Haematologica, Volume 104, Issue 4 by Haematologica

haematologica Journal of the Ferrata Storti Foundation

Editor-in-Chief Luca Malcovati (Pavia)

Managing Director Antonio Majocchi (Pavia)

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

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

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

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

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

haematologica Journal of the Ferrata Storti Foundation

Information for readers, authors and subscribers Haematologica (print edition, pISSN 0390-6078, eISSN 1592-8721) publishes peer-reviewed papers on all areas of experimental and clinical hematology. The journal is owned by a non-profit organization, the Ferrata Storti Foundation, and serves the scientific community following the recommendations of the World Association of Medical Editors (www.wame.org) and the International Committee of Medical Journal Editors (www.icmje.org). Haematologica publishes editorials, research articles, review articles, guideline articles and letters. Manuscripts should be prepared according to our guidelines (www.haematologica.org/information-for-authors), and the Uniform Requirements for Manuscripts Submitted to Biomedical Journals, prepared by the International Committee of Medical Journal Editors (www.icmje.org). Manuscripts should be submitted online at http://www.haematologica.org/. Conflict of interests. According to the International Committee of Medical Journal Editors (http://www.icmje.org/#conflicts), “Public trust in the peer review process and the credibility of published articles depend in part on how well conflict of interest is handled during writing, peer review, and editorial decision making”. The ad hoc journal’s policy is reported in detail online (www.haematologica.org/content/policies). Transfer of Copyright and Permission to Reproduce Parts of Published Papers. Authors will grant copyright of their articles to the Ferrata Storti Foundation. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties. Detailed information about subscriptions is available online at www.haematologica.org. Haematologica is an open access journal. Access to the online journal is free. Use of the Haematologica App (available on the App Store and on Google Play) is free. For subscriptions to the printed issue of the journal, please contact: Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, E-mail: info@haematologica.org). Rates of the International edition for the year 2019 are as following: Print edition

Institutional Euro 700

Personal Euro 170

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

haematologica Journal of the Ferrata Storti Foundation

Table of Contents Volume 104, Issue 4: April 2019 Cover Figure

Bone marrow smear from a patient with atypical chronic myeloid leukemia, BCR-ABL1 negative, showing abnormal chromatin clumping, hyosegmented nuclei, and agranular cytoplasm in late granulocytic cells. Courtesy of Prof. Rosangela Invernizzi.

Editorials 639

Exploitation of the neural-hematopoietic stem cell niche axis to treat myeloproliferative neoplasms Naoimh Herlihy et al.

642

Expounding on the essence of epigenetic and genetic abnormalities in blastic plasmacytoid dendritic cell neoplasms Lhara Lezama et al.

644

THROMBOTECT takes the lead Christoph Male et al.

646

Asymmetric dimethylarginine – a prognostic marker for transplant outcome? Janghee Woo and H. Joachim Deeg

Perspective Article 648

Liquid biopsy in lymphoma Davide Rossi et al.

Review Articles 653

Re-evaluation of hematocrit as a determinant of thrombotic risk in erythrocytosis Victor R. Gordeuk et al.

659

State-of-the-art review: allogeneic stem cell transplantation for myelofibrosis in 2019 Donal P. McLornan et al.

Articles Hematopoiesis

669

In vitro and in vivo evaluation of possible pro-survival activities of PGE2, EGF, TPO and FLT3L on human hematopoiesis Eva-Maria Demmerath et al.

Iron Metabolism & its Disorders

678

Gastrointestinal iron excretion and reversal of iron excess in a mouse model of inherited iron excess Courtney J. Mercadante et al.

Red Cell Biology & its Disorders

690

Hemodynamic provocation with acetazolamide shows impaired cerebrovascular reserve in adults with sickle cell disease Lena Václavuº et al.

Myelodysplastic Syndromes

700

Azacitidine with or without lenalidomide in higher risk myelodysplastic syndrome & low blast acute myeloid leukemia Melita Kenealy et al.

Myeloproliferative Neoplasms

710

The sympathomimetic agonist mirabegron did not lower JAK2-V617F allele burden, but restored nestin-positive cells and reduced reticulin fibrosis in patients with myeloproliferative neoplasms: results of phase II study SAKK 33/14 Beatrice Drexler et al.

Chronic Myeloid Leukemia

717

A new BCR-ABL1 Drosophila model as a powerful tool to elucidate the pathogenesis and progression of chronic myeloid leukemia Roberto Bernardoni et al.

Myeloid Neoplasms Haematologica 2019; vol. 104 no. 4 - April 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation 729

Blastic plasmacytoid dendritic cell neoplasm: genomics mark epigenetic dysregulation as a primary therapeutic target Maria Rosaria Sapienza et al.

Acute Lymphoblastic Leukemia

738

Autophagy inhibition as a potential future targeted therapy for ETV6-RUNX1-driven B-cell precursor acute lymphoblastic leukemia Roel Polak et al.

749

CD123 expression patterns and selective targeting with a CD123-targeted antibody-drug conjugate (IMGN632) in acute lymphoblastic leukemia Evgeniya Angelova et al.

756

THROMBOTECT – a randomized study comparing low molecular weight heparin, antithrombin and unfractionated heparin for thromboprophylaxis during induction therapy of acute lymphoblastic leukemia in children and adolescents Jeanette Greiner et al.

Non-Hodgkin Lymphoma

766

TIRAP p.R81C is a novel lymphoma risk variant which enhances cell proliferation via NF-κB mediated signaling in B-cells Regula Burkhard et al.

778

Pharmacological modulation of CXCR4 cooperates with BET bromodomain inhibition in diffuse large B-cell lymphoma Clara Recasens-Zorzo et al.

Chronic Lymphocytic Leukemia

789

Deep targeted sequencing of TP53 in chronic lymphocytic leukemia: clinical impact at diagnosis and at time of treatment Christian Brieghel et al.

797

First-line therapy in chronic lymphocytic leukemia: a Swedish nation-wide real-world study on 1053 consecutive patients treated between 2007 and 2013 Sandra Eketorp Sylvan et al.

Platelet Biology & its Disorders

806

Aerobic glycolysis fuels platelet activation: small-molecule modulators of platelet metabolism as anti-thrombotic agents Paresh P. Kulkarni et al.

Coagulation & its Disorders

819

New insight into antiphospholipid syndrome: antibodies to β2glycoprotein I-domain 5 fail to induce thrombi in rats Paolo Durigutto et al.

Stem Cell Transplantation

827

Asymmetric dimethylarginine serum levels are associated with early mortality after allogeneic stem cell transplantation Aleksandar Radujkovic et al.

835

Disability related to chronic graft-versus-host disease after alternative donor hematopoietic cell transplantation Giancarlo Fatobene et al.

Blood Transfusion

844

Related peripheral blood stem cell donors experience more severe symptoms and less complete recovery at one year compared to unrelated donors Michael A. Pulsipher et al.

Obituary 855

Francesco Lo Coco, a distinguished hematologist and a friend Paolo Corradini, Pellegrino Musto and Marco Vignetti

Letters to the Editor Letters are available online only at www.haematologica.org/content/104/4.toc

e127

Low serum haptoglobin and blood films suggest intravascular hemolysis contributes to severe anemia in hereditary hemorrhagic telangiectasia Lieze Thielemans et al. http://www.haematologica.org/content/104/4/e127

Haematologica 2019; vol. 104 no. 4 - April 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation e131

A phase II study of the efficacy and safety of an intensified schedule of azacytidine in intermediate-2 and high-risk patients with myelodysplastic syndromes: a study by the Groupe Francophone des Myelodysplasies (GFM) Lionel Ades et al. http://www.haematologica.org/content/104/4e131

e134

Positive impact of molecular analysis on prognostic scores in essential thrombocythemia: a single center prospective cohort experience Damien Luque Paz et al. http://www.haematologica.org/content/104/4/e134

e138

Antitumor synergy with SY-1425, a selective RARα agonist, and hypomethylating agents in retinoic acid receptor pathway activated models of acute myeloid leukemia Michael R. McKeown et al. http://www.haematologica.org/content/104/4/e138

e143

Phase I/II trial of cladribine, high-dose cytarabine, mitoxantrone, and G-CSF with dose-escalated mitoxantrone for relapsed/refractory acute myeloid leukemia and other high-grade myeloid neoplasms Anna B. Halpern et al. http://www.haematologica.org/content/104/4/e143

e147

Extramedullary acute myeloid leukemia presenting in young adults demonstrates sensitivity to high-dose anthracycline: a subset analysis from ECOG-ACRIN 1900 Hugo F. Fernandez et al. http://www.haematologica.org/content/104/4/e147

e151

Phase I dose-escalation study of brentuximab-vedotin combined with dexamethasone, high-dose cytarabine and cisplatin, as salvage treatment in relapsed/refractory classical Hodgkin lymphoma: The HOVON/LLPC Transplant BRaVE study Anton Hagenbeek et al. http://www.haematologica.org/content/104/4/e151

e154

Targeted next generation sequencing reveals high mutation frequency of CREBBP, BCL2 and KMT2D in high-grade B-cell lymphoma with MYC and BCL2 and/or BCL2 rearrangements Solène M. Evrard et al. http://www.haematologica.org/content/104/4/e154

e158

DUSP22-rearranged anaplastic lymphomas are characterized by specific morphological features and a lack of cytotoxic and JAK/STAT surrogate markers Arantza Onaindia et al. http://www.haematologica.org/content/104/4/e158

e163

Clinical presentation determines selection of patients for initial observation in mantle cell lymphoma Anita Kumar et al, http://www.haematologica.org/content/104/4/e163

Case Reports Case Reports are available online only at www.haematologica.org/content/104/4.toc

e167

Recognition of hemophagocytic lymphohistiocytosis in sickle cell vaso-occlusive crises is a potentially lifesaving diagnosis Orly Leiva et al. http://www.haematologica.org/content/104/4/e167

e170

Serendipity: decitabine monotherapy induced complete molecular response in a 77-year-old patient with acute promyelocytic leukemia Ramzi Abboud et al. http://www.haematologica.org/content/104/4/e170

e174

Transmission of diffuse large B-cell lymphoma by an allogeneic stem-cell transplant Shamzah Araf et al. http://www.haematologica.org/content/104/4/e174

Comments Comments are available online only at www.haematologica.org/content/104/4.toc

e178

Comment to “The outcome of peripheral T-cell lymphoma patients failing first-line therapy” Peter Dreger http://www.haematologica.org/content/104/4/e178

e179

The outcome of peripheral T-cell lymphoma patients failing first-line therapy: a report from the prospective International T-Cell Project Monica Bellei and Massimo Federico http://www.haematologica.org/content/104/4/e179

Haematologica 2019; vol. 104 no. 4 - April 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Ancient Greek

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

Scientific Latin

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

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

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

EDITORIALS Exploitation of the neural-hematopoietic stem cell niche axis to treat myeloproliferative neoplasms Naoimh Herlihy, Claire N Harrison and Donal P McLornan Department of Haematology, Guy’s and St. Thomas’ NHS Foundation Trust, London, UK E-mail: CLAIRE HARRISON - claire.harrison@gstt.nhs.uk doi:10.3324/haematol.2018.211896

yeloproliferative neoplasms (MPN) originate from a population of hematopoietic stem cells (HSCs) within the bone marrow (BM) that undergo clonal expansion as a result of factors both intrinsic and extrinsic to the cell. They are characterized by progressive marrow fibrosis, heterogeneous symptomatology, extramedullary hematopoiesis, splenomegaly, a propensity to both hemorrhage and thrombosis, and an inherent risk of transformation to acute myeloid leukemia (AML). Recently, there has been an increasing focus on the role of the BM HSC niche in MPN development and disease maintenance.1 This niche encompasses complex cellular and signaling networks with multiple interactions and ‘cross-talk’ between HSC, mesenchymal cells (MSC), perivascular cells identified as chemokine (C–X–C motif) ligand 12 (CXCL12)-abundant reticular (CAR) cells, osteolineage-derived cells and sinusoidal endothelial cells, amongst others1-3 (Figure 1). Importantly, the acquisition of gain-of-function mutations such as JAK2-V617F by MPN HSCs or other cells can result in an alteration of the niche to favor clonal expansion at the expense of background normal HSCs.4 For example, Zhan et al. recently suggested that JAK2-V617F mutant endothelial cells in the vascular niche promote clonal expansion of JAK2-V617F HSCs at the expense of wild-type progenitors.5 In chronic myeloid leukemia (CML), one study found BCR/ABL transgenic mice create a self-reinforcing leukemic niche that impairs normal hematopoiesis, favors leukemic stem cell function and contributes to development of BM fibrosis by stimulating MSCs to overproduce osteoclastic stem cells.6 Conversely, there is also mounting evidence that targeted disruption of the HSC niche may result in development of MPN, for example defective Notch activation7 and ablation of the retinoic acid receptor gamma8 and retinoblastoma genes9 have been evaluated in murine models, suggesting that genetic mutations in the niche itself can also drive the malignant process. Little was known about the role of the MSC population in MPN-pathogenesis and maintenance and data were often conflicting. Avanzini et al. described how MPN-BM-derived MSC exhibited decreased proliferative and osteogenic capacity whereas Martinaud et al. suggested an enhanced and persistent increase in osteogenic abilities coupled with an altered secretome and transcriptome in primary myelofibrosis (PMF)-derived BM-MSC.10,11 More recently, Ramos et al. demonstrated that BM-MSC derived from JAK2-mutated MPN patients favored maintenance of clonal hematopoietic cells.12 Targeting of this population, in particular increasing the ‘beneficial’ MSC subgroup, is hence of potential interest. Nearly a decade ago, Méndez-Ferrer et al. originally described how Nestin-positive(+) MSC are essential components of the HSC niche, containing all of the BM colony-forming-unit fibroblastic activity and having the ability to function as so-called ‘niche-forming’ cells.13,14 Nestin is a type VI intermediate filament protein and functions as a major component of the cytoskeleton. Méndez-Ferrer et al. described that cyclical haematologica | 2019; 104(4)

HSC trafficking is regulated by noradrenaline release via the sympathetic nervous system (SNS), transmitted to niche stromal cells by the beta(β)(3)-adrenergic receptor, resulting in reduced nuclear Sp1 transcription factor and downregulation of CXCL12. There is a close association between so-called Nestin+ MSC and adrenergic nerve fibers of the SNS and this is believed to control HSC maintenance, egress and functional capacity.15 Of additional interest, Maryanovich et al. have recently described how the aging HSC niche is associated with a distinct loss of functional SNS nerve fibers, postulating that denervation-associated remodeling would lead to ‘aged’ HSC.16 Méndez-Ferrer et al. extended their original observations and demonstrated that SNS fibers intrinsically involved in supporting Schwann cells and nestin+ MSC are reduced in the marrow of MPN patients compared to healthy individuals.14 Schwann cell death is initiated following mutant-HSC IL-1β production, and the resultant denervation ultimately leads to a reduction in so-termed ‘beneficial MSC’ and facilitates expansion of clonal mutant-HSC within the niche. During disease progression, the supportive microenviroment is further disrupted by mutantHSC-mediated hypercytokinemia and thus the MSC population and the osteoblast lineage cells continue to modulate. In the same publication, an MPN murine model was used to demonstrate that restoration of sympathetic regulation of Nestin+ MSC induced via β3-agonist exposure abrogated MPN progression and led to a reduction in disease-associated HSC.17 The β3-adrenergic agonist BRL37344 led to reductions in murine BM mutant-HSC-derived progenitors and decreases in neutrophilia, thrombocytosis and marrow fibrosis, associated with a restoration of Nestin+ MSC. In addition, BRL37344 exposure led to a significant decrease in ‘leukemic’ stem cells. This pivotal work rapidly led to conceptualization of adrenergic nervous system modulation as a novel therapeutic approach in MPN. In solid tumor oncogenesis, in vitro studies investigating neuro-biological regulation of tumor establishment, aggressiveness and metastases have emerged over the last five years.18-20 By way of example, in contrast to what has been described above in MPN, Magnon et al. studied mice bearing PC-3 prostate tumor xenografts and human prostate tumor specimens, and demonstrated that prostate cancer growth was down-regulated following chemical or surgical sympathectomy or deletion of stromal β3-adrenergic receptors; tumor samples demonstrated higher densities of adrenergic (surrounding the tumor) and cholinergic (invading the tumor) nerve fibers associated with poorer prognosis, and that the cholinergic parasympathetic nerve fibers were associated with cancer dissemination.20 Further links between neuro-modulation and prostate cancer have been suggested by a Norwegian epidemiological study suggesting β-blocker use was associated with reduced prostate cancer mortality.21 Other in vivo work has also investigated the importance of SNS modulation in breast, ovarian and melanoma tumorigenesis, amongst others, and how 639

Editorials

Figure 1. Bone marrow hematopoietic stem cell (HSC) niche in the development and disease maintenance of myeloproliferative neoplasms (MPN). This niche encompasses complex cellular and signaling networks with multiple interactions and â&#x20AC;&#x2DC;cross-talkâ&#x20AC;&#x2122; between HSC, mesenchymal cells (MSC), perivascular cells identified as chemokine ligand 12-abundant reticular cells, osteolineage-derived cells and sinusoidal endothelial cells, amongst others.1-3

this is intrinsically linked to both adaptive and innate tumor-immune responses (reviewed by Qiao et al.22). In this issue of Haematologica, Drexler et al. report a novel multi-center, phase II trial exploring the refocused use of mirabegron, an oral Î˛-3 adrenergic agonist with a lower receptor affinity than BRL37344 but commonly used for overactive bladder syndrome. This is the first study in MPN to explore the neural-HSC niche.23 Mirabegron (25 mg titrated to 50 mg once daily) was administered to 39 patients with a diagnosis of MPN fulfilling the World Health Organization (WHO) 2008 diagnostic criteria, and in whom the JAK2-V617F mutant allele burden in granulocyte DNA exceeded 20% at study entry. The primary end point was defined as the reduction in the JAK2-V617F mutant allele burden of 50% or more after 24 weeks. A sub-project, in which 20 patients participated, assessed whether mirabegron can restore the Nestin+ MSC population and alter reticulin fibrosis. The primary end point was 640

not reached in any of the patients, although a 25% reduction in JAK2-V617F allele burden at 24 weeks was recorded in one patient. Twenty-four percent of patients with polycythemia vera and 29% of patients with essential thrombocythemia showed hematologic response in accordance with European LeukemiaNet (ELN) and International Working Group Myeloproliferative Neoplasms Research and Treatment (IWG-MRT) criteria. One patient with myelofibrosis became transfusion independent. These clinical findings were considered of minor, possibly insignificant, benefit. Evaluation of BM biopsies showed an increase in Nestin+ MSC from a median of 1.09 [interquartile range (IQR) 0.38-2.37/mm2] to 3.95 (IQR 1.988.79/mm2) (P<0.0001), and a slight but significant decrease in reticulin fibrosis from a median grade of 1.0 (IQR 0-3) to 0.5 (IQR 0-2) (P=0.01) between the start and end of mirabegron treatment. The decrease in reticulin fibrosis was only observed in patients not previously treated with haematologica | 2019; 104(4)

