles associated with human T-lymphotropic virus type I infection in Japan and Jamaica. Int J Cancer. 2009;124(3):614621. Skibola CF, Berndt SI, Vijai J, et al. Genomewide association study identifies five susceptibility loci for follicular lymphoma outside the HLA region. Am J Hum Genet. 2014;95(4):462-471.

1687

ARTICLE

Plasma Cell Disorders

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1688-1697

Immunomodulatory drugs downregulate IKZF1 leading to expansion of hematopoietic progenitors with concomitant block of megakaryocytic maturation Ailing Liu,1* Shirong Li,1,2* Vera Donnenberg,3 Jing Fu,1,2 Susanne M. Gollin,4 Huihui Ma,1,5 Caisheng Lu,1,5 Donna B. Stolz,6 Markus Y. Mapara,1,2,5 Sara A. Monaghan7 and Suzanne Lentzsch1,2

Department of Medicine, Division of Hematology/Oncology, University of Pittsburgh School of Medicine and Cancer Institute, PA; 2Division of Hematology/Oncology, College of Physicians and Surgeons, Columbia University, New York, NY; 3Department of Surgery and Pharmaceutical Sciences, University of Pittsburgh School of Medicine and Cancer Institute, PA; 4Department of Human Genetics, University of Pittsburgh Graduate School of Public Health and Cancer Institute, and the University of Pittsburgh Cell Culture and Cytogenetics Facility, PA; 5Columbia Center for Translational Immunology, College of Physicians and Surgeons, Columbia University, New York, NY; 6Department of Cell Biology and Physiology, University of Pittsburgh, PA; 7Department of Pathology, University of Texas Southwestern Medical Center, Dallas, TX, USA 1

*AL and SL contributed equally to this work

ABSTRACT

Correspondence: sl3440@columbia.edu

Received: January 19, 2018. Accepted: June 25, 2018. Pre-published: June 28, 2018. doi:10.3324/haematol.2018.188227 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1688 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1688

he immunomodulatory drugs, lenalidomide and pomalidomide yield high response rates in multiple myeloma patients, but are associated with a high rate of thrombocytopenia and increased risk of secondary hematologic malignancies. Here, we demonstrate that the immunomodulatory drugs induce self-renewal of hematopoietic progenitors and upregulate megakaryocytic colonies by inhibiting apoptosis and increasing proliferation of early megakaryocytic progenitors via down-regulation of IKZF1. In this process, the immunomodulatory drugs degrade IKZF1 and subsequently down-regulate its binding partner, GATA1. This results in the decrease of GATA1 targets such as ZFPM1 and NFE2, leading to expansion of megakaryocytic progenitors with concomitant inhibition of maturation of megakaryocytes. The down-regulation of GATA1 further decreases CCND1 and increases CDKN2A expression. Overexpression of GATA1 abrogated the effects of the immunomodulatory drugs and restored maturation of megakaryocytic progenitors. Our data not only provide the mechanism for the immunomodulatory drugs induced thrombocytopenia but also help to explain the higher risk of secondary malignancies and long-term cytopenia induced by enhanced cell cycling and subsequent exhaustion of the stem cell pool.

Introduction Lenalidomide (LEN, CC-5013) and pomalidomide (POM, CC-4047) are immunomodulatory drugs (IMiDs), analogues of thalidomide, which have several cellular effects including immunomodulatory, anti-angiogenic, anti-inflammatory and anti-proliferative effects.1-3 In multiple myeloma (MM) cells, LEN binds to cereblon and thereby, is able to target two specific B-cell transcription factors, Ikaros family zinc finger proteins 1 and 3 (IKZF1 and IKZF3) for proteasomal degradation4,5 and subsequently affect transcription factors critical for multiple myeloma (MM) growth, such as CCAAT-enhancer-binding protein beta (C/EBPb)6 and IRF4.7 We have shown that IKZF1 is also expressed in CD34+ cells and undergoes degradation after ubiquitination of cereblon when cells are treated with IMiDs.8 LEN is considered a therapeutic breakthrough in the treatment of MM.9 POM is the newest IMiD, and appears to be more potent than LEN in MM.10 However, the use of IMiDs is associated with neutropenia, thrombocytopenia, bone marrow failure and stem cell mobilization.9,11,12 In addition, there is a concern of an increased risk haematologica | 2018; 103(10)

IMiDs block megakaryocytic maturation

of secondary malignancies such as myelodysplastic syndrome and acute leukemia.13-15 Our laboratory has focused on exploring the effects of IMiDs on different hematopoietic lineages. We showed that IMiDs do not exhibit direct stem cell toxicity, but affect lineage commitment.16,17 Downregulation of GATA1 by IMiDs induces a shift into myeloid lineage commitment at the expense of erythroid commitment.16 The downregulation of SPI1 (PU.1), a critical transcription factor for myeloid maturation, leads to maturational arrest with accumulation of immature myeloid precursors, resulting in neutropenia.17 Nevertheless, IMiD-induced thrombocytopenia, a major adverse side effect, is still not understood. Here, we investigated the effect of IMiDs on megakaryopoiesis after thrombopoietin (TPO) stimulation. We showed that IMiDs induce self-renewal and proliferation of megakaryocytic progenitors by down-regulating GATA1 as a consequence of the degradation of its binding partner IKZF1. This is accompanied by decreased ZFPM1/FOG-1 and NFE2 expression, leading to inhibition of megakaryocyte maturation. Our data further demonstrated that IMiD induced a decrease in CCND1/cyclin D1 accompanied by an increase in CDKN2A/p16, resulting in the maturational arrest of megakaryocytes (Mks). The effects of IMiDs on megakaryopoiesis could be abrogated by overexpression of GATA1. This study provides for the first-time mechanistic insight into how IMiDs induce thrombocytopenia and potentially contribute to secondary hematologic malignancies by sustained cell proliferation.

Colony-forming assay Colony-forming assays were performed as described previously.16, 17 For CD34+ cells self-renewal assessment, CD34+ cells were seeded in serum-free HPGM supplemented with rhIL-3, rhIL-6 and rhSCF as mentioned above and cultured in the presence of IMiDs or DMSO. After 14 days in culture, the CD34+ cells of each group (vehicle, LEN and POM) were purified using the CD34+ cell isolation kit and were plated in MethoCult H4434 medium (StemCell Technologies) for 14 days (without vehicle, LEN or POM).

Transmission electron microscopy To identify megakaryocytic precursors by transmission electron microscopy (TEM), we labeled the precursors with CD61 magnetic beads (Miltenyi Biotec). Treated cells were fixed in 2.5% glutaraldehyde in 0.1 M PBS, pH 7.4, for 1 h and post-fixed in aqueous 1% OsO4, 1% K3Fe(CN)6 for 1 h. The pellet was dehydrated through a graded series of 30–100% ethanol, 100% propylene oxide and then infiltrated in 1:1 mixture of propylene oxide/Polybed 812 epoxy resin (Polysciences) for 1 h. Ultrathin (60 nm) sections were collected and counterstained with uranyl acetate and lead citrate and observed by using a JEOL JEM 1011 transmission electron microscope (JEOL) with a bottom mount AMT 2k digital camera (Advanced Microscopy Techniques).

Statistical analyses Statistical significance of differences between group means (P<0.05) was established using Student's t test for two group comparisons. Multiple comparisons were performed using one-way ANOVA with the Bonferroni step down correction of P. Error bars on graphs reflect standard error of the mean.

Methods Results CD34+ cells isolation and culture Primary CD34+ cells were isolated from discarded peripheral blood leukapheresis products after stem cell mobilization of consenting healthy individuals and MM patients. We tested the CD34+ cells from MM patients or healthy individuals in cell proliferation and colony assays and no difference was observed. Data are not shown. The Institutional Review Boards (IRBs) of the University of Pittsburgh, Pittsburgh, PA and Columbia University, New York, NY approved all studies. Purified CD34+ cells were grown in serum-free hematopoietic growth medium (HPGM) (Lonza) supplemented with 10 ng/mL recombinant human thrombopoietin (rhTPO), 10 ng/mL recombinant human interleukin-3 (rhIL-3), 10 ng/mL recombinant human interleukin-6 (rhIL-6), and 50 ng/mL recombinant human stem cell factor (rhSCF). All cytokines were purchased from PeproTech as described previously.16, 17 LEN and POM (Sigma Aldrich) in DMSO were diluted in culture medium and added daily. Cell viability was measured by trypan blue exclusion, and cell proliferation was quantified by manual cell counting every 2 days during culture.

Megakaryocytic colony assays Megakaryocytic colony forming unit (CFU-Mk) assays were generated using the MegaCult™-C Staining Kit (StemCell Technologies) according to the manufacturer’s instructions. The number of CFU-Mk was determined using an anti-CD41 antibody, an alkaline phosphatase detection system and by counterstaining with Evan’s Blue. The total numbers of colonies were counted on day 12 of culture. The colonies were subdivided by colony size: small (3-20 cells/colony), medium (21-49 cells/ colony), or large (≥ 50 cells/colony). haematologica | 2018; 103(10)

IMiDs induce self-renewal and expansion of early myeloid and megakaryocytic progenitors by blocking apoptosis and enhancing proliferation To determine the basis for thrombocytopenia after treatment with IMiDs, we analyzed the effect of POM on megakaryocytic colony formation of CD34+ cells. Under specific conditions allowing the development of CFU-Mk, POM significantly (P<0.001) increased the numbers of CFU-Mk in comparison to vehicle, with an increase from 53.2 (± 2.56) colonies (vehicle) to 144.8 (± 4.74) colonies (POM). The up-regulation of CFU-Mk was especially evident (P<0.001) in medium/large CFU-Mk (34.4 ± 4.03 colonies in vehicle versus 113.8 ± 5.91 colonies in POM), which arose from more primitive progenitors (Figure 1A). We also studied the effects of LEN and POM on proliferation and apoptosis of hematopoietic progenitors. Using CD34+ cells, IMiDs (LEN, P<0.05; POM, P<0.001) increased the absolute cell number in the cultures up to four-fold (LEN) and 10-fold (POM) after 2 weeks (Figure 1B). The cell expansion was not only the result of decreased apoptosis (PI+ cells, day 7: vehicle 58.0%, LEN 24.2%, and POM 6.4%; day 14: vehicle 68.0%, LEN 24.3%, and POM 8.3%; P<0.05), but also of increased mitosis according to the cell cycle analysis results. The proportion of cells in S phase increased (P<0.05) after CD34+ cells were treated with IMiDs on day 7: vehicle 8.3%, LEN 16.1%, POM 19.4% and day 14: vehicle 7.0%, LEN 14.0%, POM 15.8% (Figure 1C). Investigations of the effects of IMiDs on cell cycle regulation of progenitor cells revealed that S-phase cells were increased after 1689

A. Liu et al. A

**P<0.001

*P<0.05 **P<0.001

D P<0.05 P<0.05

IMiDs treatments (day 3: vehicle 37.9%, LEN 42.2%, POM 42.5% in CD34+ cells (Figure 1D).

IMiDs inhibit megakaryocytic differentiation by blocking endomitosis and maturation To characterize the CD34+ cells that exhibit increased proliferation and decreased apoptosis in long-term cultures with IMiDs, we performed flow cytometry after 14 days of culture in the presence of IL-3/IL-6/SCF with or without TPO. Flow cytometry revealed a strong induction of early CD34+ cells (CD34+CD38-) after 14 days of treatment with POM compared to vehicle (Figure 2A). Among these early CD34+ cells, more than 40% were double positive for myeloid and megakaryocytic markers, expressing CD33+ and CD41+, after treatment with POM. The development of the immature myeloid/megakaryocytic “hybrids” (CD34+/CD38–/CD41+/CD33+) was independent of TPO (vehicle 7.4%, POM 42.6%, POM+TPO 46.7%). Interestingly, the mean fluorescence intensity 1690

Figure 1. IMiDs increase megakaryocytic colony formation and induce self-renewal and expansion of hematopoietic progenitors. (A) Purified CD34+ cells were cultured with POM (pomalidomide) or DMSO as vehicle control using MegaCult-C assay to analyze formation of megakaryocytic colonies (CFU-Mk). Data shown are mean ± SEM from triplicates; Student’s t-test was performed, P-values are two-sided, **P<0.001 (B) Proliferation profiles of CD34+ cells expanded in serum-free HPGM hematopoietic growth medium supplemented with 10 ng/mL TPO with or without 10 µM IMiDs (LEN, lenalidomide or POM). Values shown are mean ± SEM of 7 separate cultures. Pvalues were calculated via one-way ANOVA with Bonferroni post hoc test. (*P<0.05 compared to vehicle; **P<0.001 compared to vehicle). (C) CD34+ cells cultured as described above were further analyzed by flow cytometry using propidium iodide (PI) staining for apoptosis and cell cycle at days 7 and 14 of culture. Cell pellets were stained for 30 minutes with equal volumes of phosphate-buffered saline (PBS) containing 0.1 mg/mL PI and 0.6% Nonidet P-40 and 2 mg/mL RNAse. Cell cycle analysis was performed on a Beckman Coulter CyAN 9-color High Speed Flowcytometer, and data were analyzed by using Summit 4.3 software (Dako¬Cytomation) (left). P-values were calculated via one-way ANOVA with Bonferroni corrections. Right side is statistic result of triplicates. (D) CD34+ cells were treated with DMSO (0.01%), LEN, or POM (1 mM) for 3 days. The cells gated with CD34+ and analyzed by flow cytometry using propidium iodide (PI) staining for cell cycle. The result shown here is representative of two independent experiment.

(MFI) of CD33 in the CD34+CD38- immature hybrid cells strongly increased when cultured with TPO (MFI: vehicle 3.3, POM 5.9, POM+TPO 14.4). Furthermore, in the presence of POM, we were able to culture and expand CD34+ cells for up to 4 months (Figure 2B). Characterization of cells from long-term cultures by multicolor flow cytometry revealed two cell populations. First, 6.6% of the cells were CD45+34+33+11b+41+61+. These “hybrid” cells maintained CD34+ expression with concomitant expression of myeloid (CD33, CD11b) and megakaryocytic (CD41 and CD61) markers. Second, 86.6% of the cells exhibited a more mature phenotype with loss of CD34 expression, but maintained co-expression of myeloid and megakaryocytic markers, CD45+34-33+11b+41+61+. Despite the culture conditions favoring thrombopoiesis, long-term cultured hematopoietic cells notably still expressed the myeloid markers CD11b and CD33, suggesting that IMiDs induce myeloid development. To address whether IMiDs have a sustained effect on expansion and selfhaematologica | 2018; 103(10)

IMiDs block megakaryocytic maturation

C *P<0.05 **P<0.001

Figure 2. IMiDs inhibit megakaryocytic differentiation by blocking endomitosis and maturation. (A) Upper panel: Flow cytometry analysis revealed a strong induction of CD34+CD38–cells (early progenitors) after 14 days of treatment with POM +/- TPO compared to vehicle. Lower panel: Early CD34+CD38– progenitors were further analyzed for CD33+ and CD41+ expressions. Lower right panel: The mean fluorescence intensity (MFI) of CD33 in these CD34+CD38– immature hybrid cells dramatically increased with TPO. (B) In long-term cultures, CD34+ cells were maintained for up to 4 months and multicolor flow cytometry identified 2 cell populations: CD45+34+33+11b+41+61+ hybrid cells (6.6%) and a more mature population of CD45+34-41+61+ cells (86.6%). Without POM, the control cultured cells could be maintained only for up to 3 weeks and were therefore not available for comparison. Data are from one experiment. (C) Purified CD34+ cells were cultured in serum-free HPGM hematopoietic growth medium with DMSO or IMiDs. After culturing for 14 days, CD34+ cells from each group (vehicle, LEN and POM) were selected by immunomagnetic beads and were plated in MethoCult for colony formation assays (with either vehicle, LEN or POM). Data are from one experiment with colonies quantified in triplicate wells. *P<0.05 compared to Vehicle; **P<0.001 compared to Vehicle. Data were compared by one-way ANOVA with Bonferroni post-test. (D) CD34+ cells were cultured in serum-free HPGM hematopoietic growth medium with TPO to induce megakaryopoiesis with or without LEN, POM or DMSO as vehicle. After 7, 10 and 14 days, flow cytometry analysis showed an increase of immature Mks (CD41a+/CD42b–), while more mature Mks (CD41a+/CD42b+) decreased with IMIDs treatment. The result shown here is one representative experiment of triplicates. (E) Morphology of Mks, derived from CD34+ cells grown in serum-free HPGM hematopoietic growth medium with TPO with or without IMiDs at day 7 and 10 of culture. Electron micrographs of three characteristic Mks from different groups (vehicle, LEN and POM) are shown. Mks were identified by CD61+ magnetic bead labeling. Mks in LEN and POM group show features of immaturity, including scant cytoplasm, large nuclei, and a minimal demarcation membrane system. Images are from one experiment representative of two independent experiments each of which included triplicates.

renewal beyond direct exposure, we first cultured CD34+ cells in liquid cultures under the conditions described above. After 14 days, CD34+ cells were selected, and subjected to colony formation assays without adding IMiDs. Even without the continued presence of IMiDs, total colony formation was still significantly increased [vehicle 120, LEN 256 (P<0.05), POM 270 (P<0.001)]. In accordance with prior data,17 we again observed a shift to myeloid lineage commitment with increased myeloid colonies (CFU-G, CFU-GM and CFU-GEMM) at the expense of erythroid colonies (P<0.05; Figure 2C). These data suggested that IMiDs persistently affect self-renewal haematologica | 2018; 103(10)

and lineage commitment of CD34+ cells, resulting in maintenance and expansion of the progenitor cell pool. Our finding showed that IMiDs-induced development of megakaryocytic precursors appears to contrast with the thrombocytopenia in patients caused by IMiDs. Since our previous studies showed that IMiDs are not directly toxic to bone marrow hematopoietic cells,16 we hypothesized that these effects are induced by maturational arrest. To investigate the effects of IMiDs on the maturation of Mks, we evaluated different megakaryocytic cell populations by flow cytometry. CD41a is a lineage marker of Mks throughout all stages of differentiation, whereas CD42b 1691

A. Liu et al. expression is expressed by more mature Mks.18 Hence, CD41a+/CD42b– represents immature Mks, and CD41a+/CD42b+ represents more mature Mks. Flow cytometric analyses showed that the proportion of immature CD41a+/CD42b– cells was significantly increased in the presence of IMiDs (day 7: vehicle 15.8%, LEN 23.0 %, POM 22.9%; day 10: vehicle 32.1%, LEN 50.6%, POM 51.0%; day 14: vehicle 49.7%, LEN 64.7%, POM 66.0%; P<0.05). This difference was visible as soon as day 7 of culture and was maintained throughout the entire culture period (Figure 2D). To confirm the immature and dysplastic morphology, we performed transmission electron microscopy (TEM) after 10 days of IMiD treatment. Mks were identified by CD61-microbead labeling. Again, compared with vehicle, IMiD-treated Mks exhibited immature

features, including decreased size, less cytoplasm, and a heterogeneously dilated and abnormally distributed demarcation membrane system (DMS) in the cytoplasm. (Figure 2E)

IMiD decreased protein expression of GATA1 and its targets in Mk progenitors To gain insight into the mechanism of inhibition of megakaryocytic maturation by IMiDs, we evaluated several transcription factors involved in the regulation of megakaryocytic differentiation and maturation. Failure of terminal differentiation and excessive proliferation of Mks have been described to occur in the absence of GATA1.19 By RT-PCR we found significantly (P<0.05) decreased GATA1 levels in Mks treated with IMiDs

Figure 3. IMiD decreased protein expression of GATA1 and its targets in Mk progenitors. CD34+ cells were cultured in serum-free HPGM hematopoietic growth medium supplemented with 10 ng/mL TPO to initiate megakaryopoiesis with or without 10 mM IMiDs for the indicated times. (A) Real-time PCR analysis was applied to determine the levels of GATA1 mRNA. The result shown here is one representative experiment of triplicates. (B) Immunofluorescence microscopy was performed to examine the expression of GATA1 on CD34+ cells cultured in serum-free HPGM medium containing 10ng/mL TPO with 10 mM LEN, POM or 0.1% DMSO as vehicle control at day 6. Cells were stained with an antibody that recognize the N terminus of GATA1 (red) and counterstained with DAPI (blue) to visualize the nucleus. Both stainings were merged and revealed a loss of GATA1 when treated with IMiDs. Images shown here is one representative experiment of triplicates. (C) Western blot analysis was applied to determine the levels of GATA1 and IKZF1 protein. β-actin was used as the loading control. Data are representative of three independent experiments. (D) Purified CD34+ cells were cultured in serum-free HPGM hematopoietic growth medium containing 10 ng/mL TPO with 10 µM LEN, POM or 0.1% DMSO as vehicle control for 6 days and 9 days. The expression of the indicated proteins was analyzed by Western blotting. b-actin was used as the loading control. Data are representative of three independent experiments.

1692

haematologica | 2018; 103(10)

IMiDs block megakaryocytic maturation

(Figure 3A). The decrease of GATA1 by IMiDs was confirmed by immunofluorescence of megakaryocytic progenitors (Figure 3B) and Western blot analysis (Figure 3C). We have shown that IMiDs induce cereblone-ubiquitination with subsequent IKZF1 degradation in CD34+ cells8. ZFPM1 (FOG-1), similar to its interacting partner, GATA1, is required for normal differentiation of erythroid precursors and Mks.20 Moreover, most GATA1regulated events require GATA1 to bind to ZFPM1.21,22 Subsequently, we asked if IMiDs also modulate the expression of ZFPM1. Figure 3D shows that in IMiDtreated CD34+ cells, ZFPM1 is almost completely abrogated after 9 days of treatment with kinetics similar to those of GATA1. GATA1 and ZFPM1 synergistically activate the p45 NFE2 promoter,23,24 which is essential for

late maturation of Mks and platelet formation. Accordingly, we found that expression of NFE2 were down-regulated in IMiD-treated cells (Figure 3D). It is known that polyploidy formation in Mks depends on the expression of cyclin D isotypes, and that cyclin D1 is a direct target of GATA1.25 In accordance with this, we found that IMiD-induced downregulation of GATA1 was associated with decreased cyclin D1 (Figure 4D), contributing to the inhibition of maturation. The CDK inhibitor, p16, potently inhibits endomitosis of Mks25 and is decreased upon differentiation of Mks.26 Analysis of p16 by Western blot revealed that p16 was up-regulated by IMiD treatment (Figure 4E), suggesting that the loss of GATA1 induced a cascade of inhibitors of megakaryocytic maturation.

B P<0.05 P<0.05

haematologica | 2018; 103(10)

Figure 4. IKZF1 mediates the IMiD-induced inhibition of megakaryocytic maturation. (A) IKZF1 shRNA #1 (shIKZF1-1), #2 (shIKZF1-2) or control shRNA (shCNTL) transduced CD34+cell lysates were analyzed by western blotting to compare the levels of IKZF1, CRBN and GATA1. The result shown here is one representative experiment of triplicates. (B) CD34+ cells were transduced using a lentivirus carrying the control shRNA (shCNTL), IKZF1-shRNA #1 (shIKZF1-1), or IKZF1-shRNA #2 (shIKZF1-2) sequence and cultured in serum-free HPGM hematopoietic growth medium with TPO to induce megakaryopoiesis. At 6 days after transduction, the cells were sorted with GFP and subsequently lysed and analyzed for IKZF1 and GATA1 mRNA levels using qRT-PCR. The result shown here is one representative experiment of triplicates. (C) CD34+ cells were transduced using a lentivirus carrying the control shRNA (shCNTL), IKZF1-shRNA #1 (shIKZF1-1), or IKZF1-shRNA #2 (shIKZF1-2) sequence. CD34+ cells were cultured in serum-free HPGM hematopoietic growth medium with TPO to induce megakaryopoiesis. After 10 days, flow cytometry analyzed the cells gated with GFP and showed an increase of immature Mks (CD41a+/CD42bâ&#x20AC;&#x201C;) with IKZF1 knockdown. The result shown here is one representative experiment of triplicates. (D) CD34+ cells were treated with DMSO (0.01%) or POM (1 mM) for 24 h and cell lysates were analyzed by chromatin immunoprecipitation (CHIP) using IKZF1 antibody. Control IgG was used as a negative control. The precipitated DNA fragments were subjected to qRT-PCR analysis with primers amplifying the GATA1 promoter. The result shown here is representative of two independent experiments.

1693

A. Liu et al. A

*P<0.001

Figure 5. Overexpression of GATA1 reversed the effects of IMiDs on megakaryopoiesis. (A) Schematic diagram of the structure of GATA1pLenti V5 topo expression constructs. CF and NF represent C- and N-fingers of GATA1; R1, R2, R3 represent three regions of the transactivation domain of GATA1 (modified from reference45). (B) The expression of GATA1 in either EV or GATA1 transfected CD34+ cells were analyzed by Western blotting. CD34+ cells transfected with EV or GATA1 were cultured in serum-free HPGM hematopoietic growth medium containing 10 ng/mL TPO with either vehicle, LEN or POM for 6 days. b-actin was used as the loading control. Data are representative of three independent experiments. (C) Purified CD34+ cells were transfected with GATA1 or EV and cultured in serum-free HPGM hematopoietic growth medium containing 10 ng/mL TPO to induce megakaryocytic development with LEN, POM or DMSO as vehicle control for 6 days. The expression of the indicated proteins was analyzed by Western blotting. b-actin was used as the loading control. Data are representative of three independent experiments. (D) EV and GATA1 transfected CD34+ cells were subjected to colony formation assays using standard MethoCult assays with and without IMiDs. Overexpression of GATA1 abrogated the IMiD-induced upregulation of myeloid lineage commitment and rescued the development of BFU-E. Data shown are mean Âą SEM from triplicates; **P<0.001, compared with Vehicle group.

IKZF1 mediates the IMiD-induced inhibition of megakaryocytic maturation Previous studies have demonstrated that lenalidomide induces binding of IKZF1 and IKZF3 to CRBN and promotes their ubiquitination and degradation.27,28 Since IKZF1 is required for the development of the erythroid lineage,29 we were interested in the effects of IMiDs on the interaction between IKZF1 and GATA1 in CD34+cells. So, we examined the role of IKZF1 in POM-induced inhibition of megakaryocytic maturation. First, we confirmed that IKZF1 regulates GATA1 expression in CD34+ cells by knockdown of IKZF1 in CD34+ cells. Knockdown of IKZF1 resulted in decreased GATA-1 protein expression, further suggesting that GATA1 is under IKZF1 regulation (Figure 4A). In IKZF1 knockdown cells, the mRNA level of GATA-1 was significantly decreased, suggesting that GATA-1 expression is regulated by IKZF1 (Figure 4B). To examine the effects of IKZF1 in megakaryopoiesis, we performed flow cytometric analyses. In IKZF1 knockdown cells, the proportion of immature CD41a+/CD42bâ&#x20AC;&#x201C; cells was significantly increased, suggesting that IKZF1 downregulation is critical for the POM-induced maturational arrest (Figure 4C). To find how GATA1 is regulated 1694

at a transcriptional level, we analyzed GATA-1 promoter sequences and several potential IKZF1 binding sites (Figure 4D upper). Indeed, In CHIP assays, we confirmed that IKZF1 binds directly to the GATA1 promoter area. Treatment of CD34+ cells with POM completely inhibited IKZF1 binding to the GATA1 promotor due to the decreased IKZF1 levels (Figure 4D bottom). Our findings showed that IMiDs promote CRBN-dependent degradation of IKZF1 protein in CD34+ cells and decrease GATA1.

Overexpression of GATA1 abrogated the effects of IMiDs on megakaryopoiesis To further confirm the critical role of GATA1 in inhibiting maturation of Mks by IMiDs, we stably overexpressed GATA1 in human CD34+ cells using the pLenti V5 GATA1 expression vector system (Figure 5A). Overexpression (OE) of GATA1 prevented its downregulation by IMiDs compared to control cells transfected with empty vector (EV) alone (Figure 5B). Concomitant with the overexpression of GATA1, GATA1 co-factors and targets, such as ZFPM1, NFE2 and cyclin D1, remained stably expressed despite IMiDs treatment (Figure 5C). Thus, GATA1 appeared to play a critical role in the IMiDs-induced inhihaematologica | 2018; 103(10)

IMiDs block megakaryocytic maturation

Figure 6. Downregulation of GATA1 by IMiDs induces renewal and expansion of hematopoietic progenitors with a concomitant block of megakaryocytic maturation. Treatment with IMiDs induces self-renewal of hematopoietic progenitors and upregulates megakaryocytic colonies (CFU-Meg) by inhibiting apoptosis and increasing proliferation of early megakaryocytic progenitors. IMiDs down-regulate the transcription factor GATA1 and thereby affect transcription factors and cell cycle regulators controlled by GATA1. The subsequent decrease of ZFPM1 and NFE2 leads to expansion of megakaryocytic progenitors with concomitant inhibition of maturation. A decrease of cyclin D1 and an increase of p16 results in the block of megakaryocytic maturation (Adapted from Figure 7 of reference 25, with permission of Dr. John D. Crispino).

bition of megakaryocytic maturation. Most importantly, OE of GATA1 widely abrogated the effects of IMiDs on lineage commitment (Figure 5D). The numbers of BFU-E colonies increased significantly (P<0.001), whereas the numbers of CFU-G/GM colonies in the IMiDs group decreased significantly (P<0.001) when compared to EVtransfected cells. OE of GATA1 abrogated the IMiDsinduced shift of lineage commitment toward myelopoiesis at the expense of erythropoiesis/megakaryopoiesis.

Discussion Despite the fact that IMiDs are not directly cytotoxic, their use is associated with severe thrombocytopenia (grade 3/4) in up to 25% of patients with MM.10,30 The mechanism for induction of thrombocytopenia is unknown. Here, we determined that IMiDs dramatically expand CD34+ hematopoietic progenitors in liquid cultures and significantly induce megakaryopoiesis and megakaryocytic colony formation. Interestingly, IMiDs generated mainly immature CD41a+ /CD42bâ&#x20AC;&#x201C; megakaryocytes. This was further confirmed by transmission electron microscopy, revealing structural abnormalities of Mks haematologica | 2018; 103(10)

that also suggested a maturational block in megakaryopoiesis.31,32 More strikingly, we were able to maintain a small pool of CD34+ cells for as long as 4 months in liquid culture. Besides CD34+ expression, the long-term cultured cells maintained concomitant expression of myeloid (CD33, CD11b) and megakaryocytic (CD41, CD61) markers, reflecting a phenotype that is similar to acute megakaryoblastic leukemia (AMKL) with GATA1 mutations33,34. GATA1 is a transcription factor critical for development of erythroid and megakaryocytic cells. GATA1 mutations in humans act as a dominant leukemogenic oncogene in megakaryocytic progenitors and cause inherited thrombocytopenia.31 In our studies, the IMiDsinduced proliferation of hematopoietic progenitors and inhibition of maturation of Mks were associated with a loss of GATA1, ZFPM1 (FOG-1) and NFE2 expression. ZFPM1 acts as a cofactor for GATA1 and provides a paradigm for the regulation of cell type-specific gene expression by GATA1.22 GATA1 also drives the expression of another important transcription factor, NFE2 that exists as a heterodimer comprised of p45 and p18 subunits. p45Nfe2-/- mice present with an increase in Mks, a marked defect in maturation and profound thrombocytopenia.35,36 This indicates that terminal maturation of Mks depends heavily on the GATA1/NFE2 axis. In addition to promot1695

A. Liu et al.

ing differentiation of Mks, GATA1 also leads ultimately to cessation of cell proliferation.19, 37 This is consistent with our findings showing that IMiDs-induced loss of GATA1 expression in Mks not only inhibits maturation, but also leads to excessive proliferation of megakaryocytic progenitors. GATA1 also regulates numerous CDKs such as cyclin D1 and CDK inhibitors.25 Moreover, overexpression of cyclin D1/CDK4 in GATA1-deficient Mks restored their growth and polyploidization.25 Correspondingly, we observed that downregulating GATA1 in Mks by IMiDs resulted in decreased cyclin D1 expression. Our findings of the role of IKZF1 in regulating GATA1 expression are in accordance with results by Dijon et al., who reported that lentivirally induced Ik6 (a dominant negative isoform of IKZF1) overexpression resulted in decreased expression of GATA1.29 Cell-cycle regulation is an important mechanism governing the long-term selfrenewal potential of HSCs.38 Therefore, long-term treatment of patients with IMiDs might induce a pool of constantly cycling progenitors, leading to premature exhaustion of the stem cell pool. Indeed, patients receiving longterm treatment with IMiDs very often exhibit a hypocellular bone marrow associated with cytopenias.39 Interestingly, Malinge and colleagues reported that IKZF1 knockout mice showed increased megakarayopoiesis. Consistent with this phenotype, studies using mice acute megakaryoblastic leukemia (AMKL) cell line 6133 showed that IKZF1 suppresses megakaryopoiesis by negatively regulating GATA1.40 In contrast, our experiments found that the knockdown of IKZF1 resulted in decreased GATA1 protein expression in human CD34+ cells. The mechanism of IKZF1 in regulating GATA1 in mice malignant cells may be different from that in human hematopoietic progenitor cells. Interestingly, Kronke et al. reported that Lenalidomide induces ubiquitination and degradation of Casein kinase 1α (CK1α) in del(5q) MDS.41 Furthermore, in a murine model with conditional inactivation of Csnk1a1, Schneider et al. demonstrated that Csnk1a1 haploinsufficiency induces hematopoietic stem cell expansion.42 This is in accordance with our data showing that both pomalidomide and lenlidomide potently downregulate CK1α in hematopoietic progenitors, although pomalidomide exhibits a slightly lower efficiency compared to Lenalidomide (Online Supplementary Figure S1).

References 1. Bartlett JB, Dredge K, Dalgleish AG. The evolution of thalidomide and its IMiD derivatives as anticancer agents. Nat Rev Cancer. 2004;4(4):314-322. 2. Chang DH, Liu N, Klimek V, et al. Enhancement of ligand-dependent activation of human natural killer T cells by lenalidomide: therapeutic implications. Blood. 2006;108(2):618-621. 3. Quach H, Ritchie D, Stewart AK, et al. Mechanism of action of immunomodulatory drugs (IMiDS) in multiple myeloma. Leukemia. 2010;24(1):22-32. 4. Kronke J, Udeshi ND, Narla A, et al. Lenalidomide causes selective degradation of IKZF1 and IKZF3 in multiple myeloma cells. Science. 2014;343(6168):301-305.

1696

GATA1 overexpression allowed development of a more mature Mk phenotype. In addition, it preserved expression of megakaryocyte-specific transcription factors such as NFE2 and ZFPM1 and of cell cycle regulators, including cyclin D1, despite IMiDs treatment, which further indicates that an IMiDs-induced decrease in GATA1 critically affects pathways involved in self-renewal, cell cycle regulation and lineage commitment of CD34+ cells. GATA1 mutations resulting in functional silencing have been found in patients with thrombocytopenia or megakaryocytic acute leukemia.32,43 Therefore, our data suggest that the IMiDs-induced downregulation of IKZF1 and GATA1 favors myeloid lineage commitment and maintains a pool of immature cycling CD34+ cells without maturation, which may lead to stem cell exhaustion. Furthermore, downregulation of GATA1 results in an imbalance of key cellular and molecular regulators that block Mks from continued maturation, likely impeding platelet release and ultimately resulting in thrombocytopenia. sAt this moment, the clinical relevance of our in vitro findings is not entirely clear. However, prolonged treatment with IMiDs has been shown to be associated with an increased risk of MDS/AML and ALL, especially in patients treated with alkylating agents like Melphalan.44 It is therefore possible that the IMiDs-induced increased number of cycling CD34+ cells may enhance the probability of acquiring secondary DNA damage and leukemogenic events induced by other drugs. Hence, in vivo testing of various treatment combinations with IMiDs to explore and potentially predict leukemogenic effects of certain combinations is needed. Acknowledgments The authors would like to thank Mr. Dale Lewis at the University of Pittsburgh Cell Culture and Cytogenetics Facility for excellent FISH analysis. We would also like to thank Dr. Griffin P. Rodgers at the Molecular and Clinical Hematology Branch, National Heart, Lung, and Blood Institute, Bethesda, MD for providing the pLenti V5 GATA1 expression vectors and empty vectors. Funding This study was supported by a grant from the LLS and R01CA175313 (SL). MYM, CL, HM were supported in part by RO1 HL093716.

5. Chapman MA, Lawrence MS, Keats JJ, et al. Initial genome sequencing and analysis of multiple myeloma. Nature. 2011; 471(7339):467-472. 6. Li S, Pal R, Monaghan SA, et al. IMiD immunomodulatory compounds block C/EBP{beta} translation through eIF4E down-regulation resulting in inhibition of MM. Blood. 2011;117(19):5157-5165. 7. Lopez-Girona A, Heintel D, Zhang LH, et al. Lenalidomide downregulates the cell survival factor, interferon regulatory factor4, providing a potential mechanistic link for predicting response. Br J Haematol. 2011; 154(3):325-336. 8. Li S, Fu J, Mapara M, Lentzsch S. IMiD® compounds affect the hematopoiesis via CRBN dependent degradation of IKZF1 protein in CD34+ Cells. Blood. 2014;

124(21):418-418. 9. Weber DM, Chen C, Niesvizky R, et al. Lenalidomide plus dexamethasone for relapsed multiple myeloma in North America. N Engl J Med. 2007;357(21):21332142. 10. Lacy MQ, Allred JB, Gertz MA, et al. Pomalidomide plus low-dose dexamethasone in myeloma refractory to both bortezomib and lenalidomide: comparison of 2 dosing strategies in dual-refractory disease. Blood. 2011;118(11):2970-2975. 11. Mazumder A, Kaufman J, Niesvizky R, Lonial S, Vesole D, Jagannath S. Effect of lenalidomide therapy on mobilization of peripheral blood stem cells in previously untreated multiple myeloma patients. Leukemia. 2008;22(6):1280-1281; author reply 1281-1282.

haematologica | 2018; 103(10)

IMiDs block megakaryocytic maturation

12. Dimopoulos M, Spencer A, Attal M, et al. Lenalidomide plus dexamethasone for relapsed or refractory multiple myeloma. N Engl J Med. 2007;357(21):2123-2132. 13. Attal M, Lauwers-Cances V, Marit G, et al. Lenalidomide maintenance after stem-cell transplantation for multiple myeloma. N Engl J Med. 2012;366(19):1782-1791. 14. McCarthy PL, Owzar K, Hofmeister CC, et al. Lenalidomide after stem-cell transplantation for multiple myeloma. N Engl J Med. 2012;366(19):1770-1781. 15. Palumbo A, Hajek R, Delforge M, et al. Continuous lenalidomide treatment for newly diagnosed multiple myeloma. N Engl J Med. 2012;366(19):1759-1769. 16. Koh KR, Janz M, Mapara MY, et al. Immunomodulatory derivative of thalidomide (IMiD CC-4047) induces a shift in lineage commitment by suppressing erythropoiesis and promoting myelopoiesis. Blood. 2005;105(10):3833-3840. 17. Pal R, Monaghan SA, Hassett AC, et al. Immunomodulatory derivatives induce PU.1 down-regulation, myeloid maturation arrest, and neutropenia. Blood. 2010; 115(3):605-614. 18. Poirault-Chassac S, Six E, Catelain C, et al. Notch/Delta4 signaling inhibits human megakaryocytic terminal differentiation. Blood. 2010;116(25):5670-5678. 19. Vyas P, Ault K, Jackson CW, Orkin SH, Shivdasani RA. Consequences of GATA-1 deficiency in megakaryocytes and platelets. Blood. 1999;93(9):2867-2875. 20. Cantor AB, Katz SG, Orkin SH. Distinct domains of the GATA-1 cofactor FOG-1 differentially influence erythroid versus megakaryocytic maturation. Mol Cell Biol. 2002;22(12):4268-4279. 21. Wilkinson-White L, Gamsjaeger R, Dastmalchi S, et al. Structural basis of simultaneous recruitment of the transcriptional regulators LMO2 and FOG1/ZFPM1 by the transcription factor GATA1. Proc Natl Acad Sci USA. 2011;108(35):14443-14448. 22. Tsang AP, Visvader JE, Turner CA, et al. FOG, a multitype zinc finger protein, acts as a cofactor for transcription factor GATA1 in erythroid and megakaryocytic differentiation. Cell. 1997;90(1):109-119. 23. Takayama M, Fujita R, Suzuki M, et al. Genetic analysis of hierarchical regulation

haematologica | 2018; 103(10)

24.

25.

26.

27.

28.

29.

30.

31.

32.

33. 34. 35.

for Gata1 and NF-E2 p45 gene expression in megakaryopoiesis. Mol Cell Biol. 2010; 30(11):2668-2680. Wang X, Crispino JD, Letting DL, Nakazawa M, Poncz M, Blobel GA. Control of megakaryocyte-specific gene expression by GATA-1 and FOG-1: role of Ets transcription factors. EMBO J. 2002;21(19):5225-5234. Muntean AG, Pang L, Poncz M, Dowdy SF, Blobel GA, Crispino JD. Cyclin D-Cdk4 is regulated by GATA-1 and required for megakaryocyte growth and polyploidization. Blood. 2007;109(12):5199-5207. Furukawa Y, Kikuchi J, Nakamura M, Iwase S, Yamada H, Matsuda M. Lineage-specific regulation of cell cycle control gene expression during haematopoietic cell differentiation. Br J Haematol. 2000; 110(3):663-673. Kronke J, Udeshi ND, Narla A, et al. Lenalidomide causes selective degradation of IKZF1 and IKZF3 in multiple myeloma cells. Science. 2014;343(6168):301-305. Lu G, Middleton RE, Sun HH, et al. The myeloma drug lenalidomide promotes the cereblon-dependent destruction of Ikaros proteins. Science. 2014;343(6168):305-309. Dijon M, Bardin F, Murati A, Batoz M, Chabannon C, Tonnelle C. The role of Ikaros in human erythroid differentiation. Blood. 2008;111(3):1138-1146. Raza A, Reeves JA, Feldman EJ, et al. Phase 2 study of lenalidomide in transfusiondependent, low-risk, and intermediate-1 risk myelodysplastic syndromes with karyotypes other than deletion 5q. Blood. 2008; 111(1):86-93. Li Z, Godinho FJ, Klusmann JH, GarrigaCanut M, Yu C, Orkin SH. Developmental stage-selective effect of somatically mutated leukemogenic transcription factor GATA1. Nat Genet. 2005;37(6):613-619. Wechsler J, Greene M, McDevitt MA, et al. Acquired mutations in GATA1 in the megakaryoblastic leukemia of Down syndrome. Nat Genet. 2002;32(1):148-152. Choi JK. Hematopoietic disorders in Down syndrome. Int J Clin Exp Pathol. 2008;1 (5):387-395. Lorsbach RB. Megakaryoblastic disorders in children. Am J Clin Pathol. 2004;122 Suppl(S33-46). Shivdasani RA, Rosenblatt MF, Zucker-

36.

37.

38.

39.

40.

41.

42.

43.

44. 45.

Franklin D, et al. Transcription factor NF-E2 is required for platelet formation independent of the actions of thrombopoietin/ MGDF in megakaryocyte development. Cell. 1995;81(5):695-704. Lecine P, Villeval JL, Vyas P, Swencki B, Xu Y, Shivdasani RA. Mice lacking transcription factor NF-E2 provide in vivo validation of the proplatelet model of thrombocytopoiesis and show a platelet production defect that is intrinsic to megakaryocytes. Blood. 1998;92(5):1608-1616. Papetti M, Wontakal SN, Stopka T, Skoultchi AI. GATA-1 directly regulates p21 gene expression during erythroid differentiation. Cell Cycle. 2010;9(10):19721980. Orford KW, Scadden DT. Deconstructing stem cell self-renewal: genetic insights into cell-cycle regulation. Nat Rev Genet. 2008; 9(2):115-128. Fouquet G, Tardy S, Demarquette H, et al. Efficacy and safety profile of long-term exposure to lenalidomide in patients with recurrent multiple myeloma. Cancer. 2013; 119(20):3680-3686. Malinge S, Thiollier C, Chlon TM, et al. Ikaros inhibits megakaryopoiesis through functional interaction with GATA-1 and NOTCH signaling. Blood. 2013; 121(13):2440-2451. Kronke J, Fink EC, Hollenbach PW, et al. Lenalidomide induces ubiquitination and degradation of CK1alpha in del(5q) MDS. Nature. 2015;523(7559):183-188. Schneider RK, Adema V, Heckl D, et al. Role of casein kinase 1A1 in the biology and targeted therapy of del(5q) MDS. Cancer Cell. 2014;26(4):509-520. Nichols KE, Crispino JD, Poncz M, et al. Familial dyserythropoietic anaemia and thrombocytopenia due to an inherited mutation in GATA1. Nat Genet. 2000; 24(3):266-270. Badros AZ. Lenalidomide in myeloma--a high-maintenance friend. N Engl J Med. 2012;366(19):1836-1838. Zhu J, Chin K, Aerbajinai W, Trainor C, Gao P, Rodgers GP. Recombinant erythroid Kruppel-like factor fused to GATA1 up-regulates delta- and gamma-globin expression in erythroid cells. Blood. 2011;117(11): 3045-3052.

1697

ARTICLE

Stem Cell Transplantation

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1698-1707

The eGVHD App has the potential to improve the accuracy of graft-versus-host disease assessment: a multicenter randomized controlled trial

Helene M. Schoemans,1,2 Kathy Goris,1 Raf Van Durm,3 Steffen Fieuws,4 Sabina De Geest,2,5 Steven Z. Pavletic,6 Annie Im,7 Daniel Wolff,8 Stephanie J. Lee,9 Hildegard Greinix,10 Rafael F. Duarte,11 Xavier Poiré,12 Dominik Selleslag,13 Philippe Lewalle,14 Tessa Kerre,15 Carlos Graux,16 Frédéric Baron,17 Johan A. Maertens1 and Fabienne Dobbels;2 on behalf of the EBMT Transplantation Complications Working party

Department of Hematology, University Hospitals Leuven and KU Leuven, Belgium; Academic Centre for Nursing and Midwifery, KU Leuven, Belgium; 3IT Department, University Hospitals Leuven, KU Leuven, Belgium; 4L-BioStat, KU Leuven – University of Leuven & Universiteit Hasselt, Leuven, Belgium; 5Institute of Nursing Science, Department Public Health, University of Basel, Switzerland; 6Experimental Transplantation and Immunology Branch, Center for Cancer Research (CCR), National Cancer Institute (NCI), National Institutes of Health (NIH), Bethesda, MD, USA; 7 University of Pittsburgh Medical Center, Pittsburgh, PA, USA; 8Department of Hematology and Clinical Oncology, University of Regensburg, Germany; 9Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 10 Division of Hematology, Medical University of Graz, Austria; 11ICO/Hospital Duran I Reynals, Hospitalet De Llobregat, Spain; 12Cliniques Universitaires Saint-Luc, Brussels, Belgium; 13Department of Hematology, AZ Sint-Jan Brugge, Belgium; 14Institut Jules Bordet - Université Libre de Bruxelles, Belgium; 15Hematology and Stem Cell Transplantation, Ghent University Hospital, Belgium; 16Université Catholique de Louvain, CHU UCL Namur (Godinne site), Yvoir, Belgium and 17Hematology, University of Liège, GIGA-I3, Belgium 1 2

Correspondence:

ABSTRACT

helene.schoemans@uzleuven.be

Received: March 6, 2018. Accepted: June 13, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2018.190777 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1698 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1698

raft-versus-host disease (GvHD) assessment has been shown to be a challenge for healthcare professionals, leading to the development of the eGVHD App (www.uzleuven.be/egvhd). In this study, we formally evaluated the accuracy of using the App compared to traditional assessment methods to assess GvHD. Our national multicenter randomized controlled trial involved seven Belgian transplantation centers and 78 healthcare professionals selected using a 2-stage convenience sampling approach between January and April 2017. Using a 1:1 randomization stratified by profession, healthcare professionals were assigned to use either the App (“APP”) or their usual GvHD assessment aids (“No APP”) to assess the diagnosis and severity score of 10 expert-validated clinical vignettes. Our main outcome measure was the difference in accuracy for GvHD severity scoring between both groups. The odds of being correct were 6.14 (95%CI: 2.83-13.34) and 6.29 (95%CI: 4.32-9.15) times higher in favor of the “APP” group for diagnosis and scoring, respectively (P<0.001). Appassisted GvHD severity scoring was significantly superior for both acute and chronic GvHD, with an Odds Ratio of 17.89 and 4.34 respectively (P<0.001) and showed a significantly increased inter-observer agreement compared to standard practice. Despite a mean increase of 24 minutes (95%CI: 20.45-26.97) in the time needed to score the whole GvHD test package in the “APP” group (P<0.001), usability feedback was positive. The eGVHD App shows superior GvHD assessment accuracy compared to standard practice and has the potential to improve the quality of outcome data registration in allogeneic stem cell transplantation. haematologica | 2018; 103(10)

Impact of the eGVHD App on GvHD assessment

Introduction Graft-versus-host disease (GvHD) refers to the reaction of the transplanted immune system against the recipient’s tissues. This pleiotropic disease affects up to half of patients after allogeneic hematopoietic stem cell transplantation (HCT) and can damage any organ system to various degrees. It is by far the most debilitating complication of HCT, considering its major impact on morbidity and mortality.1 Yet because of the lack of widely available GvHD biomarkers, the assessment of the presence and severity of GvHD still relies mainly on the clinical evaluation of multiple organs according to a relatively complex algorithm. Moreover, the recommendations underlying this evaluation are plethoric and sometimes even contradictory, potentially leading to confusion in the HCT community.1 In fact, it has been repeatedly shown that many HCT professionals have problems implementing GvHD assessment correctly, as demonstrated by a low observed accuracy in GvHD assessment2-5 and a slow uptake of the most up-to-date guidelines.5-7 The eGVHD App is an electronic tool that we developed in collaboration with the European Group for Blood and Marrow Transplantation (EBMT) Transplantation Complications Working Party and the National Institutes of Health (NIH) to assist healthcare professionals with their GvHD assessment.4 This tool is a web application, available on mobile devices and desktop computers (see www.uzleuven.be/egvhd for a complete list of the App’s characteristics). It allows intuitive and user-friendly access to the most recent international consensus guidelines and assists the user by automatically executing the required algorithm to calculate the severity of GvHD, once the relevant clinical characteristics have been entered. Pilot testing was promising, suggesting improved GvHD assessment and good usability.4,5 Therefore, the primary aim of the present study was to compare the accuracy of the severity score of validated GvHD case-vignettes performed by healthcare professionals using the “eGVHD App” (“APP” group) with standard practice (“No APP” group). Secondary aims were to understand the characteristics that might affect the difference in accuracy between both groups and to compare the inter-observer variability in GvHD scoring results, as well as the time needed to perform the GvHD evaluation of the full test package in both groups. We also assessed current practice patterns in GvHD assessment for all participants and post-test user satisfaction and experience in the “APP” group, to allow the tool’s usability to be further improved. To evaluate the generalizability of the tool, we tested the eGVHD App in a variety of settings and with a wide range of healthcare practitioners with different professional backgrounds. We hypothesized that the eGVHD App would improve GvHD assessment by improving the accuracy of GvHD severity scoring by healthcare professionals and reducing inter-rater variability in scoring results, without increasing the time required to assess GvHD.

Methods Design This study used a hybrid design (Figure 1). The first part of the study consisted of a 2-group multicenter randomized controlled haematologica | 2018; 103(10)

trial assigning healthcare professionals 1:1 to an intervention group (“APP”) or a control group (“No APP”) to evaluate the accuracy of GvHD assessment. The second part of the study was observational and described current practice patterns in GvHD assessment (“Survey 1”) and usability aspects linked to the use of the App (“Survey 2”).

Sample and setting All Belgian hospitals performing allogeneic HCT were invited to participate (Online Supplementary Table S1) to optimize sample size and generalizability. Centers were selected on their willingness to organize a GvHD workshop on their own premises within the allocated timeframe (from January to April 2017). Healthcare professionals employed or studying at each participating hospital were recruited by convenience sampling. They were included provided they attended the workshop (see Online Supplementary Methods for workshop details) and could recall having performed at least one GvHD evaluation in the past 12 months. Information concerning data collection points, randomization procedure and blinding are available in the Online Supplementary Methods.

Outcome measures The primary aim was to assess the difference in accuracy for GvHD severity scoring between the “APP” and “No APP” groups. (See Online Supplementary Methods for the planned sub-analyses.)

Variables and measurements Demographics and practice patterns in GvHD assessment: a selfreport questionnaire (“Survey 1”) captured participant characteristics (Table 1) as well as practice pattern in GvHD assessment and pre-test technology access and acceptance data (Table 2) at baseline. Accuracy of GvHD assessment: participants were required to diagnose and score a package of 10 randomly ordered GvHD clinical vignettes based on real-life clinical cases (see Online Supplementary Methods and Online Supplementary Table S2) according to the most up-to-date international guidelines.1 Four acute GvHD (aGvHD) vignettes covered the two types of aGvHD diagnosis (‘classic aGvHD’ and ‘late aGvHD’, two vignettes each) and the four aGvHD overall severity stages (I-IV, one vignette per stage), according to the Mount Sinai Acute GvHD International Consortium (MAGIC) criteria.8 Six chronic GvHD (cGvHD) vignettes covered the two cGvHD diagnoses (‘overlap cGvHD’ and ‘classic cGvHD’, two and four vignettes, respectively) and the three severity grades of the National Institutes of Health (NIH) 2014 criteria9 (two vignettes per severity level, i.e. mild, moderate and severe). Answers were given by participants using a multiple choice form offering the following mutually exclusive options for diagnosis (‘classic aGvHD’, ‘late aGvHD’, ‘overlap cGvHD’ or ‘classic cGvHD’) and scoring (‘grade I’, ‘grade II’, ‘grade III’, ‘grade IV’, ‘Mild’, ‘Moderate’ or ‘Severe’), respectively. The individual answer of each participant was compared to the gold standard (see Online Supplementary Methods) and scored as ‘correct’ (if the answer corresponded exactly to the expert evaluation) or ‘incorrect’ (for any other answer, including missing answers) for diagnosis and severity scoring, respectively (Online Supplementary Table S3). The total number of correctly evaluated vignettes for the whole GvHD test package was also recorded per individual (score ranging from 0 to 10 correct answers), for diagnosis and scoring separately. The time needed to complete the full GvHD test package was recorded for each participant individually by study staff. Control group: participants randomized to standard practice (“No APP” control group) were allowed to use any of their usual meth1699

H.M. Schoemans et al. Table 1. Characteristics of workshop participants.

Professional background - n (%) Senior physicians Junior physicians Data managers Others Sex Median age (years) - n (%) ≤30 years - n (%) 31-40 years - n (%) 41-50 years - n (%) ≥51 years - n (%) Median experience in hematology (years) Median experience in HCT (years) Median number of HCT patients evaluated for GvHD per week very low (<1 patient/week) - n (%) low (1-6 patients/week) - n (%) moderate (7-15 patients/week) - n (%) high (>15 patients/week) - n (%) Area of expertise - n (%) Adults only Children only Both adults and children Median proficiency in English°

Whole group (n= 77)

APP (n=37)

No APP (n=40)

37 (48%) 21 (27%) 15 (19%) 4 (5%) 28 males (36%) 49 females (64%) 39 (IQR: 20; range: 22-62) 24 (31%) 18 (23%) 18 (23%) 17 (22%) 7.5 (IQR: 19; range: 0-34)$ 6 (IQR: 11; range: 0-32)$ 1 (IQR: 5; range: 0-30)$$ 25 (33%) 38 (51%) 6 (8%) 6 (8%)

18 (49%) 10 (27%) 7 (19%) 2 (5%)* 13 males (35%) 24 females (65%) 40 (IQR: 18; range: 24-62) 11 (30%) 9 (24%) 11 (30%) 6 (16%) 7 (IQR: 14; range: 0-34) 6 (IQR: 12; range: 0-32) 1 (IQR: 5; range: 0-30) 13 (35%) 17 (46%) 4 (11%) 3 (8%)

19 (48%) 11 (27%) 8 (20%) 2 (5%)** 15 males (37%) 25 females (62%) 36.5 (IQR: 22; range: 22-59) 13 (33%) 9 (23%) 7 (18%) 11 (28%) 8 (IQR: 21; range: 0-32)$ 6 (IQR: 11; range: 0-32)$ 1 (IQR: 5; range: 0-25)$$ 12 (32%) 21 (55%) 2 (5%) 3 (8%)

67 (87%) 2 (2%) 7 (9%)$ 7 (IQR: 1; range: 2-10)$$

32 (86%) 2 (5%) 3 (8%) 7.5 (IQR: 2; range: 2-10)$

35 (87%) 0 (0%) 4 (10%)$ 7 (IQR: 1; range: 3-10)$

n: number; IQR: Interquartile Range; HCT: hematopoietic stem cell transplantation.*Two nurses. **One nurse and one medical student. °Self-reported proficiency in English was reported using a Likert scale of 1 (not at all fluent) to 10 (extremely fluent). The number of $ symbols used indicates the number of missing participants.

ods to assess GvHD: their own knowledge, ‘fast facts’ sheets, scoring sheets, standard operating procedures, copies of original guideline publications, or any other chosen resource. Intervention Group: participants randomized to the “APP” group received the eGVHD App as a stand-alone GvHD assessment aid. Post-test user satisfaction and experience: post-test user satisfaction and experience was recorded in “APP” users only by “Survey 2” using a semi-structured self-report questionnaire, and two validated instruments, the “perceived usefulness” subscale of the technology acceptance model (TAM) and the Post-Study System Usability Questionnaire (PSSUQ), as described previously4 (see Online Supplementary Methods and Online Supplementary Table S4 for details).

Statistical analysis For details of the statistical analysis see the Online Supplementary Methods.

Results Seven out of the eleven Belgian allogeneic HCT centers participated in the study (response rate 64%). They were 1700

essentially academic centers, covering together more than 80% of the Belgian allogenic transplantation activity (Online Supplementary Table S1). A total of 103 individuals participated in the workshops (Figure 2). Seventy-eight professionals met the inclusion criteria and were randomized. One participant dropped-out due to a medical emergency in the clinic, hence data from 77 professionals were available for analysis: 37 in the “APP” Group and 40 in the “No APP” group. There was a median of 8 participants per center (range: 7-20) (Online Supplementary Table S1). Professional characteristics were similar in both groups (Table 1). The majority of participants were medical doctors (75%), female (64%), and had a median age of 39 years (IQR: 20, range 22-62). Professionals reported a median experience in allogeneic HCT of six years (IQR: 11, range 0-32), and evaluated a median of one allogeneic HCT patient for GvHD per week (IQR: 5, range 0-30). The majority of healthcare professionals reported having expertise in adult patient care. Selfreported proficiency in English was high with a median score of 7 (IQR: 1; range: 2-10) on a Likert scale of 1 (not at all fluent) to 10 (extremely fluent). haematologica | 2018; 103(10)

Impact of the eGVHD App on GvHD assessment

Figure 1. Study Design. APP: eGVHD App; GvHD: graft-versus-host disease. *E.g. own knowledge, 'fast facts' sheets, scoring sheets, standard operating procedures, copies of original guideline publications, or any other chosen resource.

Pre-test user current standard practice and technology access/acceptance The Glucksberg10 and the NIH 2014 criteria9 were the most frequently referenced GvHD assessment guidelines being used in clinical practice as reported by healthcare professionals (Table 2). Most professionals reported basing their usual GvHD evaluation on their own knowledge (n= 44, 57%), the NIH 2014 GvHD evaluation sheet9 (n=17, 22%), and/or a self-designed scoring paper document (n=16, 21%). The use of standard criteria to assess GvHD was reported as important (median score of 7 on a Likert scale of 1 to 10, IQR: 4, range 1-10), but performed with a relatively low level of confidence (median score of 5 on a Likert scale of 1 to 10, IQR: 4, range 1-9). The top four GvHD assessment problems spontaneously reported were: lack of knowledge or experience (n=23), time constraints (n=16), lack of data in the medical files (n=7), and the complexity of the guidelines (n=5). During the workshop, the “No APP” group planned to rely essentially on their own knowledge (n=24, 62%), the NIH 2014 GvHD evaluation sheet9 (n=9, 23%), the NIH 2005 GvHD evaluation sheet11 (n=6, 15%), a self-designed scoring document (n=6, 15%), and/or other methods (n=7, 18%) (Table 2).

Accuracy of GvHD assessment The total number of correctly evaluated clinical vignettes was higher in the “APP” group compared to the “No APP” group (Table 3). More specifically, participants in the “APP” group had a median of 10 correct answers for diagnosis (IQR 1; range 5-10), compared to a median of 6.5 (IQR 3; range 2-9) in the “No APP” group for the whole GvHD test package (the maximum obtainable score was 10). For severity assessment, the “APP” group scored a median of 9 vignettes correctly (IQR 2; range 2-10) compared to a median of 4.5 (IQR 3; range 1-7) in the “No APP” group. haematologica | 2018; 103(10)

Individual results for each vignette are shown in Online Supplementary Table S3. As a result, the odds of being correct were 6.14 (95%CI: 2.83-13.34) and 6.29 (95%CI: 4.32-9.15) times higher in favor of the “APP” group for diagnosis and scoring, respectively (P<0.001). All pre-specified sub-analyses were performed as planned. The GvHD assessment of the “APP” group remained superior for both acute and chronic GvHD separately with a significantly stronger effect in acute GvHD (OR=17.89, 95%CI: 8.47-37.79) compared to chronic GvHD (OR=4.34, 95%CI: 2.79-6.74) (P<0.001), and for all levels of severity scoring, except for aGvHD grade I. The effect of the App was more apparent for higher levels of severity (P=0.034) for both aGvHD and cGvHD. The strength of the effect did not significantly depend on center (Online Supplementary Figure S1) or professional background (Online Supplementary Figure S2). Similarly, neither the age of user (Online Supplementary Figure S3), the number of GvHD patients seen per week (Online Supplementary Figure S4), or self-reported comfort with using GvHD guidelines (Online Supplementary Figure S5) seemed to mitigate the superior performance of the “APP” group. Agreement between participant results and the expert gold standard diagnosis and severity scoring are highlighted in the diagonal of Tables 4 and 5, showing the superior performance of the “APP” group. For diagnosis, the most consistent errors of the “No APP” group were seen for casevignettes relating to ‘Overlap cGvHD’ and ‘Late aGvHD’, which both tended to be confused with ‘Classic cGvHD’. The highest discrepancies between the “No APP” group and expert acute GvHD severity scoring results were seen in ‘grade II’ (which tended to be graded according to the cGvHD criteria) and ‘grade IV’ aGvHD (which was essentially mistaken for ‘grade III’). Inconsistencies in chronic GvHD severity scoring were seen across all grades. The most frequent error in the “APP” group was a slight overes1701

H.M. Schoemans et al. Table 2. Survey 1 results: pre-test practice patterns, technology access and technology acceptance data.

Most often used International Guidelines* - n (%) Glucksberg criteria IBMTR Criteria MAGIC criteria Seattle Criteria NIH 2005 Criteria NIH 2014 Criteria Other / Does not know Median importance of the guidelines °

Whole group (n= 77)

APP (n=37)

No APP (n=40)

24 (31%) 5 (7%) 13 (17%) 13 (17%) 14 (18%) 27 (35%) 11 (14%) 7 (IQR 4 - range: 1-10)$$$$$$ 5 (IQR 3 - range: 1-9)$$$

12 (32%) 2 (5%) 4 (11%) 6 (16%) 5 (14%) 17 (46%) 7 (19%) 6 (IQR 4 - range: 1-10)$$$ 5 (IQR 4 - range: 1-9)$$

12 (30%) 3 (8%) 9 (23%) 7 (18%) 9 (23%) 10 (26%) 4 (10%) 7 (IQR 5 - range: 1-10)$$$ 5 (IQR 3 - range: 1-9)$

17 (49%) 14 (40%) 4 (11%)

14 (35%) 21 (54%) 4 (10%)

18 (50%) 7 (19%) 2 (5%) 3 (8%) 10 (27%) 8 (22%) 1 (3%)

26 (65%) 9 (23%) 3 (8%) 5 (13%) 7 (18%) 6 (15%) 1 (3%)

NA NA NA NA NA NA

24 (62%) 6 (15%) 6 (15%) 9 (23%) 7 (18%) 1 (3%)

3 (8%) 34 (92%) 17 (46%) 13 (35%) 24 (65%) 0 (0%) 1 (3%)

4 (10%) 36 (90%) 16 (40%) 18 (45%) 24 (60%) 2 (5%) 2 (5%)

12 (32%) 27 (73%) 2 (5%) 0 (0%) 0 (0%)

11 (28%) 30 (75%) 0 (0%) 1 (3%) 2 (5%)

25 (68%) 2 (5%) 3 (8%)

18 (45%) 3 (8%) 3 (8%)

Median comfort in applying the guidelines ° Level of comfort ° - n (%) Low (≤ 4) 31 (42%) Moderate (5-7) 35 (47%) High (≥ 8) 8 (11%) In my daily practice, my GvHD assessment relies on…* - n (%) Own knowledge 44 (57%) A self-designed paper form 16 (21%) A self-designed electronic file 5 (7%) 11 The official NIH 2005 paper form 8 (10%) The official NIH 2014 paper form9 17 (22%) Other 14 (18%) Not answered 2 (3%) During the study, my GvHD assessment will rely on…* - n (%) Own knowledge NA A self-designed paper form NA The official NIH 2005 paper form11 NA The official NIH 2014 paper form9 NA Other NA Not answered NA To support my daily practice, I have access to* - n (%) A desktop computer with no internet connection 7 (9%) A desktop computer with an internet connection 70 (91%) A portable device 33 (43%) A WIFI connection 31 (40%) An electronic patient medical file 48 (62%) Other 2 (3%) Not answered 3 (4%) Predicted location of use* - n (%) Bedside 23 (30%) Deskside 57 (74%) Unlikely to use 2 (3%) Other 1 (1%) Not answered 2 (3%) Predicted type of device* - n (%) Cellphone 43 (56%) Tablet 5 (7%) Laptop 6 (8%)

continued on the next page

1702

haematologica | 2018; 103(10)

Impact of the eGVHD App on GvHD assessment

continued from the previous page

Desktop Other Not answered Median importance of the availability of the app in my native language° Median reported level of likelihood of using the app°

32 (42%) 0 (0%) 2 (3%) 4 (IQR 5; range: 1-10)$$

10 (27%) 1 (0%) 0 (0%) 4 (IQR 6; range: 1-10)

22 (55%) 0 (0%) 2 (5%) 4 (IQR 5; range: 1-10)$$

8 (IQR 3; range: 1-10)$$$$$

7.5 (IQR 3; range: 1-10)$

8 (IQR 4; range: 1-10)$$$$

n: number; IQR: Interquartile Range; NA: Not applicable. *Several answers were possible. °Reported on a Likert scale of 1 (lowest) to 10 (highest). The number of $ symbols used indicates the number of missing participants.

timation of the cGvHD grade (overestimation n=34, 15%; underestimation n=20, 9%; missing/other n=4, 2%) without any misclassification, whereas the “No APP” group tended to evaluate cGvHD severity erroneously according to the aGvHD criteria (n=62, 25%), without bias for severity (overestimation n=36, 14%; underestimation n=36, 15%; missing/other n=7, 3%). Consequently, inter-observer agreement of the severity score was higher in the “APP” group compared to standard practice: the probability that 2 HCT professionals agreed on the GvHD score equaled 0.73 and 0.56 in the “App” and “No APP” group, respectively. The chance-corrected agreement was significantly higher in the “APP” group (κBP= 0.46, 95%CI: 0.23-0.68) compared to the “No APP” group (κBP =0.12, 95%CI: 0.03-0.21) (P=0.003). The time needed to complete the total test package was significantly higher in the “APP” group compared to the standard practice group, with a mean time of 48.84 minutes to complete all ten clinical vignettes in the “APP” group versus 25.27 minutes in the “No APP” group (P<0.001) (Table 3).

Post-test user satisfaction and experience No major technical issues were identified. Both “perceived usefulness” and “system usability” were considered to be good, as shown in Online Supplementary Table S4. Users reported being likely to use the eGVHD App in their daily practice and did not experience any issues with using the App in English. Spontaneously reported positive aspects of the eGVHD App were its clarity, ease of use, and its systematic approach. Users suggested some potential improvements, such as decreasing its time-consuming components, reducing the number of evaluated items, and clarifying some specific terms in more detail.

Discussion Several groups have recently advocated the use of electronic tools to improve GvHD assessment, albeit without providing formal proof of their efficacy.1,4,12-14 In this rigorous multi-center randomized trial, we unequivocally demonstrate that the accuracy of GvHD assessment of clinical vignettes by healthcare professionals is significantly higher when using the eGVHD App compared to standard practice. This effect was seen for both acute and chronic GvHD, across all severity levels (except for aGvHD grade I) and all degrees of experience and professional backgrounds, without any evidence for center effect. haematologica | 2018; 103(10)

In this study, participants in the control group were allowed to use any method of their choice to support their GvHD assessment, except for using the eGVHD App. Yet GvHD assessment results in the “APP” group, were strikingly better. We believe that the superior performance of the App users could be due to a number of factors. First, App users were provided with the most up-to-date guidelines,1 without having to look them up actively. Second, similar to using comprehensive paper data collection forms, they were encouraged to work in a systematic fashion: they had to evaluate every possible aspect of acute or chronic GvHD (to avoid overlooking less intuitive aspects of the disease) in order to select the appropriate scoring system and come to the correct severity evaluation result. Finally, the digital interface also offered users a number of advantages such as the presence of pictures and definitions to support recognition of GvHD-related features, the use of ‘skip-logic’ principles (which allows healthcare professionals to avoid wasting time on filling in information with no direct impact on diagnosis or severity scoring), the automatic computation of the resulting score, and the option of generating a report. We have to acknowledge that this superior performance was achieved at the cost of a significant increase in the time needed to score clinical vignettes, with an excess of approximately 24 minutes to score the ten clinical vignettes compared to using standard methods. This was partially due to the fact that “APP” users needed to get used to a tool they had never worked with before. Yet healthcare professionals remained open to the use of eHealth technology, both before and after actually using the App. The eGVHD App showed excellent usability, as no major technical issues were noted and user feedback was widely positive, suggesting a potential for optimal dissemination and uptake in the HCT community. Furthermore, in the event where the App-computed scores would be directly transferred into the electronic health record (eHR), the additional time spent inputting data into the App would be rewarded with potentially less time charting, and more accurate data collection. However, this integration also presupposes a number of basic pre-requisites, which still need to be developed: data cleaning methods to ensure the quality of data entry, the possibility of crosstalk between the eGVHD App and the different eHR systems, the reliability, privacy and safety of data transfer, and the option of identifying the individual who performed the data input. Consistent with prior literature, our practice pattern survey showed the lack of consensus in the HCT community as to which set of international recommendations should be used to assess GvHD, and confirmed numerous barriers to their successful dissemination and implementation.5-7 1703

H.M. Schoemans et al.