Editorials

hydroxycarbamide. A trend was also observed towards reduction in megakaryocyte cluster formation and decrease in numbers of large megakaryocytes with staghorn-like morphology, both cardinal features of MPN. Possible confounding factors are that the participants had established MPN with a median time between MPN diagnosis and trial registration of 3.6 years (range 1.6-8.6 years) and received a relatively short duration of therapy. It is possible that longer exposure to mirabegron, which was well tolerated, and considering its use earlier in the disease course may have demonstrated different results since potentially the chronicity and degree of increase in nestin+ MSC may be paramount in ultimately modulating the clinical phenotype. Mirabegron itself, as noted above, is a less potent agent and may have additional effects when compared to BRL37344, which was used in the murine work. Moreover, although significant increases in Nestin+ MSC were seen, this was limited to those patients not receiving hydroxycarbamide, an agent that most likely affected MSC senescence. As we begin to explore the complex BM HSC-neurostromal interaction through niche targeting therapies, such as that described in the novel study described above, significant thought needs to be given to the timing of such therapies within the disease course and that logical sequencing / dual agent approaches are considered when targeting of this axis is contemplated. For example, the JAK inhibitor ruxolitinib has been shown to abrogate MSCgrowth and an ability to secrete MCP and IL-6 which could potentially be beneficial through downregulation of proinflammatory MSC, yet it is unknown what effect prior or concurrent JAK inhibitor therapy would have on β3 adrenergic agonist mediated-increases in Nestin+ MSC.24 It is increasingly evident, however, that strategies with the ability to break through the self-perpetuating inflammatory and pro-oncogenic MPN microenvironment are required. Potential combination approaches with drugs acting on this axis to explore include the human pentraxin-2 protein analog PRM-151 (Promedia Pharmaceuticals), which may potentially synergistically reduce fibrosis, or with interferon, which may ‘switch on’ mutant-HSC-directed immune responses. Both of which are being assessed as single agents and in combination with ruxolitinib (reviewed by Harrison and McLornan25). Furthermore, an additional benefit of focusing on the stem cell niche could generate a surrogate marker of disease response that would facilitate more rapid evaluation of niche-targeting therapies in the clinical arena. Surrogate markers of disease response moving beyond the blood count, symptoms and spleen are urgently required in this field. An enhanced understanding of the role of the neural network within the MPN niche and how this can be successfully modulated will accelerate the design of such synergistic therapeutic approaches and help the field to move forward. This work from MéndezFerrer and Skoda is also an inspiration in following the academic trail from bench to bedside.13,14,17,23

References 1. Schmitt-Graeff AH, Nitschke R, Zeiser R. The Hematopoietic Niche in Myeloproliferative Neoplasms. Mediators Inflamm. 2015;2015:347270. 2. Schepers K, Campbell T, Passegué E. Normal and leukemic stem cell niches: insights and therapeutic opportunities. Cell Stem Cell.

haematologica | 2019; 104(4)

2015;16(3):254-267. 3. Gao X, Xu C, Asada N, Frenette PS. The hematopoietic stem cell niche: from embryo to adult. Development. 2018;145(2). 4. Mead AJ, Mullally A. Myeloproliferative neoplasm stem cells. Blood. 2017;129(12):1607-1616. 5. Zhan H, Lin CHS, Segal Y, Kaushansky K. The JAK2V617F-bearing vascular niche promotes clonal expansion in myeloproliferative neoplasms. Leukaemia. 2018;32(2):462-469. 6. Schepers K, Pietras EM, Reynaud D, et al. Myeloproliferative neoplasia remodels the endosteal bone marrow niche into a self-reinforcing leukemic niche. Cell Stem Cell. 2013;13(3):285-299. 7. Kim YW, Koo BK, Jeong HW, et al. Defective Notch activation in microenvironment leads to myeloproliferative disease. Blood. 2008;112(12):4628-4638. 8. Walkley CR, Olsen GH, Dworkin S, et al. A microenvironmentinduced myeloproliferative syndrome caused by retinoic acid receptor gamma deficiency. Cell. 2007;129(6):1097-1110 9. Wang L, Zhang H, Rodriguez S, et al. Notch-dependent repression of miR-155 in the bone marrow niche regulates hematopoiesis in an NFκB-dependent manner. Cell Stem Cell. 2014;15(1):51-65. 10. Avanzini MA, Bernardo ME, Novara F, et al. Functional and genetic aberrations of in vitro-cultured marrow-derived mesenchymal stromal cells of patients with classical Philadelphia-negative myeloproliferative neoplasms. Leukemia. 2014;28(8):1742-1745. 11. Martinaud C, Desterke C, Konopacki J, et al. Osteogenic Potential of Mesenchymal Stromal Cells Contributes to Primary Myelofibrosis. Cancer Res. 2015;75(22):4753-4765. 12. Ramos TL, Sánchez-Abarca LI, Rosón-Burgo B, et al. Mesenchymal stromal cells (MSC) from JAK2+ myeloproliferative neoplasms differ from normal MSC and contribute to the maintenance of neoplastic hematopoiesis. PLoS One. 2017;12(8):e0182470. 13. Méndez-Ferrer S, Lucas D, Battista M, Frenette PS. Haematopoietic stem cell release is regulated by circadian oscillations. Nature. 2008;452(7186):442-447. 14. Méndez-Ferrer S, Michurina TV, Ferraro F, et al. Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature. 2010; 466(7308):829-834. 15. Katayama Y, Battista M, Kao WM, et al. Signals from the sympathetic nervous system regulate hematopoietic stem cell egress from bone marrow. Cell. 2006;124(2):407-421. 16. Maryanovich M, Zahalka AH, Pierce H, et al. Adrenergic nerve degeneration in bone marrow drives aging of the hematopoietic stem cell niche. Nat Med. 2018;24(6):782-791. 17. Arranz L, Sánchez-Aguilera A, Martín-Pérez D, et al. Neuropathy of haematopoietic stem cell niche is essential for myeloproliferative neoplasms. Nature. 2014;512(7512):78-81. 18. Zahalka AH, Arnal-Estapé A, Maryanovich M, et al. Adrenergic nerves activate an angio-metabolic switch in prostate cancer. Science. 2017;358(6361):321-326. 19. Wolter JK, Wolter NE, Blanch A, et al. Anti-tumor activity of the betaadrenergic receptor antagonist propranolol in neuroblastoma. Oncotarget. 2014;5(1):161-172. 20. Magnon C, Hall SJ, Lin J, et al. Autonomic nerve development contributes to prostate cancer progression. Science. 2013; 341(6142):1236361. 21. Grytli HH, Fagerland MW, Fosså SD, et al. Association between use of β-bockers and prostate cancer-specific survival: A cohort study of 3561 prostate cancer patients with high-risk or metastatic disease. Eur Urol. 2014; 65(3):635-641. 22. Qiao G, Chen M, Bucsek MJ, et al. Adrenergic Signaling: A Targetable Checkpoint Limiting Development of the Antitumor Immune Response. Front Immunol. 2018;9:164. 23. Drexler B, Passweg JR, Tzankov A, et al. The sympathomimetic agonist mirabegron did not lower JAK2-V617F allele burden, but restored nestin-positive cells and reduced reticulin fibrosis in patients with myeloproliferative neoplasms: results of phase 2 study SAKK 33/14. Haematologica. 2019;104(4):710-716. 24. Zacharaki D, Ghazanfari R, Li H, et al. Effects of JAK1/2 inhibition on bone marrow stromal cells of myeloproliferative neoplasm (MPN) patients and healthy individuals. Eur J Haematol. 2018;101(1):57-67. 25. Harrison CN, McLornan DP. Current treatment algorithm for the management of patients with myelofibrosis, JAK inhibitors, and beyond. Hematology Am Soc Hematol Educ Program. 2017;2017(1):489-497.

641

Editorials

Expounding on the essence of epigenetic and genetic abnormalities in blastic plasmacytoid dendritic cell neoplasms Lhara Lezama, Robert S. Ohgami Stanford University, CA, USA E-mail: ROBERT S. OHGAMI - rohgami@ohgami.org doi:10.3324/haematol.2018.211557

lastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare hematologic malignancy now known to derive from immature plasmacytoid dendritic cells. However, the cell of origin for this entity was unknown a mere two decades ago when it was alternatively speculated to be of natural killer (NK)-cell or monocytic origin. Coupled with recent advancements in genetic technologies, we have been at an inflection point in time for understanding the essence of this disease.1 The recent paper by Sapienza et al., in this issue of the Journal, makes a groundbreaking and essential push forward in our understanding of BPDCN.2 The work is pivotal as: 1) it is the first to focus specifically on understanding epigenetic alterations in BPDCN; 2) it studies a large number of well annotated cases of BPDCN using broad and comprehensive genetic analyses; and 3) it assesses the therapeutic value of targeting BPDCN with epigenetic modifying therapies in vivo. However, to understand the importance of the study by Sapienza et al., it is critical to provide an overview of this neoplasm. Initially described in 1990 as a possible histiocytic or monoblastic leukemia,3,4 and then later believed to be an NK-cell tumor due to expression of both CD56 and

CD4,5,6 the cellular lineage of BPDCN was debated for more than a decade. Then, in 1999, Lucio et al.7 postulated a dendritic cell origin due to expression of CD123, a sensitive, though not specific, plasmacytoid dendritic cell marker. Further research continued to support the plasmacytoid dendritic cell origin based not only on CD123 expression but also expression of TCL1 and also cellular differentiation assays, all which pointed to plasmacytoid dendritic cells as the normal cellular counterpart for this tumor.8-10 In the 2008 World Health Organization classification, it was finally introduced as BPDCN and has continued with this designation in the revised 2016 WHO classification. BPDCN demonstrates aggressive behavior, presenting with cutaneous lesions and bone marrow involvement, and, despite a frequent good response to initial therapy, the median survival varies from 10 to 19.8 months. The genetics and molecular aspects of this entity have also been explored by several groups. An abnormal karyotype is common (>50% of patients) and particular chromosome regions are more frequently targeted in BPDCN based on classic cytogenetic analyses as well as comparative genomic hybridization studies: 4q, 5q, 6q, 9, 12p, 13q,

Figure 1. Genetic abnormalities and biological pathways important in blastic plasmacytoid dendritic cell neoplasm.

642

haematologica | 2019; 104(4)

Editorials

and 15q (Figure 1).11-13 Deletion of the 9p21.3 locus is the most recurrent event in BPDCN, found to be associated with a poor outcome when biallelic.11 More recently, the presence of a recurrent gene rearrangement involving the MYC locus, specifically t(6;8)(p21;q24) has been reported in several studies including one by our own group, and has been seen in association with older onset and shorter median survival.14-16 Whole-exome sequencing (WES) and targeted sequencing studies have identified recurrent mutations involving TET2, ASXL1, TP53, and NPM1,17,18 while separate studies have appointed E-box transcription factor TCF4 as a master regulator of the BPDCN oncogenic program as bromodomain and extra-terminal domain inhibitors (BETis) induced BPDCN apoptosis due to disruption of the TCF4 dependent regulatory network.19 Finally, aberrant activation of the NF-kB pathway has been identified through gene expression profiling.20 However, as can be seen, our understanding of the genetic aspects of BPDCN has been somewhat limited. In the current issue of the Journal, Sapienza et al.2 significantly advance our understanding of BPDCN by analyzing 14 patients, as well as the patient-derived CAL-1 cell line, by WES. This broad and detailed sequencing demonstrated that BPDCN patients were affected by mutations of genes involved in epigenetic regulation, with 25 mutated epigenetic modifier genes including those implicated in DNA methylation (TET2 and IDH2), chromatin accessibility (ARID1a, CHD8, SMARCA1), and histone modification including: methylation (ASXL1, SUZ12, MLL), demethylation (KDM4D), acetylation (EP300, EP400), ubiquitination (PHC1, PHC2), dephosphorylation (EYA2) and exchange (SRCAP) (Figure 1). This finding highlights the dysregulation of the epigenetic program in BPDCN as a hallmark of the disease indicating possible therapeutic interventions. In addition, by analyzing the transcriptome of the samples studied by WES, they examined the specific impact of these epigenetic-associated gene mutations. Gene set enrichment analysis additionally revealed two significant deregulation signatures associated with methylation of DNA: one driven by KDM5B34 histone demethylase and another by the PRMT5 methyltransferase-associated gene. The authors also detected gene set enrichment of those genes associated with response to decitabine, a DNA demethylating agent. Finally, based on the apparent significance of epigenetics in BPDCN, the authors did what few have done before, and tested in vivo the efficacy of four US Food and Drug Administration (FDA)-approved epigenetic drugs (5’-Azacytidine, decitabine, romidepsin and bortezomib) in a mouse xenograft model using the CAL-1 cell line. These drugs were used as single agents or in combination, and when used as a single agent, 5’-Azacytidine and decitabine significantly prolonged mice overall survival, while among all the combinations tested, treatment with 5’-Azacytidine in combination with decitabine achieved the best result in terms of survival. The significance of this study cannot be understated as it links genotype to advancements in epigenetic forms of treatment and fundamental biological processes. Blastic plasmacytoid dendritic cell neoplasm is a rare disease with an extremely aggressive behavior; the advances here in the genetics and molecular aspects of this entity, as well as the therapeutic approach, are quite revealing and haematologica | 2019; 104(4)

significantly open the door for future clinical trials. This work nicely provides further support for the essence of why development of new epigenetic treatment strategies are a rational approach for this aggressive disease.

References 1. Laribi K, Denizon N, Besancon A, et al. Blastic Plasmacytoid Dendritic Cell Neoplasm: From Origin of the Cell to Targeted Therapies. Biol Blood Marrow Transplant. 2016;22(8):1357-1367. 2. Sapienza MR, Abate F, Melle F, Orecchioni FF, Etebari M. Blastic plasmacytoid dendritic cell neoplasm: genomics mark epigenetic dysregulation as a primary therapeutic target. Haematologica. 2019;104(4): 729-737. 3. Tauchi T, Ohyashiki K, Ohyashiki JH, et al. CD4+ and CD56+ acute monoblastic leukemia. Am J Hematol. 1990;34(3):228-229. 4. Gattei V, Carbone A, Zagonel V, Pinto A. Expression of natural killer antigens in a subset of 'non-T, non-B lymphoma/leukaemia with histiocytic features'. Br J Haematol. 1990;76(3):444-448. 5. Kimura S, Kakazu N, Kuroda J, et al. Agranular CD4+CD56+ blastic natural killer leukemia/lymphoma. Ann Hematol. 2001;80(4):228231. 6. Kameoka J, Ichinohasama R, Tanaka M, et al. A cutaneous agranular CD2- CD4+ CD56+ "lymphoma": report of two cases and review of the literature. Am J Clin Pathol. 1998;110(4):478-488. 7. Lucio P, Parreira A, Orfao A. CD123hi dendritic cell lymphoma: an unusual case of non-Hodgkin lymphoma. Ann Intern Med. 1999;131(7):549-550. 8. Grouard G, Rissoan MC, Filgueira L, Durand I, Banchereau J, Liu YJ. The enigmatic plasmacytoid T cells develop into dendritic cells with interleukin (IL)-3 and CD40-ligand. J Exp Med. 1997;185(6):11011111. 9. Herling M, Teitell MA, Shen RR, Medeiros LJ, Jones D. TCL1 expression in plasmacytoid dendritic cells (DC2s) and the related CD4+ CD56+ blastic tumors of skin. Blood. 2003;101(12):5007-5009. 10. Petrella T, Comeau MR, Maynadie M, et al. 'Agranular CD4+ CD56+ hematodermic neoplasm' (blastic NK-cell lymphoma) originates from a population of CD56+ precursor cells related to plasmacytoid monocytes. Am J Surg Pathol. 2002;26(7):852-862. 11. Lucioni M, Novara F, Fiandrino G, et al. Twenty-one cases of blastic plasmacytoid dendritic cell neoplasm: focus on biallelic locus 9p21.3 deletion. Blood. 2011;118(17):4591-4594. 12. Dijkman R, van Doorn R, Szuhai K, Willemze R, Vermeer MH, Tensen CP. Gene-expression profiling and array-based CGH classify CD4+CD56+ hematodermic neoplasm and cutaneous myelomonocytic leukemia as distinct disease entities. Blood. 2007;109(4):17201727. 13. Leroux D, Mugneret F, Callanan M, et al. CD4(+), CD56(+) DC2 acute leukemia is characterized by recurrent clonal chromosomal changes affecting 6 major targets: a study of 21 cases by the Groupe Francais de Cytogenetique Hematologique. Blood. 2002;99(11):41544159. 14. Sakamoto K, Katayama R, Asaka R, et al. Recurrent 8q24 rearrangement in blastic plasmacytoid dendritic cell neoplasm: association with immunoblastoid cytomorphology, MYC expression, and drug response. Leukemia. 2018;32(12):2590-2603. 15. Boddu PC, Wang SA, Pemmaraju N, et al. 8q24/MYC rearrangement is a recurrent cytogenetic abnormality in blastic plasmacytoid dendritic cell neoplasms. Leuk Res. 2018;66:73-78. 16. Sumarriva Lezama L, Chisholm KM, Carneal E, et al. An analysis of blastic plasmacytoid dendritic cell neoplasm with translocations involving the MYC locus identifies t(6;8)(p21;q24) as a recurrent cytogenetic abnormality. Histopathology. 2018;73(5):767-776. 17. Menezes J, Acquadro F, Wiseman M, et al. Exome sequencing reveals novel and recurrent mutations with clinical impact in blastic plasmacytoid dendritic cell neoplasm. Leukemia. 2014;28(4):823-829. 18. Jardin F, Ruminy P, Parmentier F, et al. TET2 and TP53 mutations are frequently observed in blastic plasmacytoid dendritic cell neoplasm. Br J Haematol. 2011;153(3):413-416. 19. Ceribelli M, Hou ZE, Kelly PN, et al. A Druggable TCF4- and BRD4Dependent Transcriptional Network Sustains Malignancy in Blastic Plasmacytoid Dendritic Cell Neoplasm. Cancer Cell. 2016;30(5):764778. 20. Sapienza MR, Fuligni F, Agostinelli C, et al. Molecular profiling of blastic plasmacytoid dendritic cell neoplasm reveals a unique pattern and suggests selective sensitivity to NF-kB pathway inhibition. Leukemia. 2014;28(8):1606-1616.

643

Editorials

THROMBOTECT takes the lead Christoph Male,1 Sarah H. Oâ&#x20AC;&#x2122;Brien2 and Lesley Mitchell3 1

Department of Pediatrics, Medical University of Vienna, Austria; 2Division of Hematology & Oncology, Nationwide Childrenâ&#x20AC;&#x2122;s Hospital, Columbus, OH, USA and 3Division of Pediatric Hematology/Oncology, Department of Pediatrics, University of Alberta, AB, Canada E-mail: CHRISTOPH MALE - christoph.male@meduniwien.ac.at doi:10.3324/haematol.2018.209528

hromboembolic events in children predominantly occur as secondary complications of severe underlying diseases and their treatment, the most important risk factor being the use of central venous catheters (CVC). In spite of the relative high frequency of thromboembolic events in children with CVC, the evidence to date is equivocal as to whether there is benefit of primary thromboprophylaxis in reducing the risk of these events and whether it outweighs the risk of bleeding in sick children.1 One population at particular risk of thromboembolic events consists of children receiving induction chemotherapy for pediatric acute lymphoblastic leukemia (ALL). The mechanism implicated in the development of thromboembolic events is hypothesized to be associated with an acquired antithrombin deficiency resulting from treatment with asparaginase. The only randomized trial to date (PARKAA) assessed primary thromboprophylaxis using antithrombin replacement in children with ALL and CVC during induction chemotherapy. However, PARKAA was a feasibility study with limited power and only showed a trend to efficacy of antithrombin replacement.2 In the current issue of Haematologica, Greiner et al. report on the THROMBOTECT study which was an open-label, randomized controlled trial assessing the efficacy and safety of primary thromboprophylaxis during induction chemotherapy including asparaginase for ALL in children and adolescents, the majority of whom had a CVC.3 The study, an investigator-initiated study performed within the Berlin-Frankfurt-Munster cooperative group, recruited 949 patients who were randomized to three arms: activity-adapted antithrombin substitution, prophylactic-dose low molecular weight heparin (LMWH), or low-dose unfractionated heparin (UFH) as their standard of care. The low UFH dose was intended to prevent CVC occlusion but presumably did not achieve a systemic antithrombotic effect, so this arm might be considered a placebo arm. The primary efficacy outcome was symptomatic thromboembolic events, the principal safety outcome was bleeding, assessed during both induction and consolidation chemotherapy. Secondary safety outcomes were event-free survival and overall survival from the underlying ALL. The results of the study show a significant reduction in the incidence of thromboembolic events with use of antithrombin (1.9%) and LMWH (3.5%) compared to UFH (8.0%). Since a large proportion of children assigned to LMWH crossed over to other arms, an as-treated analysis was performed, showing approximately equal reductions in thromboembolic events risk for antithrombin and LMWH compared to UFH. The incidence of bleeding was low (0.9%) and not different between the three arms. 644