Table 3. Graft-versus-host disease (GvHD) assessment accuracy and timing results.

Results for the complete GvHD test package (median) Correctly diagnosed vignettes Correctly scored vignettes

Results for acute and chronic GvHD (median) Correctly scored acute GvHD vignettes Correctly scored chronic GvHD vignettes

Time needed to complete the whole GvHD test package Mean time to complete all vignettes (minutes)

APP (n=37)

No APP (n=40)

10 (IQR 1; range 5-10) 9 (IQR 2; range 2-10)

6.5 (IQR 3; range 2-9) 4.5 (IQR 3; range 1-7)

APP (n=37)

No APP (n=40)

4 (IQR 0; range 2-4) 5 (IQR 1; range 0-6)

2 (IQR 2; range 0-4) 3 (IQR 2.25; range 0-5)

APP (n=37)

No APP (n=40)

48.84 (Std dev: 10.3; range 31-67)

25.27 (Std dev: 9.76; range 9-54)

n: number; IQR: Interquartile Range; Std dev: standard deviation. The maximum number of correct answers for the whole package was 10 (4 for acute GvHD and 6 for chronic GvHD).

The lack of consensus and knowledge of the most recent guidelines was perhaps due to the low number of HCT patients seen per week, and probably partly explains the lower results obtained by the group using traditional methods. However, this also highlights the need to standardize GvHD evaluation within the HCT community, as recently advocated by a panel of GvHD experts.1 It is precisely in this context of lack of confidence and expertise in GvHD assessment that e-Tools, such as the eGVHD App, have the potential to increase the quality of data collection by allowing easy, reliable, user-friendly and intuitive access to the most up-to-date guidelines to any healthcare professional. Regrettably, we were unable to test the effect of the App specifically in smaller Belgian centers, as they declined the invitation to participate in this study. We are, therefore, unable to speculate on the generalizability of this tool in centers with lower transplantation volumes. The limited number of vignettes also makes it challenging to make any meaningful conclusions on specific subgroups or at the organ level. The significant difference in improved accuracy for aGvHD scoring compared to cGvHD scoring is probably simply due to the fact that each of the four aGvHD severity levels was evaluated by a single clinical vignette (instead of two per severity level for cGvHD). For instance, in the ‘late acute GvHD grade II’ clinical vignette, the largely incorrect final severity evaluation reported by the “No APP” group was partially conditioned by the fact that the distinction between acute and chronic GvHD had not been made in the first place. Moreover, the MAGIC criteria were not the standard reference for aGvHD for the majority of the participants, which could explain the exceptionally poor results for the grade IV aGvHD vignette when evaluated by the “No APP” group. The limited number of observations also restrict our ability to draw any conclusions on the potential impact of using the App in the clinical setting to decide upon starting treatment, as the threshold to start therapy is linked to much broader categories than the ones described above (typically, any grade above or equal to ‘aGvHD grade II’ or ‘cGvHD moderate’ would qualify for treatment, depending on the general health status of the patient15-17). Treatment adaptions rely also on specific response criteria,18,19 which were 1704

not investigated in this project. Future studies, therefore, need to evaluate the use and impact of the eGVHD App in the clinic. This will also allow the evaluation of the App in situations where the patient does not present with GvHD, considering that the test package studied here only evaluated the tool in the context of GvHD-afflicted patients, precluding the evaluation of detection measures such as predictive values, sensitivity and specificity. Further limitations of this study are the lack of repeated measures and the unnatural setting of clinical vignettes, which are unable to perfectly mirror the wide variations in GvHD presentation in real life and their relative incidence. This particular experimental design was chosen to simplify logistics, optimize healthcare professional participation, avoid patient stress, and keep respondent burden to a minimum. It also allowed for multiple experts to validate the GvHD assessment. Such an expert consensus is rarely obtained in clinical practice, but was considered to be the best gold standard available to date to serve as reference for the accurate scoring during GvHD assessment. So, it remains to be determined whether the App will also improve accuracy when being used in real life circumstances. Yet, even in this artificial setting, the low spontaneous GvHD scoring accuracy obtained in this evaluation with traditional methods (obtaining a median of 4.5 correctly scored vignettes out of a maximum of 10) is in line with the results of a previous validation study carried out in a more real-world setting. This study included actual patient examinations and showed that only 50-75% of freshly trained clinicians actually agreed with experts on the overall severity score of the evaluated chronic GvHD patients.6 Mitchell et al. concluded that a single training session was not sufficient to achieve consistently acceptable inter-rater agreement between novice healthcare practitioners and GvHD experts. Clinical training in GvHD physical examinations may thus be necessary to achieve reproducible severity assessment with high inter-rater reliability in practice. By ensuring the systematic assessment of all organs potentially affected by GvHD, the App can also serve as a training tool, aimed at making healthcare professionals ultimately independent of technological assistance. The eGVHD App is currently limited to a calculator funchaematologica | 2018; 103(10)

Impact of the eGVHD App on GvHD assessment Table 4. Detailed results of participants for graft-versus-host disease (GvHD) vignettes compared to the Expert Gold Standard - GvHD diagnosis.

Results from the "App" group given by 37 participants - n (%) Late Classic Overlap Missing Other acute chronic chronic

Expert Gold Standard Diagnosis

Classic acute

Classic acute GVHD °°

67 (91%) 5 (7%) 3 (2%) 0 (0%) 75 (20%)

65 (88%) 0 (0%) 0 (0%) 69 (18%)

Expert Gold Standard Diagnosis

Classic acute

Results from the "No App" group given by 40 participants - n (%) Late Classic Overlap Missing Other acute chronic chronic

Classic acute GVHD °°

76 (95%) 7 (9%) 18 (11%) 3 (4%) 104 (26%)

Late acute GVHD °° Classic chronic GVHD °°°° Overlap chronic GVHD °° Total

4 (5%)

0 (0%) 52 (65%) 9 (6%) 10 (13%) 71 (18%)

0 (0%) 1 (1%) 140 (95%) 4 (5%) 145 (39%)

1 (1%) 16 (20%) 110 (69%) 51 (64%) 178 (44%)

2 (3%) 3 (4%) 3 (2%) 69 (93%) 77 (21%)

2 (3%) 5 (6%) 23 (14%) 16 (20%) 46 (11%)

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

1 (1%) 0 (0%) 0 (0%) 0 (0%) 1 (0%)

0 (0%) 0 (0%) 0 (0%) 1 (1%) 1 (0%)

0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)

Total 74 (20%) 74 (20%) 148 (40%) 74 (20%) 370 (100%)

Total 80 (20%) 80 (20%) 160 (20%) 80 (20%) 400 (100%)

n: number; "Missing" corresponds to a lack of answer; "Other" corresponds to any answer not matching the proposed choices. The number of ° symbols used indicates the number of clinical vignettes involved. The highlighted diagonal corresponds to a perfect agreement between participants and expert results.

Figure 2. CONSORT flow diagram. APP: eGVHD App; HCPs: healthcare professionals; HCT: hematopoietic stem cell transplantation. n: number.

haematologica | 2018; 103(10)

1705

H.M. Schoemans et al. Table 5. Detailed results of participants for graft-versus-host disease (GvHD) vignettes compared to the Expert Gold Standard – GvHD Severity Scoring.

Expert Gold Standard Severity Scoring

Grade II

Results from the "App" group given by 37 participants - n (%) Grade Grade Grade Mild Moderate Severe Missing Other III IV

Grade I °

33 (89%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 33 (9%)

1 (3%) 37 (100%) 0 (0%) 1 (3%) 0 (0%) 0 (0%) 0 (0%) 39 (10%)

Grade II ° Grade III ° Grade IV ° Mild °° Moderate °° Severe °° Total

0 (0%) 0 (0%) 35 (95%) 3 (8%) 0 (0%) 0 (0%) 0 (0%) 38 (10%)

0 (0%) 0 (0%) 0 (0%) 33 (89%) 0 (0%) 0 (0%) 0 (0%) 33 (9%)

2 (5%) 0 (0%) 0 (0%) 0 (0%) 49 (66%) 0 (0%) 2 (3%) 53 (14%)

1 (3%) 0 (0%) 0 (0%) 0 (0%) 22 (30%) 61 (82%) 18 (24%) 102 (27%)

0 (0%) 0 (0%) 2 (5%) 0 (0%) 1 (1%) 11 (15%) 54 (73%) 68 (18%)

0 (0%) 0 (0%) 0 (0%) 0 (0%) 1 (1%) 1 (1%) 0 (0%) 2 (0%)

Total 37 (10%) 37 (10%) 37 (10%) 37 (10%) 74 (20%) 74 (20%) 74 (20%) 370 (100%)

Results from the "No App" group given by 40 participants - n (%) Expert Gold Standard Severity Scoring Grade I ° Grade II ° Grade III ° Grade IV ° Mild °° Moderate °° Severe °° Total

Grade I

Grade II

29 (73%) 3 (8%) 0 (0%) 1 (3%) 13 (16%) 5 (6%) 1 (1%) 52 (13%)

4 (10%) 11 (28%) 0 (0%) 8 (20%) 12 (15%) 8 (10%) 9 (11%) 52 (13%)

Grade III Grade IV Mild 0 (0%) 4 (10%) 27 (68%) 19 (48%) 0 (0%) 4 (5%) 9 (11%) 63 (16%)

0 (0%) 1 (3%) 9 (23%) 7 (28%) 0 (0%) 0 (0%) 1 (1%) 18 (4.5%)

4 (10%) 3 (8%) 0 (0%) 1 (3%) 32 (40%) 5 (6%) 8 (10%) 53 (13%)

Moderate

Severe

1 (3%) 13 (33%) 1 (3%) 2 (5%) 19 (24%) 40 (50%)

0 (0%) 4 (10%) 1 (3%) 0 (0%) 2 (1%) 15 (19%) 27 (34%) 49 (12)

23 (29%) 99 (25%)

Missing Other 0 (0%) 1 (3%) 0 (0%) 0 (0%) 1 (1%) 0 (0%) 0 (0%) 2 (0%)

2 (5%) 0 (0%) 2 (5%) 2 (5%) 1 (1%) 3 (4%) 2 (3%) 12 (3%)

Total 40 (10%) 40 (10%) 40 (10%) 40 (10%) 80 (20%) 80 (20%) 80 (20%) 400 (100%)

tion that evaluates the patient at a single point in time. Expanding on our promising accuracy results and user-feedback, future plans include the development of a module to perform longitudinal patient evaluations (with an integrated disease response evaluation according to international criteria18,19) and a module to capture patient-reported GvHD evaluation based on the Lee symptom scale.20 These added functionalities will dramatically increase the clinical usefulness of the tool in following patients over time. However, a challenging issue with eHealth tools is how to approach their constant and rapid change over time. This evolution is driven by evolving clinical practices, user feedback, and updates in computer programs and/or operating 1706

systems. The results reported in this study, for instance, have been obtained with a version of the eGVHD app which has already become obsolete, as a new version (using additional skip-logic features) has been developed to address the valid criticism expressed about the time-consuming aspect of its use. The constant evolution of the virtual world is a challenge in the current context of European regulation (EU Directive 93/42/EEC MEDDEV 2. 4/1 Rev. 9 June 2010), which requires eHealth applications to be formally validated by a tedious quality assurance process at every new adaptation of the tool. This is not practically feasible in real life, and is probably, more often than not, unnecessary. Health regulation agencies will need to adjust haematologica | 2018; 103(10)

Impact of the eGVHD App on GvHD assessment

their requirements in the near future to allow for this dynamic progress of the cyber world, even for healthcare applications. This is, in fact, probably one of the most challenging aspects of integrating eTools in modern models of care.21 Compared to other smaller-scaled initiatives, which have shown successful implementation of eHealth technologies in local electronic medical record systems14 or specific research programs12,13 to assess GvHD, the eGVHD App is now widely available (www.uzleuven.be/egvhd) for all healthcare professionals who wish to obtain bedside user-friendly assistance in their GvHD assessment, and to improve their expertise and/or the uniformity of their GvHD data collection, both in daily practice and in clinical trials. Further validation regarding its usefulness and scalability will, therefore, be able to rely on the analysis of the real-life data generated by downloads and feedback from users, based on implementation research principles. If results are convincing, the next steps could include the direct integration of eGVHD App-generated data in larger registry databases and electronic medical record systems to circumvent the need to produce separate reports and repeat data entry.

References 1. Schoemans HM, Lee SJ, Ferrara JL, et al. EBMT-NIH-CIBMTR Task Force position statement on standardized terminology & guidance for graft-versus-host disease assessment. Bone Marrow Transplant. 2018 Jun 5. [Epub ahead of print PMID: 29872128]. 2. Carpenter PA, Logan BR, Lee SJ, et al. Prednisone (PDN)/Sirolimus (SRL) Compared to PDN/SRL/Calcineurin Inhibitor (CNI) as Treatment for Chronic Graft-Versus-Host-Disease (cGVHD): A Randomized Phase II Study from the Blood and Marrow Transplant Clinical Trials Network. Biol Blood Marrow Transplant. 2016;22(3):S50-S52. 3. Weisdorf DJ, Hurd D, Carter S, et al. Prospective grading of graft-versus-host disease after unrelated donor marrow transplantation: a grading algorithm versus blinded expert panel review. Biol Blood Marrow Transplant. 2003;9(8):512-518. 4. Schoemans H, Goris K, Durm RV, et al. Development, preliminary usability and accuracy testing of the EBMT 'eGVHD App' to support GvHD assessment according to NIH criteria-a proof of concept. Bone Marrow Transplant. 2016;51(8):10621065. 5. Schoemans HM, Goris K, Van Durm R, et al. Accuracy and usability of the eGVHD app in assessing the severity of graft-versus-host disease at the 2017 EBMT annual congress. Bone Marrow Transplant. 2018;53(4):490494. 6. Mitchell SA, Jacobsohn D, Thormann Powers KE, et al. A multicenter pilot evaluation of the National Institutes of Health chronic graft-versus-host disease (cGVHD) therapeutic response measures: feasibility,

haematologica | 2018; 103(10)

10.

11.

12.

13.

Such developments will require further reflections on how to achieve optimal control of the quality of the entered data and guarantee its privacy protection according to local laws. In conclusion, the eGVHD App shows superior accuracy for the GvHD assessment of clinical vignettes compared to usual care and has, therefore, the potential to improve the quality of GvHD data in clinical research and practice. In the era of electronic medical files, â&#x20AC;&#x2DC;big dataâ&#x20AC;&#x2122; and increased connectivity, e-Tools are likely to become widespread in our daily practice and could even gradually turn the individual patient into his or her own data manager and most involved advocate. Only time and continuous research will tell whether such tools can be effectively used in clinical practice and whether healthcare professionals are ready to accept IT assistance to solve some of the practical issues. Acknowledgments The authors would like to thank all of the participating hospitals for their collaboration and enthusiasm in validating the eGVHD App. We are also very grateful for the financial support of SOFHEA vzw (Sociaal Fonds voor Hematologische Aandoeningen) for this project.

interrater reliability, and minimum detectable change. Biol Blood Marrow Transplant. 2011;17(11):1619-1629. Duarte RF, Greinix H, Rabin B, et al. Uptake and use of recommendations for the diagnosis, severity scoring and management of chronic GVHD: an international survey of the EBMT-NCI Chronic GVHD Task Force. Bone Marrow Transplant. 2014;49(1):49-54. Harris AC, Young R, Devine S, et al. International, Multicenter Standardization of Acute Graft-versus-Host Disease Clinical Data Collection: A Report from the Mount Sinai Acute GVHD International Consortium. Biol Blood Marrow Transplant. 2016;22(1):4-10. 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. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-versus-host disease in human recipients of marrow from HL-Amatched sibling donors. Transplantation. 1974;18(4):295-304. 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. Levine JE, Hogan WJ, Harris AC, et al. Improved accuracy of acute graft-versushost disease staging among multiple centers. Best Pract Res Clin Haematol. 2014;27(34):283-287. Mancini G, Frulla R, Vico M, e al. A new software for evaluating scoring and response in cGVHD according to the new NIH crite-

14.

15. 16.

17.

18.

19.

20.

21.

ria. Bone Marrow Transplant. 2016;51(Issue S1):S183. Dierov Djamilia CC, Fatmi S, Mosesso K, et al . Establishing a standardized system to capture chronic graft-versus-host disease (GVHD) data in accordance to the national institutes (NIH) consensus criteria. Bone Marrow Transplant. 2017;52 (Suppl 1):S102 (abstract O157). Deeg HJ. How I treat refractory acute GVHD. Blood. 2007;109(10):4119-4126. Martin PJ, Schoch G, Fisher L, et al. A retrospective analysis of therapy for acute graftversus-host disease: initial treatment. Blood. 1990;76(8):1464-1472. Wolff D, Gerbitz A, Ayuk F, et al. Consensus conference on clinical practice in chronic graft-versus-host disease (GVHD): first-line and topical treatment of chronic GVHD. Biol Blood Marrow Transplant. 2010;16(12): 1611-1628. Lee SJ, Wolff D, Kitko C, et al. Measuring therapeutic response in chronic graft-versushost disease. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: IV. The 2014 Response Criteria Working Group report. Biol Blood Marrow Transplant. 2015;21(6):984-999. MacMillan ML, Robin M, Harris AC, et al. A Refined Risk Score for Acute Graft-versusHost Disease that Predicts Response to Initial Therapy, Survival, and TransplantRelated Mortality. Biol Blood Marrow Transplant. 2015;21(4):761-767. Lee S, Cook EF, Soiffer R, Antin JH. Development and validation of a scale to measure symptoms of chronic graft-versushost disease. Biol Blood Marrow Transplant. 2002;8(8):444-452. Tuckson RV, Edmunds M, Hodgkins ML. Telehealth. N Engl J Med. 2017;377(16): 1585-1592.

1707

ARTICLE

Stem Cell Transplantation

Ferrata Storti Foundation

Haematologica 2018 Volume 103(10):1708-1719

Upper gastrointestinal acute graft-versus-host disease adds minimal prognostic value in isolation or with other graft-versus-host disease symptoms as currently diagnosed and treated Sarah Nikiforow,1,2 Tao Wang,3 Michael Hemmer,3 Stephen Spellman,4 Görgün Akpek,5 Joseph H. Antin,1,2 Sung Won Choi,6 Yoshihiro Inamoto,7 Hanna J. Khoury,8 Margaret MacMillan,9 David I. Marks,10 Ken Meehan,11 Hideki Nakasone,12 Taiga Nishihori,13 Richard Olsson,14 Sophie Paczesny,15 Donna Przepiorka,16 Vijay Reddy,17 Ran Reshef,18 Hélène Schoemans,19 Ned Waller,8 Daniel Weisdorf,9 Baldeep Wirk,20 Mary Horowitz,3 Amin Alousi,21 Daniel Couriel,6 Joseph Pidala,13 Mukta Arora4,9 and Corey Cutler1,2 for the GV12-02 Writing Committee on behalf of the CIBMTR® Graft-versus-Host Disease Working Committee

Division of Hematologic Malignancies, Dana-Farber Cancer Institute, Boston, MA, USA; Harvard Medical School, Boston, MA, USA; 3Center for International Blood and Marrow Transplant Research (CIBMTR), Medical College of Wisconsin, Milwaukee, WI, USA; 4 CIBMTR, Minneapolis, MN, USA; 5Rush University Medical Center, Chicago, IL, USA; 6 University of Michigan Comprehensive Cancer Center, Ann Arbor, MI, USA; 7National Cancer Hospital, Tokyo, Japan; 8Emory University School of Medicine, Atlanta, GA, USA; 9 University of Minnesota, Minneapolis, MN, USA; 10United Bristol Health Care Trust, UK; 11 Dartmouth-Hitchcock Medical Center, Lebanon, NH, USA; 12Stanford University School of Medicine, CA, USA; 13H Lee Moffitt Cancer Center, Tampa, FL, USA; 14Karolinska Institute, Huddinge, Sweden; 15Indiana University School of Medicine, Indianapolis, IN, USA; 16US Food and Drug Administration, Center for Drug Evaluation and Research, Silver Spring, MD, USA; 17University of Central Florida College of Medicine, Orlando, FL, USA; 18Columbia University Medical Center, New York, NY, USA; 19Katholieke Universiteit, Lueven, Belgium; 20University of Stony Brook, NY, USA and 21MD Anderson Cancer Research Center, Houston, TX, USA 1 2

Correspondence: ABSTRACT

sarah_nikiforow@dfci.harvard.edu

Received: October 13, 2017. Accepted: July 31, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2017.182550 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1708 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1708

pper gastrointestinal acute graft-versus-host disease is reported in approximately 30% of hematopoietic stem cell transplant recipients developing acute graft-versus-host disease. Currently classified as Grade II in consensus criteria, upper gastrointestinal acute graft-versus-host disease is often treated with systemic immunosuppression. We reviewed the Center for International Blood and Marrow Transplant Research database to assess the prognostic implications of upper gastrointestinal acute graft-versus-host disease in isolation or with other acute graft-versus-host disease manifestations. 8567 adult recipients of myeloablative allogeneic hematopoietic stem cell transplant receiving T-cell replete grafts for acute leukemia, chronic myeloid leukemia or myelodysplastic syndrome between 2000 and 2012 were analyzed. 51% of transplants were from unrelated donors. Reported upper gastrointestinal acute graft-versus-host disease incidence was 12.1%; 2.7% of recipients had isolated upper gastrointestinal acute graft-versus-host disease, of whom 95% received systemic steroids. Patients with isolated upper gastrointestinal involvement had similar survival, disease-free survival, transplant-related mortality, and relapse as patients with Grades 0, I, or II acute graft-versus-host disease. Unrelated donor recipients with isolated upper gastrointestinal acute graft-versus-host disease had less subsequent chronic graft-versus-host disease than those with Grades I or II disease (P=0.016 and P=0.0004, respectively). Upper gastrointestinal involvement added no significant prognostic information when present in addition to other manifestations of Grades I or II acute graft-versus-host disease. If upper gastrointestinal symptoms were reclassified as Grade 0 or I, 425 of 2083 haematologica | 2018; 103(10)

Upper GI acute GvHD adds minimal prognostic value

patients (20.4%) with Grade II disease would be downgraded, potentially impacting the interpretation of clinical trial outcomes. Defining upper gastrointestinal acute graft-versus-host disease as a Grade II entity, as it is currently diagnosed and treated, is not strongly supported by this analysis. The general approach to diagnosis, treatment and grading of upper gastrointestinal symptoms and their impact on subsequent acute graft-versus-host disease therapy warrants reevaluation.

Introduction Upper gastrointestinal acute graft-versus-host disease (UGI aGvHD) is a clinical syndrome of anorexia, food intolerance, nausea and vomiting, first described by Weisdorf et al. in 1990.1 In that cohort of 469 relateddonor allogeneic hematopoietic stem cell transplant (HSCT) recipients, UGI symptoms were found in at least 13% of recipients and in almost half of those cases were the only reason for initiating systemic immunosuppression. Of note, all those diagnoses were confirmed by endoscopic biopsy and histological evaluation. In subsequent studies, UGI involvement has been consistently seen in up to 27% of recipients with aGvHD, with perhaps a quarter of those having UGI symptoms in isolation.2,3 There is a large differential diagnosis for patients presenting with nausea, vomiting and/or anorexia post-allogeneic HSCT including prolonged effects of conditioning treatment and medication side effects. In addition, endoscopic appearance and histological changes such as single cell epithelial necrosis with karyorrhexis, dilation of mucosal crypts or glands, and crypt abscesses or obliteration are not specific.4-6 Thus, defining and differentiating UGI aGvHD from conditioning toxicity, cytomegalovirus (CMV) or cryptosporidium infection, mycophenolate mofetil (MMF)-related damage/ulceration, and proton-pump inhibitor use can be challenging.7-9 In the original series describing this entity, the percentage of those with UGI symptoms (in isolation or plus Grade I skin aGvHD) who developed chronic GvHD

(cGvHD) was similar to that seen after non-UGI Grade II aGvHD (74% and 65%) and higher than in those patients with no aGvHD (13%) or Grade I skin aGvHD (50%).1 Based on this finding, UGI symptoms were incorporated into the “Consensus” or modified Glucksberg grading system as a criterion for Stage 1 GI, overall Grade II aGvHD.10-12 UGI symptoms are not reflected in the International Bone Marrow Transplant Registry (IBMTR) grading schema. Many subsequent publications used the modified “Consensus” grading report incidence of Grades II-IV aGvHD without delving into the organs involved, while others defined the individual organ systems involved but did not differentiate UGI versus lower GI (LGI) symptoms.13-15 Even when the distribution of GI involvement is specified, only occasionally are aGvHD responses and long-term prognosis differentially followed based on specific involvement; however, certain centers are increasingly focusing on UGI aGvHD.2,3,16-20 The prognostic implications and corresponding staging for UGI aGvHD, either in isolation or in combination with other manifestations, have not to our knowledge been validated in a large multi-center population. The need for such evaluation is highlighted by several findings. First, one study involving routine endoscopic evaluation demonstrated that UGI aGvHD, seen in 12 of 26 subjects, uniformly resolved if treated with steroids, did not progress to symptomatic LGI aGvHD, and in almost onethird of patients resolved without alteration in baseline immunosuppression. In that study, the presence of UGI

Table 1. Incidence of acute GvHD in entire cohort.

Maximum grade of aGvHD

Entire cohort (8567) n (%)

None I II

3428 (40.0%) 2038 (48.7%) 1390 (31.7%) 1344 (15.7%) 654 (15.6%) 690 (15.7%) 2083 (24.3%) 835(19.9%) 1248 (28.5%)

III-IV

Isolated UGI Any UGI involvement

MRD (4183) n (%)

URD (4384) n (%)

1712 (20.0%) 656 (15.7%) 1056 (24.1%)

229 (2.7%) 81 (1.9%) 1039 (12.1%) 430 (10.3%)

148 (3.4%) 609 (13.9%)

Biopsy reported as obtained to confirm organ involvement* Organ Total Biopsied #Biopsies Systemic involved n n (%) positive steroid use^ Skin Skin Liver UGI LGI Skin Liver UGI LGI UGI UGI

1344 1498 252 872 914 1316 1003 167 1333 229 1039

218 (16.2%) 295 (19.7%) 25 (9.9%) 311 (35.7%) 327 (35.8%) 267 (20.2%) 138 (13.8%) 84 (50.3%) 361 (27.1%) 49 (21.3%) 395 (38.0%)

212 263 12 286 301 231 71 75 337 42 361

0 (0%) 956 (79.9%)

1896 (94.7%)

1628 (97.3%) 207 (95.4%) 966 (95.5%)

*Biopsy was reported as negative, positive, inconclusive, not tested or missing. Total number of biopsies reported excludes those not tested or missing. ^% Systemic steroid use excludes patients missing relevant data. aGvHD: acute graft-versus-host disease; MRD: matched related donor; URD: unrelated donor; UGI: upper gastrointestinal; LGI: lower gastrointestinal.

haematologica | 2018; 103(10)

1709

S. Nikiforow et al. Table 2. Demographics of subgroups with isolated UGI aGvHD or other stages without GI symptoms.

Characteristics of patients Number of patients Number of centers

No aGVHD

Grade I (Skin 1/2)

Grade II No UGI

iUGI

Grade III-IV No UGI

3428 221

1344 165

1211 165

229 61

1545 192

43 (18 - 70)

41 (18 - 71)

43 (18 - 68)

42 (18 - 72)

745 (55) 599 (45)

712 (59) 499 (41)

127 (55) 102 (45)

933 (60) 612 (40)

1179 (88)

1068 (88)

201 (88)

1304 (84)

961 (72)

830 (69)

142 (62)

991 (64)

694 (52) 282 (21) 242 (18) 126 (9)

554 (46) 276 (23) 220 (18) 161 (13)

118 (52) 49 (21) 35 (15) 27 (12)

675 (44) 338 (22) 343 (22) 189 (12)

750 (56) 288 (21) 303 (23)

699 (58) 248 (20) 258 (21)

115 (50) 66 (29) 48 (21)

777 (50) 331 (21) 426 (28)

2038 (59) 1048 (31) 342 (10) 40 (<1 - 85)

654 (49) 542 (40) 148 (11) 43 (<1 - 68)

477 (39) 541 (45) 193 (16) 41 (<1 - 70)

81 (35) 119 (52) 29 (13) 43 (14 - 70)

584 (38) 638 (41) 323 (21) 42 (3 - 74)

32 (19 - 61)

33 (19 - 60)

34 (18 - 60)

32 (19 - 59)

35 (19 - 61)

1150 (34) 909 (27) 655 (19) 712 (21)

494 (37) 361 (27) 250 (19) 238 (18)

456 (38) 291 (24) 255 (21) 208 (17)

75 (33) 55 (24) 52 (23) 47 (21)

589 (38) 346 (22) 344 (22) 266 (17)

1311 (38) 320 (9) 794 (23) 808 (24)

411 (31) 146 (11) 363 (27) 368 (27)

367 (30) 125 (10) 306 (25) 342 (28)

62 (27) 20 (9) 84 (37) 50 (22)

476 (31) 160 (10) 390 (25) 429 (28)

1936 (56)

719 (53)

636 (53)

120 (52)

777 (50)

7 (<1 - 177)

7 (1 - 279)

6 (1 - 213)

7 (<1 - 309)

413 (31) 931 (69) 708 (53) 183 (14) 109 (8)

331 (27) 880 (73) 648 (54) 115 (9) 136 (11)

56 (24) 173 (76) 113 (49) 11 (5) 19 (8)

439 (28) 1106 (72) 800 (52) 221 (14) 223 (14)

582 (43)

518 (43)

65 (28)

749 (48)

Patient-related Age at transplant, years, median (range) 42 (18 - 72) Sex Male 1805 (53) Female 1623 (47) Race Caucasian 2814 (82) Karnofsky performance score at HSCT â&#x2030;Ľ 90% 2296 (67) Disease AML 1749 (51) ALL 736 (21) CML 600 (18) MDS 343 (10) Disease status at transplant Early 1836 (54) Intermediate 772 (23) Advanced 806 (24) Donor-related Donor type HLA-identical sibling URD well-matched URD partially-matched HLA-identical sibling donor age, years, median (range) Unrelated donor age, years, median (range) D/R sex match M/M M/F F/M F/F D/R CMV status +/+ +/-/+ -/D/R ABO match Matched

Transplant-related Time from diagnosis to transplant, months, 7 (<1 - 310) median (range) Graft type Bone marrow 1057 (31) Peripheral blood 2371 (69) TBI used in conditioning regimen 1672 (49) Steroid-containing GvHD prophylaxis 392 (11) MMF-containing GvHD prophylaxis 304 (9) Year of transplant 2000-2004 1549 (45)

continued in the next page

1710

haematologica | 2018; 103(10)

Upper GI acute GvHD adds minimal prognostic value

continued from the previous page

2005-2008 2009-2012 cGVHD incidence aGvHD Therapy-related steroid use Topical steroids +/- other agents Systemic steroids +/- other agents

1192 (35) 687 (20) 1283 (37)

478 (36) 284 (21) 676 (50)

478 (39) 215 (18) 660 (55)

106 (46) 58 (25) 127 (55)

575 (37) 221 (14) 533 (34)

0 0

216 (16) 956 (71)

46 (4) 1094 (90)

7 (3) 207 (90)

15 (1) 1476 (96)

N.B., for each set of comparisons, demographics were compared and significant differences that affected a particular outcome were addressed via the respective statistical models. aGvHD: acute graft-versus-host disease; cGvHD: chronic graft-versus-host disease D: donor; R: recipient; M: male; F: female; CMV: cytomegalovirus; UGI: upper gastrointestinal; AML: acute myelogenous leukemia; ALL: acute lymphoblastic leukemia; CML: chronic myelogenous leukemia; MDS: myelodysplastic syndrome; HLA: human leukocyte antigen; URD: unrelated donor; TBI: total-body irradiation; MMF: mycophenolate mofetil; HSCT: hematopoietic stem cell transplant.

Table 3A. Clinical outcomes in patients with aGvHD: pairwise comparisons between isolated UGI aGvHD versus aGvHD without UGI symptoms.

Matched Related Donor Overall Survival Isolated UGI (baseline) Grade I Grade II Grades III/IV Disease-free Survival Isolated UGI (baseline) Grade I Grade II Grades III/IV Relapse Isolated UGI (baseline) Grade I Grade II Grades III/IV Treatment-related Mortality Isolated UGI (baseline) Grade I Grade II Grades III/IV Chronic GvHD Isolated UGI (baseline) Grade I Grade II Grades III/IV

Unrelated Donor

HR 1.00 0.78 0.95 2.06

95% CI

0.16 0.77 <0.0001

HR 1.00 0.89 1.12 2.28

0.56-1.10 0.67-1.35 1.48-2.88

0.68-1.15 0.87-1.45 1.77-2.92

0.37 0.38 <0.0001

1.00 0.81 0.93 1.66

0.59-1.11 0.67-1.28 1.22-2.28

0.19 0.67 0.0015

1.00 0.74 0.94 1.66

0.57-0.94 0.74-1.20 1.31-2.11

0.016 0.63 <0.0001

1.00 0.84 0.89 0.94

0.58-1.22 0.60-1.30 0.64-1.39

0.37 0.54 0.77

1.00 0.74 0.85 0.86

0.55-0.99 0.64-1.14 0.64-1.16

0.045 0.28 0.33

1.00 0.83 1.14 3.39

0.45-1.54 0.61-2.13 1.87-6.16

0.55 0.69 <0.0001

1.00 0.93 1.43 3.91

0.58-1.52 0.90-2.27 2.5-6.14

0.78 0.13 <0.0001

1.00 1.16 1.22 1.09

0.83-1.67 0.85-1.72 0.76-1.56

0.38 0.28 0.65

1.00 1.37 1.59 1.67

1.06-1.79 1.22-2.04 1.29-2.17

0.016 0.0004 0.0001

Bold values indicate significance at P-value <0.01. Italicized values indicate 0.1< P-value <0.05. aGvHD: acute graft-versus-host disease; UGI: upper gastrointestinal.

aGVHD did not affect development of cGVHD or survival.21 Second, other studies have found that the vast majority of patients with symptoms prompting a GI evaluation will have diffuse intestinal involvement, suggesting that symptom-directed upper endoscopy may not be necessary.22-24 Third, the reliance on biopsy confirmation in the diagnosis and reporting of UGI aGvHD varies widely, and currently the diagnosis and reporting is often based on relatively non-specific symptoms. Lastly, GvHD-related mortality and patterns of therapy in general have changed over the past two decades.25 We conducted a systematic analysis to determine: 1) the prognostic impact of isolated UGI (iUGI) aGvHD and thus verify the position of this manifestation in the haematologica | 2018; 103(10)

Consensus grading scheme when present alone, and 2) if UGI symptoms add prognostic value when present in addition to skin, LGI or hepatic aGvHD. We hypothesized that as currently diagnosed, reported and treated, the impact of UGI aGvHD on transplant-related outcomes would be less than initially reported.

Methods All patients provided informed consent to the Center for International Blood and Marrow Transplant Research (CIBMTR) research program. This study was approved by the Institutional Review Board of the National Marrow Donor Program. 1711

S. Nikiforow et al.

Figure 1. Overall survival for patients with isolated UGI aGvHD versus aGvHD without UGI symptoms. Kaplain-Meier probabilities of overall survival from time of aGvHD onset for patients with iUGI symptoms, and subsets of patients with other grades of aGvHD but without any UGI symptoms. Patients who did not develop aGvHD are not represented. A. Transplantation from a matched-related donor. B. Transplantation from a well-matched or partially-matched unrelated donor. UGI: upper gastrointestinal. aGvHD: acute graft-versus-host disease.

Patient Selection The study population included all adult patients > 18 years old who received an allogeneic HSCT from a fully human leukocyte antigen (HLA)-matched related (MRD) or well-matched or partially-matched unrelated donor (URD) following myeloablative conditioning for acute myelogenous leukemia (AML), acute lymphoblastic leukemia (ALL), chronic myelogenous leukemia (CML), or myelodysplastic syndrome (MDS) between 2000 and 2012.26,27 Only recipients of peripheral blood stem cell (PBSC) or bone marrow (BM) grafts, without ex vivo or in vivo T-cell depletion (e.g., without CD34+ cell-selection, anti-thymocyte globulin, or alemtuzumab use), who received calcineurin inhibitor-based aGvHD prophylaxis were analyzed.

Definition/diagnosis of acute GvHD CIBMTR form 2100 based on modified Glucksberg criteria, was used to collect outcome data.10 UGI aGvHD is defined as “persistent nausea with histological evidence of GvHD in stomach or duodenum” - Stage 1 GI Grade II aGvHD. However, CIBMTR guidance reads that “organ staging and overall grade of GvHD should be calculated from the clinical picture, not histology”. Thus, those with persistent nausea clinically thought to be consistent with GvHD and treated accordingly may be classified as having upper GI aGvHD. Other data included date of onset of first episode of aGvHD, whether diagnosis was based on biopsy findings, maximum organ involvement and grade of aGvHD, and specific therapy for aGvHD. Histological confirmation of UGI symptoms consisted of endoscopy and biopsy of stomach or duodenum and was reported as “negative, positive, inconclusive, not tested, or missing”. Current analyses were based on maximal reported severity and organ involvement (Table 1 and Table 2).

Statistical approach The primary endpoint of this study, when analyzed by aGvHD occurrence, was overall survival (OS), encompassing death from any cause. Secondary endpoints included treatment-related mor1712

tality (TRM) defined as death while in continuous remission; relapse, defined as a clinical recurrence, progression or persistent disease following transplantation; disease-free survival (DFS), defined as absence of death or relapse; and cGvHD.28 Variables related to patient, disease, and transplantation characteristics were reported using descriptive statistics. Patient-, disease-, and treatment-related factors were compared between related and URD groups, using the χ2 test for categorical variables and the Mann-Whitney test for continuous variables. Probabilities of OS and DFS were calculated using the Kaplan-Meier estimator, with variance estimated by Greenwood’s formula. Cumulative incidence estimates for relapse, TRM and cGvHD were calculated by treating TRM and relapse as competing risks, respectively.29 Cox proportional hazard testing models were applied in multivariable analyses. Patient-, disease-, transplant-, and aGvHD-related variables were tested via a stepwise forward-selection procedure. Patient-, disease-, and transplant-related variables are detailed in the Online Supplementary Methods. In all multivariable analyses, transplantation time was treated as the starting time point, and all GvHD-related variables were treated as time-dependent variables. Multiple time-dependent variables were defined separately at the onset of aGvHD, based on the highest aGvHD grade that a patient developed. For each patient, only a single time-dependent variable was triggered. Our primary results, as shown in Table 3A, highlighted comparisons between patients with various aGvHD manifestations +/- UGI symptoms and excluded recipients without aGvHD. The Kaplan-Meier and cumulative incidence curves were also plotted from time of aGvHD onset. In our secondary results, as shown in Table 3B, recipients without aGvHD were highlighted as the baseline comparator. Results shown in Tables 3A, 3B, and 4 and P-values referred to in the text were derived from the same multivariable models involving all patient GvHD-related groupings, although only particular comparisons are cited in each table. For each set of comparisons, demographics were compared, and significant differences that affected a particular outcome were addressed via the respective statistical models. To adjust for multiple comparisons, a 2-sided P-value of <0.01 was used as the significance threshold. haematologica | 2018; 103(10)

Upper GI acute GvHD adds minimal prognostic value

Table 3B. Clinical outcomes in patients with and without aGvHD: pairwise comparisons between no aGvHD versus isolated UGI or aGvHD without UGI symptoms.

Matched Related Donor Overall Survival No aGvHD (Baseline) Isolated UGI Grade I Grade II Grades III/IV Disease-free Survival No aGvHD (Baseline) Isolated UGI Grade I Grade II Grades III/IV Relapse No aGvHD (Baseline) Isolated UGI Grade I Grade II Grades III/IV Treatment-related Mortality No aGvHD (Baseline) Isolated UGI Grade I Grade II Grades III/IV Chronic GvHD No aGvHD (Baseline) Isolated UGI Grade I Grade II Grades III/IV

Unrelated Donor

HR 1.00 1.18 0.92 1.12 2.43

95% CI

0.33 0.25 0.16 <0.0001

HR 1.00 1.01 0.89 1.13 2.30

0.85-1.63 0.80-1.06 0.96-1.31 2.14-2.76

0.79-1.29 0.79-1.02 1.00-1.27 2.07-2.55

0.94 0.088 0.046 <0.0001

1.00 1.04 0.84 0.97 1.74

0.77-1.41 0.74-0.96 0.84-1.13 1.54-1.96

0.78 0.011 0.71 <0.0001

1.00 1.09 0.80 1.03 1.81

0.86-1.38 0.71-0.91 0.91-1.15 1.63-2.01

0.48 0.0006 0.68 <0.0001

1.00 0.93 0.79 0.83 0.88

0.66-1.33 0.67-0.92 0.68-1.00 0.73-1.06

0.70 0.0032 0.048 0.18

1.00 1.04 0.77 0.89 0.90

0.79-1.37 0.66-0.90 0.76-1.03 0.77-1.05

0.78 0.0011 0.12 0.19

1.00 1.24 1.03 1.41 4.21

0.69-2.24 0.81-1.31 1.10-1.81 3.51-5.05

0.48 0.81 0.0074 <0.0001

1.00 0.85 0.80 1.21 3.33

0.54-1.34 0.65-0.98 1.01-1.46 2.87-3.86

0.49 0.033 0.038 <0.0001

0.08 <0.0001 <0.0001 <0.0001

1.00 0.95 1.31 1.5 1.59

1.14-1.51 1.30-1.72 1.37-1.84

0.69 0.0002 <0.0001 <0.0001

1.00 1.36 1.58 1.65 1.47

0.97-1.9 1.37-1.83 1.41-1.93 1.25-1.75

Bold values indicate significance at P-value <0.01. Italicized values indicate 0.1< P-value <0.05. N.B.,Tables 3A, 3B, and 4 were derived from the same multivariable model treating time of transplantation as the starting point and each GvHD-related group as a time-dependent variable, although only particular comparisons are cited in each table. In Table 3A, patients with isolated UGI GvHD were set as the baseline comparator (HR =1.00). In Table 3B, patients without aGvHD were set as the baseline comparator (HR =1.00). Covariates, including those with significant impacts on transplant-related outcomes can be found in Online Supplementary Table S2. UGI: upper gastrointestinal. aGvHD: acute graft-versus-host disease.

Results Patient characteristics and outcomes – entire cohort A total of 8567 adult recipients of myeloablative allogeneic transplant with a T cell-replete PBSC or BM graft for AML, ALL, CML or MDS from 251 transplant centers were analyzed (Online Supplementary Table S1A). The median age of recipients was 42 years (range, 18-72). Indications for HSCT were AML/MDS (60%), ALL (21%), and CML (18%). Stem cells were from 6/6 HLA-matched siblings (49%), “well-matched” (38%), and “partiallymatched” URDs (13%).26 Donor/recipient pairings were female/male in 20%, and 71% of patients received PBSCs. All patients received a calcineurin inhibitor-based aGvHD prophylaxis regimen. Eleven and 12% received MMF or steroids for aGvHD prophylaxis, respectively. For the entire population, 1-year survival was estimated at 62%, 1-year DFS at 52%, 1-year relapse at 29%, 100haematologica | 2018; 103(10)

day TRM at 11%, and incidence of cGvHD at 1 year at 43% (Online Supplementary Table S1B). Median follow-up of survivors was 71 months (range, 1-173). Given significant differences in rates of OS, DFS, TRM and cGvHD seen between MRD and URD recipients, these groups were analyzed separately. Within the entire cohort, 2.7% of recipients had iUGI aGvHD (n=229). Overall 12.1% of recipients were documented as having any UGI aGvHD symptoms (Table 1). Rates of Grades II-IV aGvHD were 35.6% for MRD and 52.6% for URD recipients. Only 21.3% of patients recorded as having iUGI aGvHD had confirmation with a gastrointestinal biopsy; 42 of the 49 biopsies were consistent with aGvHD. Of patients with any UGI involvement, 38.0% were recorded as undergoing a GI biopsy, either upper or lower. Overall, rates of biopsies in those with UGI symptoms were higher than rates of skin, liver, or LGI biopsies. Only 11% of the 1904 patients biopsied had 1713

S. Nikiforow et al.

Figure 2. Treatment-related mortality for patients with isolated UGI aGvHD versus aGvHD without UGI symptoms. Cumulative incidence curves of TRM from time of aGvHD onset for patients with isolated UGI symptoms and subsets of patients with other grades of aGvHD without any UGI symptoms. A. Transplantation from a matched-related donor. B. Transplantation from a well-matched or partially-matched unrelated donor. UGI: upper gastrointestinal. aGvHD: acute graft-versus-host disease.

pathology reports submitted to the CIBMTR. Given the limiting numbers and reliability of biopsy-confirmed results, all subsequent analyses were conducted on the clinical grades reported. The timing of onset of UGI or any individual organ involvement was not captured in the CIBMTR database. A notable feature of this cohort was the use of systemic steroids recorded in 79.8% of patients with maximum Grade I (skin-only) disease. In patients with iUGI or UGI symptoms plus Stage I or II skin disease (both Grade II) systemic steroids were received in 95.4% and 91.6% of cases, respectively (Table 1). Further data on timing and doses of therapeutic modalities/doses and response to therapy were not available.

Prognostic impact of isolated upper GI acute GvHD Acute GvHD populations analyzed. In order to determine the optimal placement of iUGI aGvHD within the aGvHD grading system, we performed pairwise comparisons between patients with aGvHD starting from the time of transplantation: specifically, those with iUGI aGvHD versus those with Grades I, II (Stage 3 skin, Stage 1 liver or Stage 1 LGI), or III/IV aGvHD without UGI manifestations (Online Supplementary Figure S1). Notable differences in baseline characteristics included a higher percentage of patients without aGvHD receiving MRD grafts (59%) and a higher prevalence of partially-matched URD grafts among patients with Grades III/IV aGvHD (20%) versus those without aGvHD (10%), although significant differences in demographics were addressed via the respective statistical models (Table 2). Overall Survival. There were no significant differences in survival between patients with iUGI aGvHD and those with Grades I or II aGvHD without UGI symptoms in univariate or multivariable analyses (Table 3A, Figure 1). As anticipated, patients with iUGI aGvHD had better survival than those with Grade III/IV aGvHD (MRD Hazard ratio (HR) 2.06, P<0.0001; URD HR 2.28, P<0.0001). 1714

Covariates with significant impacts on survival and other transplant-related outcomes in these analyses are reported in Online Supplementary Table S2. Disease-free survival and relapse. There were no significant differences in DFS between patients with iUGI aGvHD and those with Grades I or II aGvHD, although there was a trend towards improved DFS in those with Grade I aGvHD after URD HSCT (P=0.016) (Table 3A). Patients experiencing Grade III/IV aGvHD demonstrated worse DFS (MRD HR 1.66, P=0.0015; URD HR 1.66, P<0.0001). There was no significant difference in relapse incidence between patients with iUGI aGvHD and those with other grades of aGvHD. Treatment-related mortality and chronic GvHD. TRM was similar for patients with iUGI aGvHD, and those with Grades I or II aGvHD. Patients with Grade III/IV aGvHD had more TRM (MRD HR 3.39, P<0.0001; URD HR 3.91, P<0.0001) (Table 3A, Figure 2). The incidence of cGvHD after iUGI symptoms was similar to the incidence with Grades I or II aGvHD in MRD recipients. After URD HSCT, those with iUGI aGvHD had less frequent cGvHD than patients with Grade I (HR 1.38, P=0.016) and Grade II aGvHD (HR 1.59, P=0.0004). Secondary analysis including patients without acute GvHD. In secondary analyses starting at the time of transplantation, pairwise comparisons were performed between patients without aGvHD and those with iUGI and Grades I, II and III/IV, recognizing that some patients in the “No aGvHD” group experienced early deaths related to TRM before the possible onset of aGVHD. In analyses for OS, DFS, TRM, and cGvHD incidence, outcomes after iUGI aGvHD were not significantly different from those of patients with no aGvHD (Table 3B). In comparison, patients with Grade II aGvHD without UGI symptoms trended towards worse TRM (MRD P=0.0074, URD P=0.038), and those with Grade I or II aGvHD had increased cGvHD than those without aGvHD (MRD and URD, all P-values ≤ 0.0002). Additionally, patients with haematologica | 2018; 103(10)

Upper GI acute GvHD adds minimal prognostic value

Figure 3. Overall survival for patients with aGvHD with or without UGI symptoms. Kaplan-Meier probabilities of overall survival from time of aGvHD onset for patients with various grades of aGvHD with or without any UGI symptoms. A. Transplantation from a matched-related donor. B. Transplantation from a well-matched or partially-matched unrelated donor. UGI: upper gastrointestinal. aGvHD: acute graft-versus-host disease.

Grade II aGvHD (non-UGI) symptoms tended to have inferior OS and DFS and higher TRM compared to those with Grade I aGvHD, particularly after URD HSCT (data not shown).

patients with otherwise skin-only Grade I aGvHD (MRD HR 1.00, P=1.00; URD HR 1.15, P=0.32) (Figure 4).

Prognostic impact of upper GI acute GvHD with additional GvHD involvement

In a secondary subset analysis, we analyzed patients who had Grade II manifestations other than UGI involvement, namely those who had skin-only Stage 3 disease (n=505), those who had liver involvement +/- any other non-UGI Grade II involvement (n=185), and those who had only LGI +/- skin disease (n=185). Those with liver disease tended to have been transplanted earlier, between 2000-2004; otherwise, the demographics were similar. Compared to patients with iUGI aGvHD, only URD recipients with Grade II liver involvement had inferior OS (HR 1.68, P=0.0027) and TRM (HR 2.08, P=0.011) (Online Supplementary Table S4). No differences were seen between iUGI and other Grade II subgroups for any outcomes after MRD transplant. However, URD recipients from the groups with liver, LGI or skin involvement all showed increased rates of cGvHD over those with iUGI (all P-values <0.004).

Populations Analyzed. To investigate the prognostic impact of UGI aGvHD symptoms when present in addition to other manifestations, we performed pairwise comparisons between patients with aGvHD involving various organs without UGI involvement and those with similar organ involvement plus UGI symptoms. Significant differences in demographics between those who did and did not experience UGI symptoms at a given aGvHD grade, such as year of transplantation, were addressed in the statistical models (Online Supplementary Table S3). Overall survival, disease-free survival, and relapse. There was no significant difference in OS, DFS or relapse between patients with aGvHD either with or without UGI symptoms within each aGvHD grade (Table 4, Figure 3). Inferior OS and DFS were seen after MRD HSCT when UGI symptoms were noted in addition to Grades III/IV disease, but this did not attain statistical significance (HR 1.39, P=0.027; HR 1.38, P=0.027, respectively). Of note, patients with Grade I skin-only aGvHD had similar outcomes to those with Stage 1-2 skin aGvHD plus UGI involvement (currently Grade II). Treatment-related mortality and chronic GvHD. There was no difference in TRM between patients with aGvHD with or without UGI symptoms, with the exception that more severe TRM was seen for MRD recipients when UGI symptoms occurred in addition to Grades III/IV manifestations (HR 1.78, P=0.0028). Unlike what was observed during the initial description of UGI aGvHD, there was no consistent effect of additional symptoms of UGI aGvHD on the subsequent development of cGvHD, including in haematologica | 2018; 103(10)

Prognosis of isolated UGI acute GvHD compared to Other Grade II organ involvement

Discussion We embarked upon this analysis to expand on the observations of Weisdorf et al. derived from a single institution in 1990 regarding the incidence and prognostic impact of UGI aGvHD on clinical outcomes. Our population included adult recipients of MRD and well- or partially-matched URD T-cell replete HSCTs following myeloablative conditioning for AML, ALL, CML and MDS from the CIBMTR database. Rates of Grades II-IV aGvHD seen were 44.3%, consistent with historical experience.30-32 Of the entire cohort, 2.7% experienced iUGI aGvHD (n=229) and 12.1% had UGI involvement. This 1715

S. Nikiforow et al. A

Figure 4. Cumulative incidence of cGvHD according to aGvHD grades and UGI symptoms. Cumulative incidence curves for cGvHD from time of aGvHD onset for patients with various grades of aGvHD with or without any UGI symptoms, as labeled. A. Transplantation from a matched-related donor. B. Transplantation from a well-matched or partially-matched unrelated donor. UGI: upper gastrointestinal. aGvHD: acute graft-versus-host disease.

incidence was similar to the 13% described after myeloablative MRD HSCT in 1990.1 Among those who developed aGvHD, 20% had UGI involvement similar to the rates of 24-39% observed by others. The rate of isolated UGI symptoms among those with aGvHD was 4.4% compared with 6.7%, historically.2,3,19,20,33 Using this CIBMTR cohort, we sought to address two major questions. First, we sought to determine the correct placement of iUGI aGvHD, currently defined as Stage 1 GI and overall Grade II aGvHD. We found no differences in OS, DFS, relapse, TRM or cGvHD incidence between patients with iUGI and those with Grade I or Grade II aGvHD without UGI symptoms, and noted a reduced incidence of cGvHD after iUGI among URD recipients. Furthermore, in a limited secondary analysis, we noted no difference in outcomes comparing patients with iUGI aGvHD to those without aGvHD. Thus, we did not reproduce the initial findings by Weisdorf et al. that rates of cGvHD after iUGI were above those in Grades 0 and I aGvHD but similar to those after Grade II non-UGI GvHD.1 Next, we asked whether the presence of UGI when found in conjunction with other aGvHD manifestations impacted outcomes, specifically focusing on whether UGI symptoms in addition to Grade I skin-only disease yielded outcomes similar to Stage II disease (i.e., Skin Stage 3, Liver Stage 1, and LGI Stage 1). We found no impact on outcomes in early Grades I or II aGvHD, whereas documentation of UGI symptoms in addition to Grades III/IV aGvHD correlated with worse outcomes, particularly TRM, after MRD HSCT. Whether UGI symptoms severe enough to be diagnosed in the setting of Grade III/IV manifestations comprise the same entity and share the same pathology as iUGI cannot be ascertained, and the biologic rationale for this finding is unclear. However, these findings are unlikely to change the management of this patient group. Based on these 1716

analyses, our study does not show a significant prognostic impact of UGI aGvHD, in isolation or combination, on transplant-related outcomes and suggests that the classifying of UGI as a Grade II entity as it is currently diagnosed and reported might be incorrect. The major limitations of this analysis, and any investigation of UGI aGvHD, are the difficulties surrounding diagnosis, lack of specificity of symptoms, unclear time of onset, and a seeming reluctance among transplant physicians to pursue or document biopsy confirmation of UGI involvement.34 Anecdotally, individual transplant centers vary in UGI biopsy performance and reporting, especially when LGI symptoms are present.35,36 In the study herein, only 61 of 251 participating centers reported a case of iUGI aGvHD. Among the 8567 patients included, only 1737 biopsies (20.3%) were documented. In patients with iUGI aGvHD specifically, 69% did not have a biopsy recorded. Therefore, the incidence may be under-reported in our database or confused with non-specific GI inflammation. Perhaps most importantly, the above results are in the context of systemic steroid administration to 90% of patients with UGI aGvHD, with dose, duration of, and response to therapy not specified in this data set. It is highly possible that widespread use of systemic steroids is impacting outcomes in this group, especially as many have demonstrated that UGI aGvHD has higher response rates than other Grade II manifestations.1,2,33 Therefore, we can only state that UGI aGvHD, when diagnosed, reported to the CIBMTR, and treated according to current standards of care across multiple institutions, does not impact prognosis. This is in contrast to all other manifestations of aGvHD, particularly Grades II-IV for which, despite treatment with systemic corticosteroids, occurrence is historically associated with worse outcomes.10-12,33,37 In our primary analyses, utilizing a multivariate model including all patient groups and starting from the time of haematologica | 2018; 103(10)

Upper GI acute GvHD adds minimal prognostic value

Table 4. Comparison of clinical outcomes in patients with aGvHD: pairwise comparisons between those without or with UGI symptoms.

Overall Survival

Matched Related Donor 95% CI

Unrelated Donor 95% CI

UGI + Skin Grade I (currently Grade II) vs. Skin Grade I only Grade II plus UGI symptoms vs. Grade II without UGI Grades III/IV plus UGI symptoms vs. Grades III/IV without UGI Disease-free Survival UGI + Skin Grade I (currently Grade II) vs. Skin Grade I only Grade II plus UGI symptoms vs. Grade II without UGI Grades III/IV plus UGI symptoms vs. Grades III/IV without UGI Relapse UGI + Skin Grade I (currently Grade II) vs. Skin Grade I only Grade II plus UGI symptoms vs. Grade II without UGI Grades III/IV plus UGI symptoms vs. Grades III/IV without UGI Treatment-related Mortality UGI + Skin Grade I (currently Grade II) vs. Skin Grade I only Grade II plus UGI symptoms vs. Grade II without UGI Grades III/IV plus UGI symptoms vs. Grades III/IV without UGI Chronic GvHD UGI + Skin Grade I (currently Grade II) vs. Skin Grade I only Grade II plus UGI symptoms vs. Grade II without UGI Grades III/IV plus UGI symptoms vs. Grades III/IV without UGI

1.10

0.77-1.58

0.58

1.04

0.80-1.37

0.76

1.13

0.89-1.44

0.33

0.88

0.72-1.07

0.20

1.39

1.04-1.86

0.027

1.13

0.88-1.45

0.33

1.02

0.73-1.44

0.90

1.24

0.95-1.62

0.11

1.25

0.99-1.58

0.056

0.92

0.76-1.12

0.40

1.38

1.04-1.84

0.027

1.12

0.86-1.47

0.39

1.01

0.68-1.52

0.94

1.30

0.95-1.78

0.11

1.07

0.79-1.45

0.64

0.87

0.67-1.12

0.27

1.22

0.77-1.94

0.39

0.86

0.53-1.38

0.53

1.05

0.56-1.98

0.87

1.05

0.65-1.71

0.84

1.43

0.99-2.07

0.059

1.06

0.78-1.44

0.72

1.78

1.22-2.60

0.0028

1.38

0.99-1.93

0.058

1.00

0.70-1.43

1.00

1.15

0.87-1.51

0.32

0.79

0.67-1.01

0.064

0.86

0.70-1.07

0.17

1.20

0.77-1.86

0.43

0.93

0.61-1.43

0.75

Bold values indicate significance at P-value <0.01. Italicized values indicate 0.01< P-value <0.05. N.B.,Tables 3A, 3B, and 4 were derived from the same multivariable model treating time of transplantation as the starting point and each GvHD-related group as a time dependent variable, although only particular comparisons are cited in each table. UGI: upper gastrointestinal. GvHD: graft-versus-host disease.

HSCT, we did not cite patients who had no aGvHD as our primary baseline comparator, since that group included individuals who died of early TRM and were never at risk of aGvHD. Instead, all GvHD-related variables were treated as time-dependent variables. Kaplan-Meier and incidence curves for these groups were plotted from time of aGvHD onset. However, in a secondary analysis (Table 3B) we did show outcomes compared to patients without aGvHD. A landmark analysis to eliminate the confounder/competing risk of early death performed at 30 or 60 days from transplantation could have addressed this haematologica | 2018; 103(10)

issue; however, many cases of aGvHD would have been excluded by doing so. Given the incidence of <3% for iUGI and multiple comparisons required, our cohort would not have had enough power to demonstrate statistically significant differences if we chose this statistical approach. While one might argue that the current analyses involving the iUGI group are underpowered, there is unlikely to be a larger dataset in which to perform such comparisons. Additional limitations of this dataset and analysis include the exclusion of pediatric HSCT recipients, 1717

S. Nikiforow et al.

patients with lymphoma and multiple myeloma, recipients of umbilical cord or haploidentical graft sources, and those undergoing reduced-intensity conditioning. These populations have different rates and potentially different manifestations of aGvHD and merit separate analyses. Our data can lead in two directions. One, given current non-standardized reporting and treatment patterns, reclassification of patients with UGI GvHD as Grade I could be considered, with Grade I aGvHD generally considered to have few prognostic implications. In our cohort, that reclassification would impact 425 (20.4%) of the 2083 patients currently graded as Grade II. While this would not impact cases of “severe” aGvHD (Grades III-IV), patients with Grades II-IV aGvHD would decrease from 44.3% to 39.9%. Whether such a shift would significantly impact perceived incidence or outcomes of patients Grades II-IV aGvHD, for example, could be investigated by retrospective reanalysis of multicenter studies with GvHD as a major outcome, such as the BMT CTN trials 0201 or 0402 (clinicaltrials.gov Identifiers: 00075816 and 00406393).31,38 Alternatively, an approach which downgrades UGI symptoms could be included in secondary analyses in the prospective PROGRESS 1 and 2 trials currently testing novel aGvHD prophylaxis strategies (clinicaltrials.gov Identifiers: 02208037 and 02345850), to assess impact. The implications of a change in grading include powering of future studies as well as interpretation of efficacy, given that major endpoints currently include seemingly non-informative events. Even just relabeling Stage 1 into, for example, 1a for UGI and 1b for LGI, might facilitate the tracking of UGI and LGI symptoms in future analyses. Alternatively, the general HSCT field could move to a more standardized approach to diagnosis using regular endoscopy biopsies and more consistent pathologic reporting, with or without more detailed organ system reporting, such as in the Minnesota risk-adapted acute GvHD risk score.2,39 Mehta et al. recently published a retrospective single center study in which all UGI aGvHD was confirmed by biopsy and treated in a similar manner with systemic steroids.40 Patients with Grade II aGvHD consisted of 10% with iUGI, 33% with UGI + other organ involvement, and 57% with no UGI symptoms. In this study, although comparisons with Grade I aGvHD were not performed, all subsets of Grade II aGvHD had similar outcomes in terms of OS, DFS, non-relapse mortality (NRM), relapse and cGvHD. In contrast to our more “real-world” data set, this highly controlled analysis supports maintaining UGI aGvHD as a Grade 2 event. Analysis of this type of controlled data set which includes details on kinetics of individual manifestations, response to therapy, and infectious complications of therapy would be enlightening. In summary, we challenge the field to revisit how UGI aGvHD is diagnosed, reported, graded and treated given

References 1. Weisdorf D, Snover D, Haake R, et al. Acute upper gastrointestinal graft-versushost disease: clinical significance and response to immunosuppressive therapy. Blood. 1990;76(3):624-629. 2. MacMillan ML, Robin M, Harris AC, et al. A refined risk score for acute graft-versus-

1718

that its current prognostic utility within the Consensus criteria is extremely limited. We would recommend highly standardized prospective trials involving endoscopic biopsies to explore whether system steroid therapy is required for these symptoms in isolation or with Grade I skin-only aGvHD. However, redefining UGI manifestations, especially as currently reported across multiple institutions, to a Grade I-defining entity and evaluating the impact on outcomes of large therapeutic trials of aGvHD prophylaxis and therapy should be considered. Funding SN was supported by ASH Fellow Scholar Award, ASBMT New Investigator Award, Jock and Bunny Adams Research and Education Fund, Farmor Fund Grant, Simberg Grant from Friends of Dana Farber. CC was supported by the Stem Cell Cyclists of the Pan-Mass Challenge. The CIBMTR is supported primarily by Public Health Service Grant/Cooperative Agreement 5U24-CA076518 from the National Cancer Institute (NCI), the National Heart, Lung and Blood Institute (NHLBI) and the National Institute of Allergy and Infectious Diseases (NIAID); a Grant/Cooperative Agreement 5U10HL069294 from NHLBI and NCI; a contract HHSH250201200016C with Health Resources and Services Administration (HRSA/DHHS); two Grants N00014-15-1-0848 and N00014-16-1-2020 from the Office of Naval Research; and grants from Actinium Pharmaceuticals, Inc.; Alexion; Amgen, Inc.; Anonymous donation to the Medical College of Wisconsin; Astellas Pharma US; AstraZeneca; Atara Biotherapeutics, Inc.; Be the Match Foundation; Bluebird Bio, Inc.; Bristol Myers Squibb Oncology; Celgene Corporation; Cellular Dynamics International, Inc.; Cerus Corporation; Chimerix, Inc.; Fred Hutchinson Cancer Research Center; Gamida Cell Ltd.; Genentech, Inc.; Genzyme Corporation; Gilead Sciences, Inc.; Health Research, Inc. Roswell Park Cancer Institute; HistoGenetics, Inc.; Incyte Corporation; Janssen Scientific Affairs, LLC; Jazz Pharmaceuticals, Inc.; Jeff Gordon Children’s Foundation; The Leukemia & Lymphoma Society; Medac, GmbH; MedImmune; The Medical College of Wisconsin; Merck & Co, Inc.; Mesoblast; MesoScale Diagnostics, Inc.; Miltenyi Biotec, Inc.; National Marrow Donor Program; Neovii Biotech NA, Inc.; Novartis Pharmaceuticals Corporation; Onyx Pharmaceuticals; Optum Healthcare Solutions, Inc.; Otsuka America Pharmaceutical, Inc.; Otsuka Pharmaceutical Co, Ltd. – Japan; PCORI; Perkin Elmer, Inc.; Pfizer, Inc; Sanofi US; Seattle Genetics; Spectrum Pharmaceuticals, Inc.; St. Baldrick’s Foundation; Sunesis Pharmaceuticals, Inc.; Swedish Orphan Biovitrum, Inc.; Takeda Oncology; Telomere Diagnostics, Inc.; University of Minnesota; and Wellpoint, Inc. The views expressed in this article do not reflect the official policy or position of the National Institutes of Health, the Department of the Navy, the Department of Defense, the US Food & Drug Administration, or any other agency of the U.S. Government.

host disease that predicts response to initial therapy, survival, and transplant-related mortality. Biol Blood Marrow Transplant. 2015;21(4):761-767. 3. Alousi AM, Weisdorf DJ, Logan BR, et al. Etanercept, mycophenolate, denileukin, or pentostatin plus corticosteroids for acute graft-versus-host disease: a randomized phase 2 trial from the Blood and Marrow

Transplant Clinical Trials Network. Blood. 2009;114(3):511-517. 4. Snover DC, Weisdorf SA, Vercellotti GM, et al. A histopathologic study of gastric and small intestinal graft-versus-host disease following allogeneic bone marrow transplantation. Hum Pathol. 1985;16(4):387-392. 5. Nomura K, Iizuka T, Kaji D, et al. Clinicopathological features of patients

haematologica | 2018; 103(10)

Upper GI acute GvHD adds minimal prognostic value

7. 8.

10.

11.

12.

13.

14.

15.

16.