Regarding leukemia outcome, there was an increased relapse rate in children randomized to antithrombin when compared to those randomized to UFH in the intentionto-treat analysis, but no difference in the as-treated analysis. The authors conclude that thromboprophylaxis should be recommended during ALL induction therapy and, given the unclear effect of antithrombin substitution on leukemia outcome, they recommend LMWH as the primary choice at present. The THROMBOTECT study is an important breakthrough, as it is the first adequately powered randomized trial of primary thromboprophylaxis in pediatric patients. The study shows that thromboprophylaxis with antithrombin or LMWH is effective at preventing thromboembolic events without increasing the risk of bleeding. The THROMBOTECT collaborators can be commended for their outstanding effort in completion of this important study, which will improve the care of children with ALL. Moreover, the study serves as proof-of-concept for thromboprophylaxis in children in other clinical settings. The completion of the study will not only have a significant impact on clinical management, but will also demonstrate that pediatric clinical trials of anticoagulation can be completed. As with all clinical trials, particularly in children, there are limitations to the study. First, the study was not masked for practical and ethical reasons, as this would have required placebo subcutaneous injections which would have been unacceptable to children and caregivers. The lack of masking increased the potential for cross-over between treatment arms, diminishing the distinction between arms. Of the patients assigned to LMWH, 33% refused the intervention after randomization because of the subcutaneous injections, of whom approximately two-thirds received UFH or no thromboprophylaxis and one-third were given antithrombin substitution. The study design is problematic in that patients who crossed over were allowed to choose between treatment arms, creating an additional source of selection bias. However, the intention-to-treat and the as-treated analyses are reasonably concordant, so the reduction in risk of thromboembolic events with antithrombin and LMWH thromboprophylaxis is still valid. Second, the open-label study treatment, in combination with the primary outcome being clinically symptomatic thromboembolic events, implies a risk of diagnostic suspicion bias in outcome assessment. Although clinically suspected thromboembolic events were required to be confirmed by objective radiographic imaging, neither attending physicians nor radiologists were masked to treatment allocation. Moreover, there was no central independent adjudication of outcome events. haematologica | 2019; 104(4)

Editorials

Third, THROMBOTECT provides information regarding only symptomatic thromboembolic events. However, previous studies in children with ALL and CVC have shown relatively low frequencies of symptomatic thromboembolic events but substantial frequencies of asymptomatic thromboembolism detected by systematic radiographic screening.2 While some believe that symptomatic thromboembolic events are clinically most relevant, many of the patients with asymptomatic thromboembolism in PARKAA had significant degrees of venous occlusion. The THROMBOTECT study did not include radiographic screening (e.g. ultrasound) which would have achieved a more complete identification of both symptomatic and asymptomatic thromboembolism. Moreover, using objective radiographic screening would have reduced the potential for observer bias. To what extent thromboprophylaxis affects asymptomatic thromboembolism, remains open. Fourth, while the patients received thromboprophylaxis only during the induction phase they were followed for thromboembolic events and bleeding outcomes into the consolidation phase. One fifth of thromboembolic events and half of the bleeds occurred during induction consolidation. Therefore, as the study interventions had already been discontinued, their association with these outcome events cannot be definitively determined. The manuscript presents an exploratory subgroup analysis based on age. In children >6 years, frequencies of thromboembolic events were higher (6.4%) and differences between treatment arms more pronounced, while in younger children, thromboembolic events were observed less frequently (2.7%) and not significantly different between arms. This possible age effect should be interpreted with caution, because thromboembolic events may not be detected in younger children as symptoms may be reported to a lesser extent. Treatment effects were qualitatively not different for younger children, and a benefit from anticoagulant prophylaxis, even if smaller, may be extrapolated from older children. Given that there were few bleeding events, using thromboprophylaxis in children of all ages appears reasonable. One notable observation made in the THROMBOTECT study was the large proportion of children/families that refused LMWH due to subcutaneous injections. Of eligible patients at participating centers, 38% would not consent to enter the study. Among consenting participants, of those randomised to LMWH, 33% refused this treatment due to subcutaneous injections. While the difficulty in treating children with subcutaneous drugs is well known by pediatricians, the THROMBOTECT study provides solid evidence documenting the magnitude of the problem and concludes that there are problems with compliance with anticoagulant drugs administered subcutaneously. Although antithrombin was effective at decreasing the incidence of thromboembolic events with no additional risk of bleeding, the unexpected finding that patients receiving antithrombin had an increased rate of relapse is an issue. As this association was not constant over a num-

haematologica | 2019; 104(4)

ber of analyses, it may well be a chance finding. However, a biological effect of antithrombin substitution on leukemia outcome cannot be completely excluded and so the use of antithrombin for thromboprophylaxis cannot be recommended until more evidence is available. Moreover, antithrombin concentrate is expensive, and substitution requires monitoring of antithrombin levels and intravenous infusion which is a burden to the patient. Therefore, while THROMBOTECT has shown that both antithrombin and LMWH are effective at preventing thromboembolic events, there remain challenges with these choices for thromboprophylaxis. The authors of the paper conclude that the THROMBOTECT results provide the rationale to develop new studies to further determine best practice in preventing thromboembolic events in pediatric ALL. An ongoing clinical trial, the PREVAPIX-ALL (NCT02369653) study is a randomized controlled trial determining the efficacy and safety of primary prophylaxis with apixaban in prevention of thromboembolic events in pediatric patients with ALL/lymphoblastic lymphoma during induction chemotherapy. A total of 500 participants are randomized to apixaban (intervention) or no systemic anticoagulation (control). Subjects are followed for symptomatic thromboembolic events and all patients are screened for thromboembolism by ultrasound and echocardiography at the end of the induction phase. Apixaban is a direct oral anticoagulant and has been shown in adults to require no monitoring, making it an attractive option in children. The importance of availability of an oral anticoagulant is underscored by the results of THROMBOTECT with respect to the limited acceptance of subcutaneously injected LMWH. While the PREVAPIX-ALL study is open label, bias is minimized by the screening of all participants at the end of the study using standardized imaging tests and a blinded central adjudication committee. In conclusion, THROMBOTECT has established a positive benefit-risk balance for primary thromboprophylaxis in children with ALL. PREVAPIX-ALL will add to these findings by assessing the efficacy and safety of a direct oral anticoagulant in this population. These studies will determine the optimum clinical approach for the prevention of thromboembolic events in pediatric ALL, and provide the basis for further studies of thromboprophylaxis in children in other settings.

References 1. Vidal E, Sharathkumar A, Glover J, Faustino EV. Central venous catheter-related thrombosis and thromboprophylaxis in children: a systematic review and meta-analysis. J Thromb Haemost. 2014;12(7):1096-1109. 2. Mitchell L, Andrew M, Hanna K, et al. Trend to efficacy and safety using antithrombin concentrate in prevention of thrombosis in children receiving l-asparaginase for acute lymphoblastic leukemia. Results of the PAARKA study. Thromb Haemost. 2003;90(2):235-244. 3. Greiner J, Schrappe M, Claviez A, et al. THROMBOTECT - a randomized study comparing low molecular weight heparin, antithrombin and unfractionated heparin for thromboprophylaxis during inudction therapy of acute lymphoblastic leukemia in children and adolescents. Haematologica. 2019;104(4):756-765.

645

Editorials

Asymmetric dimethylarginine â&#x20AC;&#x201C; a prognostic marker for transplant outcome? Janghee Woo1,2 and H. Joachim Deeg1,2 1

Fred Hutchinson Cancer Research Center and 2University of Washington School of Medicine, Seattle, WA, USA

E-mail: H. JOACHIM DEEG - jdeeg@fredhutch.org doi:10.3324/haematol.2018.212191

esults of allogeneic hematopoietic cell transplantation (HCT) have improved progressively over the past two decades.1 However, disease relapse, graftversus-host disease (GvHD), and various other causes of non-relapse mortality continue to be hurdles to greater success. These challenges have been tackled from various angles, both clinically and in the laboratory. One focus of research has been on the role of endothelial cells in the pathophysiology and manifestations of GvHD, sinusoidal obstruction syndrome, diffuse alveolar hemorrhage, and transplant-associated microangiopathy. There is a substantial body of literature on the effects of pro-inflammatory cytokines, derived from endothelial cells or other cellular compartments, on the interactions between donor-derived cells and host tissues and organs.2,3 In this issue of Haematologica, Radujkovic and colleagues examine a potential role of asymmetric dimethylarginine (ADMA), an endogenous compound derived from endothelial cells, as a pre-transplant marker for post-transplant complications.4 For this purpose, they studied data from 938 patients transplanted at two German centers who had serum samples collected within 4 weeks before transplantation and that were available for determination of ADMA levels. The results of their analysis indicate that higher levels of ADMA before HCT were associated with an increased risk of non-relapse mortality within the first year after transplantation. There was no association with relapse or GvHD. However, overall survival and progression-free survival during the first year after transplantation were negatively affected by higher pre-HCT levels of ADMA. Higher AMDA levels were also associated with shortened overall survival, shortened progression-free survival, and a higher incidence of non-relapse mortality within 1 year after the onset of acute GvHD. As ADMA is an endogenous molecule, the authors conclude that their findings underscore the importance of endothelial cell function for posttransplant outcome. ADMA occurs naturally as a metabolic byproduct of protein modification processes in human cells, as first described by Vallance et al. in 1992.5 One important function is its interference with L-arginine â&#x20AC;&#x201C; via nitric oxide (NO) synthase6 â&#x20AC;&#x201C; in the production of NO, a molecule wellknown to be involved in endothelial function. The presence of elevated ADMA levels has, therefore, broad implications, including interference with vasodilation, facilitation of atherogenesis, insulin resistance, the development of autoimmune disorders, and rejection of kidney allografts, among others.7-10 It should not, therefore, be surprising that ADMA plays a significant role in disorders such as cardiovascular disease, diabetes mellitus, certain forms of renal disease, and in erectile dysfunction.6 ADMA levels increase substantially in response to native or oxidized low density lipoprotein cholesterol, thereby further inhibiting NO production. Studies in animal models have indicated that 646

ADMA increases at a time when vascular disease may not be clinically evident. These data suggest that ADMA, originating from endothelial cells, may also directly affect endothelial function, which is relevant to transplant outcomes as reported by the authors. Reduced NO levels will alter tissue perfusion, resulting in organ dysfunction and, conceivably, greater susceptibility of lungs, intestinal tract, liver or kidneys to the effects of HCT conditioning regimens and the cytokines released as a consequence of donor/host interactions.2 As ADMA has also been suggested to be a potential biomarker for insulin resistance,10 it might be involved in altered blood glucose regulation, a problem for many patients after HCT. Furthermore, NO is a potent inhibitor of platelet aggregation and adhesion, as well as leukocyte adhesion, and thereby reduces the risk of thrombotic events. Thus, reduction of NO production in the presence of elevated levels of ADMA might facilitate the formation of microthrombi and contribute to microvascular dysfunction.11 The reference of the authors to post-transplant microangiopathy is, therefore, relevant, and it is unfortunate that no data on specific causes of death, in particular no histological data, were available. Clearly, this paper raises many questions. As the authors note, they present an observational study, which does not allow any cause-and-effect relationship to be determined, and they remain undecided in their discussion. For example, it is puzzling that nitrate levels were elevated along with ADMA prior to transplantation, as were, incidentally, thrombomodulin levels, which were also associated with increased non-relapse mortality in a previous report.3 Since ADMA levels were analyzed in pre-HCT samples, a logical assumption would be that pre-HCT events, such as exposure to chemotherapy or infectious agents or, possibly, genetic factors, were responsible for raised levels of ADMA. There are data showing that a polymorphism in dimethylarginine dimethylaminohydrolase 2 (DDAH2), the enzyme that promotes ADMA metabolism, is associated with elevated ADMA levels,12 and it is tempting to speculate that this polymorphism might affect post-HCT nonrelapse mortality. Limited studies on polymorphism of the NO synthase gene by the authors failed to show a correlation with outcome, and such a polymorphism (or mutation) would not by itself explain elevated ADMA levels. It is of interest to see that the authors provide outcome data, not only relative to the day of HCT, but also relative to the onset of GvHD, considering the possibility that the development of GvHD might have affected the subsequent course in an ADMA-dependent way. However, there was no significant impact of ADMA levels on post-GvHD outcome, relapse, overall survival, or progression-free survival, and the results were not different from those given in relation to the transplant date. What might one have expected? The authors had reported previously that patients with haematologica | 2019; 104(4)

Editorials

(steroid-refractory) GvHD were those with elevated preHCT angiopoietin 2 levels.3 While the overall picture in that setting is more complex, are there potential interactions between elevated ADMA and angiopoietin 2 levels? Or are both elevated because of the same insult? It would have been of interest to see (in the present study) pre-transplant angiopoietin 2 levels and whether there was a correlation with ADMA concentrations. Additional studies will be necessary to further dissect interactions between ADMA and other molecules, to follow ADMA levels longitudinally (for example, do levels change with the development of GvHD?), and to determine whether the time of onset of GvHD might have an impact (there was a wide time span during which GvHD developed). Furthermore, while a global comparison of highintensity and reduced-intensity conditioning regimens is provided, and early and late disease stages were considered, it would have been interesting to have had more specific data on the conditioning regimens and, for example, treatment of diseases prior to transplantation. This is not to diminish the authorsâ&#x20AC;&#x2122; accomplishments; however, some of these studies will be necessary in order to provide a mechanistic explanation for the data. In conclusion, the data must be interpreted with caution. This is particularly important when considering possible interventions. As suggested by the authors, high doses of citrulline may be useful to raise L arginine levels, counterbalancing the effect of ADMA. Could phosphodiesterase-5inhibitors (such as sildenafil) be useful to force the induction of NO synthase? Additional work should generate interesting data.

haematologica | 2019; 104(4)

References 1. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Eng J Med. 2010;363(22):2091-2101. 2. Antin JH, Ferrara JLM. Cytokine dysregulation and acute graft-versushost disease. Blood. 1992;80:2964-2968. 3. Luft T, Dietrich S, Falk C, et al. Steroid-refractory GVHD: T-cell attack within a vulnerable endothelial system. Blood. 2011;118(6):1685-1692. 4. Radujkovic A, Dai H, Kordelas L, et al. Asymmetric dimethylarginine serum levels are associated with early mortality after allogeneic stem cell transplantation. Haematologica. 2019;104(4):827-834. 5. Vallance P, Leone A, Calver A, Collier J, Moncada S. Accumulation of an endogenous inhibitor of nitric oxide synthesis in chronic renal failure. Lancet. 1992;339(8793):572-575. 6. Forstermann U, Sessa WC. Nitric oxide synthases: regulation and function. Eur Heart J. 2012;33(7):829-837, 837a-837d. 7. Matsuguma K, Ueda S, Yamagishi S, et al. Molecular mechanism for elevation of asymmetric dimethylarginine and its role for hypertension in chronic kidney disease. J Am Soc Nephrol. 2006;17(8):2176-2183. 8. Franceschelli S, Ferrone A, Pesce M, Riccioni G, Speranza L. Biological functional relevance of asymmetric dimethylarginine (ADMA) in cardiovascular disease. Int J Mol Sci. 2013;14(12):24412-24421. 9. Sibal L, Agarwal SC, Home PD, Boger RH. The role of asymmetric dimethylarginine (ADMA) in endothelial dysfunction and cardiovascular disease. Curr Cardiol Rev. 2010;6(2):82-90. 10. Lee W, Lee HJ, Jang HB, et al. Asymmetric dimethylarginine (ADMA) is identified as a potential biomarker of insulin resistance in skeletal muscle. Sci Rep. 2018;8(1):2133. 11. Jodele S, Dandoy CE, Myers KC, et al. New approaches in the diagnosis, pathophysiology, and treatment of pediatric hematopoietic stem cell transplantation-associated thrombotic microangiopathy. Transfus Apher Sci. 2016;54(2):181-190. 12. Xuan C, Xu LQ, Tian QW, et al. Dimethylarginine dimethylaminohydrolase 2 (DDAH 2) gene polymorphism, asymmetric dimethylarginine (ADMA) concentrations, and risk of coronary artery disease: a case-control study. Sci Rep. 2016;6:33934.

647

PERSPECTIVE ARTICLE Ferrata Storti Foundation

Liquid biopsy in lymphoma

Davide Rossi,1,2 Valeria Spina,1 Alessio Bruscaggin1 and Gianluca Gaidano3 Experimental Hematology, Institute of Oncology Research, Bellinzona, Switzerland; Hematology, Oncology Institute of Southern Switzerland, Bellinzona, Switzerland and 3 Division of Hematology, Department of Translational Medicine, University of Eastern Piedmont, Novara, Italy 1 2

Haematologica 2019 Volume 104(4):648-652

Correspondence: DAVIDE ROSSI davide.rossi@eoc.ch Received: January 22, 2019. Accepted: February 20, 2019 Pre-published: March 7, 2019. doi:10.3324/haematol.2018.206177 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/648 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

648

Background The term “liquid biopsy” means accessing tumor DNA through a blood sampling, without the need of an invasive tissue biopsy. Cell-free fragments of DNA (cfDNA) are shed into the bloodstream by cells undergoing apoptosis and circulate at a low concentration in plasma as double-stranded DNA fragments that are predominantly short (<200 base pairs).1 In healthy subjects, cfDNA primarily derives from the apoptosis of cells of hematopoietic lineage, with minimal contributions from other tissues, and circulates in concentrations of 1-10 ng/mL of plasma.2-8 In lymphoma patients, a proportion of cfDNA derives from apoptotic tumor cells.5 The total amount of cfDNA in lymphoma patients is always increased compared with age- and gender-matched healthy subjects, with a mean concentration of 30 ng/mL of plasma.9-12 Levels of circulating tumor DNA (ctDNA) vary across different lymphoma subtypes, being higher in aggressive lymphomas than in indolent lymphomas. Beside lymphoma type, tumor volume also affects cfDNA levels, which are higher in advanced stage disease than in limited stage disease, and in overt progressive disease than in a disease that is clinically responding to treatment.9,11 This perspective aims at describing the unmet needs in the field of diagnosis, genotyping, and assessment of treatment response in lymphomas that can be addressed by ctDNA technologies, as well as current evidence, and/or further investigations or actions that would be needed before transferring ctDNA technologies into the clinic.

Technologies for ctDNA identification and measurement By using the tumor mutation profile or the immunoglobulin gene rearrangement as lymphoma fingerprints, normal cfDNA can be discriminated from cfDNA derived from tumor cells, also called circulating tumor DNA (ctDNA).9,12-15 ctDNA fraction in the pool of cfDNA originating from hematopoietic cells is frequently very small. Therefore, the test used for ctDNA detection and quantification must suppress both the technical noise (i.e. reduce the background errors) and the biological noise (i.e. suppress true mutations originating from an underlying clonal hematopoiesis by sequencing paired granulocytes genomic DNA) in order to reach the required analytical sensitivity and specificity.15,16 Finally, sensitivity strongly relies on input material quantity and quality. For example, a single gene test can only achieve a sensitivity of 1 in 10,000 (i.e. 10-4) if the input material matches or exceeds this threshold. When the mutation profile is used as tumor fingerprint, the type of genetic aberrations being detected guide the choice of the molecular technique to be used for ctDNA identification and quantification. A single, trunk, fully clonal, stereotypic genetic variant, that occurs in most patients, characterizes a few lymphoma types [eg. the MYD88 L265P mutation in lymphoplasmacytic lymphoma and primary central nervous system lymphoma (PCNSL)].17 Such mutations can be detected and quantified by PCR-based methods like mutation-specific droplet digital PCR.17 Molecular aberrations of most lymphomas, however, are heterogeneous. Ultra-deep next-generation sequencing (NGS) methods can overcome the limitations of assays covering single somatic variants by detecting a large spectrum of genetic alterations, including single nucleotide variants, insertions/deletions, chromosomal rearrangements, and copy number changes.18-20 The Cancer Personalized Profiling by Deep Sequencing (CAPP-seq) is a targeted capture ultra-deep NGS method for ctDNA detection and quantification in molecular heterogeneous tumors (Figure 1).18,19 CAPP-seq utilizes a disease-specific “selector”, which is a set of exonic and intronic targets chosen to cover regions of haematologica | 2019; 104(4)

ctDNA analysis in Lymphoma

known recurrent mutations for a particular cancer type. Those targets are then amplified and sequenced in a patient’s cfDNA sample, allowing quantification of ctDNA based on the detection of tumor-specific mutations, and simultaneous determination of an individual’s specific tumor mutation profile. This method can simultaneously assay all classes of mutations, including single nucleotide variants, insertion/deletions, copy number alterations and rearrangements.18-20 The “selector” is tumor specific and requires detailed knowledge of the underlying genetic landscape of the tumor, a limitation that is currently overcome by the availability of from dozens to hundreds of genomes across all types of lymphoma. The clonoSEQ Assay is a diagnostic test validated and approved for measuring minimal residual disease (MRD) on genomic DNA from bone marrow samples in leukemias and myeloma.21,22 In the assay, genomic DNA is amplified by a set of locus-specific multiplex PCR using V, D and J gene primers covering all possible rearranged IgH (VDJ), IgH (DJ), IgK, and IgL receptor gene sequences. The amplicon library is then subjected to ultra-deep NGS. The tumor-specific clonotype is first identified in a tumor-enriched biological sample and

then tracked within the repertoire of IgH, IgK and IgL rearrangements amplified and sequenced in post-treatment samples. By leveraging on the advantage that the IgH, IgK and IgL rearrangements represent a stable and tumor specific fingerprint, the clonoSEQ Assay has also been applied to ctDNA quantification in lymphomas.9,10,12 However, tracking IgH, IgK and IgL sequences has some shortcomings when applied to cfDNA, including the need for lymphoma clonotype assignment through the analysis of the tissue biopsy, limited sensitivity in low tumor burden settings, and reduced applicability because of somatic hypermutation (SHM), which is ongoing in some lymphoma types such as diffuse large B-cell lymphoma of the germinal center type and follicular lymphoma, leading to difficulties in identifying clonotypic sequences. Overall, although methods for ctDNA identification and quantification are becoming more common, they are not yet widely used in clinical laboratories and are not, therefore, prominently featured in disease management guidelines. Methodological challenges, both in molecular biology and bioinformatics analyses, must be overcome, standardized and harmonized as these methods become more routinely used.