17.

with acute graft-versus-host disease of the upper digestive tract. J Gastroenterol Hepatol. 2014;29(11):1867-1872. Xu C-F, Zhu L-X, Xu X-M, Chen W-C, Wu D-P. Endoscopic diagnosis of gastrointestinal graft-versus-host disease. World J Gastroenterol. 2008;14(14):2262-2267. Washington K, Jagasia M. Pathology of graft-versus-host disease in the gastrointestinal tract. Hum Pathol. 2009;40(7):909-917. Parfitt JR, Jayakumar S, Driman DK. Mycophenolate mofetil-related gastrointestinal mucosal injury: variable injury patterns, including graft-versus-host diseaselike changes. Am J Surg Pathol. 2008; 32(9):1367-1372. Washington K, Bentley RC, Green A, et al. Gastric graft-versus-host disease: a blinded histologic study. Am J Surg Pathol. 1997; 21(9):1037-1046. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant. 1995;15(6):825-828. Rowlings PA, Przepiorka D, Klein JP, et al. BMTR severity index for grading acute graft-versus-host disease: retrospective comparison with Glucksberg grade. Br J Haematol. 1997;97(4):855-864. Martino R, Romero P, Subira M, et al. Comparison of the classic Glucksberg criteria and the IBMTR Severity Index for grading acute graft-versus-host disease following HLA-identical sibling stem cell transplantation. International Bone Marrow Transplant Registry. Bone Marrow Transplant. 1999;24(3):283-287. Mielcarek M, Furlong T, O’Donnell PV, et al. Posttransplantation cyclophosphamide for prevention of graft-versus-host disease after HLA-matched mobilized blood cell transplantation. Blood. 2016;127(11):15021508. Ganetsky A, Shah A, Miano TA, et al. Higher tacrolimus concentrations early after transplant reduce the risk of acute GvHD in reduced-intensity allogeneic stem cell transplantation. Bone Marrow Transplant. 2016;51(4):568-572. Devine SM, Owzar K, Blum W, et al. Phase II study of allogeneic transplantation for older patients with acute myeloid leukemia in first complete remission using a reducedintensity conditioning regimen: results from cancer and leukemia group B 100103 (Alliance for Clinical Trials in Oncology)/Blood and Marrow Transplant Clinical Trial Network 0502. J Clin Oncol. 2015;33(35):4167-4175. Socié G, Vigouroux S, Yakoub-Agha I, et al. A phase 3 randomized trial comparing inolimomab vs usual care in steroid-resistant acute GVHD. Blood. 2017;129(5):643-649. Armand P, Kim HT, Sainvil M-M, et al. The addition of sirolimus to the graft-versus-

haematologica | 2018; 103(10)

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

host disease prophylaxis regimen in reduced intensity allogeneic stem cell transplantation for lymphoma: a multicentre randomized trial. Br J Haematol. 2016;173 (1):96-104. Leisenring WM, Martin PJ, Petersdorf EW, et al. An acute graft-versus-host disease activity index to predict survival after hematopoietic cell transplantation with myeloablative conditioning regimens. Blood. 2006;108(2):749-755. Bejanyan N, Rogosheske J, DeFor T, et al. Higher dose of mycophenolate mofetil reduces acute graft-versus-host disease in reduced-intensity conditioning double umbilical cord blood transplantation. Biol Blood Marrow Transplant. 2015;21(5):926933. Bolaños-Meade J, Logan BR, Alousi AM, et al. Phase 3 clinical trial of steroids/mycophenolate mofetil vs steroids/placebo as therapy for acute GVHD: BMT CTN 0802. Blood. 2014;124(22):3221-3227. Wakui M, Okamoto S, Ishida A, et al. Prospective evaluation for upper gastrointestinal tract acute graft-versus-host disease after hematopoietic stem cell transplantation. Bone Marrow Transplant. 1999;23(6):573-578. Ross W, Ghosh S, Dekovich A, et al. Endoscopic biopsy diagnosis of acute gastrointestinal graft-versus-host disease: rectosigmoid biopsies are more sensitive than upper gastrointestinal biopsies. Am J Gastroenterol. 2008;103(4):982-989. Kreisel W, Dahlberg M, Bertz H, et al. Endoscopic diagnosis of acute intestinal GVHD following allogeneic hematopoietic SCT: a retrospective analysis in 175 patients. Bone Marrow Transplant. 2012;47(3):430-438. Aslanian H, Chander B, Robert M, et al. Prospective evaluation of acute graft-versus-host disease. Dig Dis Sci. 2011; 57(3):720-725. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Engl J Med. 2010;363(22):2091-2101. Weisdorf D, Spellman S, Haagenson M, et al. Classification of HLA-matching for retrospective analysis of unrelated donor transplantation: revised definitions to predict survival. Biol Blood Marrow Transplant. 2008;14(7):748-758. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant. 2009;15(12):16281633. Lee SJ, Klein JP, Barrett AJ, et al. Severity of chronic graft-versus-host disease: association with treatment-related mortality and relapse. Blood. 2002;100(2):406-414.

29. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Amer Statist Assn. 1958;53:457-481. 30. Hamilton BK, Rybicki L, Dean R, et al. Cyclosporine in combination with mycophenolate mofetil versus methotrexate for graft versus host disease prevention in myeloablative HLA-identical sibling donor allogeneic hematopoietic cell transplantation. Am J Hematol. 2015;90(2):144-148. 31. Cutler C, Logan B, Nakamura R, et al. Tacrolimus/sirolimus vs tacrolimus/methotrexate as GVHD prophylaxis after matched, related donor allogeneic HCT. Blood. 2014;124(8):1372-1377. 32. Kharfan-Dabaja M, Mhaskar R, Reljic T, et al. Mycophenolate mofetil versus methotrexate for prevention of graft-versus-host disease in people receiving allogeneic hematopoietic stem cell transplantation. Cochrane Database Syst Rev. 2014; 25(7):CD010280. 33. MacMillan ML, Weisdorf DJ, Wagner JE, et al. Response of 443 patients to steroids as primary therapy for acute graft-versus-host disease: Comparison of grading systems. Biol Blood Marrow Transplant. 2002; 8(7):387-394. 34. Abraham J, Janin A, Gornet J-M, et al. Clinical severity scores in gastrointestinal graft-versus-host disease. Transplantation. 2014;97(9):965-971. 35. Ip S, Marquez V, Schaeffer DF, Donnellan F. Sensitivities of biopsy sites in the endoscopic evaluation of graft-versus-host disease: retrospective review from a tertiary center. Dig Dis Sci. 2016;61(8):2351-2356. 36. Thompson B, Salzman D, Steinhauer J, Lazenby AJ, Wilcox CM. Prospective endoscopic evaluation for gastrointestinal graftversus-host disease: determination of the best diagnostic approach. Bone Marrow Transplant. 2006;38(5):371-376. 37. Gratwohl A, Hermans J, Apperley J, et al. Acute graft-versus-host disease: grade and outcome in patients with chronic myelogenous leukemia. Working Party Chronic Leukemia of the European Group for Blood and Marrow Transplantation. Blood. 1995; 86(2):813-818. 38. Anasetti C, Logan BR, Lee SJ, et al. Peripheral-blood stem cells versus bone marrow from unrelated donors. N Engl J Med. 2012;367(16):1487-1496. 39. MacMillan ML, DeFor TE, Weisdorf DJ. What predicts high risk acute graft-versushost disease (GVHD) at onset?: identification of those at highest risk by a novel acute GVHD risk score. Br J Haematol. 2012;157(6):732-741. 40. Mehta RS, Cao Q, Holtan S, MacMillan ML, Weisdorf DJ. Upper GI GVHD: similar outcomes to other grade II graft-versushost disease. Bone Marrow Transplant. 2017;52(8):1180-1186.

1719

ARTICLE

Cell Therapy & Immunotherapy

Ferrata Storti Foundation

CD16+NK-92 and anti-CD123 monoclonal antibody prolongs survival in primary human acute myeloid leukemia xenografted mice

Brent A. Williams,1,2 Xing-Hua Wang,1 Jeffrey V. Leyton,3 Sonam Maghera,1,2 Bishoy Deif,1 Raymond M. Reilly,4,5,6 Mark D. Minden7,8 and Armand Keating1,2,8,9,10

Haematologica 2018 Volume 103(10):1720-1729

1 Cell Therapy Program, Princess Margaret Cancer Centre, Toronto, Ontario; 2Institute of Medical Science, University of Toronto, Ontario; 3Department of Nuclear Medicine and Radiobiology, Université de Sherbrooke, Quebec; 4Department of Medical Imaging, University of Toronto, Ontario; 5Department of Pharmaceutical Sciences, University of Toronto, Ontario; 6Toronto General Research Institute, University Health Network, Toronto, Ontario; 7Department of Medical Biophysics, University of Toronto, Ontario; 8 Department of Medicine, University of Toronto, Ontario; 9Institute of Biomaterials and Biomedical Engineering, University of Toronto and 10Krembil Research Institute, University Health Network, Toronto, Ontario, Canada

ABSTRACT

Correspondence: brentw@uhnres.utoronto.ca or brentwilliams.brent@gmail.com Received: January 17, 2018. Accepted: July 3, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2017.187385 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1720

atients with acute myeloid leukemia (AML) often relapse after initial therapy because of persistence of leukemic stem cells that frequently express the IL-3 receptor alpha chain CD123. Natural killer (NK) cell-based therapeutic strategies for AML show promise and we explore the NK cell lines, NK-92 and CD16+NK-92, as a treatment for AML. NK-92 has been tested in phase I clinical trials with minimal toxicity; irradiation prior to infusion prevents risk of engraftment. The CD16 negative NK-92 parental line was genetically modified to express the high affinity Fc gamma receptor, enabling antibody-dependent cellmediated cytotoxicity, which we utilized in combination with an antiCD123 antibody to target leukemic stem cells. NK-92 was preferentially cytotoxic against leukemic stem and progenitor cells compared with bulk leukemia in in vitro assays, while CD16+NK-92 in combination with an anti-CD123 mAb mediated antibody-dependent cell-mediated cytotoxicity against CD123+ leukemic targets. Furthermore, NK-92 infusions (with or without prior irradiation) improved survival in a primary AML xenograft model. Mice xenografted with primary human AML cells had a superior survival when treated with irradiated CD16+NK-92 cells and an anti-CD123 monoclonal antibody (7G3) versus treatment with irradiated CD16+NK-92 cells combined with an isotype control antibody. In this proof-of-principle study, we show for the first time that a CD16+NK92 cell line combined with an antibody that targets a leukemic stem cell antigen can lead to improved survival in a relevant pre-clinical model of AML. Introduction

1720

Acute myeloid leukemia (AML) accounts for the majority of acute leukemias in adults and a minority in children.1,2 While up to 70-85% of AML patients treated with current chemotherapy protocols achieve morphological remission,1,3 many relapse because of recurrence from residual leukemic stem cells (LSCs) resulting in an overall 5-year survival of approximately 40%.2 AML was the first malignancy with clear evidence of a stem cell hierarchy, with the LSCs being enriched in the CD34+CD38– fraction.4,5 In addition, they often express the IL-3 receptor alpha chain (CD123), a marker not highly expressed on normal hematopoietic stem cells.6 AML patients with a greater than 1% burden of CD34+CD38–CD123+ LSCs at diagnosis have a reduced disease-free and overall survival rate, directly implicating CD123 as a relevant target antigen.7 Natural killer (NK)-cell-based approaches are under development for the treatment of AML, such as the use of haploidentical NK-cell infusions.8,9 While this shows promise, there is inherent variability in the haematologica | 2018; 103(10)

CD16+NK-92 and anti-CD123 antibody therapy for AML

NK-cell preparations. Another approach is to use a permanent NK cell line, such as NK-92 which was derived from a patient with an NK-cell lymphoma,10 and demonstrates enhanced cytotoxicity over endogenously-derived NK cells against a variety of human leukemia cell lines and primary leukemic blasts.11 However, this cell line lacks the Fc gamma receptor IIIA (CD16), typically expressed by NK cells and, therefore, cannot mediate antibody-dependent cell-mediated cytotoxicity (ADCC). NK-92 has been tested in three published phase I clinical trials, including one clinical trial by our group for relapsed and refractory hematologic cancers (lymphoma and multiple myeloma), which all demonstrated minimal toxicity.12-14 However, to prevent potential engraftment of NK-92 and generate a NK malignancy, the cells are irradiated with 1000 cGy which does not significantly decrease in vitro cytotoxicity.15-17 Natural killer cells typically express CD16 and are able to mediate ADCC against antibody-coated targets, enabling both adaptive and innate immune responses. Since the parental NK-92 cell line lacks CD16, and cannot mediate ADCC, a high-affinity allelic variant (valine at position 176 instead of phenylalanine) of the CD16A Fcγ receptor was transduced into the NK-92 cell line. These gene-modified CD16+NK-92 cells (NK-92.176V and NK92.176V.GFP) demonstrate ADCC in vitro18 but have not been tested in vivo. Here, we show that NK-92 preferentially kills leukemic stem cells compared with bulk leukemia cells and can prolong survival with or without prior radiation. Moreover, gene-modified NK-92 expressing the high affinity CD16 receptor (NK-92.176V.GFP) more effectively killed CD123+ targets in vitro and demonstrated an enhanced ability to target LSCs. Finally, irradiated CD16+NK-92 combined with the anti-CD123 antibody, 7G3, enhanced survival in a primary AML xenograft model compared with control arms.

Methods Cell lines and primary samples K562 was obtained from the American Type Culture Collection (Manassas, VA, USA) and maintained in IMDM+10% FBS. OCI/AML2, OCI/AML3 and OCI/AML5 were generously provided by Dr. Mark Minden and maintained in MEMalpha+ 10%FBS (OCI/AML2 and OCI/AML3) or MEMalpha+10% FBS and 10% 5637 bladder carcinoma conditioned medium (OCI/AML5). NK92 was originally kindly provided by Dr. Hans Klingemann, expanded and was maintained in X-VIVO 10 medium (Lonza) supplemented with 450 U/mL of IL-2 and 2.5% human AB serum (GM1). Four primary AML samples were obtained from the Princess Margaret Hospital Leukemia Tissue Bank, Toronto, Canada, according to an approved institutional protocol. NK-92 and NK-92.176V GFP (hereafter referred to as CD16+NK-92) was obtained from Conkwest under a Material Transfer Agreement (MTA) and maintained as described for NK-92. Frozen master cell banks for cell lines were established and new vials utilized to establish new cultures every six weeks. Mycoplasma testing by PCR was conducted periodically with all cultures testing negative.

Chromium release assay We used a chromium release assay (CRA) as previously described by our group19 and detailed in the Online Supplementary Methods. haematologica | 2018; 103(10)

Flow cytometry and cell sorting Immunophenotyping of bone marrow (BM) was performed using an FC500 or Facscalibur flow cytometer. FACS buffer was made with PBS+2mM EGTA+2% FBS. Primary AML and leukemic stem cell fractions were detected using the following antibodies (company; product #; clone): anti-CD34 PE (BD biosciences; 348057; 8G12), anti-CD34 FITC (BD Pharmingen; 555821; 581), anti-CD38 APC (ebiosciences; 17-0389-42; HIT2) and anti-CD123 PE (BD Pharmingen; 558714; 7G3), anti-CD45 APC (BD Pharmingen; 557513; TU116) and anti-class I HLA A, B, C (Biolegend; 311404; W6/32). NK-92 cells lines were assessed for CD16 expression using CD16 PE (Biolegend; 302008; 3G8). Leukemia cell lines were evaluated using anti-CD123 PerCy5.5 (BD Biosciences; 560904; 7G3). Cell sorting was performed using a FacsAria cell sorter as described in the Online Supplementary Methods.

Methylcellulose cytotoxicity assay We used a methylcellulose cytotoxicity assay as previously described19 and included in the Online Supplementary Methods and in Supplementary Figure S1. Briefly, cell line or primary AML cells were incubated with and without NK-92 in a 4-hour assay prior to infusion into methylcellulose. Colonies were assessed at 2-4 weeks and percent colony inhibition calculated. NK-92 did not grow colonies under these conditions.

Animals NOD/SCID gammanull (NSG) mice from The Jackson Laboratory were bred and maintained in the Ontario Cancer Institute animal facility according to protocols approved by the Animal Care Committee. Mice were fed irradiated food and Baytril containing water ad libitum during experimental periods. Prior to infusion with AML, NSG mice were irradiated with 325 or 225 cGy to facilitate engraftment. We developed a primary AML xenograft model utilizing a patient-derived AML sample (details in the Online Supplementary Methods and Supplementary Figure S2). Mice were sacrificed when humane end points were reached as per our animal use protocol (2791). Primary AML xenografted mice were treated with NK-92 (kindly provided by Hans Klingemann) or CD16+NK-92 (provided by Conkwest), with or without murine antibody therapy as follows: 7G3 (provided to RMR under an MTA with CSL Ltd., Parkville, Australia), BM4 isotype control (under MTA to RMR) and MG2a-53 isotype control (Biolegend; 401502; MG2a-53).

Statistical analysis Survival analysis was carried out with Kaplan-Meier survival curves using the log rank rest (P<0.05) with Medcalc software. Comparison of cytotoxicity was made with the two-tailed Student t-test (P<0.05) to compare in vitro cytotoxicity and engraftment data using Medcalc software.

Results NK-92 preferentially kills leukemic stem cells compared with bulk leukemia cells We initially set out to determine the cytotoxicity of NK92 against primary AML cells using chromium release as a measure of bulk tumor cell kill. A panel of 4 primary AML blast samples treated with NK-92 yielded a dose-dependent response and moderate degrees of cytotoxicity against 4 samples at a 25:1 E:T ratio (% lysis): 080179 (42.3±3.6%), 080078 (29.8±3.6%), 08008 (43.9±1.47%), 0909 (42.6±0.1%) (Figure 1A). Primary AML cells were 1721

B.A. Williams et al. A

Patient sample ID

Effector:Target Ratio

killed in a dose-dependent manner and killing was abrogated in the presence of the calcium chelator EGTA that blocks granule exocytosis (Figure 1B). To determine the effect of NK-92 on LSCs, we sorted a primary AML sample into CD34+CD38– and CD34+CD38+ fractions for further testing using the CRA. Primary AML-derived CD34+CD38– cells were more sensitive to killing than CD34+CD38+ blasts by NK-92 in a 4hour CRA at E:T ratios of 1:1 (58.9±11.5%, 20.3±1.7%), 5:1 (78.3±9.7%, 43.5+11.1%) and 10:1 (72.9±5.6%, 38.5±2.4%); this difference was not significant at a 25:1 E:T ratio (Figure 2A). To test the effect of NK-92 against LSCs relative to bulk tumor, we compared the CRA with a methylcellulose cytotoxicity assay (MCA) designed to measure the killing of clonogenic primary AML cells during the 4-hour coincubation. We also employed a control to correct for the effect of NK-92 against leukemia targets during the 4week exposure in methylcellulose by enumerating colonies that arose from infusing NK-92 and targets in 1722

Figure 1. Chromium release assay of NK92 against primary acute myeloid leukemia (AML) samples. (A) Four freshly thawed primary AML blast samples were labeled with 100 mCi of Na251CrO4 prior to treatment with NK-92 at four E:T ratios. (B) AML blast sample 080078 was tested in a separate experiment at four E:T ratios with and without calcium chelator EGTA 4 (mM) and MgCl2 (3 mM). Data are presented as the mean percent lysis of triplicate samples (+/-Standard Deviation) from a representative experiment carried out three times.

methylcellulose without prior incubation. The MCA showed that NK-92 at a 25:1 E:T eliminated clonogenic growth of most primary AML blast samples yielding % colony inhibition values of: 86.3±2.3%, 98.4±2.8%, 100±0% and 100±0%, demonstrating much higher cytotoxicity than obtained with the CRA, which was performed on the same day (Figure 2B and C). To determine if the difference in cytotoxicity measured by the CRA and MCA was not simply due to methodological reasons, we screened additional cell line targets using this methodology. While OCI/AML3 clonogenic cells were more sensitive to NK-92 than bulk tumor, OCI/AML2 bulk and clonogenic cells were equivalently sensitive to NK-92 (Online Supplementary Figure S3). OCI/AML2 was the only target tested which did not have a differential cytotoxicity measured by the CRA and MCA. This demonstrates that the enhanced cytotoxicity measured in the MCA compared with the CRA is, for most targets, not intrinsically related to the method of data comparison. haematologica | 2018; 103(10)

CD16+NK-92 and anti-CD123 antibody therapy for AML

Figure 2. NK-92 cytotoxicity against sorted leukemic stem cells and clonogenic leukemic cells relative to bulk leukemia cells. (A) Primary acute myeloid leukemia (AML) samples were sorted into CD34+CD38â&#x20AC;&#x201C; and CD34+CD38+ fractions for subsequent testing in a chromium release assay with NK-92 at a 25:1 E:T ratio. Data are presented as the mean percent (%) lysis of triplicate samples (+/-Standard Deviation) representative of two separate experiments. (B) Four primary AML samples were incubated with or without NK-92 at a 25:1 E:T ratio for 4 hours in 96-well U bottom plates and utilized in either a chromium release assay (CRA) or a methylcellulose cytotoxicity assay (MCA) conducted on the same day. The % lysis values and % colony inhibition values are plotted together (B). An example of the methylcellulose cytotoxicity assay (C) shows a representative assay for one sample (080179) with a control (AML only) (i), low density control (AML + NK-92 infused into methylcellulose only) (ii), and treatment group (AML + NK-92 co-incubated together in a 96-well plate well and then infused in methylcellulose) (iii).

Irradiated NK-92 reduces leukemic stem cell fraction in the secondary transplantation assay

NK-92 prolongs survival in a primary AML xenograft model

To assess the cytotoxic effect of irradiated NK-92 (iNK-92) on LSCs in the in vivo setting, secondary transplantation experiments were conducted to evaluate total engraftment and fraction of LSCs in secondary recipients. Primary AML cells (3x106) were injected by tail vein into two cohorts of 4 mice and treated with or without iNK-92 from day 2 (15x106 iNK-92 cells) twice weekly to a total dose of 75x106 iNK-92 cells. At six weeks, mice were sacrificed and BM (1x106 cells) from each of the 4 primary recipients (donor mice) in control or treatment groups transplanted into 4 secondary recipient NSG mice (Figure 3A). Evaluation of BM from secondary recipients inoculated with BM from AML-infused mice untreated by iNK92 revealed a high proportion of human CD45+ cells (80.8, 93.3, 80.4, 96.4 Av=87.7%), while one mouse from AML infused iNK-92 treatment group was leukemia-free with engraftment at background levels of non-injected mice (96.4, 94.7, 1.8, 95.7 Av=72.2%). There was no significant difference between secondary engraftment of AML between groups receiving BM from AML-infused mice in the control and iNK-92 therapy groups (Figure 3B). However, the proportion of CD34+CD38â&#x20AC;&#x201C;CD123+ cells in secondary transplanted mice for the AML control group was 7.85% (8.01, 9.48, 8.66, 5.25) and for the AML + iNK92 group was 3.66% (7.13, 3.46, 0.03, 4.00), which was significantly lower (P=0.05) (Figure 3C).

We next sought to assess the impact of NK-92 on survival of mice inoculated with primary human AML. NSG mice inoculated with 3x106 primary AML cells received 10x106 non-irradiated NK-92 weekly for three doses (Figure 4A). This treatment increased median survival from 57 to 72 days (log rank test P<0.01), although most ultimately succumbed to disease (Figure 4B). Autopsy revealed enlarged spleens and pale fragile bones compared with controls. Flow cytometry of BM from NSG mice inoculated with AML only (Online Supplementary Figure S4A), or AML + NK-92 treatments that became symptomatic (Online Supplementary Figure S4B), had 99% engraftment by human leukemia in the bone marrow, while the mouse that survived long term (~9 months) was healthy at sacrifice and did not have evidence of leukemic infiltration in the marrow (Online Supplementary Figure S4C) or splenomegaly. To determine the impact of irradiation on the in vivo activity of NK-92, the cells were irradiated with 1000 cGy prior to infusion into NSG mice inoculated ten days before with 3x106 primary AML cells. NSG mice were administered 20x106 iNK-92 [intraperitoneal injection (i.p.)] weekly x 5 doses and monitored for signs of leukemia (Figure 5A). Survival was improved in the treatment group (26-48 days) to near statistical significance (P=0.0566), but all mice ultimately succumbed to disease (Figure 5B).

haematologica | 2018; 103(10)

1723

B.A. Williams et al. A

% Engraftment

% Leukemic stem cells

Figure 3. Effect of iNK-92 on secondary bone marrow (BM) engraftment of acute myeloid leukemia (AML) cells and leukemic stem cells (LSCs). (A) 3x106 AML cells were also infused intravenously (i.v.) into two cohorts of 4 mice and treated with and without iNK-92 from day 2 and given 15x106 cells twice weekly to a total dose of 75x106 (A). BM (1x106 cells) from each of 4 primary recipients in control and treatment was serially transplanted 1:1 into 4 new NOD/SCID gammanull (NSG) mice. These mice were sacrificed at six weeks and BM assayed for overall leukemic engraftment as determined by presence of % human CD45+ cells (B) and LSC engraftment as determined by % human CD34+CD38â&#x20AC;&#x201C;CD123+ cells (C) (*P=0.05).

CD16+NK-92 mediates ADCC in vitro and in vivo and prolongs survival in an AML xenograft model by targeting leukemic stem cells To develop a strategy to enhance killing of LSCs, we utilized a gene-modified CD16+NK-92 transduced with the high affinity CD16 receptor (NK-92.176V GFP), which is capable of mediating ADCC against antibody-coated targets. The proportion of cells expressing CD16 was 2.3% for parental NK-92 and 27.9% for CD16+ NK-92 (Online Supplementary Figure S5). The chromium release assay was modified to measure ADCC by coating target cells with antibodies prior to the 2-hour chromium incubation. The CD123+ leukemia cell line OCI/AML5 was pretreated with anti-CD123 (7G3) antibody at a dose of 10 Âľg/mL prior to use in a chromium release assay. CD16+NK-92 showed cytotoxicity against OCI/AML5 cells at all E:T ratios, and was significantly enhanced (2-6x) when targets were coated with anti-CD123 mAb (Figure 6), demonstrating effective ADCC. To enhance the approach of irradiated NK-92 against human AML in the xenograft NSG model, we treated the mice with a CD16+NK-92 cell line, in combination with an 1724

anti-CD123 mAb (7G3), given on the same days to facilitate targeting of leukemic stem cells by antibody-dependent cell-mediated cytotoxicity. A blocking dose of isotype control antibody BM4 (200 mg) was given prior to the administration of 7G3 at a low dose (8 mg), due to the known sequestration of 7G3 in the spleen (Online Supplementary Figure S6A). In this pilot experiment, we demonstrated that 7G3 (8 mg) could enhance the therapeutic efficacy of iCD16+ NK-92, as determined by an improvement of median survival by 13 days (P=0.0173) (Online Supplementary Figure S6B). We conducted a more rigorously controlled experiment of CD16+NK-92 and anti-CD123 mAb therapy utilizing an isotype control antibody (BM4) in the control arms (Figure 7A). The cohorts included: no therapy, 7G3, BM4, iCD16+NK-92, iCD16+NK-92 + 7G3, iCD16+NK-92 + BM4. The dosing schedule was iCD16+NK-92, with or without 7G3 or BM4, given on days 3, 5, 7, 10, and 12 after AML inoculation (day 0). NSG mice without therapy had a median survival of 32 days. iCD16+NK-92 alone significantly improved survival to a median of 37 days (P<0.001). Treatment with BM4 did not enhance survival haematologica | 2018; 103(10)

CD16+NK-92 and anti-CD123 antibody therapy for AML

Figure 4. NK-92 therapy of primary acute myeloid leukemia (AML) xenografted NOD/SCID gammanull (NSG) mice. 3x106 primary AML cells were injected intravenously (i.v.) via tail vein into irradiated NOD/SCID gamma null mice to establish disease in control (n=5) and therapy mice (n=10). 10x106 NK-92 were infused via tail vein weekly for three weeks starting on the day of AML inoculation in treatment group (A). Mice were monitored for signs of leukemia and sacrificed at humane end points. Kaplan-Meier survival curves were generated to compare survival in control and treatment groups (P<0.01) (B).

over control with both groups having a median survival of 32 days (P=0.619), but 7G3 significantly improved median survival to 35 days compared with the AML only control (P<0.001), but not the BM4 isotype control (P=0.1509). A combination of 7G3 and iCD16+NK-92 produced the best survival outcome, with a median survival of 42 days, which was significantly enhanced over mice infused with AML only and received no therapy (+10 days; P<0.001), 7G3 (+ 7 days; P<0.0025) and iCD16+NK-92 + BM4 (+ 10 days; P<0.0025) (Figure 7B).

Discussion The overall long-term survival for patients with AML is approximately 40%,2 demonstrating a need for novel treatment strategies, particularly for patients in remission with detectable residual disease. We recently published a phase I clinical trial of NK-92 in relapsed and refractory hematologic malignancies (lymphoma and multiple myeloma) with minimal toxicities despite cumulative doses as high as 150 billion cells, and showed clinical responses.14 Prior studies implicating NK cells as therapeutically relevant in haplotype transplantation for AML20 prompted us to further investigate NK-92 as therapy for AML. We have studied the mechanism of NK-92 cytotoxicity against primary AML samples and its efficacy in a primary haematologica | 2018; 103(10)

human AML xenograft model. We confirmed initial reports that NK-92 mediates cytotoxicity in vitro against primary AML11 and demonstrated that killing was due primarily to granule exocytosis rather than ligand-mediated cytotoxicity (e.g. via Fas ligand) as shown by inhibiting cytotoxicity with the calcium chelator, EGTA. We noted that classically defined, sorted CD34+CD38â&#x20AC;&#x201C; LSCs4 were more sensitive to NK-92 killing than leukemia blast cells at low E:T ratios in a standard chromium release assay. Given the conflicting reports in the literature of the definitive immunophenotype of the LSC in AML,4,5,21,22 we opted to use a clonogenic assay to assess the effect of immune effector cells against LSCs in a larger set of samples. Primary AML grows well in methylcellulose and leads to the generation of visually detectable single cell-derived colonies that identifies the frequency of individual leukemic stem and progenitor cells. Specifically, we used a methylcellulose cytotoxicity assay (MCA) established previously by our lab19 that enables a comparison of the degree of killing of bulk leukemia blast cells versus colony inhibition in a 4-hour period. This approach provides another means to assay differential cytotoxicity against bulk and LSC populations. The MCA demonstrated a 2-3-fold higher % colony inhibition than the % lysis measured by the CRA. These results support our initial finding using cell sorted LSCs, which showed that NK-92 can preferentially recognize and kill LSCs over bulk leukemia. 1725

B.A. Williams et al. A

Figure 5. iNK-92 therapy of primary acute myeloid leukemia (AML) xenografted NOD/SCID gammanull (NSG) mice. 3x106 primary AML cells were injected intravenously (i.v.) via tail vein into irradiated NGS to establish disease in control (n=5) and therapy (n=5) mice. iNK-92 given intraperitoneally (i.p.) 20x206 weekly for six weeks were used to treat AML xenografted mice starting ten days after inoculation (A). Mice were monitored for signs of leukemia and sacrificed at humane end points (B). KaplanMeier survival curves were generated to compare survival in control and treatment groups (P=0.0566).

Only two studies to date have looked at the in vitro sensitivity of CD34+CD38â&#x20AC;&#x201C; LSCs to immune effector cell killing. In the first, lymphokine-activated killer (LAK) cells and allogeneic lymphocytes exerted a modest cytotoxic effect on AML LSCs comparable to the effect on the nonstem cell fraction.23 In a more recent study, endogenous single killer immunoglobulin-like receptor (KIR)-expressing NK cells, mismatched for the HLA of primary AML targets, showed equivalent killing of LSCs and blasts with either the chromium release or methylcellulose-based cytotoxicity assays.24 Here, we demonstrate preferential killing of LSCs versus bulk leukemia by NK-92, not shown by these other studies using a more rigorously controlled methylcellulose cytotoxicity assay than that used by Lankencamp et al. This is consistent with our work on NK-92 treatment of multiple myeloma (MM) cell lines showing that NK-92 preferentially kills clonogenic MM cells over bulk tumor cells.25 To pursue more in-depth studies of the effectiveness of NK-92 in killing LSCs, we developed an animal model of primary human AML by using NSG mice infused with a primary AML sample containing a small fraction of CD34+CD38â&#x20AC;&#x201C; cells. Secondary transplantation is the current gold standard to determine the effect of small molecules on LSCs26 but is rarely used to evaluate cellular therapies for leukemia. We attempted this assay by transplanting bone marrow from control and NK-92 treated mice into new recipients to assess the impact on individual mice rather than on pooled cells. BM engraftment occurred in all AML-only 1726

cohort secondary mice, while one mouse from the iNK-92 group was leukemia free, with engraftment at the background levels of non-injected mice. While the average BM engraftment of secondary transplant mice in the therapy groups was less than the control, this was not statistically significant. However, the LSC fraction was significantly decreased in secondary recipients, providing some evidence of NK-92 cytotoxicity against LSCs in the secondary transplant assay. AML-xenografted NSG mice were effectively treated with NK-92 infusions, leading to improvement in survival versus controls, confirming previous work.11 We accomplished this with lower doses of NK-92 on a less compressed schedule than the original study, and without the use of IL-2 in the regimen. Irradiated NK-92 could prolong survival in mice, but was less effective than the non-irradiated cells. We postulate that the reason for this reduction in therapeutic efficacy is the lack of ability for the cells to proliferate in vivo. We have demonstrated for the first time that irradiated NK-92 improves survival in an AML xenograft model, which has translational relevance given that only irradiated NK-92 is administered to patients in phase I trials. However, recognizing that the effect of iNK-92 on improving survival in vivo is modest, we wished to add the mechanism of ADCC to cell killing by using the genetically modified CD16+NK-92 in combination with a monoclonal antibody. CD16+NK-92 has been combined with rituximab to enhance killing of CD20+ malignant cells, showing its potential to enhance the killing of cells haematologica | 2018; 103(10)

CD16+NK-92 and anti-CD123 antibody therapy for AML

Figure 6. CD16+NK-92 in vitro ADCC assay against primary acute myeloid leukemia (AML) and OCI/AML5. OCI/AML5 cells were labeled with 100 mCi of Na251CrO4 for 2 hours +/- 10 mg/mL of 7G3 (anti-CD123 mAb) prior to treatment with CD16+NK-92 in 96-well plates in a standard chromium release assay. Data are presented as the mean percent lysis of triplicate samples (+/Standard Deviation) from a representative experiment carried out twice.

expressing a tumor-associated antigen.18 CD16+NK-92 cytotoxicity against OCI/AML5 was enhanced via ADCC when target cells were coated with a murine IgG2a antihuman CD123 mAb (7G3), indicating the ability to redirect CD16+NK-92 against a LSC-associated antigen. Cross-species interaction between murine Fc gamma receptors and human immune effectors cells has been reported, with murine IgG2a being the most effective isotype subclass in facilitating ADCC.27,28 However, the enhancement in CD16+NK-92 cytotoxicity seen here with 7G3 is likely an underestimate of the full potential of these cells to mediate ADCC, as the antibodies were murine rather than human in origin. We then sought to combine iCD16+NK-92 and 7G3 therapy in our AML xenograft model and built upon prior work which demonstrated that systemic treatment with 7G3 alone in an AML NOD/SCID xenograft model reduced BM engraftment of human leukemia.29 We initially used a very small quantity of 7G3 in a single dose, preceded by blocking with a non-specific isotypematched antibody to block Fc receptors, and improve circulation and binding of 7G3 to CD123, as established by Leyton et al.30 This approach worked in increasing the efficacy of the iCD16+NK-92 cells and improved survival. In a follow-up experiment, we did not use an Fc blocking pre-dose strategy, but gave 100 mg of 7G3 or isotype control antibody BM4 for five doses with or without the iCD16+NK-92 cells. In this experiment, iCD16+NK-92 alone prolonged survival over control. The BM4 antibody had no therapeutic effect and did not enhance iCD16+NK92, while 7G3 alone had a modest survival benefit above control, which was not statistically significant above BM4 isotype control group. Of note, the best outcome was in the iCD16+NK-92 + 7G3 treated group which had a 10day improvement in median survival compared with the iCD16+NK-92 + BM4 treatment group. Our data, therefore, demonstrate that the combination of iCD16+NK-92 and 7G3 can improve survival by antibody-dependent cell-mediated cytotoxicity. Furthermore, this represents the first demonstration of in vivo efficacy of the CD16+NK-92 cell line alone and in combination with antibody, which has only previously been tested in vitro.18 The haematologica | 2018; 103(10)