Figure 1. Schematic representation of liquid biopsy unmet needs in diagnostic, genotyping and minimal residual disease (MRD) monitoring fields with relative actions to overcome these limitations. PCNSL: primary central nervous system lymphoma; PET / CT: positron emission tomography/computed tomography; DLBCL: diffuse large B-cell leukemia; cHL: classic Hodgkin lymphoma; CAPP-seq: Cancer Personalized Profiling by Deep Sequencing.

haematologica | 2019; 104(4)

649

D. Rossi et al.

Lymphoma diagnosis by ctDNA ctDNA cannot substitute tissue biopsy for lymphoma diagnosis. Only one single, rare, special scenario, namely the non-invasive diagnosis of PCNSL in those patients whose brain masses are surgically inaccessible, might one day be able to take advantage of the diagnostic potential of ctDNA. The diagnostic procedure of choice to establish the diagnosis of PCNSL is a stereotactic biopsy; if ocular or cerebrospinal fluid (CSF) involvement is evident, vitrectomy or CSF cytology may be sufficient. If a biopsy of the brain lesion is not possible, and CSF or ocular involvement is ruled out, histological diagnosis can be difficult at both initial stages and at relapse. The MYD88 L265P mutation occurs in up to 85% of tissue biopsies from PCNSL patients but never in those from non-hematologic brain tumors, suggesting that this mutation is a fairly sensitive and highly specific biomarker for differential PCSNL among central nervous system cancers.23-30 Droplet digital PCR assays probing the MYD88 L265P mutation in cfDNA samples from PCNSL patients known to harbor the MYD88 L265P have a 60% true positive rate.17 However, droplet digital PCR assays for detecting the MYD88 L265P mutation in cfDNA are far from being a validated non-invasive diagnostic test of PCNSL. Indeed, apart from standardization of the technique to suppress the false positive rate originating from the methodology, there are no data on the biological false positive rate of this assay. The MYD88 L265P mutation occurs in pre-malignant conditions such as monoclonal gammopathies of undetermined significance (MGUS) and monoclonal B-cell lymphocytosis (MBL). Both are relatively common in the older adult, and thus can co-occur by chance with a brain mass in the same subject, raising the issue of false positive results originating from a biological background (Figure 1).31,32 Plasma samples from large cohorts of patients diagnosed with a brain mass should be tested with standardized droplet digital PCR assays for the MYD88 L265P mutation to precisely define its diagnostic accuracy before bringing this test into diagnostic routine practice for PCNSL.

Tumor genotyping by ctDNA Tumor genotyping of lymphomas lacking a leukemic phase has so far relied on the analysis of the diagnostic tissue biopsy. However, multiregional sequencing showed that the diagnostic tissue biopsy might be subject to a selection bias resulting from spatial heterogeneity and, therefore, might not be representative of all the tumor genetics.33 Indeed, in follicular lymphoma, different areas of the same tumor may show different genetic profiles (i.e. intratumoral heterogeneity).34 A biopsy from one part of a tumor may miss mutations occurring in subclones residing in anatomically distant sites, including clinically relevant genetic biomarkers for treatment tailoring or anticipation of resistance. 33 Furthermore, serial sampling of tumor material through repeat biopsies is not usually feasible in lymphomas lacking a leukemic phase, hampering efforts to understand patterns of genomic evolution during disease progression and the development of treatment emergent resistant mutations. On the basis of this, lymphoma genotyping on ctDNA can complement, though not entirely substitute, the analysis of the diagnostic tissue biopsy in order to deal with the clinical need of a com650

prehensive and easily accessible tumor genotyping. ctDNA is representative of the entire lymphoma heterogeneity, thus bypassing the bias imposed by tissue biopsies in the reconstruction of the entire cancer clonal architecture, and identifying resistant clones that are dormant in non-accessible tumor sites. Accessing the blood stream has also a clear advantage for sampling in the serial monitoring of treatment emergent resistant mutations in real time.35 Independent studies have assessed the sensitivity and specificity of targeted gene mutation analysis in ctDNA versus tumor biopsy as gold standard from untreated DLBCL patients by using CAPP-Seq (Figure 1).15,36,37 The recovery rate of confirmed mutations (i.e. true positive rate) in the tumor biopsy ranges from 95% to 99%. The mutations confirmed by biopsy that were missed in ctDNA (i.e. false negative rate) range from 1% to 5% and are mostly of low allelic abundance in the tumor. After suppressing the biological background originating from clonal hematopoiesis by the sequencing of matched granulocyte DNA, such a false positive rate is represented by somatic variants recovered in cfDNA but absent in the tumor biopsy due to tumor mutations restricted to clones that are anatomically distant from the biopsy site.15,36,37 CAPP-seq of ctDNA thus stands as a robust and validated technology for accurate DLBCL genotyping. Genotyping of ctDNA by CAPP-seq allows recovery of 100% of tumor biopsy-confirmed actionable mutations of DLBCL, like EZH2, MYD88, CD79B, and longitudinal monitoring in the blood of the emergence of ibrutinib-resistant mutations.15,36-38 These data support the implementation in the clinic of this noninvasive technique in both settings. CAPP-seq standardization is, however, required before bringing this test into diagnostic routine practice for DLBCL (Figure 1). ctDNA is an alternative source of tumor DNA when representation of lymphoma cells is insufficient in the tissue biopsy, as in classic Hodgkin lymphoma (cHL).16,39 The rarity of neoplastic Hodgkin and Reed-Sternberg cells in the biopsies is a limit to the genetic characterization of cHL, which can only be overcome by complex techniques for tumor cell enrichment that are beyond the budget of a diagnostic lab. By CAPP-seq, biopsyconfirmed tumor mutations are detectable in ctDNA samples with a true positive rate of 87% in cHL patients.16 Though clinical application is still a long way off, CAPP-seq of ctDNA opens up the opportunity of genotyping large cohorts of cHL patients for the identification of genetic prognostic biomarkers and, within clinical trials, for the identification of biomarkers predictive of response to treatment.

Residual disease quantification by ctDNA Due to the lack of a leukemic dissemination, MRD monitoring has so far been limited to tissue-born lymphomas without bone marrow (BM) involvement, such as DLBCL and cHL. MRD monitoring in lymphomas is defined as any approach aimed at detecting, and possibly quantifying, residual tumor cells beyond the sensitivity level of routine imaging techniques. Whenever a patient achieves complete clinical remission, a number of different scenarios may actually be taking place, including full eradication of the neoplastic clone or persistence of residual tumor cells capable of giving rise to a full clinical relapse within months or years. According haematologica | 2019; 104(4)

ctDNA analysis in Lymphoma

to the Lugano criteria, positron emission tomography (PET)/computed tomography (CT) has become the recommended imaging strategy for sensitive disease response assessment in DLBCL and cHL.40 The best classification of patients with good versus poor prognosis is reached by the end-of-treatment PET/CT. However, this timepoint would be rather late to adapt treatment strategies according to the quality and depth of response. Interim PET/CT performed after two cycles of treatment has been tested for the early identification of chemorefractory patients, as they are candidates for treatment intensification to maximize the chances of cure, as well as to identify good-risk patients early, as they are candidates for treatment de-escalation to avoid both short- and long-term complications of chemoradiotherapy.41 The accuracy of interim PET/CT has been considered adequate to inform early treatment intensification or de-escalation in both limited and advanced stage cHL.42 However, even in the ideal technical and analytical setting, interim PET/CT results are inconsistent with the final outcome in approximately 20-30% of patients, who are thus still exposed to overor under-treatment.41,42 In DLBCL, interim PET/CT does not correctly inform on the subsequent outcome in a larger number of patients than in cHL. Indeed, the positive predictive value of interim PET/CT in DLBCL is 50%.43 This means that half DLBCL patients are misclassified by interim PET/CT as being R-CHOP resistant, but ultimately are converted to a negative PET/CT at the end of treatment and cured by R-CHOP. The negative predictive value of interim PET/CT is 70%.43 This means that 30% of DLBCL patients are misclassified by interim PET/CT as R-CHOP sensitive, but ultimately relapse after R-CHOP. On the basis of this, interim PET/CT can not yet be adopted for clinical use to guide treatment decisions in individual DLBCL patients and remains a subject for research. Minimal residual disease can be measured in tissueborn lymphomas without BM involvement and lacking a leukemic component by using ctDNA technologies. Compared to genomic DNA extracted from circulating mononuclear cells, plasma cfDNA harbors a 150-fold higher representation of tumor DNA, which makes cfDNA more reliable than genomic DNA from circulating cells for MRD monitoring in DLBCL.12 By using the immunoglobulin gene rearrangements to quantify ctDNA in plasma, DLBCL patients with undetectable ctDNA after two chemotherapy courses show a superior progression-free survival compared with patients with positive ctDNA.37 Despite its value as a prognostic tool, using the immunoglobulin gene rearrangement to quantify ctDNA in DLBCL has several shortcomings. This includes limited sensitivity in low tumor burden settings and reduced applicability because of the somatic hypermutation process, leading to difficulties in identifying clonotypic sequences.12 In addition, as for PET/CT, also

haematologica | 2019; 104(4)

for high throughput sequencing of the immunoglobulin genes the best informative timepoint is end of treatment.37 By covering a large spectrum of genetic lesions, ctDNA quantification by CAPP-seq is cross-validated by multiple tumor tags, and avoids false negative results caused by treatment-induced clonal shift. In both DLBCL and cHL, the change in ctDNA measured by CAPP-seq after two cycles of therapy associates with both event-free and overall survival.15,16,44 A drop of 100fold (or 2-log drop) in ctDNA levels after two chemotherapy courses is associated with an eventual complete response and cure. Conversely, a drop of less than 2-log in ctDNA after two treatment courses is associated with an eventual progression.16,44 Quantification of ctDNA coupled with PET/CT improves the accuracy of residual disease assessment at the interim time compared to the sole PET/CT in both DLBCL and cHL. Indeed, patients inconsistently judged as interim PET/CT positive, but having a negative (i.e. >2-log drop in ctDNA) liquid biopsy, are actually cured, while patients inconsistently judged as interim PET/CT negative, but having a positive (i.e. <2-log drop in ctDNA) liquid biopsy, are actually not cured.16,44 These results generate the hypothesis that ctDNA may complement interim PET/CT in informing on DLBCL and cHL patientsâ&#x20AC;&#x2122; outcome (Figure 1). Before translating this technology into the management of DLBCL, the precise cumulative sensitivity and specificity of PET/CT and ctDNA monitoring in anticipating the clinical course of patients should be precisely defined in clinical trials.

Further investigations At the clinical level, it is critical that well-designed trials validate current concepts and further explore applications of ctDNA for interim monitoring, surveillance monitoring, and response assessment in lymphomas. The most immediate implementation of ctDNA technology in lymphoma clinical trials includes: i) non-invasive diagnostics of PCNSL; ii) baseline screening for the identification of patients harboring actionable mutations; iii) early and accurate identification of non-responding patients; iii) monitoring the development of resistance mutations against targeted agents (Figure 1). At the technological level, standardization and harmonization projects, like those performed before the implementation of clinical MRD assessment in leukemias, should be designed and implemented also in lymphoma in order to meet clinical standards, and allow accurate, robust and reproducible results of ctDNA genotyping and quantification. Acknowledgments Work by the authors has been supported by Grant n. KFS3746-08-2015, Swiss Cancer League, Bern, Switzerland; AIRC 5 x 1000, "Metastatic disease: the key unmet need in oncology", project code 21198, Milan, Italy

651

D. Rossi et al.

References 1. Fleischhacker M, Schmidt B. Circulating nucleic acids (CNAs) and cancer a survey. Biochim Biophys Acta. 2007;775(1):181-232. 2. Thierry AR, El Messaoudi S, Gahan PB, Anker P, Stroun M. Origins, structures, and functions of circulating DNA in oncology. Cancer Metastasis Rev. 2016;35(3):347-376. 3. Thakur BK, Zhang H, Becker A, et al. Double-stranded DNA in exosomes: a novel biomarker in cancer detection. Cell Res. 2014;24(6):766-769. 4. Kahlert C, Melo SA, Protopopov A, et al. Identification of double-stranded genomic DNA spanning all chromosomes with mutated KRAS and p53 DNA in the serum exosomes of patients with pancreatic cancer. J Biol Chem. 2014;289(7):3869-3875. 5. Jahr S, Hentze H, Englisch S, et al. DNA fragments in the blood plasma of cancer patients: quantitations and evidence for their origin from apoptotic and necrotic cells. Cancer Res. 2001;61(4):1659-1665. 6. Stroun M, Lyautey J, Lederrey C, OlsonSand A, Anker P. About the possible origin and mechanism of circulating DNA: apoptosis and active DNA release. Clin Chim Acta. 2001;313(1-2):139-142. 7. Mouliere F, El Messaoudi S, Pang D, Dritschilo A, Thierry AR. Multi-marker analysis of circulating cell-free DNA toward personalized medicine for colorectal cancer. Mol Oncol. 2014;8(5):927-941. 8. Mouliere F, Robert B, Arnau Peyrotte E, et al. High fragmentation characterizes tumourderived circulating DNA. PLoS One. 2011;6(9):e23418. 9. Roschewski M, Dunleavy K, Pittaluga S, et al. Circulating tumour DNA and CT monitoring in patients with untreated diffuse large B-cell lymphoma: a correlative biomarker study. Lancet Oncol. 2015;16(5): 541-549. 10. Armand P, Oki Y, Neuberg DS, et al. Detection of circulating tumour DNA in patients with aggressive B-cell non-Hodgkin lymphoma. Br J Haematol. 2013;163(1):123126. 11. Hohaus S, Giachelia M, Massini G, et al. Cell-free circulating DNA in Hodgkin's and non-Hodgkin's lymphomas. Ann Oncol. 2009;20(8):1408-1413. 12. Kurtz DM, Green MR, Bratman SV, et al. Noninvasive monitoring of diffuse large Bcell lymphoma by immunoglobulin highthroughput sequencing. Blood. 2015;125 (24):3679-3687. 13. Forshew T, Murtaza M, Parkinson C, et al. Noninvasive identification and monitoring of cancer mutations by targeted deep sequencing of plasma DNA. Sci Transl Med. 2012;4(136):136ra68. 14. Ladetto M, BrĂźggemann M, Monitillo L, et al. Next-generation sequencing and realtime quantitative PCR for minimal residual disease detection in B-cell disorders. Leukemia. 2014;28(6):1299-1307. 15. Rossi D, Diop F, Spaccarotella E, et al. Diffuse large B-cell lymphoma genotyping

652

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

on the liquid biopsy. Blood. 2017;129(14): 1947-1957. Spina V, Bruscaggin A, Cuccaro A, et al. Circulating tumor DNA reveals genetics, clonal evolution and residual disease in classical Hodgkin lymphoma. Blood. 2018;131 (22):2413-2425. Hattori K, Sakata-Yanagimoto M, Suehara Y, et al. Clinical significance of disease-specific MYD88 mutations in circulating DNA in primary central nervous system lymphoma. Cancer Sci. 2018;109(1):225-230. Kurtz, DM, Scherer, F, Newman, AM, et al. Prediction of therapeutic outcomes in DLBCL from circulating tumor DNA dynamics. J Clin Oncol. 2016;34(15 Suppl): 7511. Newman AM, Bratman SV, To J, et al. An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage. Nat Med. 2014;20(5):548-554. Scherer F, Kurtz DM, Newman AM, et al. Development and validation of biopsy-free genotyping for molecular subtyping of diffuse large B-cell lymphoma. Blood. 2016;128(22):1089-1089. Wood B, Wu D, Crossley B, et al. Deepsequencing approach for minimal residual disease detection in acute lymphoblastic leukemia. Blood. 2012;120(26):5173-5180. Martinez-Lopez J, Lahuerta JJ, Pepin, et al. Prognostic value of deep sequencing method for minimal residual disease detection in multiple myeloma. Blood. 2014;123(20): 3073-3079. Gonzalez Aguilar A, Idbaih A, Boisselier B, et al. Recurrent mutations of MYD88 and TBL1XR1 in primary central nervous system lymphomas. Clin Cancer Res. 2012;18(19): 5203-5211. Bruno A, Boisselier B, Labreche K, et al. Mutational analysis of primary central nervous system lymphoma. Oncotarget. 2014;5(13):5065-5075. Braggio E, Van Wier S, Ojha J, et al. Genome-wide analysis uncovers novel recurrent alterations in primary central nervous system lymphomas. Clin Cancer Res. 2015;21(17):3986-3994. Vater I, Montesinos Rongen M, Schlesner M, et al. The mutational pattern of primary lymphoma of the central nervous system determined by whole exome sequencing. Leukemia. 2015;29(3):677-685. Yamada S, Ishida Y, Matsuno A, Yamazaki K. Primary diffuse large B-cell lymphomas of central nervous system exhibit remarkably high prevalence of oncogenic MYD88 and CD79B mutations. Leuk Lymphoma. 2015;56(7):2141-2145. Fukumura K, Kawazu M, Kojima S, et al. Genomic characterization of primary central nervous system lymphoma. Acta Neuropathol. 2016;131(6):865-875. Hattori K, Sakata Yanagimoto M, Okoshi Y, et al. MYD88 (L265P) mutation is associated with an unfavourable outcome of primary central nervous system lymphoma. Br J Haematol. 2017;177(3):492-494. Nakamura T, Tateishi K, Niwa T, et al.

31. 32.

33.

34.

35. 36.

37.

38.

39.

40.

41.

42.

43.

44.

Recurrent mutations of CD79B and MYD88 are the hallmark of primary central nervous system lymphomas. Neuropathol Appl Neurobiol. 2016;42(3):279-290. Landgren O, Staudt L. MYD88 L265P somatic mutation in IgM MGUS. N Engl J Med. 2012;367(23):2255-2256. Kalpadakis C, Pangalis GA, Vassilakopoulos TP, et al. Detection of L265P MYD-88 mutation in a series of clonal B-cell lymphocytosis of marginal zone origin (CBL-MZ). Hematol Oncol. 2017;35(4):542-547. Gerlinger M, Rowan AJ, Horswell S, al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med. 2012;366(10):883-892. Araf S, Wang J, Korfi K, et al. Genomic profiling reveals spatial intra-tumor heterogeneity in follicular lymphoma. Leukemia. 2018;32(5):1258-1263. Diaz LA Jr, Bardelli A. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol. 2014;32(6):579-586. Scherer F, Kurtz DM, Newman AM, et al. Non invasive genotyping and assessment of treatment response in diffuse large B cell lymphoma. Blood. 2015;126(23):114. Scherer F, Kurtz DM, Newman AM, et al. Distinct biological subtypes and patterns of genome evolution in lymphoma revealed by circulating tumor DNA. Sci Transl Med. 2016;8(364):364ra155. Scherer F, Kurtz DM, Newman AM, et al. Noninvasive Detection of Ibrutinib Resistance in Non-Hodgkin Lymphoma Using Cell-Free DNA. Blood. 2016;128(22): 1752. Camus V, Stamatoullas A, Mareschal S, et al. Detection and prognostic value of recurrent exportin 1 mutations in tumor and cell-free circulating DNA of patients with classical Hodgkin lymphoma. Haematologica. 2016; 101(9):1094-1101. Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014;32(27):3059-3068. Moghbel MC, Mittra E, Gallamini A, et al. Response Assessment Criteria and Their Applications in Lymphoma: Part 2. J Nucl Med. 2018;58(1):13-22. Johnson, P.W. Response-adapted frontline therapy for Hodgkin lymphoma: are we there yet? Hematology Am Soc Hematol Educ Program. 2016;2016(1):316-322. Mamot C, Klingbiel D, Hitz F, et al. Final Results of a Prospective Evaluation of the Predictive Value of Interim Positron Emission Tomography in Patients With Diffuse Large B-Cell Lymphoma Treated With R-CHOP-14 (SAKK 38/07). J Clin Oncol. 2015;33(23):2523-2529. Kurtz DM, Scherer F, Jin MC, et al. Circulating Tumor DNA Measurements As Early Outcome Predictors in Diffuse Large B-Cell Lymphoma. J Clin Oncol. 2018;36 (28):2845-2853.

haematologica | 2019; 104(4)

REVIEW ARTICLE

Re-evaluation of hematocrit as a determinant of thrombotic risk in erythrocytosis

Ferrata Storti Foundation

Victor R. Gordeuk,1 Nigel S. Key2 and Josef T. Prchal3

Division of Hematology and Oncology, University of Illinois at Chicago, IL; 2Division of Hematology-Oncology and UNC Hemophilia and Thrombosis Center, UNC, Chapel Hill, NC and 3Division of Hematology and Hematologic Malignancies, University of Utah and Huntsman Cancer Center, Salt Lake City, UT, USA 1

ABSTRACT

Haematologica 2019 Volume 104(4):653-658

ere we critically evaluate the role of elevated hematocrit as the principal determinant of thrombotic risk in polycythemia and erythrocytosis, defined by an expansion of red cell mass. Since red cell volume determination is no longer readily available, in clinical practice, polycythemia and erythrocytosis are defined by elevated hemoglobin and hematocrit. Thrombosis is common in Chuvash erythrocytosis and polycythemia vera. Although the increased thrombotic risk is assumed to be due to the elevated hematocrit and an associated increase in blood viscosity, thrombosis does not accompany most types of erythrocytosis. We review studies indicating that the occurrence of thrombosis in Chuvash erythrocytosis is independent of hematocrit, that the thrombotic risk is paradoxically increased by phlebotomy in Chuvash erythrocytosis, and that, when compared to chemotherapy, phlebotomy is associated with increased thrombotic risk in polycythemia vera. Inherited and environmental causes that lead to polycythemia and erythrocytosis are accompanied by diverse cellular changes that could directly affect thrombotic risk, irrespective of the elevated hematocrit. The pressing issue in these disorders is to define factors other than elevated hematocrit that determine thrombotic risk. Defining these predisposing factors in polycythemia and erythrocytosis should then lead to rational therapies and facilitate development of targeted interventions.

Correspondence: VICTOR R. GORDEUK vgordeuk@uic.edu JOSEF T. PRCHAL josef.prchal@hsc.utah.edu

Introduction

Received: November 1, 2018. Accepted: January 28, 2019. Pre-published: March 14, 2019.

Polycythemia and erythrocytosis There are several different parameters for diagnosis of polycythemia and erythrocytosis based on a blood count: the number of red blood cells, the hematocrit, and the hemoglobin concentration. Elevations in these measures can occur on a primary or secondary basis (Table 1).1 Primary polycythemia results from functional abnormalities intrinsic to erythroid progenitors, causing them to be hypersensitive to or independent of erythropoietin. This category includes polycythemia vera (PV), which is associated with acquired somatic mutations in the Janus kinase 2 gene (JAK2), dominantly inherited primary familial and congenital polycythemia or erythrocytosis, caused by germline gain-of-function erythropoietin receptor (EPOR) mutations,2 and erythrocytosis due to SH2B3 mutations.3,4 Primary familial and congenital polycythemia or erythrocytosis predisposes patients to cardiovascular disorders, perhaps due to chronic augmented erythropoietin signaling in all tissues bearing EPOR.2 In contrast, in secondary erythrocytosis, functionally normal erythroid progenitors are exposed to increased levels of circulating erythropoiesis-stimulating factors. In most instances, the erythropoiesis-stimulating factor is erythropoietin, but cobalt, insulin growth factor 1, increased angiotensin signaling and manganese may also stimulate erythropoiesis.1,5,6 Acquired causes of secondary erythrocytosis include erythrocytosis of pulmonary disease, high altitude erythrocytosis, Eisenmenger syndrome, smoking, carboxyhemoglobinemia, erythropoietin-producing tumors, doping with erythropoietin, posthaematologica | 2019; 104(4)

doi:10.3324/haematol.2018.210732 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/653 ÂŠ2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

653

V.R. Gordeuk et al.

renal transplant erythrocytosis, exogenous testosterone use, and cobalt and manganese toxicities.1,5,6 Congenital secondary erythrocytosis can be caused by high oxygen affinity hemoglobin variants, inherited low 2,3-diphosphoglycerate leading to high hemoglobin oxygen affinity, congenital methemoglobinemia, and a recently described gain-offunction mutation of the gene encoding erythropoietin (EPO).7 Other congenital conditions include rare germline mutations in hypoxia sensing pathway genes, including loss of function mutations of VHL encoding von Hippel Lindau (VHL) protein and EGLN1 encoding prolyl hydroxylase 2 (PHD2), and gain-of-function mutations of EPAS1 encoding hypoxia inducible factor (HIF)-2α.1 Chuvash erythrocytosis (CE) is an autosomal recessive condition, endemic to Chuvashia in Russia and Ischia in Italy, which results from homozygosity for a C→T missense mutation of VHL (VHL c.598C>T or VHLR200W).8-10 The mutated protein impairs interactions of VHL with the HIFα subunits, thereby reducing the rate of ubiquitin-mediated HIF-α degradation by the proteasome. As a result, the levels of HIF-1 and HIF-2 heterodimers increase, leading to increased expression of their target genes, including EPO, vascular endothelial growth factor (VEGF), glucose transporter 1 (GLUT1), tissue factor (F3) and a plethora of other genes.9,11,12 In endothelial cells, more than 3% of genes are upregulated by HIF-1.13 CE erythroid progenitors are hypersensitive to erythropoietin, a feature of primary polycythemia, but affected subjects also have increased erythropoietin levels mediated by increased HIF-2, a feature of secondary erythrocytosis.9,14 Similar combined features of both primary and secondary elevations in hematocrit are seen in certain other germline mutations of VHL (loss-of-function mutations) and EPAS1 (gain-of-function mutations).1

Viscosity, hematocrit and blood volume Both PV and erythrocytosis secondary to hypoxia or upregulated hypoxia sensing are characterized by an increased red cell mass and total blood volume, but the two conditions may at times be divergent with regard to plasma volume. The plasma volume is increased in PV, potentially causing the hematocrit to underestimate the degree of erythrocytosis, whereas the plasma volume may not be increased in all types of erythrocytosis secondary to hypoxia or to upregulated hypoxia sensing.15,16 Some clinical manifestations of erythrocytosis, such as headaches and tinnitus, appear to be related to increased viscosity of blood resulting from the expanded red cell mass and elevated hematocrit. An increase in blood viscosity at higher hematocrits with blood volume in the normal range impairs blood flow and reduces the transport of oxygen.17 In vitro, the viscosity of blood increases exponentially with an increase in hematocrit. However, mitigating factors in patients with erythrocytosis serve to improve oxygen transport, a process that is dependent on both cardiac output and hemoglobin concentration.18 Most importantly, the increase in blood volume accompanying erythrocytosis enlarges the vascular bed, decreases peripheral resistance and increases cardiac output. In addition, the blood flow is axial, with a central core of circulating red cells sliding over a peripheral layer of lubricating plasma. Therefore, optimum oxygen transport with increased blood volume occurs at a higher hematocrit value than with normal blood volume,18,19 and a moderate increase in hematocrit may be beneficial despite the increased viscosity. This may not hold true when there is a more pronounced increase in hematocrit, a circum654

stance in which high viscosity causes reduced blood flow19,20 that may be responsible for cerebral and cardiovascular impairment in some high-altitude dwellers21 or in patients with severely elevated hematocrit.22,23 In those instances, hematocrit has been reported to reach extreme values, sometimes exceeding 90%.24 In normovolemic individuals, cerebral blood flow decreases at a certain point of hematocrit elevation.25 However, blood flow is also influenced by the oxygen demand of tissues through incompletely understood mechanisms26 and cerebral blood flow remains high at high hematocrits when oxygen delivery is impaired. This was elegantly illustrated in six patients with high hemoglobin oxygen-affinity variants whose cerebral blood flow was 81% higher than that of 11 subjects of comparable age, matched for hematocrit and viscosity, but without the hemoglobin variant.27 Furthermore, cerebral blood flow decreases at much higher levels of hematocrit with any accompanying increased percentage of fetal hemoglobin,28 which also has high oxygen-affinity.29

Elevated hematocrit and thrombosis Thrombotic events are well documented in patients with PV and CE, apparently less so in those with primary familial and congenital polycythemia or erythrocytosis and HIF-2α gain-of-function mutations, but not in patients with secondary erythrocytosis such as Eisenmenger syndrome,30,31 other cyanotic heart disorders,32,33 high altitude dwellers,

Table 1. Classification of polycythemia and erythrocytosis. Primary - functional abnormalities expressed in erythroid progenitors Acquired Polycythemia vera (JAK2 mutations) Familial Primary familial & congenital polycythemia or erythrocytosis (EPOR mutations) Erythrocytosis due to SH2B3 mutations Secondary to increased erythropoietin Acquired Carboxyhemoglobinemia Erythropoietin doping Erythropoietin-secreting tumor High altitude Lung or heart disease Smoking Familial Left-shifted oxygen dissociation curve 2,3-diphosphoglycerate deficiency High O2 affinity hemoglobins Methemoglobinemia Mutations in hypoxia-sensing pathway genes EGLN1 (PHD2) mutations EPAS1 (HIF-2α) mutations VHL mutations (includes Chuvash erythrocytosis), typically homozygous or compound heterozygous Gain-of-function mutation of the EPO gene Secondary to increased exposures other than erythropoietin Acquired Cobalt Insulin growth factor 1 Manganese Post-renal transplant (increased angiotensin signaling) Testosterone haematologica | 2019; 104(4)

Hematocrit and thrombotic risk in erythrocytosis

and subjects with high oxygen-affinity hemoglobins. Several lines of evidence suggest that an isolated elevation in hematocrit does not, per se, lead to thrombosis. For example, cerebral infarction in young children with cyanotic heart disease is attributed to iron deficiency and relative anemia rather than to erythrocytosis.34,35 In the Framingham study hematocrit was associated with risk of stroke but this association disappeared in multivariate analysis when smoking, a well-established risk factor for stroke,36 was removed.37 In a UK study of 7,346 men, an increased risk of stroke was not seen at higher hematocrit levels (≥51%) in normotensive men but was apparent in hypertensive individuals.38 Coronary blood flow is decreased in secondary erythrocytosis,22 but there is equivocal evidence as to whether the risk of coronary thrombosis is increased in patients with a high hematocrit.23,39,40 Secondary erythrocytosis reportedly does not pose a thrombotic risk in surgical patients.41 Studies of the influence of elevated hematocrit on the risk of thrombosis in animal models of PV and erythrocytosis secondary to elevated erythropoietin have failed to find a consistent positive relationship.42-45 A study of a murine model in which erythrocytosis was induced by transfusing packed red blood cells, with evaluation of thrombotic risk 24 hours later, found that an elevated hematocrit promoted arterial thrombus formation.46 However, acute erythrocytosis induced by transfusion may not reflect the physiology of the chronic elevation of hematocrit seen in PV and secondary erythrocytosis.47 Furthermore, it is not certain how well the ferric chloride-induced thrombosis model in mice reflects thrombosis formation in humans. Thus, in this review, we focus on thrombosis in human conditions of chronic elevation in hematocrit.

Chuvash erythrocytosis and polycythemia vera share thrombosis as the principal cause of morbidity and mortality Chuvash erythrocytosis The propensity to thrombosis is even higher in CE than in PV.48 Although endemic in Chuvashia and Ischia, CE is distributed worldwide.8,9,49 This form of erythrocytosis is characterized by a high risk of both arterial and venous thrombosis in subjects living near sea level. It protects from anemia in heterozygotes50 but causes augmented hypoxia sensing with elevated hematocrit in homozygotes.12,51 The VHLR200W variant is not associated with tumors characteristic of the VHL tumor predisposition syndrome. Thrombosis largely accounts for the morbidity and mortality of CE although affected individuals have lower body mass index, systolic blood pressure, glucose and HbA1c levels, and white blood cell and platelet counts compared to controls.48,52,53 The high rate of thrombosis in CE begins in childhood51 and increases with age.48 However, higher hematocrit is not an independent predictor of thrombotic risk in either children or adults.48,51 Furthermore, a history of therapeutic phlebotomy in CE is associated with an increased risk of thrombosis.48 Thus, the thrombotic risk in CE appears to be independent of viscosity, but rather to be related to changes in the upregulated hypoxic responses associated with the homozygous VHL598C>T mutation. We found many HIF-regulated transcripts to be differentially upregulated in CE peripheral blood mononuclear cells, including IL1B, encoding interleukin 1β (2.1-fold), TSP1, haematologica | 2019; 104(4)

encoding thrombospondin-1 (1.5-fold), NLRP3, encoding NLR family pyrin domain containing 3 (1.4-fold), SERPINE1, encoding plasminogen activator inhibitor-1 (PAI-1) (1.2-fold), and F3 encoding tissue factor (1.1-fold).11 We also found differential gene expression in granulocytes and reticulocytes, and increased TSP-1 concentrations in plasma.48 Thus, increased HIF may cause a pro-thrombotic milieu in CE.54-56 The positive association of phlebotomy with thrombosis in CE parallels observations in the Polycythemia Vera Study Group (PVSG) 01 and 05 studies.57 We postulate that the heightened thrombotic risk is likely due to upregulation of HIF-controlled prothrombotic genes such as tissue factor54-56,58 and thrombospondin.48 It is likely that other HIFregulated plasma or vascular factors also play contributory roles.59 In aggregate, these data demonstrate that the thrombotic risk in CE is independent of hematocrit.

Polycythemia vera Thrombosis is the most common complication of PV.60-62 One-half to three-quarters of these events are arterial.63 Ischemic strokes and transient ischemic attacks account for the majority of thrombotic complications, followed in frequency by myocardial infarction, deep vein thrombosis, and pulmonary embolism. Cerebral venous thrombosis and splanchnic thrombosis, including Budd-Chiari syndrome, occur with increased frequency in PV. While it is not unusual for Budd-Chiari syndrome to present as the first indicator of PV, we have been unable to find the exact prevalence of this complication in any large published study of PV. Endogenous erythroid colony formation and the JAK2V617F mutation may be found in patients with splanchnic thrombosis years before an increase in hematocrit.64,65 In fact, the majority of “idiopathic” Budd-Chiari syndrome patients have the JAK2V617F mutation despite a normal hematocrit.64 The association of Budd-Chiari syndrome and PV is so strong that many experts advocate screening for PV with JAK2V617F mutation analysis in all patients who present with hepatic vein or portal/mesenteric thrombosis, regardless of hematocrit.66,67 It should be noted that in PV the hematocrit may be normal despite a marked elevation in red cell mass and total blood volume and that the hematocrit in the splanchnic veins may not be the same as that in the peripheral veins from where the blood sample is drawn. Furthermore, the peripheral hematocrit may be deceptively normal due to an increase in plasma volume in the presence of splenomegaly.15,68 The rationale for phlebotomy in PV was provided by a retrospective analysis of 69 patients in whom elevated hematocrit was controlled by phlebotomy and thrombocytosis by busulfan or other forms of chemotherapy.69 Over 15 years of observation, the incidence of thrombosis was proportional to the elevation in hematocrit,69 but it is not clear how much of the control of the hematocrit was related to phlebotomy versus chemotherapy-related suppression of hematopoiesis. The prospective, randomized PVSG 01 and 05 studies demonstrated that phlebotomy to control hematocrit was associated with a higher thrombotic risk compared to chemotherapy.57 The PVSG 01 study was the first randomized trial of PV patients.57 Enrollment in the study occurred between 1967 and 1974. All patients (n = 431) were initially treated with phlebotomy to reduce the hematocrit to <45% and then randomized to treatment with phlebotomy alone (n = 134), chlorambucil (n = 141) or 32P (n = 156) to maintain the hematocrit <45%. Phlebotomy was administered in the 655

V.R. Gordeuk et al.

chlorambucil and 32P arms if the hematocrit was >45% despite the chemotherapy regimen. In 1987, with a maximal follow-up of 19 years, 37.8% of the patients had experienced thrombosis as a major study outcome and 14.8% had died from thrombosis. Overall, therapeutic phlebotomy was independently and significantly associated with an increased risk of thrombosis compared to chemotherapy, but hematocrit level was not independently associated with thrombotic risk. The increased risk of thrombosis in patients undergoing phlebotomy compared to that in patients treated with myelosuppressive therapy seemed to be limited to the first 3 years of therapy.57 The increased thrombotic risk did not seem to be related to poorer disease control as reflected by hematocrit and platelet count: in a retrospective analysis that paired patients with thrombosis to those without thrombosis within the same treatment group, neither hematocrit nor platelet count was associated with thrombosis.70 As of 1987, 10.2% of the patients in the PVSG 01 study had developed acute leukemia and 11.8% had died from a hematologic malignancy. Acute leukemia was much more common in the 32P arm (9.6%) and the chlorambucil arm (13.5%) than in the phlebotomy alone arm (1.5%), and this contributed to the finding that the overall survival of patients treated with phlebotomy was comparable to that of patients treated with 32P and slightly better than that of patients treated with chlorambucil.70,71 The increased risk of thrombosis with phlebotomy compared to chemotherapy observed in the PVSG 01 study was followed up in the PVSG 05 study. Patients were initially phlebotomized to achieve a hematocrit â&#x2030;¤40% and then randomized to treatment with phlebotomy and the combination of aspirin (300 mg) and dipyridamole (75 mg) three times daily (n = 88) versus 32P (n = 90) to maintain the hematocrit <45%.57 The study was stopped at a median followup of <2 years when seven (8.0%) patients in the phlebotomy, aspirin and dipyridamole group had experienced a major thrombosis versus two (2.2%) in the 32P group, providing further evidence of a higher rate of thrombosis with therapeutic phlebotomy versus chemotherapy for PV. The European Collaboration on Low-Dose Aspirin in the Polycythemia Vera study (ECLAP), which included 1,638 patients from 12 countries and 94 centers, found no difference in thrombotic complications for patients with hematocrits within the range of 40-55%; however, there were not enough subjects with hematocrits >55% for evaluation.72 Evaluation of a cohort of 1,042 patients with PV in the ECLAP trial demonstrated an advantage of hydroxyurea therapy over phlebotomy with respect to the proportion of fatal/nonfatal cardiovascular events: 13.2% in the phlebotomy group versus 7.9% in the hydroxyurea group (P=0.006).73 An important attempt to clarify this issue was a prospective study by the Cytoreductive Therapy in Polycythemia Vera (CYTO-PV) Collaborative Group of the effect of hematocrit on thrombosis in PV patients. This study showed that patients treated with phlebotomy and hydroxyurea to a hematocrit <45% (n = 182) had a lower rate of thrombosis compared to that of patients treated to a target hematocrit of 45-50% (n = 183).74 Hydroxyurea is the most common myelosuppressive agent used in the treatment of PV;75,76 it is effective at controlling erythrocyte, leukocyte, and platelet counts without inducing acute leukemia, and it decreased the risk of thrombosis during the first few years of therapy compared to that in a historical cohort treated with phlebotomy alone.57 By 6 months into the CYTO-PV study,74 fewer patients in the high-hematocrit group were receiving 656

hydroxyurea (47% versus 59%) and among those receiving hydroxyurea the mean daily dose was 12% lower in the high-hematocrit group. This eventuated in a higher white blood cell count in the high-hematocrit group throughout the study (P<0.001). Although absence of hydroxyurea therapy77 and high leukocyte counts78 are independent correlates of thrombotic risk in PV, the rate of thrombosis was greater in the higher hematocrit group whether or not the patient was treated with chemotherapy and whether or not the white blood cell count was elevated.74 Thus, we cannot rule out the possibility that hematocrit may contribute to increased risk of thrombosis in PV along with other PVassociated prothrombotic factors. However, some of the authors of the CYTO-PV and ECLAP studies re-analyzed the study population73 and concluded that there is a â&#x20AC;&#x153;greater antithrombotic protection of hydroxyurea over phlebotomy against arterial thrombosis while the two treatments produce similar results in the protection from venous thrombosis.â&#x20AC;?79 In the PVSG 01 study, a history of previous thrombosis and older age were independent risk factors for thrombosis after controlling for therapeutic phlebotomy versus chemotherapy.57,70 Currently, the age of the patient (>60 years) and previous thrombotic events are universally acknowledged risk factors for major vascular complications in PV.60 The proportion of activated neutrophils is increased in PV,80 and it is possible that neutrophils may be an important factor in PV-associated thrombosis.81 In a multivariate analysis of the relationship of peripheral blood cell counts with thrombosis in PV subjects, an increased number of leukocytes was the most significant correlate of increased thrombotic risk.78 A study of 1,545 patients by the International Working Group - Myeloproliferative Neoplasms Research and Treatment (IWG-MRT) found that survival of PV patients correlated negatively with leukocytosis, older age, venous thrombosis, and atypical karyotype.82 It was also reported that PV may be associated with tissue factor expression in polymorphonuclear leukocytes in the absence of any in vitro challenge, and that expression is decreased after treatment with hydroxyurea.83 An additional risk for thrombotic events in PV may be environmental hypoxia. We found that PV patients residing in Salt Lake City at approximately 1,400 meters have a higher rate of arterial and venous thromboses than that of patients residing at sea level in Baltimore,84 even though they are only exposed to modest hypoxia.85 In a multivariate analysis, living in Salt Lake City was an independent thrombotic risk factor in PV.84 This may be explained by the recent observation that hypoxia decreases protein S levels in normal subjects by an HIF-1-mediated mechanism.86