Fc optimized anti-CD123 humanized monoclonal antibody CSL362 (derived from 7G3) can facilitate ADCC from peripheral blood-derived allogeneic NK cells against primary AML and CD123-expressing cell-line targets31 and is currently being tested in several clinical trials for AML. A recent study demonstrated potent CSL362-mediated NK cell ADCC against primary AML blasts and LSCs with comparable results seen with NK cells from AML patients and healthy donors.32 CD16+NK-92 cells which express the high affinity CD16 receptors could be combined with CSL362 as a potentially potent treatment of AML. Our strategy of targeting LSCs by anti-CD123-facilitated ADCC in vivo had a comparable improvement in median survival to CD123 CAR T-cell therapy in an AML NSG xenograft model.33 The latter, however, used the KG1a line and not primary AML cells, as in our case. Using a different CAR vector, another group tested CD123 CAR T cells in a primary AML model with no improvement in median survival, but had approximately 40% long-term survivors at day 100 (approx. 70 days after control median survival), which was statistically significant.34 One other study of CD123 CAR T cells demonstrated efficacy against LSCs using primary and secondary engraftment models, but did not have a survival end point.35 Another group used CD123 CAR-transduced cytokine-induced killer (CIK) cells, and showed in vitro efficacy against a CD123+ cell line and primary AML targets.36 A limitation of CD123 CAR T-cell therapy is that it can target HSCs with low CD123 expression, making it a potentially myeloablative therapy, suitable only in conjunction with stem cell transplantation.34 In contrast, a clinical study of antibody based targeting of CD123 using CSL360 did not exhibit myeloablation, despite having the potential for NK-mediated ADCC against LSCs with low expression of CD123.37 Therefore, we anticipate that our approach utilizing irradiated CD16+NK-92 in combination with anti-CD123 antibodies may have a reduced risk of causing life-threatening myelosuppression, but this remains to be shown in clinical studies of both these approaches. In summary, we have shown that NK-92 preferentially targets LSCs over bulk leukemia blasts in vitro and irradiated NK-92 can improve survival in an AML xenograft 1727

B.A. Williams et al. A

Figure 7. iCD16+NK-92 +/- 7G3 or isotype control treatment of primary acute myeloid leukemia (AML) xenografted mice. NOD/SCID gammanull (NSG) mice were inoculated intravenously (i.v.) with 3x106 passage human AML spleen-derived cells (day 0) and treated with iCD16+NK-92 +/- 7G3 or BM4 x 5 doses [intraperitoneal injection (i.p.)] (3x/week) starting on day 3 (A). Controls included no therapy and antibodies alone (n=5 for all groups). Survival was determined using Kaplan-Meier survival analysis with a log rank test (B).

model, which can be enhanced using CD16+NK-92 combined with an anti-CD123 monoclonal antibody. This provides the first proof-of-principle for the targeting of LSCs by combining an antibody and a standardized cellular therapy. A humanized version of 7G3 (CLS362) is being tested in clinical trials for AML, but is reliant upon a patientâ&#x20AC;&#x2122;s endogenous NK-cell function for efficacy. Combination therapy with ADCC capable NK cell lines such as CD16+NK-92 or the new haNK platform (CD16+IL-2+NK-92) with CSL362 would enable the therapeutic translation of our approach into a clinical trial in the future. haNK cells are currently in clinical trials for solid tumors in combination with FDA-approved monoclonal antibodies. The approach we have demonstrated can be readily applied to enhance the targeting of any antigen, 1728

and its particular novelty here is in demonstrating the targeting of a cancer stem cell marker and a consequent improvement in survival in a murine AML xenograft model. Funding BW was supported by a Terry Fox Foundation Award from the National Cancer Institute of Canada, CIHR clinician scientist training award, as well as grants from the Hospital for Sick Children, Ontario Cancer Institute and University of Toronto. AK was supported by the Gloria and Seymour Epstein Chair in Cell Therapy and Transplantation at the University Health Network and University of Toronto. We would like to acknowledge the generous provision of 7G3 for some experiments by Angel Lopez. haematologica | 2018; 103(10)

CD16+NK-92 and anti-CD123 antibody therapy for AML

References 1. Hurwitz CA, Mounce KG, Grier HE. Treatment of patients with acute myelogenous leukemia: review of clinical trials of the past decade. J Pediatr Hematol Oncol. 1995;17(3):185-197. 2. Lowenberg B, Downing JR, Burnett A. Acute myeloid leukemia. N Engl J Med. 1999;341(14):1051-1062. 3. Ribeiro RC, Razzouk BI, Pounds S, Hijiya N, Pui CH, Rubnitz JE. Successive clinical trials for childhood acute myeloid leukemia at St Jude Children's Research Hospital, from 1980 to 2000. Leukemia. 2005; 19(12):2125-2129. 4. Lapidot T, Sirard C, Vormoor J, et al. A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature. 1994;367(6464):645-648. 5. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(7):730-737. 6. Jordan CT, Upchurch D, Szilvassy SJ, et al. The interleukin-3 receptor alpha chain is a unique marker for human acute myelogenous leukemia stem cells. Leukemia. 2000; 14(10):1777-1784. 7. Vergez F, Green AS, Tamburini J, et al. High levels of CD34+CD38low/-CD123+ blasts are predictive of an adverse outcome in acute myeloid leukemia: a Groupe OuestEst des Leucemies Aigues et Maladies du Sang (GOELAMS) study. Haematologica. 2011;96(12):1792-1798. 8. Miller JS, Soignier Y, Panoskaltsis-Mortari A, et al. Successful adoptive transfer and in vivo expansion of human haploidentical NK cells in patients with cancer. Blood. 2005;105(8):3051-3057. 9. Rubnitz JE, Inaba H, Ribeiro RC, et al. NKAML: a pilot study to determine the safety and feasibility of haploidentical natural killer cell transplantation in childhood acute myeloid leukemia. J Clin Oncol. 2010;28(6):955-959. 10. Gong JH, Maki G, Klingemann HG. Characterization of a human cell line (NK92) with phenotypical and functional characteristics of activated natural killer cells. Leukemia. 1994;8(4):652. 11. Yan Y, Steinherz P, Klingemann HG, et al. Antileukemia activity of a natural killer cell line against human leukemias. Clin Cancer Res. 1998;4(11):2859-2868. 12. Arai S, Meagher R, Swearingen M, et al. Infusion of the allogeneic cell line NK-92 in patients with advanced renal cell cancer or melanoma: a phase I trial. Cytotherapy. 2008;10(6):625-632. 13. Tonn T, Schwabe D, Klingemann HG, et al. Treatment of patients with advanced cancer with the natural killer cell line NK-92. Cytotherapy. 2013;15(12):1563-1570. 14. Williams BA, Law AD, Routy B, et al. A

haematologica | 2018; 103(10)

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

phase I trial of NK-92 cells for refractory hematological malignancies relapsing after autologous hematopoietic cell transplantation shows safety and evidence of efficacy. Oncotarget. 2017;8(51):89256-89268. Klingemann H, Wong E, Maki G. A cytotoxic NK-cell line (NK-92) for ex vivo purging of leukemia from blood. Bone Marrow Transplant. 1996;2(2):68-75. Tonn T, Becker S, Esser R, Schwabe D, Seifried E. Cellular immunotherapy of malignancies using the clonal natural killer cell line NK-92. J Hematother Stem Cell Res. 2001;10(4):535-544. Tam YK, Miyagawa B, Ho VC, Klingemann HG. Immunotherapy of malignant melanoma in a SCID mouse model using the highly cytotoxic natural killer cell line NK-92. J Hematother. 1999;8(3):281-290. Binyamin L, Alpaugh RK, Hughes TL, Lutz CT, Campbell KS, Weiner LM. Blocking NK cell inhibitory self-recognition promotes antibody-dependent cellular cytotoxicity in a model of anti-lymphoma therapy. J Immunol. 2008;180(9):6392-6401. Williams BA, Wang XH, Keating A. Clonogenic assays measure leukemia stem cell killing not detectable by chromium release and flow cytometric cytotoxicity assays. Cytotherapy. 2010;12(7):951-960. Ruggeri L, Capanni M, Urbani E, et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science. 2002;295(5562):20972100. Taussig DC, Miraki-Moud F, Anjos-Afonso F, et al. Anti-CD38 antibody-mediated clearance of human repopulating cells masks the heterogeneity of leukemia-initiating cells. Blood. 2008;112(3):568-575. Goardon N, Marchi E, Atzberger A, et al. Coexistence of LMPP-like and GMP-like leukemia stem cells in acute myeloid leukemia. Cancer Cell. 2011;19(1):138-152. Costello RT, Mallet F, Gaugler B, et al. Human acute myeloid leukemia CD34+/CD38- progenitor cells have decreased sensitivity to chemotherapy and Fas-induced apoptosis, reduced immunogenicity, and impaired dendritic cell transformation capacities. Cancer Res. 2000; 60(16):4403-4411. Langenkamp U, Siegler U, Jorger S, et al. Human acute myeloid leukemia CD34+CD38- stem cells are susceptible to allorecognition and lysis by single KIRexpressing natural killer cells. Haematologica. 2009;94(11):1590-1594. Swift BE, Williams BA, Kosaka Y, et al. Natural killer cell lines preferentially kill clonogenic multiple myeloma cells and decrease myeloma engraftment in a bioluminescent xenograft mouse model. Haematologica. 2012;97(7):1020-1028. Skrtic M, Sriskanthadevan S, Jhas B, et al. Inhibition of mitochondrial translation as a

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

therapeutic strategy for human acute myeloid leukemia. Cancer Cell. 2011; 20(5):674-688. Kipps TJ, Parham P, Punt J, Herzenberg LA. Importance of immunoglobulin isotype in human antibody-dependent, cell-mediated cytotoxicity directed by murine monoclonal antibodies. J Exp Med. 1985;161(1):1-17. Biddle WC, Pancook J, Goldrosen M, Han T, Foon KA, Vaickus L. Antibody-dependent, cell-mediated cytotoxicity by an anticlass II murine monoclonal antibody: effects of recombinant interleukin 2 on human effector cell lysis of human B-cell tumors. Cancer Res. 1990;50(10):29912996. Jin L, Lee EM, Ramshaw HS, et al. Monoclonal antibody-mediated targeting of CD123, IL-3 receptor alpha chain, eliminates human acute myeloid leukemic stem cells. Cell Stem Cell. 2009;5(1):31-42. Leyton JV, Hu M, Gao C, et al. Auger electron radioimmunotherapeutic agent specific for the CD123+/CD131- phenotype of the leukemia stem cell population. J Nucl Med. 2011;52(9):1465-1473. Busfield SJ, Biondo M, Wong M, et al. Targeting of acute myeloid leukemia in vitro and in vivo with an anti-CD123 mAb engineered for optimal ADCC. Leukemia. 2014;28(11):2213-2221. Xie LH, Biondo M, Busfield SJ, et al. CD123 target validation and preclinical evaluation of ADCC activity of anti-CD123 antibody CSL362 in combination with NKs from AML patients in remission. Blood Cancer J. 2017;7(6):e567. Mardiros A, Dos Santos C, McDonald T, et al. T cells expressing CD123-specific chimeric antigen receptors exhibit specific cytolytic effector functions and antitumor effects against human acute myeloid leukemia. Blood. 2013;122(18):3138-3148. Gill S, Tasian SK, Ruella M, et al. Preclinical targeting of human acute myeloid leukemia and myeloablation using chimeric antigen receptor-modified T cells. Blood. 2014; 123(15):2343-2354. Pizzitola I, Anjos-Afonso F, Rouault-Pierre K, et al. Chimeric antigen receptors against CD33/CD123 antigens efficiently target primary acute myeloid leukemia cells in vivo. Leukemia. 2014;28(8):1596-1605. Tettamanti S, Marin V, Pizzitola I, et al. Targeting of acute myeloid leukaemia by cytokine-induced killer cells redirected with a novel CD123-specific chimeric antigen receptor. Br J Haematol. 2013; 161(3):389-401. He SZ, Busfield S, Ritchie DS, et al. A Phase 1 study of the safety, pharmacokinetics and anti-leukemic activity of the anti-CD123 monoclonal antibody CSL360 in relapsed, refractory or high-risk acute myeloid leukemia. Leuk Lymphoma. 2015; 56(5):1406-1415.

1729

ARTICLE

Coagulation & its Disorders

Ferrata Storti Foundation

Cytoprotective and pro-angiogenic functions of thrombomodulin are preserved in the C loop of the fifth epidermal growth factor-like domain

Xiangmin Wang,1,2 Bin Pan,1,2 Goichi Honda,3 Xintao Wang,2 Yuko Hashimoto,4 Hiroshi Ohkawara,2 Kailin Xu,1 Lingyu Zeng1 and Takayuki Ikezoe2 Department of Hematology, The Affiliated Hospital of Xuzhou Medical University, Xuzhou, China; 2Department of Hematology, Fukushima Medical University, Japan; 3 Medical Affairs Department, Asahi Kasei Pharma, Kanda Jinbocho, Chiyoda-ku, Tokyo, Japan and 4Department of Diagnostic Pathology, Fukushima Medical University, Japan 1

Haematologica 2018 Volume 103(10):1730-1740

ABSTRACT

Correspondence: ikezoet@fmu.ac.jp or zengly2000@163.com

Received: December 18, 2017. Accepted: June 13, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2017.184481 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1730 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1730

e previously found that the fifth epidermal growth factor-like domain of thrombomodulin (TME5) exerts cytoprotective and pro-angiogenic functions via G-protein coupled receptor 15 (GPR15). TME5 is comprised of three S-S bonds that divide it into three loops: A (TME5A), B (TME5B), and C (TME5C). Herein we identified the minimum structure of TME5 that produces favorable effects in vascular endothelial cells (ECs). We found that TME5C, composed of 19 amino acids, but not TME5A or TME5B, stimulated the proliferation of human umbilical vein endothelial cells (HUVECs) and human hepatic sinusoidal endothelial cells (HHSECs). Matrigel plug assays showed that TME5C stimulates in vivo angiogenesis. In addition, TME5C counteracted calcineurin inhibitor-induced apoptosis and vascular permeability in HUVECs and HHSECs. Western blot analysis indicated that exposure of either HUVECs or HHSECs to TME5C increased the levels of anti-apoptotic myeloid cell leukemia-1 protein in association with the activation of signal transduction pathways, including extracellular signal-regulated kinase, AKT, and mitogen-activated protein kinase p38. Importantly, TME5C did not affect the coagulation pathway in vitro. The cytoprotective function of TME5C was mediated by cell surface-expressed GPR15, as TME5C was not able to protect vascular ECs isolated from Gpr15 knock-out (KO) mice. Strikingly, TME5C successfully ameliorated sinusoidal obstruction syndrome in a murine model by counteracting the reduction of sinusoidal EC numbers. Taken together, the cytoprotective and pro-angiogenetic functions of TM are preserved in TME5C. The use of TME5C may be a promising treatment strategy to prevent or treat lethal complications, such as sinusoidal obstruction syndrome, whose pathogenesis is based on endothelial insults.

Introduction Hepatic sinusoidal obstruction syndrome (SOS) is a potentially life-threatening complication after hematopoietic stem cell transplantation (HSCT).1 The incidence of SOS ranges from 5% to 60%, depending on the conditioning regimen and transplantation type.2 The clinical manifestation of SOS includes rapid and unexplained weight gain, ascites, painful hepatomegaly, and jaundice.3 The development of SOS after HSCT is associated with injury to sinusoidal endothelial cells (ECs) and hepatocytes via a variety of factors, including a hepatotoxic conditioning regimen, immunosuppressive treatments with calcineurin inhibitors, and lipopolysaccharide (LPS) released by gram-negative bacteria.4,5 Due to the paucity of glutathione content in zone III of the liver, sinusoidal ECs are more vulnerable to toxic agents than hepatocytes.5 SOS is now referred to as an endothelial syndrome together with transplant-associated thrombotic microangiopathy (TA-TMA) and engraftment syndrome (ES).2,6 As a result of endothelial injury, a hypercoagulable state is caused in patients with ES.7 As of haematologica | 2018; 103(10)

TME5C exerts cytoprotective and angiogenic functions

yet, clinical trials evaluating the efficacy of anticoagulants or thrombolytics for treatment of endothelial syndrome have not been conducted.8-10 Since 2008, recombinant human soluble thrombomodulin (rTM) has been used to treat disseminated intravascular coagulation in Japan. rTM binds thrombin and converts protein C to activated protein C (APC), which inhibits activated factor V and VIII and acts as an anticoagulant.11-14 APC is well known to protect various cell types, including endothelial cells and podocytes, via protease-activated receptor 1 and endothelial protein C receptor.15 rTM counteracted capillary leakage in a patient who developed ES after HSCT.16 In addition, rTM rescues individuals with SOS and TA-TMA developed after HSCT.17-21 We previously showed that rTM possesses the ability to protect vascular ECs in an APC-dependent and ABC-independent manner from various insults, including the calcineurin inhibitor cyclosporine.22,23 In addition, we found that the cytoprotective and pro-angiogenic functions of rTM are localized in the fifth epidermal growth factor-like domain of thrombomodulin (TME5), which does not possess the ability to produce APC, although it retains some binding capacity towards thrombin.24 Furthermore, we found that G protein-coupled receptor 15 (GPR15) expressed on vascular ECs is indispensable for the cytoprotective functions of TME5.25,26 TME5 consists of 40 amino acids, including six cysteine residues that form three disulfide bonds, making TME5 a structure with three separate disulfide-bonded loops: the A loop (residues C390 to C395), B loop (residues C399 to C407), and C loop (residues C409 to C421).27 In contrast with the A and B loops, the structure of the C-loop is similar to that of epidermal growth factor (EGF).27 In the study herein, we identified the minimum structure of TME5 that exerts its cytoprotective and pro-angiogenic activities in vitro and in vivo.

Methods

Table 1. Amino acid sequences of TME5A, TME5B, and TME5C.

Name TME5A TME5B TME5C TME5C mutant

Amino acid sequence QMFCNQTACPA DCDPNTQASCE ECPEGYILDDGFICTDIDE ECPEAYILDDGFICTDIDE

TM: thrombomodulin.

used for experiments. Female C57BL/6 mice (8-week-old) were purchased from Japan SLC, Inc. (Hamamatsu, Japan). Female BALB/c (H-2Kd, donor) and female C57BL/6 (H-2Kb, recipient) mice, aged ten weeks and weighing 20-25 g, were purchased from Japan SLC, Inc. All procedures were performed according to the animal care guidelines of Fukushima Medical University. During invasive operations, animals were anesthetized by inhaling isoflurane.

Reagents TM mutants TME5A (residues C387-C397), TME5B (residues C398-C408), and TME5C (residues C408-C426) were synthesized by the Peptide Institute Inc. (Osaka, Japan). The TME5C mutant with a single amino acid substitution was synthesized by GL Biochem (Shanghai, China). The amino acid sequences are listed in Table 1. Cyclophosphamide (CY) was purchased from Shionogi & Co., Ltd (Osaka, Japan). Busulfan (BU) and tacrolimus (FK506) were purchased from Sigma-Aldrich, Tokyo, Japan. TME5 and rTM were provided by Asahi Kasei Pharma (Tokyo, Japan).

Proliferation assays HUVECs (5×103 cells/well), HHSECs (5×103 cells/well), or murine ECs (5×103 cells/well) were cultured in 96-well plates containing TME5C (25, 50, 250, 500, 1000 nM), TME5A (500 nM), TME5B (500 nM), or TME5 (30 nM) with or without FK506 (10 mg/ml) for 24 h. Bromodeoxyuridine (BrdU, 10 mM/well) was added and incubated for an additional 4 h. The quantity of BrdU incorporated into cells was assessed in accordance with the manufacturer’s protocol (Roche, Basel, Switzerland).

Cell culture HUVECs were purchased from Lonza Walkersville Inc. (Walkersville, MD, USA) and cultured in endothelial cell growth basal medium-2 (EBM-2) culture medium supplemented with endothelial cell growth factors (EGM-2; Lonza Walkersville Inc.). Human hepatic sinusoidal endothelial cells (HHSECs) were purchased from ScienCell (San Diego, CA, USA) and cultured (37°C, 5% CO2) in endothelial cell medium (ECM, containing 5% fetal bovine serum (FBS; ScienCell). Murine thoracic aorta vascular ECs were isolated from mice as previously described.28 Briefly, mice were anesthetized, and the thoraces were opened to expose the heart and lungs. The aorta was dissected out and immersed in 20% FBS dulbecco's modified eagle medium (FBS-DMEM; Wako, Tokyo, Japan) in the presence of collagenase type II (Sigma-Aldrich, Tokyo, Japan) for 45 min at 37°C. The cells were then collected and cultured with DMEM supplemented with endothelial cell growth supplement (Sigma-Aldrich, Tokyo, Japan). Five days later, the cells were harvested and utilized for further experiments.

Mice Gpr15 knockout (Gpr15 KO) mice (129/SvEv; 129P2-Gpr15 tm1.1Litt/J, stock number 008769) were purchased from Jackson Laboratory (Bar Harbor, ME, USA). This strain had been backcrossed to C57BL/6 for three generations before being haematologica | 2018; 103(10)

Vascular permeability assay The effects of FK506 and TM mutants on vascular permeability were measured by a vascular permeability assay kit (Millipore, Billerica, MA). Briefly, HUVECs were plated onto collagen-coated inserts and cultured for 72 h until confluence. After starvation for 24 h, cells were treated with TME5A/B/C (500 nM) or TME5 (30 nM) with or without FK506 (10 mg/ml) for 12 h. Then fluorescein isothiocyanate-dextran was added. The extent of permeability was determined by measuring the fluorescence of the plate well solution (excitation: 485 nm, emission: 535 nm).

In vitro vascular tube formation assay To evaluate the pro-angiogenetic effects of TME5C, TME5A, and TME5B in vitro, HUVECs or HHSECs were plated on growth factor-reduced matrigel (Corning corporation, NY, USA) precoated 24-well plates (2.0×104 cells/well) and incubated with control diluent, TME5 (30 nM), TME5A/B/C (500 nM), or vascular endothelial growth factor (VEGF, 0.5 nM, positive control). After 8 h, the endothelial cell-derived tube-like structure was photographed using an inverted microscope (KEYENCE BZX700, Osaka, Japan) (magnification ×40). The tube length in three randomly chosen fields from each well was measured using NIH ImageJ software (NIH, Bethesda, MA, USA). 1731

X. Wang et al.

Figure 1. TME5C stimulates proliferation of endothelial cells. (A). Amino acid sequence alignment of TME5A, TME5B, and TME5C. (B, C). BrdU incorporation assay. HUVECs or HHSECs were cultured with TME5C (25, 50, 250, 500, 1000 nM), TME5A (500 nM), TME5B (500 nM), TME5C mutant (500 nM), or TME5 (30 nM) for 24 h. Proliferation was measured by BrdU incorporation assays. Experiments were performed three times in triplicate plates. Results represent the mean Âą SD. *P<0.05. BrdU: bromodeoxyuridine; HUVECs: human umbilical vein endothelial cells; HHSECs: human hepatic sinusoidal endothelial cells; TM: thrombomodulin; N.S.: not significant.

Murine angiogenesis assay To assess the pro-angiogenetic effects of TM mutants in vivo, growth factor-reduced matrigel (0.3 mL, containing 40 U/mL heparin) with control diluent, TME5A/B/C (500 nM), TME5 (30 nM), VEGF (0.5 nM, positive control) was subcutaneously injected into C57BL/6 mice (8-week-old, female) near the abdominal midline. Four days later, mice were euthanized, and the matrigel plugs were dissected out and photographed.

Hemoglobin determination of matrigel plugs Matrigel plugs were mixed with 1 ml distilled water and put on ice for 5-10 min. After centrifugation for 6 min at 8000 g, the supernatants were mixed with drabkinâ&#x20AC;&#x2122;s reagent (Sigma-Aldrich, Tokyo, Japan) and hemoglobin was measured as previously described.24 Absorbance was measured with a microplate reader at 540 nm. Methemoglobin (Sigma-Aldrich, Tokyo, Japan) was used to obtain a standard curve.

Apoptosis Assays The ability of TM mutants to rescue HUVECs, HHSECs, or murine ECs from FK506-induced apoptosis was measured using the Annexin V Apoptosis Detection Kit (K129, BioVision, Milpitas, CA, USA) and propidium iodide (PI) as previously described.22 Briefly, cells were exposed to FK506 (10 mg/ml) with or without TME5A/B/C (500 nM) or TME5 (30 nM). After 36 h, cells were harvested and subjected to PI and PE-Cy5 anti-annexin V. Early apoptosis cells are annexin V positive and PI negative. Late apoptosis cells are annexin V positive and PI positive.

Western blot analysis Western blot analysis was performed as described previously.22 The following antibodies were used: anti-p-ERK (T202/Y204) (Cell Signaling Technology, Danvers, MA, USA), 1732

anti-ERK (Cell Signaling Technology; 9102), anti-p-AKT (Ser473) (Cell Signaling Technology; 9271), anti-AKT (Cell Signaling Technology; 9272), anti-p-Stat5 (Tyr694) (Cell Signaling Technology; 9351), anti-Stat5 (Cell Signaling Technology; 9363), anti-p38 (Cell Signaling Technology; 9212 ), anti-p-p38 (Tyr180/182) (Cell Signaling Technology; 9216), anti-Mcl-1 (Cell Signaling Technology; 4572), and anti-GAPDH (Cell Signaling Technology; 5174).

Prothrombin time (PT) and activated partial thromboplastin time (APTT)

For PT detection, 200 ml PT reagent (Sysmex Corporation, Kobe, Japan) was mixed with 100 ml human plasma with or without rTM, TME5, TME5A/B/C, or TME5C mutant (10 ml). For APTT detection, 100 ml APTT reagent (Sysmex Corporation) was mixed with 100 ml human plasma with or without rTM, TME5, TME5A/B/C, or TME5C mutant (10 ml). After incubation for 120 s, 100 ml CaCl2 was added to this mixture. The clotting time was measured using a KC1 Delta coagulometer (Tcoag, Co. Wicklow, Ireland).

SOS murine model C57BL/6 mice were randomly divided into three groups (n=16 in each group): mice that received vehicle phosphate buffered saline (PBS) without bone marrow transplantation (BMT) were defined as the control group. Mice that received BU/CY followed by BMT and treated with vehicle PBS were defined as the BMT group. Mice that received BU/CY followed by BMT and treated with TME5C were defined as the BMT treated with TME5C group. SOS was induced by previously reported BU/CY reconditioning treatment followed by BMT with some modification;29 in brief, BU (25 mg/kg/day for 4 days) followed by CY (100 mg/kg/day for 2 days) were given to haematologica | 2018; 103(10)

TME5C exerts cytoprotective and angiogenic functions

Figure 2. TME5C stimulates angiogenesis in endothelial cells. (A, B, E). In vitro vascular tube formation assays. HUVECs or HHSECs were plated on growth factorreduced matrigel-precoated 24-well plates (2.0x104 cells/well) and incubated with control diluent, TME5 (30 nM), TME5A/B/C (500 nM), TME5C mutant (500 nM) or VEGF (0.5 nM, positive control). After 8 h, the endothelial cell-derived tube-like structure was photographed. (C, D, F). The tube length in three randomly chosen fields from each well was measured using NIH ImageJ software. (G, H). In vivo angiogenesis assays. Growth factor-reduced matrigel (0.3 ml, containing 40 U/ml heparin) with control diluent, TME5 (30 nM), TME5A/B/C (500 nM), or VEGF (0.5 nM, positive control) was subcutaneously injected into C57BL/6 mice near the abdominal midline (n=3 in each group). Four days later, mice were euthanized, and the matrigel plugs were dissected out and photographed. The matrigel plugs were homogenized in the presence of 1 ml distilled water and mixed with drabkin’s reagent. The hemoglobin levels were then measured using a microplate reader. Results represent the mean ± SD. *P<0.05. VEGF: vascular endothelial growth factor; HUVECs: human umbilical vein endothelial cells; HHSECs: human hepatic sinusoidal endothelial cells; TM: thrombomodulin; N.S.: not significant.

the mice intraperitoneally from day -7 to day -4 and from day 3 to day -2, respectively. Two days later, the mice were intravenously infused with bone marrow cells harvested from BALB/C mice (5×106 per mouse). The day of BMT was set as day 0. Intraperitoneal administration of either TME5C (500 mg/kg) or vehicle PBS was initiated on day -7 and continued to day 13. Each agent was given to mice every other day. Blood was withdrawn, and plasma levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were measured on days 7, 14, and 20.

scope. Some liver slices were treated with 3% H2O2 and blocked with 1% bovine serum albumin (BSA). The slices were then incubated with primary pan-endothelial cell monoclonal antibody (MECA-32, Novus Biologicals, Littleton, MA, USA) followed by incubation with biotinylated goat anti-rat secondary antibody and ABC HRP reagent. Color was developed with 3,3′diaminobenzidine. Quantification of MECA-32-positive stained sinusoidal ECs was performed using NIH ImageJ software and expressed as the number of positive stained cells/analyzed area. Masson staining was carried out in accordance with the manufacturer’s protocol (Sigma-Aldrich, Tokyo, Japan).

Hematoxylin-eosin (H&E), immunohistochemistry (IHC), and Masson staining

SOS score

On days 7, 14, and 20 after BMT, some of the mice were sacrificed. Livers were surgically removed and fixed with formaldehyde solution. Specimens were dehydrated, waxed, and sliced into 4-mm thickness by an RM2126 microtome. After H&E staining, pathologic changes were evaluated under a light micro-

Histological slices after H&E, IHC, or Masson staining were blindly evaluated according to the scoring system modified from that described by DeLeve et al.30 Based on the total score, the observed SOS was ranked as mild, moderate, or severe as previously described.31

haematologica | 2018; 103(10)

1733

X. Wang et al.

TUNEL staining

ELISA

Apoptosis of hepatocytes and sinusoidal ECs was assessed using an in situ cell death detection kit (Roche) according to the manufacturer’s protocol. Briefly, paraffin-embedded liver tissue sections were pretreated with dewaxation, rehydration, and proteinase K working solution, and subsequently the terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) reaction mixture was added before adding the converter-POD. After the substrate solution was added, the slides were evaluated under a light microscope (×400).

The plasma of mice was collected and analyzed with an enzyme-linked immunosorbent assay (ELISA) kit to measure the concentrations of TM, fibrinogen/fibrin degradation product (FDP), and plasminogen activator inhibitor-1 (PAI-1) according to the manufacturer’s protocol (Cloud-Clone Corp. Wuhan, China).

Statistical analysis Statistical analyses were performed to assess the differences

Figure 3. TME5C blocks apoptosis of FK506-treated endothelial cells. (A). BrdU incorporation assay. HUVECs or HHSECs were cultured with TME5C (25, 50, 250, 500, 1000 nM), TME5A (500 nM), TME5B (500 nM), TME5C mutant (500 nM), or TME5 (30 nM) in combination with FK506 (10 mg/ml) for 24 h. Proliferation was measured by BrdU incorporation assays (n=3). (B, C). Apoptosis assays. HUVECs or HHSECs were exposed to FK506 (10 mg/ml) with or without TME5A/B/C (500 nM) or TME5 (30 nM). After 36 h, cells were harvested, stained with anti-annexin V and PI, and subjected to FACS. Annexin V+PI- and Annexin V+PI+ indicate early and late apoptosis, respectively. (D, E). Quantitative analysis of the apoptotic cells in each group (n=3). (F). Vascular permeability assays. HUVEC monolayers were exposed to TME5 (30 nM) or TME5A/B/C (500 nM) with or without FK506 (10 mg/ml) for 12 h, and then fluorescein isothiocyanate-dextran was added. The fluorescence of the plate well solution was measured to quantify the extent of permeability. Experiments were performed three times. Results represent the mean ± SD. *P<0.05. BrdU: bromodeoxyuridine; HUVECs: human umbilical vein endothelial cells; HHSECs: human hepatic sinusoidal endothelial cells; PI: propidium iodide. FK506: tacrolimus; TM: thrombomodulin; N.S.: not significant.

1734

haematologica | 2018; 103(10)

TME5C exerts cytoprotective and angiogenic functions

between two groups under multiple conditions using one-way analysis of variance (ANOVA) followed by Bonferroni multiple comparison tests using GraphPad Software (La Jolla, CA, USA). Differences in animal survival (Kaplan-Meier survival curves) were analyzed by log-rank test. A P-value < 0.05 was considered statistically significant.

Results TME5C, but not TME5A or TME5B, stimulates proliferation of endothelial cells Figure 1A shows the amino acid sequences of TM mutants TME5A, TME5B, and TME5C. We first examined the effects of each TM mutant on the proliferation of

HUVECs or HHSECs. Exposure to TME5C (25-1000 nM) but not molar equivalents of TME5A or TME5B stimulated their proliferation in a dose-dependent manner, as assessed by BrdU incorporation assays. For example, 500 nM TME5C stimulated proliferation of HUVEC and HHSECs by nearly 1.5-fold. The highest dose of TME5C (1000 nM) did not further stimulate the proliferation of HUVECs and HHSECs (Figure 1B). TME5 also stimulated the proliferation of endothelial cells in a dose-dependent manner, which was consistent with our previous study.25 TME5 produced the maximum pro-proliferative effect at a concentration of 30 nM (Online Supplementary Figure S1). We conducted further experiments with a TME5C mutant with a single amino acid substitution (Gâ&#x2020;&#x2019;A, Table 1). This mutant form of TME5C lost the ability to

Figure 4. TME5C increases the levels of p-ERK, p-AKT, p-p38, and Mcl-1 in endothelial cells. (A and C). HUVECs or HHSECs were exposed to control diluents (PBS as control) or TM mutants (500 nM). After 48 h, proteins were extracted and subjected to western blot analyses. The membrane was sequentially probed with the indicated antibodies. (B and D). Relative quantifications of p-ERK, p-AKT, p-p38, and Mcl-1. ImageJ software was used to measure the band intensities after western blotting. All experiments were performed three times. Results represent the mean Âą SD. *P<0.05. TM: thrombomodulin; HUVECs: human umbilical vein endothelial cells; HHSECs: human hepatic sinusoidal endothelial cells.

haematologica | 2018; 103(10)

1735

X. Wang et al.

stimulate the proliferation of HUVECs, suggesting that the pro-proliferative effect of TME5C was dependent on the specific amino acid sequence of this peptide (Figure 1C).

TME5C stimulates angiogenesis The potential role of TME5C to stimulate angiogenesis was examined in vitro and in vivo. TME5C (500 nM) but not TME5A or TME5B stimulated vascular tube formation of HUVECs and HHSECs by nearly 3-fold compared with control diluent-treated cells (Figure 2A-D). On the other hand, no remarkable proangiogenic effect was noted in the mutant form of TME5C (Figure 2E,F). Furthermore, in vivo angiogenesis assays with matrigel plugs revealed that TME5C (500 nM) stimulated angiogenesis in C57BL/6 mice (Figure 2G). In addition, the hemoglobin concentration was significantly increased in matrigel containing TME5C (Figure 2H) compared to matrigel containing control diluent. Consistent with the results of the proliferation assays, 30 nM TME5 produced an almost identical angiogenic effect with 500 nM TME5C (Figure 2A-D, G, H).