Conclusion Certain disorders with elevated hematocrit, such as PV, CE, primary familial and congenital polycythemia or erythrocytosis (EPOR mutation), and EPAS1 gain-of-function mutations, are associated with thrombotic complications. These conditions are characterized by diverse cellular and metabolic changes that could be directly associated with thrombotic risk, irrespective of hematocrit level. The challenge in these conditions is to elucidate factors for the thrombotic risk other than the elevated hematocrit, and to define what, if any, role that viscosity plays in thrombotic risk. Defining these thrombosis-predisposing factors would provide the basis for idenhaematologica | 2019; 104(4)

Hematocrit and thrombotic risk in erythrocytosis

tifying and developing novel targeted therapies for these disorders. The evidence we have presented here points to favoring the use of myelosuppressive therapy for intermediate- and high-risk PV, as this approach has been proven to decrease the risk of thrombosis in PV. Furthermore, we trust that the urge to correct any abnormal laboratory data by a therapeutic intervention should be tempered by consideration of the risk-benefit ratio of any such intervention. The routine practice of phlebotomy for elevated hematocrit, with its inevitable iron deficiency (which leads to inhibition of PHD2, increased HIF, and increased erythropoietin) and potential detrimental thrombotic effects, should be re-evaluated. We hope that this review will encourage more studies to pursue the

References 1. Prchal JT. Primary and secondary erythrocytosis. In: Lichtman MA, Williams WJ, eds. Williams Hematology 9th Edition. New York: McGraw Hill Medical; 2015. 2. Prchal JT, Semenza GL, Prchal J, Sokol L. Familial polycythemia. Science. 1995;268 (5219):1831-1832. 3. McMullin MF, Wu C, Percy MJ, Tong W. A nonsynonymous LNK polymorphism associated with idiopathic erythrocytosis. Am J Hematol. 2011;86(11):962-964. 4. Giani FC, Fiorini C, Wakabayashi A, et al. Targeted application of human genetic variation can improve red blood cell production from Stem Cells. Cell Stem Cell. 2016;18 (1):73-78. 5. Tuschl K, Mills PB, Parsons H, et al. Hepatic cirrhosis, dystonia, polycythaemia and hypermanganesaemia--a new metabolic disorder. J Inherit Metab Dis. 2008;31(2):151163. 6. Quadri M, Federico A, Zhao T, et al. Mutations in SLC30A10 cause parkinsonism and dystonia with hypermanganesemia, polycythemia, and chronic liver disease. Am J Hum Genet. 2012;90(3):467-477. 7. Zmajkovic J, Lundberg P, Nienhold R, et al. A gain-of-function mutation in EPO in familial erythrocytosis. N Engl J Med. 2018;378(10):924-930. 8. Polyakova LA. [Familial erythrocytosis among inhabitants of the Chuvash ASSR. Problem Gematolog I Pereliv Krovi]. 1974;10:30-36. 9. Ang SO, Chen H, Hirota K, et al. Disruption of oxygen homeostasis underlies congenital Chuvash polycythemia. Nat Genet. 2002;32(4):614-621. 10. Perrotta S, Nobili B, Ferraro M, et al. Von Hippel-Lindau-dependent polycythemia is endemic on the island of Ischia: identification of a novel cluster. Blood. 2006;107(2): 514-519. 11. Zhang X, Zhang W, Ma SF, et al. Iron deficiency modifies gene expression variation induced by augmented hypoxia sensing. Blood Cells Mol Dis. 2014;52(1):35-45. 12. Gordeuk VR, Sergueeva AI, Miasnikova GY, et al. Congenital disorder of oxygen sensing: association of the homozygous Chuvash polycythemia VHL mutation with thrombosis and vascular abnormalities but not tumors. Blood. 2004;103(10):3924-3932. 13. Manalo DJ, Rowan A, Lavoie T, et al.

haematologica | 2019; 104(4)

14.

15. 16.

17. 18. 19.

20.

21. 22.

23.

24.

25. 26.

27.

challenge of defining the specific molecular basis of thrombosis in diverse types of polycythemia and erythrocytosis. Improved knowledge of the pathophysiology of these entities should be extended to the development of targeted approaches for the prevention and therapy of thrombotic complications. A review of potential molecular mechanisms contributing to thrombosis in myeloproliferative neoplasms was published at the time of the submission of this manuscript.87 Acknowledgments This work was supported in part by a grant from the National Institutes of Health, R01HL137991 (to JP), and institutional funds from the University of Illinois at Chicago.

Transcriptional regulation of vascular endothelial cell responses to hypoxia by HIF-1. Blood. 2005;105(2):659-669. Hickey MM, Lam JC, Bezman NA, Rathmell WK, Simon MC. von Hippel-Lindau mutation in mice recapitulates Chuvash polycythemia via hypoxia-inducible factor2alpha signaling and splenic erythropoiesis. J Clin Invest. 2007;117(12):3879-3889. Spivak JL. Myeloproliferative neoplasms. N Engl J Med. 2017;376(22):2168-2181. Sergeyeva A, Gordeuk VR, Tokarev YN, Sokol L, Prchal JF, Prchal JT. Congenital polycythemia in Chuvashia. Blood. 1997;89(6): 2148-2154. Dintenfass L. A preliminary outline of the blood high viscosity syndromes. Arch Intern Med. 1966;118(5):427-435. Thorling EB, Erslev AJ. The "tissue" tension of oxygen and its relation to hematocrit and erythropoiesis. Blood. 1968;31(3):332-343. Fan FC, Chen RY, Schuessler GB, Chien S. Effects of hematocrit variations on regional hemodynamics and oxygen transport in the dog. Am J Physiol. 1980;238(4):H545-522. Pearson TC, Humphrey PRD, Thomas DJ, al. e. Hematocrit, blood viscosity, cerebral blood flow, and vascular occlusion. In: Lowe GDO, ed. Clinical Aspects of Blood Viscosity and Cell Deformability,. New York: Springer-Verlag; 1981. Monge C. Life in the Andes and chronic mountain sickness. Science. 1942;95(2456): 79-84. Kershenovich S, Modiano M, Ewy GA. Markedly decreased coronary blood flow in secondary polycythemia. Am Heart J. 1992;123(2):521-523. Conley CL, Russell RP, Thomas CB, Tumulty PA. Hematocrit values in coronary artery disease. Arch Intern Med. 1964;113: 170-176. Jefferson JA, Escudero E, Hurtado ME, et al. Excessive erythrocytosis, chronic mountain sickness, and serum cobalt levels. Lancet. 2002;359(9304):407-408. Thomas DJ, du Boulay GH, Marshall J, et al. Cerebral blood-flow in polycythaemia. Lancet. 1977;2(8030):161-163. Borzage MT, Bush AM, Choi S, et al. Predictors of cerebral blood flow in patients with and without anemia. J Appl Physiol (1985). 2016;120(8):976-981. Wade JP, du Boulay GH, Marshall J, et al. Cerebral blood flow, haematocrit and viscosity in subjects with a high oxygen affini-

28.

29.

30. 31.

32. 33.

34.

35. 36. 37.

38. 39. 40.

41.

ty haemoglobin variant. Acta Neurol Scand. 1980;61(4):210-215. Cui MH, Billett HH, Suzuka S, et al. Fetal hemoglobin improves cerebral blood flow and decreases brain inflammation in transgenic-sickle Mice. Blood. 2016;128(22):3639. Papassotiriou I, Kister J, Griffon N, et al. Modulating the oxygen affinity of human fetal haemoglobin with synthetic allosteric modulators. Br J Haematol. 1998;102(5): 1165-1171. Vongpatanasin W, Brickner ME, Hillis LD, Lange RA. The Eisenmenger syndrome in adults. Ann Intern Med. 1998;128(9):745-755. Martin-Garcia AC, Arachchillage DR, Kempny A, et al. Platelet count and mean platelet volume predict outcome in adults with Eisenmenger syndrome. Heart. 2018; 104(1):45-50. Thorne SA. Management of polycythaemia in adults with cyanotic congenital heart disease. Heart. 1998;79(4):315-316. Perloff JK, Marelli AJ, Miner PD. Risk of stroke in adults with cyanotic congenital heart disease. Circulation. 1993;87(6):19541959. Phornphutkul C, Rosenthal A, Nadas AS, Berenberg W. Cerebrovascular accidents in infants and children with cyanotic congenital heart disease. Am J Cardiol. 1973;32(3): 329-334. Cottrill CM, Kaplan S. Cerebral vascular accidents in cyanotic congenital heart disease. Am J Dis Child. 1973;125(4):484-487. Shinton R, Beevers G. Meta-analysis of relation between cigarette smoking and stroke. BMJ. 1989;298(6676):789-794. Kannel WB, Gordon T, Wolf PA, McNamara P. Hemoglobin and the risk of cerebral infarction: the Framingham study. Stroke. 1972;3(4):409-420. Wannamethee G, Perry IJ, Shaper AG. Haematocrit, hypertension and risk of stroke. J Intern Med. 1994;235(2):163-168. Mayer GA. Hematocrit and coronary heart disease. Can Med Assoc J. 1965;93(22):11511153. Hershberg PI, Wells RE, McGandy RB. Hematocrit and prognosis in patients with acute myocardial infarction. JAMA. 1972;219(7):855-860. Lubarsky DA, Gallagher CJ, Berend JL. Secondary polycythemia does not increase the risk of perioperative hemorrhagic or thrombotic complications. J Clin Anesth. 1991;3(2):99-103.

657

V.R. Gordeuk et al. 42. Paffett-Lugassy N, Hsia N, Fraenkel PG, et al. Functional conservation of erythropoietin signaling in zebrafish. Blood. 2007;110(7): 2718-2726. 43. Shibata J, Hasegawa J, Siemens HJ, et al. Hemostasis and coagulation at a hematocrit level of 0.85: functional consequences of erythrocytosis. Blood. 2003;101(11):44164422. 44. Lamrani L, Lacout C, Ollivier V, et al. Hemostatic disorders in a JAK2V617F-driven mouse model of myeloproliferative neoplasm. Blood. 2014;124(7):1136-1145. 45. Strassel C, Kubovcakova L, Mangin PH, et al. Haemorrhagic and thrombotic diatheses in mouse models with thrombocytosis. Thromb Haemost. 2015;113(2):414-425. 46. Walton BL, Lehmann M, Skorczewski T, et al. Elevated hematocrit enhances platelet accumulation following vascular injury. Blood. 2017;129(18):2537-2546. 47. Prchal JT. Secondary polycythemia erythrocytosis. Chapter 57. In: Kaushansky K, Lichtman MA, Prchal JT, et al., eds. Williams Hematology 9th Edition. New York, NY: McGraw Hill; 2015:871-888. 48. Sergueeva A, Miasnikova G, Shah BN, et al. Prospective study of thrombosis and thrombospondin-1 expression in Chuvash polycythemia. Haematologica. 2017;102(5): e166-e169. 49. Liu E, Percy MJ, Amos CI, et al. The worldwide distribution of the VHL 598C>T mutation indicates a single founding event. Blood. 2004;103(5):1937-1940. 50. Miasnikova GY, Sergueeva AI, Nouraie M, et al. The heterozygote advantage of the Chuvash polycythemia VHLR200W mutation may be protection against anemia. Haematologica. 2011;96(9):1371-1374. 51. Sergueeva AI, Miasnikova GY, Polyakova LA, Nouraie M, Prchal JT, Gordeuk VR. Complications in children and adolescents with Chuvash polycythemia. Blood. 2015;125(2):414-415. 52. McClain DA, Abuelgasim KA, Nouraie M, et al. Decreased serum glucose and glycosylated hemoglobin levels in patients with Chuvash polycythemia: a role for HIF in glucose metabolism. J Mol Med (Berl). 2013;91(1):59-67. 53. Yoon D, Okhotin DV, Kim B, et al. Increased size of solid organs in patients with Chuvash polycythemia and in mice with altered expression of HIF-1alpha and HIF2alpha. J Mol Med. 2010;88(5):523-530. 54. Stavik B, Espada S, Cui XY, et al. EPAS1/HIF2 alpha-mediated downregulation of tissue factor pathway inhibitor leads to a prothrombotic potential in endothelial cells. Biochim Biophys Acta. 2016;1862(4):670678. 55. Sun L, Liu Y, Lin S, et al. Early growth response gene-1 and hypoxia-inducible factor-1alpha affect tumor metastasis via regulation of tissue factor. Acta Oncol. 2013;52(4):842-851. 56. Narita I, Shimada M, Yamabe H, et al. NFkappaB-dependent increase in tissue factor expression is responsible for hypoxic podocyte injury. Clin Exp Nephrol. 2016;20(5):679-688.

658

57. Berk P, Wasserman L, Fruchtman S. Treatment of polycythemia vera. A summary of clinical trials conducted by the Polycythemia Study Group. In: Wasserman L, Berk P, Berlin N, eds. Polycythemia Vera and the Myeloproliferative Disorders. Philadelphia: WB Saunders; 1995. 58. Reeves BN, Song J, Kim SJ, et al. Upregulation of tissue factor may contribute to thrombosis in PV and ET. American Society of Hematology Annual Meeting. San Diego, CA; 2018. 59. Gordeuk VR, Chung DW, Shah BN, et al. Thrombosis and von Willebrand factor in Chuvash polycythemia. Blood. 2017;130 (Suppl 1):2377. 60. Marchioli R, Finazzi G, Landolfi R, et al. Vascular and neoplastic risk in a large cohort of patients with polycythemia vera. J Clin Oncol. 2005;23(10):2224-2232. 61. Falanga A, Marchetti M. Thrombotic disease in the myeloproliferative neoplasms. Hematology Am Soc Hematol Educ Program. 2012;2012:571-581. 62. Wehmeier A, Daum I, Jamin H, Schneider W. Incidence and clinical risk factors for bleeding and thrombotic complications in myeloproliferative disorders. A retrospective analysis of 260 patients. Ann Hematol. 1991;63(2):101-106. 63. Landolfi R, Marchioli R, Kutti J, et al. Efficacy and safety of low-dose aspirin in polycythemia vera. N Engl J Med. 2004;350(2):114-124. 64. De Stefano V, Fiorini A, Rossi E, et al. Incidence of the JAK2 V617F mutation among patients with splanchnic or cerebral venous thrombosis and without overt chronic myeloproliferative disorders. J Thromb Haemost. 2007;5(4):708-714. 65. De Stefano V, Teofili L, Leone G, Michiels JJ. Spontaneous erythroid colony formation as the clue to an underlying myeloproliferative disorder in patients with Budd-Chiari syndrome or portal vein thrombosis. Semin Thromb Hemost. 1997;23(5):411-418. 66. Colaizzo D, Amitrano L, Tiscia GL, et al. The JAK2 V617F mutation frequently occurs in patients with portal and mesenteric venous thrombosis. J Thromb Haemost. 2007;5(1):55-61. 67. Reikvam H, Tiu RV. Venous thromboembolism in patients with essential thrombocythemia and polycythemia vera. Leukemia. 2012;26(4):563-571. 68. Spivak JL. Polycythemia vera: myths, mechanisms, and management. Blood. 2002;100(13):4272-4290. 69. Pearson TC, Wetherley-Mein G. Vascular occlusive episodes and venous haematocrit in primary proliferative polycythaemia. Lancet. 1978;2(8102):1219-1222. 70. Berk PD, Goldberg JD, Donovan PB, Fruchtman SM, Berlin NI, Wasserman LR. Therapeutic recommendations in polycythemia vera based on Polycythemia Vera Study Group protocols. Semin Hematol. 1986;23(2):132-143. 71. Berlin NI, Wasserman LR. Polycythemia vera: a retrospective and reprise. J Lab Clin Med. 1997;130(4):365-373. 72. Di Nisio M, Barbui T, Di Gennaro L, et al.

73.

74.

75.

76. 77.

78.

79.

80.

81.

82.

83.

84.

85. 86.

87.

The haematocrit and platelet target in polycythemia vera. Br J Haematol. 2007;136(2): 249-259. Barbui T, Vannucchi AM, Finazzi G, et al. A reappraisal of the benefit-risk profile of hydroxyurea in polycythemia vera: a propensity-matched study. Am J Hematol. 2017;92(11):1131-1136. Marchioli R, Finazzi G, Specchia G, et al. Cardiovascular events and intensity of treatment in polycythemia vera. N Engl J Med. 2013;368(1):22-33. Dingli D, Tefferi A. Hydroxyurea: the drug of choice for polycythemia vera and essential thrombocythemia. Curr Hematol Malig Rep. 2006;1(2):69-74. Barbui T, Finazzi G. Evidence-based management of polycythemia vera. Best Pract Res Clin Haematol. 2006;19(3):483-493. Fruchtman SM, Mack K, Kaplan ME, Peterson P, Berk PD, Wasserman LR. From efficacy to safety: a Polycythemia Vera Study Group report on hydroxyurea in patients with polycythemia vera. Semin Hematol. 1997;34(1):17-23. Landolfi R, Di Gennaro L, Barbui T, et al. Leukocytosis as a major thrombotic risk factor in patients with polycythemia vera. Blood. 2007;109(6):2446-2452. Barbui T, De Stefano V, Ghirardi A, Masciulli A, Finazzi G, Vannucchi AM. Different effect of hydroxyurea and phlebotomy on prevention of arterial and venous thrombosis in polycythemia vera. Blood Cancer J. 2018;8(12):124. Falanga A, Marchetti M, Evangelista V, et al. Polymorphonuclear leukocyte activation and hemostasis in patients with essential thrombocythemia and polycythemia vera. Blood. 2000;96(13):4261-4266. Vannucchi AM. Insights into the pathogenesis and management of thrombosis in polycythemia vera and essential thrombocythemia. Intern Emerg Med. 2010;5(3):177184. Tefferi A, Rumi E, Finazzi G, et al. Survival and prognosis among 1545 patients with contemporary polycythemia vera: an international study. Leukemia. 2013;27(9):18741881. Maugeri N, Giordano G, Petrilli MP, et al. Inhibition of tissue factor expression by hydroxyurea in polymorphonuclear leukocytes from patients with myeloproliferative disorders: a new effect for an old drug? J Thromb Haemost. 2006;4(12):2593-2598. Zangari M, Fink L, Tolomelli G, et al. Could hypoxia increase the prevalence of thrombotic complications in polycythemia vera? Blood Coagul Fibrinolysis. 2013;24(3):311-316. Ruiz-Arguelles GJ. Altitude above sea level as a variable for definition of anemia. Blood. 2006;108(6):2131; author reply 2131-2132. Pilli VS, Datta A, Afreen S, Catalano D, Szabo G, Majumder R. Hypoxia downregulates protein S expression. Blood. 2018;132(4):452-455. Bar-Natan M, Hoffman R. New insights into the causes of thrombotic events in patients with myeloproliferative neoplasms raise the possibility of novel therapeutic approaches. Haematologica. 2019;104(1):3-6.

haematologica | 2019; 104(4)

REVIEW ARTICLE

State-of-the-art review: allogeneic stem cell transplantation for myelofibrosis in 2019

Ferrata Storti Foundation

Donal P. McLornan,1,2 Ibrahim Yakoub-Agha,3 Marie Robin,4 Yves Chalandon,5 Claire N. Harrison1,2* and Nicolaus Kroger6*

Guy’s and St.Thomas’ NHS Foundation Trust, Department of Haematology, Guy’s Tower, Great Maze Pond, London, UK; 2Comprehensive Cancer Centre, King’s College, London, UK; 3CHU de Lille, LIRIC, INSERM U995, Universite de Lille, France; 4Hôpital Saint-Louis, Service d'Hématologie-Greffe, Assistance Publique Hôpitaux de Paris, University Paris 7, INSERM 1131, France; 5Geneva University Hospitals, Division of Hematology, Rue GabriellePerret-Gentil 4 and Faculty of Medicine, University of Geneva, Switzerland and 6University Hospital Eppendorf, Hematology Department, Hamburg, Germany 1

Haematologica 2019 Volume 104(4):659-668

CNH and NK: joint senior authors

ABSTRACT

dvances in understanding the pathogenesis and molecular landscape of myelofibrosis have occurred over the last decade. Treating physicians now have access to an ever-evolving armamentarium of novel agents to treat patients, although allogeneic hematopoietic stem cell transplantation remains the only curative approach. Improvements in donor selection, conditioning regimens, disease monitoring and supportive care have led to augmented survival after transplantation. Nowadays, there are comprehensive guidelines concerning allogeneic hematopoietic stem cell transplantation for patients with myelofibrosis. However, it commonly remains difficult for both physicians and patients alike to weigh up the risk-benefit ratio of transplantation given the inherent heterogeneity regarding both clinical course and therapeutic response. In this timely review, we provide an up-to-date synopsis of current transplantation recommendations, discuss usage of JAK inhibitors before and after transplantation, examine donor selection and compare conditioning platforms. Moreover, we discuss emerging data concerning the impact of the myelofibrosis mutational landscape on transplantation outcome, peri-transplant management of splenomegaly, poor graft function and prevention/management of relapse.

Correspondence: DONAL P MCLORNAN donal.mclornan@nhs.net Received: September 9, 2018. Accepted: November 16, 2018. Pre-published: March 14, 2019.

doi:10.3324/haematol.2018.206151

Introduction Myelofibrosis is a heterogeneous disease as regards both disease phenotype and mutational landscape. Following the discovery of the JAK2V617F mutation in 2005 and subsequent studies confirming the clinical efficacy of JAK inhibitors, the treatment paradigm has been revolutionized.1-6 Worldwide experience with JAK inhibitor therapy continues to grow and a considerable proportion of patients will gain beneficial symptom and/or splenic responses, albeit heterogeneous and of variable duration. Furthermore, given that many other novel therapeutics, such as anti-fibrotic and immunomodulatory agents, have been used to treat myelofibrosis,7 the majority of patients moving forward with allogeneic hematopoietic stem cell transplantation (SCT) have had at least one prior treatment, making ‘realworld’ transplant decisions increasingly complex. Data from the European Society for Blood and Marrow Transplantation (EBMT) suggest a year-on-year increase in transplants for myelofibrosis. In this review we focus on current indications for allogeneic SCT, prognostic scoring models to aid decision-making, donor selection, conditioning regimens, the role of splenectomy and prevention and management of relapse. haematologica | 2019; 104(4)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/659 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

659

D.P. McLornan et al.

The transplant decision: how to decide who should be considered for allogeneic stem cell transplantation Utilization of established and novel prognostic scoring systems in myelofibrosis Clinical prognostic scoring systems play a pivotal role in decisions regarding allogeneic SCT. Until recently, the most commonly applied was the International Prognostic Scoring System (IPSS) which estimates survival from time of diagnosis.8 The score is based upon five factors: age >65 years, hemoglobin <100 g/L, leukocyte count >25x109/L, circulating blasts ≥1% and constitutional symptoms, permitting stratification into four groups: low risk (0 risk factors; estimated median survival, 135 months), intermediate risk-1 (1 risk factor; median survival, 95 months), intermediate risk2 (2 risk factors, median survival, 48 months) and high risk (3+ risk factors; median survival, 27 months). The Dynamic IPSS (DIPSS), utilizing the same five factors, permits application of the scoring system at any stage in the disease course.9 Finally, the DIPSS-plus incorporates three additional adverse factors – transfusion dependency; platelet count <100x109/L and unfavorable cytogenetics.10,11 We follow current European LeukemiaNet/EBMT expert consensus whereby “Patients with intermediate-2- or high-risk disease according to the IPSS, DIPSS or DIPSS-plus and age <70 years should be considered potential candidates for allogeneic SCT”.12 Patients with “intermediate-1-risk disease and age <65 years should be considered as candidates if they present with either refractory, transfusion-dependent anemia, or a percentage of blasts in peripheral blood >2%, or adverse (as defined by the DIPSS-plus classification) cytogenetics”. Decisions regarding transplantation for intermediate-1-risk disease remain complex and are discussed below. A retrospective, comparative multicenter outcome analysis of 438 patients with primary myelofibrosis aged <65 years at diagnosis who underwent allogeneic SCT (n=190) or conventional therapy in the era before JAK inhibitors (n=248), utilizing DIPSS scoring, demonstrated that allogeneic SCT clearly benefited intermediate-2 or high-risk patients whereas for low-risk disease, transplantation is not immediately indicated and is held in reserve for progressive disease.13 For intermediate-1 disease, individual counseling was recommended. Despite evident utility, marked heterogeneity may be observed within each allocated IPSS/DIPSS subgroup. Grinfeld et al. performed genomic and clinical phenotype analyses of 2,041 patients with myeloproliferative neoplasms, including 311 with myelofibrosis, and combined the findings into a unifying patient-specific predictive model.14 For myelofibrosis, this model predicted event-free survival better than did either the DIPSS or IPSS (81% versus 69% versus 77% concordance) and an online calculator is being developed for predicting patient-specific outcomes (https://jg738.shinyapps.io/mpn_app/). It will be of interest to apply this model to a population of myelofibrosis patients undergoing allogeneic SCT. A collaborative group recently studied 685 molecularlyannotated patients with secondary myelofibrosis to assess whether an independent prognostic scoring system could be derived.15 The so-called MYelofibrosis SECondary to PV and ET-Prognostic Model (MYSEC-PM) allocated individuals into four prognostic categories based on negative predictors of survival whereby two points were attributed to hemoglobin <110 g/dL, a CALR-unmutated phenotype and circulating blasts ≥3%, one point to thrombocytopenia 660

<150x109/L and constitutional symptoms and 0.15 points to any year of age. Median survival estimates for each group ranged from not reached in the low-risk cohort to 2 years in high-risk cohorts. Akin to IPSS/DIPSS, recent analyses suggested that transplant-specific age-adjustment of the MYSEC-PM provided prognostic predictive power as regards overall survival following allogeneic SCT.16-18

The role of mutational profiling in allogeneic stem cell transplant decisions Comprehensive mutational profiling has helped identify heterogeneous somatic mutations in patients with myelofibrosis. Delineation of this mutational landscape confers prognostic significance as regards overall survival and risk of disease progression/transformation and increasingly influences therapeutic decisions (Table 1).19-22 In the nontransplant setting, it is well established that myelofibrosis patients with CALR type-1/like mutations survive longer than patients with CALR type-2/like and MPL or JAK2 mutations.21 ‘Triple negativity’, i.e. lacking a detectable JAK2, MPL or CALR mutation, is associated with more adverse outcomes. Previous analyses of 617 myelofibrosis patients revealed that the median survival was only 3.2 years for those who were ‘triple negative’.19 Conventionally, high molecular risk myelofibrosis is defined by the presence of at least one of EZH2, ASXL1, IDH1/2 and SRSF2 mutations and is associated with worse overall and leukemiafree survival.22 These data raise the question of whether earlier transplantation should be considered for those who have ‘triple negative’ disease, particularly with high molecular risk mutations, and need to be taken into consideration when counseling patients, particularly transplant-eligible individuals with intermediate-1-risk disease and 'good' donors. Recently, Guglielmelli and colleagues described the utility of both the mutation-enhanced IPSS ‘MIPSS70’ and ‘MIPSS70-plus’ (including cytogenetic evaluation) scoring systems for ‘transplant-age’ patients ≤70 years old with either pre-fibrotic or overt, primary myelofibrosis who were considered candidates for transplantation.23 Significant risk factors for overall survival were leukocyte count >25x109/L, platelet count <100x109/L, presence of >2 high molecular risk mutations, hemoglobin <100 g/L, peripheral blood blasts ≥2%, constitutional symptoms, high molecular risk category, fibrosis grade >2 and absence of CALR type1/like mutations. These scoring systems enabled three discrete prognostic risk categories to be delineated: low-risk, intermediate-risk and high-risk with 5-year survival rates of 95%; 70% and 29%, respectively. The MIPSS70-plus included cytogenetics in the multivariable analysis. Recent updates to ‘MIPSS70-plus version 2’ occurred with recognition of U2AF1Q157 as a high molecular risk mutation and the scoring system uses new sex- and severity-adjusted hemoglobin thresholds.24 Importantly these scores incorporate current molecular data and up-to-date WHO 2016 disease classification and will aid decisions regarding allogeneic SCT. It can be difficult when a clinician is managing a young, fit patient with myelofibrosis who has intermediate risk-1 disease and a fully matched donor. Should we transplant upfront when the patient is fit, and the donor is available or should we wait for further disease upstaging before considering allogeneic SCT? Collectively, we feel that this is a very individualized choice, and has many strata including the patient’s wishes and donor type, disease-associated haematologica | 2019; 104(4)

Allogeneic SCT for myelofibrosis Table 1. Key characteristics of pivotal non-JAK2/CALR or MPL mutations in myelofibrosis.

Chromosome

Characteristics

Reference

Spliceosome mutations

SRSF2

17q25.1

• Mutations reported in up to 17% of PMF • Commonly monoallelic mutations affecting residue P95 • More common in older age, higher DIPSS-plus group, accompanying IDH mutations • Associated with worse outcome: OS and LFS

U2AF1

21q22.3

• Involved in pre-mRNA splicing • Mutations reported in up to 16% of PMF • Mutations associated with older age; normal karyotype; anemia and thrombocytopenia

SF3B1

2q33.1

• Incidence of around 6% • More common with bulky splenomegaly • Presence does not appear to influence survival

ASXL1

20q11

• Frequently frameshift mutations – majority associated with similar outcomes • Associated with worse OS • Poor risk category CALR [-]/ ASXL1 [+]

EZH2

7q36.1

• Loss of function mutations associated with poor OS • May have higher white cell counts and frequently co-exist with JAK2V617F

• Higher frequency of mutations in blast phase disease • Clusters with older age • ? more often associated with a normal karyotype

71,72

Chromatin modifications

Other IDH1 IDH2

2q33.3 15q26.1

DIPSS: Dynamic International Prognostic Scoring System; PMF: primary myelofibrosis; OS: overall survival; LFS: leukemia-free survival.

symptom burden and objective quality of life assessment. Given the marked intra-category heterogeneity, refinement of prognostication should occur with cytogenetic and mutational data and the patient should be counseled appropriately, particularly if high molecular risk features are present. Certainly, acquisition of high-risk karyotypes, transfusion dependence or steadily increasing peripheral blood blast counts would suggest a trigger to move towards allogeneic SCT.

Recipient age: does this play a role in the transplant decision? The majority of prognostic scoring systems incorporate age as a risk factor; however, multiple dynamic factors determine post-transplant outcome (Figure 1). One of the most common questions is how old is ‘too old’? Transplantation may be more challenging for elderly recipients and there is marked variation globally in the arbitrary age ‘cut-off’ for allogeneic SCT for myelofibrosis. Historically, many earlier studies suggested worse outcome with increasing age. For example, early myeloablative conditioning (MAC) studies involving total body irradiation demonstrated worse outcome for those patients >45 years old, and other small studies suggested worse outcomes for those >50 and >60 years old.25-27 In contrast, the Seattle group reported on a highly selected group of myelofibrosis patients undergoing allogeneic SCT with a median age of 65 years (range, 60-78 years), many of whom had multiple co-morbidities.28 Time to engraftment and graft-versus-hostdisease (GvHD) rates did not differ significantly from those haematologica | 2019; 104(4)

in younger cohorts. Moreover, no significant outcome differences were identified when comparing patients <65 years old with those >65 years old. Focusing on chronological age alone may not be correct and well-selected elderly patients with minimal co-morbidities and good organ function can benefit significantly from allogeneic SCT. However, real-life decisions can be more difficult in those >70 years old or in those individuals aged 65-70 years who have co-morbidities. In some cases, potential candidates may be frail because of disease symptom burden and have a worse Performance Status, a history of thrombosis or significant potential for hepatic dysfunction. These factors need careful consideration and optimization where possible. As described above, an upper age limit of 70 years was cautiously suggested but this does not mean that ‘fit’ individuals within a few years above this threshold should be excluded. In general, the transplant decision is heavily influenced by careful assessment of the patient and a multi-disciplinary approach. Moreover, factors incorporated in the comprehensive geriatric assessment should be considered and this requires validation in the setting of patients with myelofibrosis undergoing allogeneic SCT.29

Is there a role for splenectomy before transplantation for myelofibrosis? Evidence on the beneficial role of splenectomy prior to allogeneic SCT remains somewhat sparse although in theory splenectomy may be an attractive strategy as it could 661

D.P. McLornan et al.

Figure 1. Factors determining outcomes following allogeneic stem cell transplantation in myelofibrosis. The determinants of outcomes following allogeneic stem cell transplantation can be divided into: pre-transplantation, transplant-specific, and post-transplant strategies and relapse management. IPSS: International Prognostic Scoring System; DIPSS: dynamic IPSS; IST: immunosuppressive therapy; DLI: donor lymphocyte infusion; MRD: minimal residual disease; Sib: sibling; MUD: matched unrelated donor; MMUD: mismatched unrelated donor; MAC: myeloablative conditioning; RIC: reduced Intensity conditioning; CMV: cytomegalovirus

aid engraftment and perhaps be associated with improved post-transplant outcome.30 However, this procedure is not without risk. Over a decade ago, the Mayo clinic group reported on 314 myelofibrosis patients, albeit not transplant candidates, undergoing elective splenectomy: the intervention was associated with significant perioperative complications in nearly 28% of cases.31 Despite major improvements with minimally invasive approaches, there are still potential risks of both thrombosis and hemorrhage. Although a prospective study of 103 patients undergoing allogeneic SCT coordinated by the Chronic Malignancies Working Party of the EBMT suggested more rapid neutrophil engraftment in those who had undergone splenectomy (n=14) compared to those who had not, both univariate and multivariate analyses suggested a significantly higher rate of relapse at 3 years for the former.32 Moreover, effects on immune reconstitution and GvHD modulation are unclear. In contrast, a French group retrospectively reported on 85 myelofibrosis allogeneic SCT patients from a single center, 39 of whom had undergone pre-transplant splenectomy.33 Of note, one half of those patients undergoing splenectomy had surgical or post-surgical complications, most frequently of a thrombotic or hemorrhagic nature. Following Cox adjustment analyses, there was no association between pre-transplant splenectomy and either nonrelapse mortality (NRM) or relapse risk, in fact there was a suggestion towards improved overall survival and eventfree survival. This evidently requires evaluation in a larger cohort and ideally in a controlled study as suggested by the authors. In contrast to these findings, initial analyses by 662

McLornan et al. found no significant effect of splenectomy on overall survival or NRM in a large cohort of splenectomized patients undergoing allogeneic SCT with either reduced intensity conditioning (RIC) or MAC (n=180) registered in the EBMT registry database and nor did the retrospective Center for International Blood and Marrow Transplant Research (CIBMTR) study.34,35 Moreover, given the potential splenic effects of JAK inhibitor therapy, alternative methods of reducing bulky splenomegaly are possible. Lastly, it is unknown what effect pre-transplant spleen removal will have on immune reconstitution. In general, it is the view of the authors that pre-transplant splenectomy cannot be routinely recommended although it is clear that individual patient-stratified assessment should occur. Lowdose splenic irradiation prior to transplantation has been explored in small cohorts of patients, but overall there is insufficient evidence on this strategy.36,37

What is the role of JAK inhibitors before transplantation for myelofibrosis? Many questions arise from the use of JAK inhibitors in the myelofibrosis transplant algorithm. Following the phase III trials, COMFORT-I and â&#x20AC;&#x201C;II, confirming the efficacy of the JAK1/JAK2 inhibitor ruxolitinib (Novartis, Switzerland) in myelofibrosis, many potential allogeneic SCT recipients have been treated with this agent.5,6 JAK inhibitors collectively are attractive agents given that they may improve Performance Status, reduce splenomegaly and potentially shorten time to engraftment and may dampen an inherently pro-inflammatory milieu. Earlier and more recent studies haematologica | 2019; 104(4)

Allogeneic SCT for myelofibrosis Table 2. Summary of outcomes in the main studies on reduced intensity and myeloablative conditioning in myelofibrosis.