TME5C, but not TME5A or TME5B, blocks FK506-induced growth inhibition and apoptosis in endothelial cells We examined whether TME5C counteracted the growth inhibition of HUVECs and HHSECs induced by the calcineurin inhibitor FK506. FK506 inhibited the proliferation of HUVECs and HHSECs by 80-90% and 7080%, respectively. This growth inhibition was significantly attenuated by the presence of TME5C in a dosedependent manner up to 500 nM (Figure 3A). The highest dose of TME5C (1000 nM) was not as potent as 500 nM TME5C. Neither TME5A (500 nM) nor TME5B (500 nM) counteracted the effect of FK506 on proliferation of HUVECs and HHSECs (Figure 3A). We therefore chose 500 nM of TME5C for subsequent experiments. Importantly, the mutant form of TME5C lost the ability to block FK506-induced growth inhibition (Figure 3A). We next examined whether TME5C could block FK506induced apoptosis in HUVECs and HHSECs. FK506 induced more than 40% of HUVECs and HHSECs to be apoptotic. Interestingly, when these cells were cultured in the presence of both FK506 (10 mg/mL) and TME5C (500 nM), the population of apoptotic cells significantly

Figure 5. TME5C exerts cytoprotective function in murine ECs in a GPR15-dependent manner. (A). BrdU incorporation assays. WT or Gpr15 KO murine ECs were cultured with TME5C (500 nM) with or without FK506 (10 mg/ml) for 24 h. Proliferation was measured by BrdU incorporation assays. (B). Apoptosis assay. WT or Gpr15 KO murine ECs were exposed to FK506 (10 mg/ml) and/or TME5C (500 nM). After 36 h cells were harvested and stained with anti-annexin V and PI. FACS was used to analyze apoptotic cells. Annexin V+PI- and Annexin V+PI+ indicate early and late apoptosis, respectively. (C). Quantitative analysis of apoptotic cells in each group (n=3). (D, E). Vascular tube formation assays in vitro. WT or Gpr15 KO murine ECs were plated on growth factor-reduced matrigel-precoated 24-well plates (2.0x104 cells/well) and incubated with control diluent, TME5C (500 nM), or VEGF (0.5 nM, positive control). After 8 h, the endothelial cell-derived tube-like structure was photographed using an inverted microscope. The tube length in three randomly chosen fields from each well was measured using NIH ImageJ software. Results represent the mean Âą SD. *P<0.05. BrdU, bromodeoxyuridine; KO: knock out; GPR15: G-protein coupled receptor; WT: wild-type; FK506: tacrolimus; TM: thrombomodulin; N.S.: not significant; VEGF: vascular endothelial growth factor.

1736

haematologica | 2018; 103(10)

TME5C exerts cytoprotective and angiogenic functions

decreased (Figure 3B-E). By comparison, neither TME5A (500 nM) nor TME5B (500 nM) showed cytoprotective effects in HUVECs and HHSECs. In parallel with the induction of apoptosis, vascular permeability was profoundly induced in HUVECs after exposure to FK506 (10 mg/ml) for 12 h. Of note, FK506-induced vascular permeability was significantly attenuated in the presence of TME5C (500 nM), but not TME5A or TME5B (Figure 3F). Once more, a lower dose of TME5 (30 nM) produced a cytoprotective effect comparable to 500 nM TME5C (Figure 3).

TME5C upregulates p-ERK, p-AKT, p-p38, and Mcl-1 in endothelial cells We examined whether TME5C acts through the intracellular signal transduction pathways in HUVECs and HHSECs. Western blot analysis with different antibodies against intracellular signal transduction pathways found that exposure of either HUVECs or HHSECs to TME5C (500 nM) but not TME5A (500 nM) or TME5B (500 nM) for 48 h significantly increased the levels of phospho (p)-ERK, p-AKT, p-p38, and Mcl-1 in these cells (Figure 4A-D).

GPR15 is indispensable for the effects of TME5C Further experiments were carried out to test whether GPR15 also mediated the cytoprotective function of TME5C as it did for TME5. BrdU incorporation assays found that TME5C stimulated the proliferation of ECs isolated from WT C57BL/6 mice by nearly 1.5-fold compared with ECs treated with control diluent. In contrast, TME5C was not able to stimulate the proliferation of vascular ECs isolated from Gpr15 KO mice (Figure 5A). In

addition, TME5C significantly rescued the ECs isolated from WT C57BL/6 mice, but not Gpr15 KO mice, from FK506-induced growth inhibition and apoptosis (Figure 5A-C). Moreover, TME5C stimulated vascular tube formation in WT C57BL/6 murine ECs but not Gpr15 KO murine ECs (Figure 5D,E).

TME5C does not affect thrombin-mediated coagulation The fourth, fifth and sixth region of EGF-like domain of TM (TME456) binds thrombin and converts protein C to APC.32 The present study explored whether TME5C binds thrombin by measuring PT and APTT. rTM (500 nM) prolonged PT and APTT by approximately 200% and 165%, respectively. A much higher concentration of rTM (5000 nM) prolonged PT and APTT by more than 400% and 900%, respectively. Interestingly, TME5 (5000 nM) also prolonged PT and APTT by 137% and 225%, respectively. Of note, even the highest concentration of 5000 nM TME5C did not prolong either PT or APTT (Figure 6).

TME5C ameliorates SOS in a murine model To induce SOS, we used a murine BMT model preconditioned with BU and CY (Figure 7A). BMT recipients treated with PBS showed a decrease in food intake, curled hairs, and abdominal distention (data not shown). Dark brown colored livers indicating congestion and massive ascites were noted in BMT recipients treated with PBS on day 7 after BMT (Figure 7B,C). In addition, liver enzymes, including ALT and AST, were significantly elevated in BMT recipients treated with PBS at day 7 (Figure 7D). On the other hand, these indicators were less significant in

Figure 6. Effects of TME5C on PT and APTT. Plasma was obtained from a healthy volunteer and mixed with PT or APTT reagent with or without a series of TM fragments at various concentrations. (A, B). PT assays of TM fragments. (C, D). APTT assays of TM fragments. Experiments were performed three times. Results represent the mean Âą SD. *P<0.05 compared with control. PT: prothrombin time; APTT: activated partial thromboplastin time; TM: thrombomodulin; rTM: recombinant human soluble thrombomodulin.

haematologica | 2018; 103(10)

1737

X. Wang et al. A

G H

I J

BMT recipients treated with TME5C (Figure 7B-D). Strikingly, 9 out of 16 (56.3%) BMT recipient mice treated with PBS died by day 30 after BMT (Figure 7E), while none of the BMT recipient mice treated with TME5C died during the experimental period (Figure 7E). Pathological examination of livers removed from BMT mice treated with PBS on day 7 showed severe hemorrhagic necrosis with partial obstruction of liver sinusoids, infiltration of inflammatory cells, subendothelial hemorrhage, and loss of liver sinusoidal lining cells (Figure 7F). IHC staining of ECs with MECA-32 indicated a decrease in the number of sinusoid ECs, loss of integrity of sinusoid walls, and liver sinusoidal EC detachment (Figure 7G,H). These liver abnormalities were partly recovered by the 20th day after BMT in PBS-treated mice. In addi1738

Figure 7. TME5C ameliorates SOS in mice. C57BL/6 mice were randomly divided into three groups (n=16 in each group): untreated mice without BMT were defined as the control group, BMT recipient mice treated with vehicle PBS were defined as the BMT group, and BMT recipient mice treated with TME5C were defined as the BMT+TME5C group. (A). Schematic diagram of SOS model with the schedule of treatment. (B). Liver macroscopic picture of mice. On day 7 after BMT, some mice were sacrificed, and their livers were photographed. The images show representative livers of mice in each group. (C). Volume of ascites. On day 7 after BMT, ascites was harvested from mice (n=3 per group) and weighed. (D). Plasma levels of liver enzymes. On days 7, 14, and 20 after BMT, blood was collected from mice (n=3 per group), and plasma levels of AST and ALT were measured. (E). Survival of mice. Survival of mice was monitored every day. (F, G, I). H&E, IHC, and Masson staining (H&E and Masson x100, IHC x200). On days 7, 14, and 20 after BMT, some mice were sacrificed. Livers were removed and fixed with formaldehyde solution for H&E staining (blue arrow shows necrosis of hepatocytes and obstruction of liver sinusoid). Some liver slices were incubated with primary pan-endothelial cell monoclonal antibody (MECA-32) for IHC staining. Yellow arrows indicate the sinusoidal ECs. Masson staining was also carried out to evaluate the sinusoidal fibrosis of liver. Blue indicates collagen. (H). Quantification of MECA-32 positive-stained sinusoidal ECs in each group (n=3 in each group). (J). SOS score. Scoring of light microscopy histological slices stained with H&E or Masson stain were blindly evaluated according to the scoring system. (K). TUNEL stain (x400). Apoptosis of hepatocytes were assessed using an in situ cell death detection kit. Red arrows indicate apoptotic hepatocytes and sinusoidal ECs. (L). ELISA. Plasma was collected for ELISA to measure the concentrations of TM, FDP, and PAI-1. Results represent the mean Âą SD. *P<0.05. BU: busulfan; CY: cyclophosphamide; BMT: bone marrow transplantation; H&E, hematoxylin-eosin; IHC, immunohistochemistry; TM: thrombomodulin; FDP: fibrin degradation product; PAI1: plasminogen activator inhibitor-1; PBS: phosphate buffered saline.

tion, Masson staining demonstrated massive sinusoidal fibrosis with collagen deposition in livers removed from BMT recipients treated with PBS (Figure 7I). Importantly, all liver damage and associated findings were less severe in BMT recipients treated with TME5C. Careful pathological examination of livers with the SOS scoring system also found that the severity of SOS was significantly less in BMT recipients treated with TME5C than in those treated with PBS throughout the experimental period (Figure 7J). Moreover, TUNEL assays identified fewer apoptotic hepatocytes and sinusoid ECs in livers removed from TME5C-treated BMT mice compared with livers removed from BMT recipients treated with PBS (Figure 7K). We additionally measured plasma levels of TM, FDP, haematologica | 2018; 103(10)

TME5C exerts cytoprotective and angiogenic functions

and PAI-1, which are recognized markers of endothelial cell damage and coagulopathy in these mice. All of these markers steeply increased in PBS-treated BMT mice at the 7th day after BMT. Levels of these markers were significantly lower in BMT recipients treated with TME5C than in those treated with PBS (Figure 7L).

Discussion Our previous study found that the cytoprotective effects of TM were preserved in TME5.24,26 TME5 consists of three loops: A, B, and C (Figure 1A). The C-loop in the C-terminal subdomain is formed by a stretch of amino acids between the fifth and sixth cysteine residues, and its amino acid sequence is longer than those of the other two loops.27 The C loop contains a short tri-stranded bsheet structure,33 and is more similar to EGF when compared to the A or B loops.27 The study herein found that TME5C, but not A or B, exerted pro-angiogenetic and cytoprotective effects in a GPR15-dependent manner both in vitro and in vivo. Strikingly, TME5C ameliorated HSCT-associated SOS in a murine model. We previously showed that TME5 did not produce APC,24 but retained some binding capacity towards thrombin (Figure 6). The present study found that TME5C lost the ability to interact with thrombin, as TME5C did not affect PT or APTT (Figure 6). The ratio of concentration of TME5 that affect cytoprotection (30 nM) to coagulation (5000 nM) was approximately 1:166. The concentration of TME5C that produced cytoprotection was 500 nM. However, even the 166-fold higher concentration of TME5C (83 mM) did not prolong APTT (data not shown). Thus, the use of TME5C may be safe for BMT recipients as well as SOS patients who are at risk of bleeding due to low platelet counts and/or coagulopathy. We have recently identified GPR15 as a binding partner of TME5 by performing a pull-down assay with membrane protein isolated from HUVECs followed by matrix assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) analysis.25 We found that the cytoprotective and pro-angiogenic effects of TME5 were mediated by GPR15, as neither cytoprotection nor angiogenesis is noted in vascular endothelial cells isolated from Gpr15 KO mice after exposure to TME5.25 The study herein found that TME5C exerts cytoprotective and proangiogenic effects in vascular ECs isolated from wild-type C57BL/6 mice, but not in ECs isolated from Gpr15 KO mice, suggesting that TME5C also produces favorable effects in ECs via GPR15. GPR15 is also expressed on T lymphocytes and is required for infection of HIV as a co-receptor.34 Accumulating evidence suggests the involvement of GPR15 in the regulation of inflammation; GPR15 expressed on murine TH1 and TH17 effector cells is

References 1. Richardson PG, Ho VT, Cutler C, et al. Hepatic veno-occlusive disease after hematopoietic stem cell transplantation: novel insights to pathogenesis, current status of treatment, and future directions. Biol

haematologica | 2018; 103(10)

implicated in the development of colitis.35 However, the expression of GPR15 on regulatory T cells is associated with the accumulation of these cells in the intestines and the alleviation of colitis.36 GPR15 is also required for the homing of dendritic epidermal T cells into epidermal tissues.34 Intriguingly, TME5 inhibits mixed lymphocyte reactions in vitro in association with a decrease in the production of inflammatory cytokines such as interleukin-6 and tumor necrosis factor Îą.37 The anti-inflammatory effect of TME5 is also mediated by GPR15, as TME5 was not able to inhibit the mixed lymphocyte reaction when we used lymphocytes isolated from Gpr15 KO mice.37 Of note, the use of TME5 significantly alleviated graft-versushost disease (GvHD) in a murine model.37 TME5 also rescues mice from LPS-induced sepsis,26 further confirming the anti-inflammatory function of TME5C. Liver sinusoidal ECs are not only the supporting cells of quiescent hepatocytes, but also the source of vascularization during liver regeneration.38 Hepatocytes undergo apoptosis in parallel with a decrease in vascularization when sinusoidal ECs are damaged.31 EC insults cause a hypercoagulable state through the upregulation of tissue factor and downregulation of TM on ECs, leading to the formation of fibrin clots in the microvasculature.31 As a consequence, the fibrinolytic system is activated to form various FDP fragments. Of note, PAI-1, an important antifibrinolytic regulator mainly produced by ECs, impedes fibrinolysis by inhibiting tissue plasminogen activator and augments thrombus formation.39 Interestingly, the levels of PAI-1 are elevated in SOS patients but not in patients with GvHD and other liver injuries, suggesting that PAI-1 might be a marker capable of discriminating SOS from other BMT-related complications.40 Increases in plasma levels of TM, FDP, and PAI-1 were noted in the murine SOS model, and they were significantly counteracted by the use of TME5C (Figure 7L), indicating cytoprotective roles of TME5C in vivo. The optimum effective concentration of TME5C in our experiment (500 nM) was higher than that of TME5 (30 nM). TME5 was produced by the yeast Pichia, while TME5C is a chemically synthesized peptide, therefore we cannot precisely compare the potency of these two compounds. However, it is possible that the C-loop alone is not enough to fully produce the pro-proliferative and proangiogenic effect of TM. Taken together, TME5C, which does not affect the coagulation system, exerts pro-angiogenetic and cytoprotective activities both in vitro and in vivo. Hence, the use of TME5C may be a promising strategy to prevent and/or treat HSCT-related complications, including SOS. Funding This study was supported by Asahi Kasei Pharma (Tokyo, Japan), Uehara Memorial Foundation, and KAKENHI (26461406).

Blood Marrow Transplant. 2013;19(1 Suppl):S88-90. 2. Mohty M, Malard F, Abecassis M, et al. Sinusoidal obstruction syndrome/venoocclusive disease: current situation and perspectives-a position statement from the European Society for Blood and Marrow

Transplantation (EBMT). Bone Marrow Transplant. 2015;50(6):781-789. 3. Senzolo M, Germani G, Cholongitas E, et al. Veno occlusive disease: update on clinical management. World J Gastroenterol. 2007;13(29):3918-3924. 4. Palomo M, Diaz-Ricart M, Carbo C, et al.

1739

X. Wang et al.

10.

11.

12.

13.

14. 15.

16.

17.

1740

Endothelial dysfunction after hematopoietic stem cell transplantation: role of the conditioning regimen and the type of transplantation. Biol Blood Marrow Transplant. 2010;16(7):985-993. Vion AC, Rautou PE, Durand F, et al. Interplay of inflammation and endothelial dysfunction in bone marrow transplantation: focus on hepatic veno-occlusive disease. Semin Thromb Hemost. 2015; 41(6):629-643. Carreras E, Diaz-Ricart M. The role of the endothelium in the short-term complications of hematopoietic SCT. Bone Marrow Transplant. 2011;46(12):1495-1502. Tuncer HH, Rana N, Milani C, et al. Gastrointestinal and hepatic complications of hematopoietic stem cell transplantation. World J Gastroenterol. 2012;18(16):18511860. Bearman SI, Lee JL, Baron AE, et al. Treatment of hepatic venocclusive disease with recombinant human tissue plasminogen activator and heparin in 42 marrow transplant patients. Blood. 1997;89(5):15011506. Haussmann U, Fischer J, Eber S, et al. Hepatic veno-occlusive disease in pediatric stem cell transplantation: impact of preemptive antithrombin III replacement and combined antithrombin III/defibrotide therapy. Haematologica. 2006;91(6):795-800. Bearman SI, Shen DD, Hinds MS, et al. A phase I/II study of prostaglandin E1 for the prevention of hepatic venocclusive disease after bone marrow transplantation. Br J Haematol. 1993;84(4):724-730. Esmon CT, Esmon NL, Harris KW. Complex formation between thrombin and thrombomodulin inhibits both thrombin-catalyzed fibrin formation and factor V activation. J Biol Chem. 1982;257(14):7944-7947. O'Brien LM, Mastri M, Fay PJ. Regulation of factor VIIIa by human activated protein C and protein S: inactivation of cofactor in the intrinsic factor Xase. Blood. 2000; 95(5):1714-1720. Kalafatis M, Rand MD, Mann KG. The mechanism of inactivation of human factor V and human factor Va by activated protein C. J Biol Chem. 1994;269(50):31869-31880. Dahlback B, Villoutreix BO. The anticoagulant protein C pathway. FEBS Lett. 2005; 579(15):3310-3316. Schuepbach RA, Madon J, Ender M, et al. Protease-activated receptor-1 cleaved at R46 mediates cytoprotective effects. J Thromb Haemost. 2012;10(8):1675-1684. Ikezoe T, Takeuchi A, Taniguchi A, et al. Recombinant human soluble thrombomodulin counteracts capillary leakage associated with engraftment syndrome. Bone Marrow Transplant. 2011;46(4):616-618. Ikezoe T, Togitani K, Komatsu N, et al.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

Successful treatment of sinusoidal obstructive syndrome after hematopoietic stem cell transplantation with recombinant human soluble thrombomodulin. Bone Marrow Transplant. 2010;45(4):783-785. Nakamura D, Yoshimitsu M, Kawada H, et al. Recombinant human soluble thrombomodulin for the treatment of hepatic sinusoidal obstructive syndrome post allogeneic hematopoietic SCT. Bone Marrow Transplant. 2012;47(3):463-464. Ohwada C, Takeuchi M, Kawaguchi T, et al. Successful treatment with recombinant soluble thrombomodulin of two cases of sinusoidal obstructive syndrome/hepatic venoocclusive disease after bone marrow transplantation. Am J Hematol. 2011; 86(10):886888. Yamamoto S, Yagawa A, Toyama D, et al. Successful treatment of hepatic sinusoidal obstructive syndrome after hematopoietic stem cell transplantation in a child using recombinant thrombomodulin. Acta Haematol. 2013;129(1):62-64. Fujiwara H, Maeda Y, Sando Y, et al. Treatment of thrombotic microangiopathy after hematopoietic stem cell transplantation with recombinant human soluble thrombomodulin. Transfusion. 2016;56(4): 886-892. Ikezoe T, Yang J, Nishioka C, et al. Thrombomodulin protects endothelial cells from a calcineurin inhibitor-induced cytotoxicity by upregulation of extracellular signal-regulated kinase/myeloid leukemia cell1 signaling. Arterioscler Thromb Vasc Biol. 2012;32(9):2259-2270. Ikezoe T, Yang J, Nishioka C, et al. Thrombomodulin blocks calcineurin inhibitor-induced vascular permeability via inhibition of Src/VE-cadherin axis. Bone Marrow Transplant. 2017;52(2):245-251. Ikezoe T, Yang J, Nishioka C, et al. The fifth epidermal growth factor-like region of thrombomodulin exerts cytoprotective function and prevents SOS in a murine model. Bone Marrow Transplant. 2017;52(1):73-79. Pan B, Wang X, Nishioka C, et al. G-protein coupled receptor 15 mediates angiogenesis and cytoprotective function of thrombomodulin. Sci Rep. 2017;7(1):692. Pan B, Wang X, Kojima S, et al. The fifth epidermal growth factor like region of thrombomodulin alleviates LPS-induced sepsis through interacting with GPR15. Thromb Haemost. 2017;117(3):570-579. Sampoli Benitez BA, Hunter MJ, Meininger DP, et al. Structure of the fifth EGF-like domain of thrombomodulin: An EGF-like domain with a novel disulfide-bonding pattern. J Mol Biol. 1997;273(4):913-926. Kobayashi M, Inoue K, Warabi E, et al. A simple method of isolating mouse aortic

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

endothelial cells. J Atheroscler Thromb. 2005;12(3):138-142. Qiao J, Fu J, Fang T, et al. Evaluation of the effects of preconditioning regimens on hepatic veno-occlusive disease in mice after hematopoietic stem cell transplantation. Exp Mol Pathol. 2015;98(1):73-78. DeLeve LD, McCuskey RS, Wang X, et al. Characterization of a reproducible rat model of hepatic veno-occlusive disease. Hepatology. 1999;29(6):1779-1791. Zeng L, An L, Fang T, et al. A murine model of hepatic veno-occlusive disease induced by allogeneic hematopoietic stem cell transplantation. Cell Biochem Biophys. 2013;67(3):939-948. Stearns DJ, Kurosawa S, Esmon CT. Microthrombomodulin. Residues 310-486 from the epidermal growth factor precursor homology domain of thrombomodulin will accelerate protein C activation. J Biol Chem. 1989;264(6):3352-3356. Tejero R, Bassolino-Klimas D, Bruccoleri RE, et al. Simulated annealing with restrained molecular dynamics using CONGEN: energy refinement of the NMR solution structures of epidermal and type-alpha transforming growth factors. Protein Sci. 1996;5(4):578-592. Lahl K, Sweere J, Pan J, et al. Orphan chemoattractant receptor GPR15 mediates dendritic epidermal T-cell recruitment to the skin. Eur J Immunol. 2014;44(9):25772581. Nguyen LP, Pan J, Dinh TT, et al. Role and species-specific expression of colon T cell homing receptor GPR15 in colitis. Nat Immunol. 2015;16(2):207-213. Kim SV, Xiang WV, Kwak C, et al. GPR15mediated homing controls immune homeostasis in the large intestine mucosa. Science. 2013;340(6139):1456-1459. Pan B, Wang X, Kojima S, et al. The fifth epidermal growth factor-like region of thrombomodulin alleviates murine graftversus-host disease in a G-protein coupled receptor 15 dependent manner. Biol Blood Marrow Transplant. 2017; 23(5):746-756. el Mouelhi M, Kauffman FC. Sublobular distribution of transferases and hydrolases associated with glucuronide, sulfate and glutathione conjugation in human liver. Hepatology. 1986;6(3):450-456. Park YD, Yasui M, Yoshimoto T, et al. Changes in hemostatic parameters in hepatic veno-occlusive disease following bone marrow transplantation. Bone Marrow Transplant. 1997;19(9):915-920. Salat C, Holler E, Kolb HJ, et al. Plasminogen activator inhibitor-1 confirms the diagnosis of hepatic veno-occlusive disease in patients with hyperbilirubinemia after bone marrow transplantation. Blood. 1997;89(6):2184-2188.

haematologica | 2018; 103(10)

ARTICLE

Blood transfusion

A subset of anti-HLA antibodies induces FcγRIIa-dependent platelet activation

Ferrata Storti Foundation

Maaike Rijkers,1 Anno Saris,2 Sebastiaan Heidt,3 Arend Mulder,3 Leendert Porcelijn,4 Frans H.J. Claas,3 Ruben Bierings,1 Frank W.G. Leebeek,5 A.J. Gerard Jansen,1,5 Gestur Vidarsson,6 Jan Voorberg1,7 and Masja de Haas3,4,8

Department of Plasma Proteins, Sanquin-AMC Landsteiner Laboratory, Amsterdam; Department of Immunopathology, Sanquin-AMC Landsteiner Laboratory, Amsterdam; 3 Department of Immunohaematology and Blood Transfusion, Leiden University Medical Center; 4Department of Immunohaematology Diagnostics, Sanquin Diagnostic Services, Amsterdam; 5Department of Hematology, Erasmus University Medical Center, Rotterdam; 6 Department of Experimental Immunohematology, Sanquin-AMC Landsteiner Laboratory, Amsterdam; 7Department of Vascular Medicine, Amsterdam UMC, University of Amsterdam and 8Center for Clinical Transfusion Research, Sanquin, Leiden, the Netherlands 1 2

Haematologica 2018 Volume 103(10):1741-1752

ABSTRACT

LA antibodies are associated with refractoriness to platelet transfusion, leading to rapid platelet clearance, sometimes coinciding with clinical side effects such as fever and chills. The presence of HLA antibodies is not always manifested by clinical symptoms. It is currently unclear why refractoriness to platelet transfusion is only observed in a subset of patients. Here, we utilized the availability of a unique panel of human monoclonal antibodies to study whether these were capable of activating platelets. Three out of eight human HLA-specific monoclonal antibodies induced activation of HLA-matched platelets from healthy donors as evidenced by enhanced α-granule release, aggregation, and αIIbb3 activation. The propensity of HLA monoclonal antibodies to activate platelets was independent of the HLA subtype to which they were directed, but was dependent on the recognized epitope. Activation was fully inhibited either by blocking FcγRIIa, or by blocking FcγRIIa-dependent signaling with Syk inhibitor IV. Furthermore, activation required the presence of the IgG-Fc part, as F(ab’)2 fragments of HLA monoclonal antibodies were unable to induce platelet activation. Mixing experiments revealed that activation of platelets occurred in an intra-platelet dependent manner. Accordingly, a proportion of sera from refractory patients with HLA antibodies induced FcγRIIa-dependent platelet activation. Our data show that a subset of HLA antibodies is capable of crosslinking HLA and FcγRIIa thereby promoting platelet activation and enhancing these cells’ phagocytosis by macrophages. Based on these findings we suggest that FcγRIIa-dependent platelet activation may contribute to the decreased platelet survival in platelet-transfusion-dependent patients with HLA antibodies.

Correspondence: m.rijkers@sanquin.nl

Received: February 9, 2018. Accepted: May 30, 2018. Pre-published: June 1, 2018.

doi:10.3324/haematol.2018.189365 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/10/1741 ©2018 Ferrata Storti Foundation

Introduction Antibodies against human leukocyte antigen (HLA) can be induced by pregnancy, blood transfusion or transplantation.1–3 During or after a pregnancy, 15-50% (depending on the number of pregnancies) of women develop HLA antibodies.4–6 Platelet reactive alloantibodies commonly directed toward HLA of the donor platelets develop in 20-30% of chronic platelet transfusion recipients.2,3,7 Although platelet refractoriness is more commonly caused by non-immune factors2,7,8, 30-50% of platelet-transfusion-dependent recipients with HLA antibodies become refractory to platelet transfusions due to alloimmunization.2,7 HLA antibodies in this setting are primarily composed of immunoglobulin G (IgG) and are directed toward HLAhaematologica | 2018; 103(10)

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.

1741

M. Rijkers et al.

A and B.3 Binding of antibodies to HLA class I on donor platelets results in the formation of IgG-opsonized platelets which are rapidly cleared from the circulation. Several parameters may contribute to the efficacy of HLA antibody-induced platelet clearance. Firstly, HLA density on platelets may differ between individuals. A recent study showed that platelets from donors with consistently low HLA-B8, B12 or B35 displayed a strongly reduced antibody-mediated internalization by macrophages.9 Furthermore, low levels of HLA antibodies were not associated with platelet refractoriness in the TRAP (Trial to Reduce Alloimmunization to Platelets) study.10 Results from the same study revealed that high levels of anti-HLA antibodies were clearly related to refractoriness to platelet transfusion.10 Transfusions with HLA-compatible platelets have been shown to be effective in patients with preexisting HLA antibodies.11 In an early clinical trial no beneficial effect of treatment with HLA-matched platelet concentrates was observed.12 These findings indicate that platelet refractoriness is of non-immune origin in a significant number of patients and, collectively, suggest that HLA antibodies, dependent on their titer and the HLA density on donor platelets, can induce platelet refractoriness. Whether additional mechanisms contribute to the observed clinical effects of anti-HLA antibodies has not yet been clearly delineated. Apart from the transfusion setting, a pathogenic role for platelet-specific antibodies has been described in several diseases. Patients with immune thrombocytopenia often have autoantibodies against glycoprotein (GP)Ib/IX or GPIIbIIIa, frequently coinciding with refractoriness.13–15 Anti-GPIbα has been associated with Fcγ receptor IIa (FcγRIIa)-independent platelet activation, through loss of sialic acid and subsequent clearance via the Ashwell Morell receptor localized on hepatocytes.16 Alternatively, in heparin-induced thrombocytopenia, antibodies directed to platelet factor 4/heparin complex induce platelet clearance and FcγRIIa-dependent platelet activation.17,18 Previous studies have described on FcγRIIa-dependent activation of platelets by (non-physiological) crosslinking of the murine pan-HLA class I antibody W6/32.19 Complement-dependent platelet aggregation induced by HLA antibodies has also been reported.20 Based on these findings we hypothesized that a subset of human HLA antibodies may be able to activate platelets. To address this issue we tested a panel of well-characterized human monoclonal HLA antibodies and HLA antibody-containing sera from platelet-transfusion refractory patients for their ability to activate platelets.

Methods HLA monoclonal antibodies and patients’ sera Human HLA-specific monoclonal antibodies, all of IgG1 isotype, were produced by hybridoma technology as described previously.21,22 Blood samples of patients refractory to platelet transfusion were sent to the Department of Immunohematology Diagnostic Services, Sanquin, Amsterdam, the Netherlands. Leftover material was used according to the Dutch established codes of conduct for responsible use of patients’ material and as approved by our institute.23 HLA antibody specificities in patients’ sera were determined by a single antigen bead assay (Luminex). Thirteen sera positive for HLA antibodies and negative for other platelet-specific antibodies were used. 1742

Human platelets Citrated whole blood was obtained from healthy human volunteers with known HLA type (second field) in accordance with Dutch regulations and after approval from the Sanquin Ethical Advisory Board in accordance with the Declaration of Helsinki. Written informed consent was given by all participants. Platelets were isolated and washed as described elsewhere,24 and resuspended in platelet assay buffer (10 mM HEPES, 140 mM NaCl, 3 mM KCl, 0.5 mM MgCl2, 10 mM glucose and 0.5 mM NaHCO3, pH 7.4).

Platelet activation Washed platelets (2.5x108 platelets/mL) were incubated with HLA monoclonal antibodies or patients’ sera (1:50) containing HLA antibodies for 1 h at room temperature. Where appropriate, platelets were pre-incubated with FcγRIIa blocking antibody IV.3, Syk inhibitor IV or intravenous immunoglobulin.

Flow cytometry For flow cytometry measurements, platelets were fixed in 1% PFA and diluted in platelet assay buffer. Anti-CD62P, anti-PAC-1 and anti-IgG antibodies were used to stain the platelets. Analysis was performed using a FACSCanto II (Becton Dickinson) flow cytometer.

Internalization of opsonized platelets by macrophages Internalization of platelets by macrophages was determined as previously described.9 In short, PKH26-labeled platelets were opsonized with HLA monoclonal antibodies in the presence or absence of Syk inhibitor IV and incubated with monocyte-derived macrophages. Platelet internalization was quantified by imaging flow cytometry (ImageStream®X Mark II Imaging Flow Cytometer, Merck Millipore, Amsterdam, the Netherlands).

Data and statistical analysis Flow cytometry data were analyzed using FlowJo version 10 (Ashland, OR, USA). Data are represented as either mean ± standard deviation (SD) or all data points are shown. Statistical analyses were performed using GraphPad Prism 7 version 7.02 (La Jolla, CA, USA), with the analyses used specified in the respective figure legends. Differences were considered statistically significant when P values were <0.05. Further details on materials and methods can be found in the Online Supplementary Data.

Results HLA monoclonal antibodies induced platelet α-granule release Eight human HLA-specific monoclonal antibodies were used to study the effect of HLA antibodies on platelets from healthy donors (Table 1). These antibodies all recognize different HLA epitopes, of which some are specific for a particular HLA antigen (e.g. GV2D5 binds only HLAA1) and others are broadly reactive (e.g. WIM8E5, binding to HLA-A1/A10(A25/A26/A34/A43/A66)/A11/A9(A23/ A24)/A29/A30/A31/A33/A28(A68/A69)).21,25 Donors were selected in such a way that platelets expressed an HLA type matching the specificity of antibodies used in each experiment. A similar level of binding of HLA monoclonal antibodies to the matched platelets was obtained in all experiments, as verified by flow cytometry (Figure 1A). The ability of HLA monoclonal antibodies to induce αgranule release was assessed by measuring CD62P expohaematologica | 2018; 103(10)

Platelet activation by HLA antibodies

sure on the platelet surface. At low concentrations (2.5 mg/mL), a subset of HLA monoclonal antibodies induced α-granule release, as shown by significantly increased CD62P membrane exposure (Figure 1B). Statistically significant enhanced CD62P exposure was observed only for the broadly reactive WIM8E5 and HLA-A2/A28-specific SN607D8, whereas GV5D1 (anti-HLA A1/A23/A24) showed a trend of enhanced CD62P exposure, but this was not statistically significant. SN230G6, with a specificity similar to that of SN607D8 and GV2D5, binding to HLA-A1 like GV5D1, did not induce platelet activation. VTM1F11, HDG8D9 and BRO11F9 also did not induce enhanced CD62P membrane exposure on platelets. Increased CD62P exposure was observed upon incubation with higher concentrations of WIM8E5, SN607D8 and GV5D1 (Figure 1C). Addition of 10 mg/mL of the other HLA monoclonal antibodies included in this study did not induce significantly increased CD62P membrane exposure (Figure 1C). Additionally, incubation of platelets with two human monoclonal anti HPA-1a antibodies did not induce CD62P exposure despite their efficient binding to platelets (Online Supplementary Figure S1A).To further confirm the release of α-granules from platelets, the release of von Willebrand factor (VWF) and SPARC (secreted protein acidic and rich in cysteine), both proteins residing in αgranules,26 was measured in the supernatant of platelets incubated with HLA monoclonal antibodies. VWF (Figure 1D) and SPARC (Figure 1E, Online Supplementary Figure S2A) were released by platelets upon incubation with either WIM8E5 or SN607D8. Incubation with SN230G6 resulted in significant, but low levels of released VWF and SPARC. Together these results indicate that a subset of HLA monoclonal antibodies can induce α-granule release in platelets.

HLA monoclonal antibodies induce integrin αIIbb3 activation and platelet agglutination

We subsequently studied the activation of GPIIb/IIIa (integrin αIIbb3) using the PAC-1 antibody, which recognizes the active configuration of this integrin.27 PAC-1 binding was increased significantly upon incubation with WIM8E5 and SN607D8, while SN230G6, an antibody not potent in inducing α-granule release, did not lead to activation of integrin αIIbb3 (Figure 2A). We then performed light aggregometry to study whether platelets aggregate upon incubation with HLA monoclonal antibodies. In agreement with the results for α-granule release and integrin αIIbb3 activation, WIM8E5 and SN607D8 induced

Table 1. Human HLA-specific monoclonal antibodies used in this study.

Antibody name

HLA specificity

WIM8E5

A1/A10(A25/A26/A34/A43/A66)/A11/A9(A23/A24)/A29/ A30/A31/A33/A28(A68/A69) A2/A28(A68/A69) A2/B57/B58 A1/A23/A24 {not A*2403; A80 weak} A1 B27/B7/B60 B51/B35 A3/A11/A24

SN607D8* SN230G6* GV5D1$ GV2D5$ VTM1F11 HDG8D9 BRO11F6

* SN607D8 and SN230G6 originate from the same patient. $GV5D1 and GV2D5 originate from the same patient.

haematologica | 2018; 103(10)

dose-dependent platelet aggregation. No platelet aggregation was observed upon incubation with SN230G6 (Figure 2B). A combination of suboptimal concentrations of PAR1 activating peptide and HLA monoclonal antibodies did not induce enhanced CD62P exposure. In combination with low suboptimal concentration of PAR1 activating peptide or collagen, both activating (WIM8E5 and SN607D8) and non-activating HLA monoclonal antibodies (SN230G6) significantly enhanced platelet aggregation, which was most pronounced for WIM8E5 (Online Supplementary Figure S1B,C). WIM8E5 also induced exposure of phosphatidylserine, as measured by annexin V binding. No significantly increased levels of annexin V binding were observed for SN607D8 and SN230G6, suggesting that only strongly activating HLA antibodies can induce phosphatidylserine exposure on platelets (Online Supplementary Figure S2B). These results indicate that the subset of HLA antibodies which induces α-granule release also stimulates integrin αIIbb3 activation, platelet aggregation and phospatidylserine exposure.

Activation of platelets by HLA monoclonal antibodies is FcγRIIa dependent Next, we tested whether the platelet-Fc receptor, FcγRIIa,28,29 is involved in anti-HLA mediated platelet activation. The pathway of FcγRIIa-dependent platelet activation by IgG is well described in, for instance, heparininduced thrombocytopenia.17 FcγRIIa-mediated platelet activation via its ITAM motif has been shown to be dependent on the tyrosine kinase Syk.30 We tested whether HLA antibody-mediated activation of platelets required Syk,31 employing the extensively characterized Syk inhibitor IV.30 This inhibitor has no effect on activation via PAR1 receptor (Online Supplementary Figure S2C), confirming its specificity. CD62P exposure induced by WIM8E5 or SN607D8 was inhibited in a dose-dependent manner by Syk inhibitor IV (Figure 3A). Involvement of Syk was further supported by blocking experiments employing Syk inhibitor I and II (Online Supplementary Figure S2D). Similarly, VWF release (Figure 3B) and SPARC release (Figure 3C) were blocked by Syk inhibitor IV. Activation of integrin αIIbb3 (Figure 3D) and platelet aggregation (Figure 3E) were completely abrogated following the addition of Syk inhibitor IV. These results indicate that HLA monoclonal antibody-induced platelet activation is dependent on Syk, which acts downstream of FcγRIIa. To further substantiate the involvement of FcγRIIa, blocking monoclonal anti-FcγRIIa-antibody IV.3 was used to prevent binding of the Fc tail of the HLA monoclonal antibodies to FcγRIIa.29 Analysis of CD62P surface exposure and of VWF and SPARC secretion revealed that α-granule release was completely blocked upon pre-incubation with IV.3 (Figure 3F-H). In addition, IV.3 completely blocked WIM8E5- and SN607D8-induced platelet aggregation (Figure 3I), integrin αIIbb3 activation (Figure 3J) and annexin V binding (Online Supplementary Figure S2B). The R131H polymorphism in FcγRIIa did not affect platelet activation by the HLA monoclonal antibodies used in this study (Online Supplementary Figure S3). Together these results indicate that platelet α-granule release, activation and aggregation induced by HLA monoclonal antibodies are FcγRIIa dependent. To confirm the involvement of the Fc-tail of the HLA monoclonal antibodies in platelet activation, F(ab’)2 frag1743

M. Rijkers et al.

ments of WIM8E5, SN607D8 and SN230G6 were generated (Online Supplementary Figure S4A). The lack of an Fc-tail (Online Supplementary Figure S4B) and binding (Online Supplementary Figure S4C) of the F(ab’)2 fragments to HLA molecules on platelets was confirmed. Staining with antiIgG directed to the Fc-tail of human IgG was negative for F(ab’)2 fragments and positive for IgG. For WIM8E5, F(ab’)2

binding was lower than that of the corresponding IgG, but significant binding was still observed (Online Supplementary Figure S4C). None of these F(ab’)2 fragments induced CD62P membrane exposure (Online Supplementary Figure S4D), indicating that crosslinking of an HLA molecule and FcγRIIa by an intact anti-HLA IgG is crucial to induce platelet activation.

Figure 1. HLA monoclonal antibodies induce platelet α-granule release. (A) Platelets were matched for HLA type with the specificity of eight HLA monoclonal antibodies (mAbs) directed at different epitopes. Mean fluorescent intensity (MFI) upon staining with anti-human IgG was measured with flow cytometry for the control (buffer only, no HLA antibodies) and 2.5 mg/mL of the HLA mAbs. Right panel: representative flow cytometry plot of 10 mg/mL SN607D8 with a not matching donor and a matching donor. (B,C) CD62P surface expression of platelets incubated with 2.5 mg/mL (B) or 10 mg/mL (C) HLA mAbs compared to control (buffer only). Representative flow cytometry plots of WIM8E5, SN607D8 and SN230G6. (D) VWF release in platelet supernatant upon incubation with HLA mAbs WIM8E5, SN607D8 and SN230G6 measured by enzyme-linked immunosorbent assay. (E) Representative western blot of SPARC release in platelet supernatant upon incubation with HLA mAbs WIM8E5, SN607D8 and SN230G6. Paired t-tests (A, B and C) or paired ANOVA with the Tukey multiple comparison test (D). Each line represents a separate experiment with a separate donor (A, B and C). Mean ± SD (D). *P<0.05, **P<0.01, ***P<0.005, ****P<0.001.

1744

haematologica | 2018; 103(10)

Platelet activation by HLA antibodies

FcγRIIa-dependent platelet activation has been firmly implicated in the pathogenesis of heparin-induced thrombocytopenia.17 In vitro experiments have demonstrated that platelet activation by heparin-platelet factor 4 can be inhibited by intravenous immunoglobulin.18 To study whether the mechanism of platelet activation by monoclonal antibodies is similar to that described for heparininduced thrombocytopenia, platelets were pre-incubated with intravenous immunoglobulin and subsequently incubated with the HLA monoclonal antibodies WIM8E5 or SN607D8. Intravenous immunoglobulin diminished CD62P surface exposure in a dose-dependent manner and completely blocked α-granule release at a concentration of 2 mg/mL (Online Supplementary Figure S5A). These results suggest that high levels of IgG can compete with HLA monoclonal antibodies for binding to FcγRIIa.

Platelet activation by HLA monoclonal antibodies occurs through intra-platelet binding to FcγRIIa FcγRIIa-dependent platelet activation can theoretically occur in either an inter-platelet-dependent manner (the HLA molecule of one platelet is crosslinked with the FcγRIIa of another platelet) or intra-platelet-dependent manner (an HLA molecule and FcγRIIa on a single platelet

are crosslinked by the antibody). Rubinstein and co-workers showed that antibodies directed to beta-2-microglobulin can bind to FcγRIIa on other platelets (inter-platelet binding) resulting in their activation.32 Activation of platelets in patients with heparin-induced thrombocytopenia is considered to occur both in an inter- and intraplatelet manner.18 To elucidate whether platelet activation by HLA monoclonal antibodies occurs in an inter-and/or intra-platelet manner, we studied whether platelets missing the binding epitope of the activating HLA monoclonal antibody WIM8E5 (“nonmatching”) could be activated in the presence of platelets which were able to bind WIM8E5 (“matching”). In the case of inter-platelet activation, the HLA of the “matching” platelets can theoretically be crosslinked with the FcγRIIa on the “nonmatching” platelets (Figure 4A). Platelets from a donor with an HLA type not matching with WIM8E5 did not show increased levels of CD62P upon incubation with WIM8E5. When these platelets were mixed with platelets from a donor with an HLA type matched for WIM8E5, CD62P surface expression remained unaltered (Figure 4B). Platelets from a WIM8E5 “matching” donor did show enhanced CD62P surface exposure, and levels did not change when platelets were mixed with platelets from a “nonmatching” donor

Figure 2. Integrin αIIbb3 activation and platelet agglutination are induced by HLA monoclonal antibodies. A) Integrin αIIbb3 activation, derived from PAC-1 binding, upon incubation with 10 mg/mL WIM8E5, SN607D8 or SN230G6 compared to control (buffer only, no HLA antibodies). Flow cytometry plots are representative of more than eight independent experiments with different donors. (B) Platelet agglutination upon addition of HLA monoclonal antibodies (mAbs), measured by light transmission aggregometry. Mean ± SD of percentage maximum aggregation. Paired ttests (A) or paired ANOVA with the Tukey multiple comparison test (B). *P<0.05, **P<0.01, ***P<0.005, ****P<0.001.

haematologica | 2018; 103(10)

1745

M. Rijkers et al.

Figure 3. HLA monoclonal antibodies induce FcγRIIa-dependent platelet activation. (A) CD62P exposure upon incubation with WIM8E5 and SN607D8, inhibited by pre-incubation with Syk inhibitor IV compared to control (buffer only, no HLA antibodies). (B) VWF release, measured by enzyme-linked immunosorbent assay, induced by WIM8E5 and SN607D8, inhibited by Syk inhibitor IV. (C) SPARC release in platelet supernatant induced by WIM8E5 and SN607D8, inhibited by Syk inhibitor IV. (D) Integrin αIIbb3 activation (as measured by PAC-1 binding) induced by WIM8E5 and SN607D8, inhibited by Syk inhibitor IV. (E) Agglutination induced by WIM8E5, inhibited by Syk inhibitor IV. (F) CD62P exposure induced by WIM8E5 and SN607D8, blocked by the FcγRIIa blocking antibody IV.3. (G) VWF release induced by WIM8E5 and SN607D8, inhibited by IV.3 (H) Release of SPARC in platelet supernatant inhibited by IV.3. (I) Agglutination induced by WIM8E5 and SN607D8 inhibited by pre-incubation with IV.3. (J) PAC-1 binding induced upon incubation with WIM8E5 and SN607D8 inhibited by IV.3. Data are given as mean ± SD. Paired ANOVA with the Tukey multiple comparison test. *P<0.05, **P<0.01, ***P<0.005, ****P<0.001.

1746

haematologica | 2018; 103(10)

Platelet activation by HLA antibodies

Figure 4. HLA monoclonal antibodies activate platelets in an intra-platelet-dependent manner. (A) Schematic representation of theoretically possible inter-platelet activation and intra-platelet activation induced by the HLA monoclonal antibody WIM8E5. (B) Platelet donors were selected as either “WIM8E5 matching” or “WIM8E5 nonmatching” and their platelets were stained with calcein-green or calcein-violet, respectively. Platelets from two donors were mixed in a 1:1 ratio. Platelets from a single donor or the mixed platelets were incubated with control (buffer only, no HLA antibodies) or WIM8E5. By gating for calcein-green or calceinviolet, CD62P exposure was determined for “WIM8E5 matching” and “WIM8E5 nonmatching” platelets. (C) Platelets from a “WIM8E5 matching” donor (stained with calcein green) were mixed 1:1 with platelets from a donor with either nonmatching WIM8E5 or matching WIM8E5 (stained with calcein violet). Samples were incubated with control (buffer only, no HLA antibodies), WIM8E5 or PAR1 activating peptide (PAR1 AP). Percentage double-positive events of calcein-green/calcein-violet-stained platelets are given, representing the ability of HLA antibodies to crosslink the HLA molecule with the FcγRIIa in an inter- or intra-platelet manner. Representative flow cytometry plots are shown for a mix of WIM8E5 matching + WIM8E5 nonmatching platelets and platelets from two different donors with both matching HLA typing. Paired ANOVA with the Tukey multiple comparison test. *P<0.05, **P<0.01, ***P<0.005, ****P<0.001.

haematologica | 2018; 103(10)

1747

M. Rijkers et al.

(Figure 4B). To further study interactions between platelets from “matching” and “nonmatching” donors, platelets from the “matching” donor were stained with calcein-green and the platelets of the “nonmatching” donor with calcein-violet (or vice versa), and double-positive events were measured. Upon incubation with WIM8E5, double-positive events, although significant, barely increased from 1 to 2%, while activation induced by PAR1 activating peptide (used as the control activating agonist) led to 8% positive events (Figure 4C). When platelets from two “matching” donors were combined, which were both activated upon incubation with WIM8E5, double-positive events increased from 1 to

6.5% (Figure 4C). As no enhanced CD62P exposure was observed on WIM8E5 “nonmatching” platelets and WIM8E5 had a very minor effect on the percentage of double-positive events when “matching” and “nonmatching” platelets were mixed, these results suggest that activation of platelets is dependent on the ability of HLA antibodies to interact in cis (intra-platelet) with FcγRIIa.

HLA antibodies from patients’ sera can activate platelets To establish whether the findings observed for human HLA monoclonal antibodies were physiologically relevant, we tested whether sera from patients, refractory to

1748

Figure 5. Patients’ sera with HLA antibodies induce FcγRIIa-dependent activation of platelets from a subset of donors. Thirteen sera containing HLA alloantibodies were incubated with platelets from two different donors. AB serum tested negative for HLA and other specific platelet antibodies was used as a control. Background CD62P and IgG binding are indicated by the gray background. Activation was inhibited by preincubation with 10 mg/mL IV.3 or 5 mM Syk inhibitor IV. (A) CD62P exposure on platelets from donor 1 (expressing HLA A1 A2 B51). (B) IgG binding to platelets from donor 1. (C) CD62P exposure on platelets from donor 2 (expressing HLA A1 A2 B35 B62). (D) IgG binding to platelets from donor 2. Antibody specificities of the sera are reported in the Online Supplementary Data.

haematologica | 2018; 103(10)

Platelet activation by HLA antibodies

platelet transfusions and containing HLA antibodies, were capable of activating platelets (Online Supplementary Table S1). Thirteen sera were tested with platelets from two donors. In each serum HLA antibodies matching the HLA type of the donor platelets were present. Addition of control serum lacking HLA antibodies induced low levels of activation in both experiments (Figure 5A,C). Four of the 13 sera induced CD62P exposure on platelets from donor 1. Activation of platelets could be inhibited by IV.3 and Syk inhibitor to background levels as observed for platelets incubated with control serum (Figure 5A,B). When tested with platelets from donor 2, pronounced activation was observed for sera 1, 3, 7, 12 and 13 (Figure 5C,D). Sera 1, 12 and 13 were capable of activating platelets of both donor 1 and 2; serum 8 exclusively activated platelets of donor 1 and sera 3 and 7 only activated platelets from donor 2. Levels of IgG binding were relatively higher for sera which induced enhanced CD62P exposure, although some sera induced significant activation despite relatively low levels of IgG binding. Together,

these results suggest that HLA antibodies in sera from refractory patients can induce FcγRIIa-dependent platelet activation.

HLA monoclonal antibodies induce phagocytosis by macrophages To study the effect of FcγRIIa-dependent platelet activation on platelet clearance, monocyte-derived macrophages were incubated with platelets opsonized with either WIM8E5, SN607D8, SN230G6 or anti-HPA-1a antibody (as a positive control) and phagocytosis was studied employing imaging flow cytometry (Figure 6A-C). Opsonization by WIM8E5 significantly enhanced phagocytosis of platelets as shown by the increase in PKHlabeled platelets that were internalized by macrophages (Figure 6B). Incubation with Syk inhibitor IV significantly reduced phagocytosis of platelets opsonized by WIM8E5 (Figure 6C). Quantitative assessment of the effect of activating (WIM8E5 and SN607D8) and non-activating (SN230G6) revealed that enhanced phagocytosis of

Figure 6. Phagocytosis of platelets opsonized by HLA monoclonal antibodies and the effect of FcγRIIa-dependent signaling. Platelets were incubated with 10 mg/mL WIM8E5, SN607D8 or SN230G6 in the presence or absence of 5 mM Syk inhibitor IV. Opsonized platelets were incubated for 1 h with monocyte-derived macrophages and internalization was analyzed through the use of imaging flow cytometry. (A-C) Representative images of imaging flow cytometry. BF: bright field; CD61: extracellular platelet staining; PKH: platelet staining; HLA-DR: macrophage staining. (A) Control without Syk inhibitor, (B) WIM8E5 without Syk inhibitor, (C) WIM8E5 with Syk inhibitor. (D) Intracellular platelet (PKH) fluorescence quantifies the amount of platelets taken up by macrophages. Data are given as mean ± SD, *P<0.05, **P<0.01. Control: buffer only, no HLA antibodies added. MF: macrophage.

haematologica | 2018; 103(10)

1749

M. Rijkers et al.

WIM8E5 and SN607D8 was significantly decreased in the presence of Syk inhibitor IV (Figure 6D). The results obtained were derived from three independent experiments employing platelets from different donors (Online Supplementary Figure S7). Together, these results show that FcγRIIa-dependent activation of a subset of HLA antibodies promotes uptake of platelets by macrophages.

Discussion The presence of HLA antibodies in patients receiving platelet transfusions is in some cases associated with rapid platelet clearance, which may be accompanied by transfusion reactions, including chills and fever.2,7 Here, we studied whether HLA antibodies can activate platelets. We showed that a subset of human HLA monoclonal antibodies induced platelet α-granule release, integrin αIIbb3 activation and platelet aggregation. All these effects could be fully inhibited by blocking FcγRIIa-dependent signaling in platelets. This indicates that activation by HLA antibodies is induced upon crosslinking of HLA molecules with the platelet Fc receptor FcγRIIa. Employing sera from HLAantibody positive and refractory patients we confirmed that FcγRIIa-dependent platelet activation can be induced on donor platelets. We also showed that FcγRIIa-dependent activation enhances phagocytosis of platelets by macrophages. Although all the monoclonal antibodies tested in this study bound efficiently to HLA-matched platelets, only three of them significantly induced CD62P surface exposure. It has been reported that high levels of HLA antibodies in patients correlate with a higher risk of refractoriness,10 however, we did not observe a correlation between the level of IgG bound to platelets and the ability of HLA monoclonal antibodies or patients’ sera to induce platelet activation. No correlation between platelet activation and binding of HLA monoclonal antibodies to specific HLA alleles was found. Antibodies WIM8E5, GV5D1 and GV2D5 all bind to HLA-A1, but only WIM8E5 and GV5D1 induce significantly increased CD62P membrane exposure on platelets. Similarly, whereas SN607D8 and SN230G6 both bind to HLA-A2, only SN607D8 induces platelet activation. Recently the affinities of SN607D8 and SN230G6 for HLA-A2 were reported: SN607D8 has a KD of 1.2x10-8 M and SN230G6 has a KD of 5.9x10-10 M.33 It is likely that the affinity of HLA antibodies affects their ability to activate platelets to some extent. However, as the non-activating antibody SN230G6 has a significantly higher affinity than that of the activating antibody SN607D8, the propensity of HLA monoclonal antibodies to induce FcγRIIa-dependent platelet activation is not exclusively dependent on their affinity. For SN607D8 and SN230G6, residues critical for their binding to HLA have been determined (149A and 152V25, and 62GE34 respectively). These residues are located on opposite sides of the peptide-binding groove. We, therefore, hypothesize that the location of the binding site on HLA determines whether the Fc tail of an HLA antibody can bind and crosslink with FcγRIIa, causing activation through Syk. We showed that FcγRIIa-dependent signaling enhanced platelet phagocytosis by macrophages. The degree of Sykdependent phagocytosis correlated with the ability of the HLA monoclonal antibodies to activate and induce phosphatidylserine exposure on the platelets (Online 1750

Supplementary Figure S7). Similarly, differences in platelet clearance via activating and non-activating antibodies have been described in human FcγRIIa transgenic mice, in which an anti-CD9 antibody inducing FcγRIIa-platelet activation led to more rapid development of thrombocytopenia compared to a non-activating platelet specific antibody.35 Besides, shock and thrombosis were observed only in the presence of platelet-activating antibodies and not with non-activating antibodies.35 The activating potential of HLA antibodies has not been correlated with a risk of adverse effects, possibly because HLA-incompatible platelets are usually given to thrombocytopenic patients with low platelet counts, reducing the risk of thrombosis. Previous studies on the effect of HLA antibodies on platelets have suggested that they are either inert36 or induce platelet activation through complement activation.20 Our experiments were performed with washed platelets in the absence of complement, so we cannot rule out that complement activation plays a role as well. Rubinstein et al. showed that crosslinking of a murine pan-anti-HLA antibody induced platelet activation.19 However, the effect of the anti-HLA mouse monoclonal antibody that they used was only observed by artificial crosslinking of the HLA antibodies with goat antimouse IgG. Employing a mouse model, Waterman et al. showed that Fcγ receptors were critical for mice to develop platelet refractoriness induced by major histocompatibility complex alloantibodies.37 Previous studies have shown that both murine and human HLA antibodies can induce release of Weibel-Palade bodies and subsequent CD62P exposure on endothelial cells, inducing enhanced monocyte adherence to endothelial cells via both CD62P binding and crosslinking of endothelial HLA with monocyte Fc receptors.38 It has been described previously that platelets can be activated by monoclonal antibodies directed to b2-microglobulin in an FcγRIIa-dependent manner.32 Similar to our observations, Rubinstein et al. reported differences in the platelet-activating abilities of several monoclonal antibodies directed towards different epitopes on b2-microglobulin.32 However, activation by anti-b2-microglobulin monoclonal antibodies was proposed to occur in an inter-platelet fashion, while our data provide evidence for intra-platelet activation by HLA monoclonal antibodies. The mechanism of platelet activation by HLA monoclonal antibodies is similar to that of the platelet activation described in heparin-induced thrombocytopenia, in which antibodies directed to the platelet factor 4/ heparin complexes develop and induce FcγRIIa-dependent platelet activation.17 Intravenous immunoglobulin has been shown to have a beneficial effect in patients with heparin-induced thrombocytopenia;39,40 here we have shown that intravenous immunoglobulin inhibits HLA monoclonal antibodyinduced FcγRIIa-dependent platelet activation in vitro. Based on our observations it may be worth exploring whether intravenous immunoglobulin can be used to reduce the rapid clearance of transfused platelets in a subset of refractory patients with anti-platelet antibodies. Li et al. described a mechanism in which platelets are activated by GPIbα antibodies in an FcγRIIa-independent way.16 Recently, Quach et al. showed that this activation was dependent on mechanomolecular signaling.41 Similar to our data, Li et al. and Quach et al. documented enhanced CD62P exposure on platelets. In addition, their data suggested that activation of platelets coincides with haematologica | 2018; 103(10)

Platelet activation by HLA antibodies

Figure 7. Proposed mechanism of platelet activation. When an activating HLA antibody (ab) binds to HLA on the platelet surface it crosslinks with FcγRIIa. This induces FcγRIIa-dependent signaling leading to the activation of Syk via the ITAM motif on FcγRIIa. Downstream signaling leads to platelet activation: α-granules are released (followed by e.g. CD62P exposure), integrin αIIbb3 is activated and platelets start to aggregate. This activation pathway can be inhibited by IV.3 (which blocks crosslinking with FcγRIIa) or Syk inhibitor IV (which blocks signaling via Syk). HLA antibodies that bind to an epitope on HLA preventing interaction with FcγRIIa do not induce platelet activation. FcγRIIa-dependent signaling also leads to phosphatidylserine exposure and induces enhanced phagocytosis by macrophages.

the release of neuraminidase which cleaves sialic acid from platelet surface receptors thereby enhancing clearance of platelets through the Ashwell Morell receptors expressed by hepatocytes.42 We show that HLA antibodies activate platelets in an FcγRIIa-dependent manner. Whether this also results in release of sialidase and subsequent clearance of platelets in patients with HLA antibodies has not been studied. However, our macrophage internalization experiments show that platelet activation induced by HLA alloantibodies may have an impact on platelet survival. Incubation of healthy donor platelets with patients’ sera containing HLA antibodies revealed that FcγRIIa-dependent platelet activation could be induced by polyclonal HLA antibodies from approximately one third of the tested sera from refractory patients. Some sera, such as sera 7 and 8, only induced activation of platelets in one of the two donors. This suggests that only a subset of antibodies present in the polyclonal sera is responsible for FcγRIIadependent platelet activation. Apparently, these antibodies only matched HLA antigens present on one of the two donors. HLA antibodies present in sera which did not induce FcγRIIa-dependent activation with the donor platelets tested might potentially have a different effect when tested with platelets derived from a larger panel of donors. As observed for the panel of monoclonal HLA

References 1. Hickey MJ, Valenzuela NM, Reed EF, Reed EF. Alloantibody generation and effector function following sensitization to human leukocyte antigen. Front Immunol. 2016;7:1-13. 2. Stanworth SJ, Navarrete C, Estcourt L, Marsh J. Platelet refractoriness – practical

haematologica | 2018; 103(10)

antibodies, the results obtained for the patients’ sera indicate that only a subset of HLA antibodies is capable of inducing FcγRIIa-dependent platelet activation. Prediction models based on the three-dimensional structure of HLA epitopes (eplets), rather than the HLA antigens itself, have been developed as a basis for matched platelet transfusions.34,43 If the eplets targeted by plateletactivating HLA antibodies were known, it is possible that incorporation of this knowledge in the selection of HLAmatching donor platelets could help to further reduce side effects of platelet transfusions in refractory patients. In conclusion, we have shown that platelets are activated in an FcγRIIa-dependent manner by a subset of HLA antibodies (Figure 7). This mechanism may contribute to the enhanced clearance of platelets in refractory patients, as suggested by the increased phagocytosis of platelets opsonized by a subset of activating HLA antibodies. This suggests that testing the capacity of patients’ sera to induce platelet activation could be used to further stratify patients with HLA antibodies who need platelet transfusions. Acknowledgments Supported by grants PPOC-2013-019 and PPOC-2015024P (Netherlands Ministry of Health). AJGJ is supported by a Clinical Fellowship of the European Hematology Association.

approaches and ongoing dilemmas in patient management. Br J Haematol. 2015;171(3):297-305. 3. Pavenski K, Freedman J, Semple JW. HLA alloimmunization against platelet transfusions: pathophysiology, significance, prevention and management. Tissue Antigens. 2012;79(4):237-245. 4. De Clippel D, Baeten M, Torfs A, et al.

Screening for HLA antibodies in plateletpheresis donors with a history of transfusion or pregnancy. Transfusion. 2014;54(12): 3036-3042. 5. Triulzi DJ, Kleinman S, Kakaiya RM, et al. The effect of previous pregnancy and transfusion on HLA alloimmunization in blood donors: implications for a transfusion-related acute lung injury risk reduction strategy.

1751

M. Rijkers et al. Transfusion. 2009;49(9):1825–1835. 6. Middelburg RA, Porcelijn L, Lardy N, Briët E, Vrielink H. Prevalence of leucocyte antibodies in the Dutch donor population. Vox Sang. 2011;100(3):327-335. 7. Rebulla P. A mini-review on platelet refractoriness. Haematologica. 2005;90(2):247253. 8. Doughty H, Murphy M, Metcalfe P, et al. Relative importance of immune and nonimmune causes of platelet refractoriness. Vox Sang. 1994;66(3):200-205. 9. Saris A, Tomson B, Brand A, et al. Platelets from donors with consistent low HLA-B8, B12 or B35 expression do not undergo antibody-mediated internalization. Blood. 2018;131(1):144-152. 10. Jackman RP, Deng X, Bolgiano D, et al. Lowlevel HLA antibodies do not predict platelet transfusion failure in TRAP study participants. Blood. 2013;121(16):3261-3267. 11. Jia Y, Li W, Liu N, et al. Prevalence of platelet-specific antibodies and efficacy of crossmatch-compatible platelet transfusions in refractory patients. Transfus Med. 2014; 26(6):406-410. 12. Messerschmidt G, Makuch R, Appelbaum F, et al. A prospective randomized trial of HLA-matched versus mismatched singledonor platelet transfusions in cancer patients. Cancer. 1988;62(4):795-801. 13. Zeng Q, Zhu L, Tao L, et al. Relative efficacy of steroid therapy in immune thrombocytopenia mediated by anti-platelet GPIIbIIIa versus GPIba antibodies. Am J Hematol. 2011;87(2):206-208. 14. Peng J, Ma S, Liu J, et al. Association of autoantibody specificity and response to intravenous immunoglobulin G therapy in immune thrombocytopenia: a multicenter cohort study. J Thromb Haemost. 2014;12(4):497-504. 15. Berchtold P, Wenger M. Autoantibodies against platelet glycoproteins in autoimmune thrombocytopenic purpura: their clinical significance and response to treatment. Blood. 1993;81(5):1246-1250. 16. Li J, van der Wal DE, Zhu G, et al. Desialylation is a mechanism of Fc-independent platelet clearance and a therapeutic target in immune thrombocytopenia. Nat Commun. 2015;6:7737. 17. Arepally GM. Heparin-induced thrombocytopenia. Blood. 2017;129(21):2864-2872. 18. Rollin J, Pouplard C, Sung HC, et al. Increased risk of thrombosis in FcγRIIA 131RR patients with HIT due to defective control of platelet activation by plasma IgG2. Blood. 2015;125(15):2397-2404. 19. Rubinstein BE, Urso I, Boucheix C, Carroll RC. Platelet activation by cross-linking HLA

1752

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

class I molecules and Fc receptor. Blood. 1992;79(11):2901-2908. Brandt J, Julius C, Osborne J, Anderson C. The mechanism of platelet aggregation induced by HLA-related antibodies. Thromb Haemost. 1996;76(5):774-779. Mulder A, Eijsink C, Kardol MJ, et al. Identification, isolation, and culture of HLAA2-specific B lymphocytes using MHC class I tetramers. J Immunol. 2003;171(12):65996603. Mulder A, Kardol M, Regan J, Buelow R, Claas F. Reactivity of twenty-two cytotoxic human monoclonal HLA antibodies towards soluble HLA class I in an enzymelinked immunosorbent assay (PRA- STAT). Hum Immunol. 1997;56(1-2):106-113. Nightingale MJ, Ceulemans J, Ágoston S, et al. The value to blood establishments of supplier quality audit and of adopting a European Blood Alliance collaborative approach. Blood Transfus. 2014;12(1):91-98. Rijkers M, van der Meer PF, Bontekoe IJ, et al. Evaluation of the role of the GPIb-IX-V receptor complex in development of the platelet storage lesion. Vox Sang. 2016;111(3):247-256. Mulder A, Eijsink C, Kester MGD, et al. Impact of peptides on the recognition of HLA class I molecules by human HLA antibodies. J Immunol. 2005;175(9):5950–5957. Maynard DM, Heijnen HFG, Horne MK, White JG, Gahl WA. Proteomic analysis of platelet alpha-granules using mass spectrometry. J Thromb Haemost. 2007;5(9): 1945-1955. Shattil SJ, Hoxie JA, Cunningham M, Brass LF. Changes in the platelet membrane glycoprotein IIb- IIIa complex during platelet activation. J Biol Chem. 1985;260(20):1110711114. Qiao J, Al-Tamimi M, Baker RI, Andrews RK, Gardiner EE. The platelet Fc receptor, FcγRIIa. Immunol Rev. 2015;268(1):241252. Arman M, Krauel K. Human platelet IgG Fc receptor FcγRIIA in immunity and thrombosis. J Thromb Haemost. 2015;13(6):893– 908. Yamamoto N, Takeshita K, Shichijo M, et al. The orally available spleen tyrosine kinase inhibitor 2-[7-(3,4-dimethoxyphenyl)-imidazo[1,2-c]pyrimidin-5-ylamino]-nicotinamide dihydrochloride (BAY 61-3606) blocks antigen- induced airway inflammation in rodents. J Pharmacol Exp Ther. 2003;306(3): 1174-1181. Yanaga F, Poole A, Asselin J, et al. Syk interacts with tyrosine-phosphorylated proteins in human platelets activated by collagen and cross-linking of the Fcγ-llA receptor.

Biochem J. 1995;311(Pt2):471-478. 32. Rubinstein E, Boucheix C, Urso I, Carroll RC. Fc gamma receptor-mediated interplatelet activation by a monoclonal antibody against beta 2 microglobulin. J Immunol. 1991;147(9):3040-3046. 33. Daga S, Moyse H, Briggs D, et al. Direct quantitative measurement of the kinetics of HLA-specific antibody interactions with isolated HLA proteins. Hum Immunol. 2018;79(2):122-128. 34. Duquesnoy RJ, Marrari M, Jelenik L, et al. Structural aspects of HLA class I epitopes reacting with human monoclonal antibodies in Ig-binding, C1q-binding and lymphocytotoxicity assays. Hum Immunol. 2013;74(10): 1271-1279. 35. Taylor SM, Reilly MP, Schreiber AD, et al. Thrombosis and shock induced by activating antiplatelet antibodies in human FcγRIIA transgenic mice: the interplay among antibody, spleen, and Fc receptor. Blood. 2000;96(13):4254-4260. 36. Kretschmer V, Huss B, Feller K, Ponath N, Kiefel V. Influence of HPA-1a and HLA antibodies on primary hemostasis. Semin Thromb Hemost. 1995;21(suppl 2):96-102. 37. Waterman HR, Kapp LM, Munday A, Odem-Davis K, Zimring JC. Transfusioninduced alloimmunization and platelet refractoriness in a mouse model: mechanisms and interventions. Transfusion. 2016;56(1):91-100. 38. Valenzuela NM, Mulder A, Reed EF. HLA class I antibodies trigger increased adherence of monocytes to endothelial cells by eliciting an increase in endothelial P-selectin and, depending on subclass, by engaging FcγRs. J Immunol. 2013;190(12):6635-6650. 39. Jones CG, Pechauer SM, Curtis BR, et al. IgG in normal plasma inhibits HIT antibodymediated platelet activation: implications for plasma exchange in HIT. Blood. 2018;131(3):703-706. 40. Padmanabhan A, Jones CG, Pechauer SM, et al. IVIg for treatment of severe refractory heparin-induced thrombocytopenia. Chest. 2017;152(3):478-485. 41. Quach ME, Dragovich MA, Chen W, et al. Fc-independent immune thrombocytopenia via mechanomolecular signaling in platelets. Blood. 2018;15(131):787-796. 42. Hoffmeister KM, Falet H. Platelet clearance by the hepatic Ashwell-Morrell receptor: mechanisms and biological significance. Thromb Res. 2016;141 Suppl 2:S68-S72. 43. Duquesnoy R, Marrari M, Mulder A, et al. First report on the antibody verification of HLA-ABC epitopes recorded in the websitebased HLA Epitope Registry. Tissue Antigens. 2014;83(6):391-400.

haematologica | 2018; 103(10)