Conditioning intensity

Conditioning regimen

GvHD rates

Overall survival

Comments

Reference

RIC

Flu Mel (sibling) Flu Mel + ATG (URD)

75% sibling 32% URD

24% graft failure in the URD group

RIC

103

Flu Bu

Acute grade II-IV Sibling 38% URD 41% Acute grade II-IV 26% Chronic L:24%; E:24% Acute Chronic 43% 40% 40% 32% 24% 26%

67% at 5 years

Low rates of graft failure and timely engraftment Heterogeneous cohort; diseasefree survival long-term in approximately 1/3

MAC

Predominantly MAC

170 Sibling 117 MUD 33 Other 104

Various TBI based (n=15) Busulphan based (n=80); RIC (n=9)

RIC

233

Flu Bu (38%) Flu Mel (28%) Flu TBI. (22%) Flu Bu FBM Flu Mel

RIC

160

Flu Bu (105) Flu Mel (55)

MAC RIC

760 1423

Common regimens BuCy or TBI based Flu Bu; Flu Mel

39% at 5 years 31% at 5 years 31% at 5 years

Acute grade II-IV : 64% Chronic L+E: 84%

61% at 7 years

Acute grade II-IV : 37% Chronic at 1 year 42%

56% for MSD, 37% URD, 34% MMUD Similar OS, NRM and relapse rates

Acute grade II-IV 47% 68% 68 %

Improved survival with targeted busulfan dosing in BuCy Donor type most important determinant of outcome

100% donor chimerism was seen more frequently at day +30 and day +100 in patients who received FBM or Flu Mel than Flu Bu. Acute Chronic 7-year OS was 52% for Flu Mel regimen appears 31% 62% the Flu Mel group and to induce more NRM than 53 % 49% 59% for the Flu Bu group the Flu Bu regimen; but with augmented disease control; similar outcomes. Acute grade I-IV Chronic L/E Primary analyses; full 29% 23/27% Median OS= 6.6 years analysis in preparation 32% 20/32% Median OS =5.3 years No differences in NRM between MAC/RIC Worse outcome with MMUD and poor Performance Status

ATG: antithymocyte globulin; Bu: busulfan; Cy: cyclophosphamide; E: extensive; FBM: fludarabine, bis-chlorethyl-nitroso-urea/carmustine, melphalan; Flu: fludarabine; GvHD: graft-versus-host disease; L: limited; MAC: myeloablative conditioning; Mel: melphalan; MMUD: mismatched unrelated donor; MUD: matched unrelated donor; NRM: non-relapse mortality; OS: overall survival; RIC: reduced intensity conditioning; TBI: total body irradiation; URD: unrelated donor.

demonstrated that, in general, there were no serious adverse effects following the use of JAK inhibitors prior to allogeneic SCT with appropriate tapering.38-40 Of note the French phase II JAK ALLO trial, investigating the use of ruxolitinib before allogeneic SCT, was temporarily halted because of two cases of febrile cardiogenic shock and one of tumor lysis syndrome following ruxolitinib discontinuation.41 Subsequent experience has not demonstrated that either of these adverse events is common. Outcomes of 100 patients who had been exposed to JAK inhibitor therapy and underwent allogeneic SCT, between 2009-2014, were determined in a retrospective, multicenter study.42 The cohort was divided into five groups defined by clinical status/response to JAK inhibition: groups (i) clinical improvement (n=23), (ii) stable disease (n=31), (iii) new cytopenia/increasing blasts/JAK inhibitor intolerance (n=15), (iv) progressive disease: splenomegaly (n=18), and (v) leukemic transformation (n=13). The overall survival rate at 2 years after allogeneic SCT for the entire cohort was 61% (95% CI: 49-71%). The cumulative incidence of grade II-IV acute GvHD by day 100 was 37% (95% CI: 27-47%) and that of chronic GvHD by 2 years was 48%. Survival analyses performed based on response to JAK inhibition prior to allogeneic SCT revealed a 2-year overall survival haematologica | 2019; 104(4)

rate of 91% (95% CI: 69-97%) for patients with clinical improvement. However, it remains unclear whether the better post-transplant outcome in those achieving a clinical improvement was drug-related or simply reflected the presence of an inherently more favorable disease phenotype. A multicenter, German study reported on 159 patients, 46 of whom had received pre-transplant ruxolitinib at any point with a median treatment duration of 4.9 months (range, 0.4-39.1 months).43 There was a trend towards a lower rate of relapse in the ruxolitinib group (9% versus 17%, P=0.2), with similar disease-free and overall survival rates. The hypothesis that pre-exposure to JAK inhibitors may modulate the relapse risk requires further exploration. In a recently published study of a small cohort of patients (n=12) undergoing allogeneic SCT for myelofibrosis, ruxolitinib (5 mg BID) was continued until stable engraftment.44 There was no graft failure and timely neutrophil engraftment, although the drug had to be discontinued in two patients, on days +17 and +18, because of post-engraftment cytopenia. Of particular relevance, the rate of grade II-IV acute GvHD within the first 100 days was low at 8% although the cytomegalovirus reactivation rate was 41%. The use of peri-transplantation JAK inhibition to reduce the risk of GvHD remains an area of great interest. In conclusion, the 663

D.P. McLornan et al.

majority of patients in 2018 who proceeded to allogeneic SCT had been exposed to JAK inhibitor therapy. The JAK inhibitor should be tapered down over a 10- to 14-day period so that JAK inhibition therapy is ceased just before the commencement of conditioning or prior to stem cell return. Whether the best strategy for patients receiving JAK inhibitor therapy with a view to allogeneic SCT is to undergo the transplant at the time of best response or whether the procedure should be delayed until a further trigger occurs remains difficult to determine and hence patientstratified approaches are required. Of note, JAK inhibitors are additionally becoming established in the treatment of steroid refractory GvHD, although this is outside the purpose of this review.45

Conditioning intensity and regimen? Myeloablative conditioning for allogeneic stem cell transplantation in patients with myelofibrosis Historically, the majority of MAC platforms consisted of total body irradiation with or without high-dose cyclophosphamide and early toxicity, NRM and GvHD rates were high, especially for older individuals.25,35 (Table 2). The large CIBMTR study reported on 289 patients who underwent allogeneic SCT between 1989 and 2002, 79% of whom received MAC â&#x20AC;&#x201C; predominantly busulfan + cyclophosphamide and total body irradiation with or without cyclophosphamide.25 The day +100 transplant-related mortality rate was 18% for those undergoing allogeneic SCT from an HLA-matched sibling and 35% for those receiving a graft from an unrelated donor. Approximately one-third of the patients achieved long-term (5 years) relapse-free survival. Kerbauy et al. reported on 104 patients, with a median age of 45 years (range, 18-70 years), 95 of whom received MAC. The NRM rate at 5 years was 34% and there were significant rates of acute grade II-IV GvHD (64%) and chronic GvHD (84%).46 The estimated 5year survival rate was 61% (95% CI: 43-65%) for the entire cohort. Those undergoing MAC with targeted levels of busulfan in combination with cyclophosphamide (120 mg/kg) had a significantly higher probability of overall survival (68%). For younger recipients planned for MAC allogeneic SCT, our preference is towards busulfan + cyclophosphamide in an attempt to reduce therapy-related morbidity.

Reduced intensity conditioning for myelofibrosis allogeneic stem cell transplantation â&#x20AC;&#x201C; what evidence exists to guide the choice of regimen? Due to age distribution, the vast majority of patients undergoing transplantation will receive RIC. However, the term RIC covers a highly heterogeneous group of treatments as regards both immunosuppressive and moderate myeloablative properties. According to the EBMT consensus, RIC regimens are strictly defined as those which do not fit the definition for MAC or non-myeloablative regimens.47 Historically, published RIC platforms were admixed and reflected local practice (Table 2). The first prospective, EBMT multicenter phase II trial of RIC allogeneic SCT in myelofibrosis enrolled 103 patients between 2002-2007.32 Conditioning was uniform and consisted of busulfan (10 mg/kg) orally (or equivalent IV dose)/fludarabine (180 mg/m2) and T-cell depletion achieved with antithymocyte globulin (Fresenius, Graefeling, Germany) at a dose of 3 x 10 mg/kg (for related transplantation) or 3 x 20 mg/kg (for 664

unrelated donor transplantation). Primary graft failure rates were low (~2%) and timely engraftment of both neutrophils (median, 18 days; range, 10-84 days) and platelets >20 x 109/L occurred (median, 22 days; range, 8 -145 days). Rates of acute GvHD were acceptable at 27% for grade IIIV and 11% for grade III-IV. The cumulative incidence of NRM at 1 year was 16% (95% CI: 9-23%) and was significantly higher in the mismatched donor setting than in the fully matched donor setting (38% versus 12%; P=0.003). The cumulative incidences of relapse at 3 and 5 years were acceptable at 22% and 29%, respectively. Multivariate analysis suggested a higher 3-year relapse incidence with splenectomy (HR: 3.6) and Lille HR score (HR: 5.23). HLAmismatched transplantation was a risk factor for both therapy-related mortality and adverse overall survival. Updated survival analyses revealed estimated 5- and 8-year overall survival rates of 68% and 65%, respectively. More recently, the two most frequently used RIC regimens for myelofibrosis - fludarabine-busulfan (FB) and fludarabine-melphalan (FM) were compared in a retrospective study of 160 patients (FB: n=105, FM: n=55) with a median follow up >5 years.48 Conditioning protocols were uniform, the FM regimen consisted of fludarabine 90 mg/m2 and melphalan 140 mg/m2 and the FB regimen consisted of intravenous busulfan (or oral equivalent) 8 mg/kg, fludarabine 180 mg/m2 and antithymocyte globulin-FreseniusÂŽ. After statistical weighting, the incidence of acute GvHD was 62% for the FM group and 31% for the FB group while chronic GvHD rates were 49% and 53%, respectively. Although the FM regimen had more pronounced early toxicity, the long-term outcome of patients treated with the two regimens was similar. Multivariate analysis failed to demonstrate significant differences regarding overall survival or disease-free survival although individuals undergoing FM conditioning had relapse rates of <5%. In both groups, the use of a HLAmismatched unrelated donor was associated with worse outcomes as regards NRM, overall survival and progression-free survival. A prospective, multicenter, phase II MPD-RC study investigated the use of FM conditioning in 66 patients recruited between 2007-2011.49 Those with sibling donors (n=32) received FM whereas those with unrelated donors (n=34) received FM + antithymocyte globulin. Significantly inferior results were seen with unrelated donors than with sibling donors: with a median follow up of 25 months, the overall survival rate was 75% in the sibling group compared to only 32% in the unrelated donor group (HR: 3.9; 95% CI: 1.8-8.9; P<0.001). Moreover, considerable NRM rates were observed in the unrelated donor setting. Of note, this small study did not determine outcome differences between matched or mismatched unrelated donors, in contrast to other studies.32,34

Can we realistically compare myeloablative versus reduced intensity condition in allogeneic transplantation for myelofibrosis? There are no current, prospective trials comparing RIC with MAC for myelofibrosis and conclusions from retrospective studies are not always straight forward. In a welldescribed cohort of 92 patients undergoing allogeneic SCT between 1982-2009, 40 patients received MAC and 52 RIC regimens.27 No differences existed with regards to day +100 transplant-related mortality. The probability of survival at 5 years was 49% for patients given MAC and 59% for those given RIC (P=0.125). However, RIC platforms were associated with markedly improved outcome for those <60 years old compared to those who were older (estimated 5-year haematologica | 2019; 104(4)

Allogeneic SCT for myelofibrosis

survival of 75% versus 20%) and there was a lower incidence of acute GvHD with RIC regimens than with MAC. Almost 10 years ago, Patriarca et al. reported on the 20-year experience of the Gruppo Italiano Trapianto Di Midollo Osseo (GITMO) in which 48 patients underwent MAC and 52 were given RIC regimens.50 Intensity of the conditioning regimen did not significantly influence either overall survival or relapse incidence. More recently, a preliminary analysis from the CMWP of the EBMT by McLornan et al. reported on a total of 2,183 patients who underwent allogeneic SCT for myelofibrosis; regimen intensity was assessed by standard EBMT criteria.34 Conventional MAC regimens were utilized in 760 patients while 1,423 received RIC regimens. Donor sources were similar: MAC cohort, 309 (41%) matched sibling donors and 451 (59%) unrelated donors; RIC cohort, 543 (38%) matched sibling donors and 880 (62%) unrelated donors. In the preliminary analyses, no statistically significant differences in engraftment, GvHD, NRM, progression-free survival and overall survival rates were found between these two large RIC and MAC cohorts; the final multivariate analysis is awaited. Lastly, the exact role of T-cell depletion strategies in allogeneic SCT for myelofibrosis requires clarification. In addition to the above discussions, retrospective analyses by Robin et al., on behalf of EBMT, have demonstrated that in the matched sibling donor setting, use of antithymocyte globulin decreased acute GvHD rates without increasing the relapse risk.51

Alternative donors Mismatched related donor transplantation in myelofibrosis Until recently, overall experience remained restricted because of historical fears of graft rejection and GvHD and exclusion from clinical trials. The EBMT group recently described a retrospective study of 56 myelofibrosis patients (median age, 57 years; range, 38-72 years) who underwent mismatched related donor transplantation.52 In this cohort, 70% received MAC and 30% received RIC; the source of hematopoietic stem cells was bone marrow in 66% of cases and peripheral blood in 34%. Conditioning approaches varied, reflecting the heterogeneous nature of this cohort, but the most commonly used was thiotepa, busulfan and fludarabine (TBF) and post-transplant cyclophosphamide. Encouragingly, neutrophil engraftment by day 28 was achieved by 82% of the cohort, at a median time of 21 (range, 19-23) days. Primary graft failure occurred in five patients and secondary graft failure in seven, with two patients dying before engraftment. The cumulative incidence of acute GvHD at day +100 was 28% for grade II-IV and 9% for grade III-IV disease. The cumulative incidence of chronic GvHD at 1 year was 45%. The 1- and 2-year overall survival rates were 61% (range, 48-74%) and 56% (range, 41-70%), respectively. The cumulative incidence of relapse was 19% (range, 7-31%) and the NRM was 38% (range, 24-51%) at 2 years. The exact role of transplantation from mismatched related donors in myelofibrosis does, therefore, require further clarification but this study demonstrates that engraftment is feasible with acceptable rates of GvHD and overall survival, although cumulative NRM rates remain somewhat high.

Umbilical cord blood stem cell transplantation In the setting of myelofibrosis, Takagi et al. initially reported on 14 patients who had undergone RIC umbilical haematologica | 2019; 104(4)

cord blood transplantation, predominantly high-risk candidates with secondary acute myeloid leukemia.53 Neutrophil engraftment was achieved in the vast majority (92%) at a median of 23 days and full donor chimerism, where evaluable, was rapidly achieved. Survival was poor with a 4-year overall survival rate of only 28%, perhaps not surprisingly given the high-risk population. More recently, a series of 35 umbilical cord blood transplant recipients (median age, 54 years), registered under EUROCORD, was reported: seven had developed blast phase myelofibrosis and almost half underwent splenectomy.54 The most common conditioning regimen was total body irradiation, cyclophosphamide and fludarabine (TCF). Cord blood units were 5/6 (23%) and 4/6 (77%) HLAmatched. Neutrophil engraftment was achieved in 28/35 patients at a median time of 30 days and a total of 14 patients displayed graft failure (4 underwent a second transplant procedure). The 2-year overall and event-free survival rates were 44% and 30% respectively. In the RIC setting, all recipients undergoing TCF conditioning achieved both neutrophil and platelet engraftment and there was an association between utilization of this regimen and improved event-free survival.

Blast phase myelofibrosis and allogeneic stem cell transplantation Historically, the outcome of patients with blast phase myeloproliferative neoplasms has been extremely poor. A recent retrospective analysis of 410 patients with blast phase myeloproliferative neoplasms, not focusing on allogeneic SCT, revealed a sobering median survival of 3.5 months.55 Intensive chemotherapy resulted in complete remission or complete remission with incomplete count recovery rates of 35% and 24%, respectively, and even following allogeneic SCT the 3-year survival rate was 32%. This retrospective work suggested that allogeneic SCT has a role in improving short-term survival but durability of response remains under question although the authors did acknowledge that the overall cohort of patients who underwent allogeneic SCT was small. The EBMT group published data on 46 patients who underwent allogeneic SCT for blast phase myelofibrosis and confirmed the importance of achieving complete remission prior to transplantation, which was also demonstrated in a retrospective collaborative French study.56,57 The transplant-related mortality rate at 1 year was acceptable at 28% and the 3-year progression-free and overall survival rates were 26% and 33%, respectively. The impact of both karyotype and mutational landscape on determining outcomes following transplantation for blast phase myelofibrosis is an area of active research and consideration should be given to post-transplant maintenance strategies.

Post-transplant outcomes and complications Impact of mutations on transplant outcome Early studies investigating the impact of the mutational landscape on the outcomes of allogeneic SCT yielded conflicting results.32,58 Retrospective studies revealed a significant adverse impact of lack of the JAK2V617F mutation on overall survival after allogeneic SCT for myelofibrosis whereas a CALR mutation was associated with improved overall survival and less NRM.59,60 The impact of more extensive mutational profiling was recently investigated in 665

D.P. McLornan et al.

a large retrospective study.61 A targeted 16-gene panel was used to analyze a total of 169 patients (110 with primary myelofibrosis, 46 with secondary myelofibrosis and 13 in transformation). Multivariate analysis revealed that the presence of a CALR mutation was associated with favorable overall survival (HR: 0.448; P=0.03) and progressionfree survival (HR: 0.393; P=0.01), and reduced NRM (HR, 0.415; P=0.05). Additionally, both IDH2 and ASXL1 mutations were independent prognostic risk factors for reduced progression-free survival (HR: 5.451; P=0.002 and HR: 1.53; P=0.008, respectively). Interestingly, ‘triple negativity’ did not appear to have a significant effect on posttransplant outcomes in this cohort although the numbers analyzed were small. Overall, where possible and dependent on local facilities, we recommend that a targeted gene panel is used to determine the mutational risk profile.

molecular profiling revealed JAK2V617F (n=101), MPL (n=4) and CALR (n=31) mutations.66 Individuals with a detectable mutation at either day +100 or day +180 had a significantly higher risk of clinical relapse at 5 years compared to those in whom molecular studies were negative (62% versus 10%, P<0.001 and 70% versus 10%, P<0.001, respectively). Multivariate modeling revealed that detectable MRD at day +180 and high-risk disease status were significant factors associated with higher relapse rates. MRD status and chimerism studies should be used in tandem to determine whether donor lymphocyte infusions are required. Kroger et al. described the use of both pre-emptive (n=8) and salvage donor lymphocyte infusion (n=9) regimens in 17 myelofibrosis patients undergoing allogeneic SCT, highlighting the utility of MRD monitoring for guiding donor lymphocyte infusions and that pre-emptive strategies take precedence over salvage approaches.67

Poor graft function Despite initial engraftment, it is well established that poor graft function can be problematic and patients may remain transfusion dependent and/or require growth factor support for a considerable period. Alchalby et al. reported that the cumulative incidence of poor graft function was 17% in a cohort of 100 RIC allogeneic SCT patients and that the median onset of this complication was day +49 (range, 24-99 days).62 Poor graft function was defined by either two or three cytopenias (hemoglobin <100 g/L, neutrophil count <1.0×109/L, platelet count <30×109/L) at day +30 after allogeneic SCT, with transfusion requirements in the presence of complete donor chimerism and an absence of severe GvHD/disease relapse. Persistence of significant splenomegaly at day +30 remained a significant risk factor for poor graft function and univariate analysis revealed an association with older recipient age, perhaps reflecting age-related changes in a hostile bone marrow niche. Poor graft function did not appear to influence survival. Of relevance, Hart et al. recently compared engraftment kinetics in a small cohort of acute myeloid leukemia and myelofibrosis allogeneic SCT recipients.63 Compared to the group with acute myeloid leukemia, the myelofibrosis patients had marked early clearance of hematopoietic stem cells due to early splenic pooling accompanied by a significant reduction in VCAM1 expression in the bone marrow niche which may well explain, in part, the observed early poor graft function. Whether JAK inhibitor-mediated reductions in splenomegaly will result in lower incidences of poor graft function remains undetermined. Judicious monitoring of chimerism and appropriate growth factor and transfusion support are necessitated by the recognition that some patients may require stem cell top up or consideration given to splenectomy if bulky splenomegaly persists.

Measurable residual disease monitoring and donor lymphocyte infusion strategies Measurable residual disease (MRD) monitoring to guide weaning of immunosuppression with or without utilization of immunotherapeutic strategies has become well established in the setting of allogeneic SCT for myelofibrosis. Alchalby et al. demonstrated that clearance of detectable JAK2V617F after allogeneic SCT was associated with a significantly reduced risk of relapse.59 Both MPL and CALR mutations can be used as markers of MRD.64-65 More recently, the significance of MRD was evaluated in a large, single-center cohort (n=136), in which pre-transplant 666

Relapse following allogeneic stem cell transplantation: is this still a significant clinical problem? Relapse remains a significant issue following allogeneic SCT for myelofibrosis. Longer term follow-up of the EBMT prospective study described above suggested relapse rates of up to 25% at 5 years.32 A more recent EBMT retrospective study by McLornan et al. investigated the management and outcomes of 202 relapsed patients.68 The overall median time to relapse was relatively short at 7 months (range, 1.4-111). Patients who relapsed early had significantly worse outcomes than those of patients who relapsed later. Management approaches to the relapse episode were heterogeneous and direct comparisons were not possible; however, there was a suggested benefit from adoptive immunotherapeutic approaches with donor lymphocyte infusions and/or a second allograft. There is some experience on the use of using JAK inhibitors following relapse to bridge to a second transplant and for symptom control but no improvements in donor chimerism or reductions in JAK2 allelic burden, where applicable, were seen.69 Pre-emptive use of JAK inhibitors in the post-transplant setting to reduce relapse incidence is of interest and requires prospective studies.

Conclusions Rapid advances in the availability of novel therapeutic agents for myelofibrosis has made it increasingly complex to gauge the timing and sequencing of allogeneic SCT in the patient’s care process. Nonetheless, allogeneic SCT remains the only curative approach for transplant-eligible patients. All available prognostic information and recent scoring systems should be utilized, including comprehensive mutational profiling where available, to stratify realistic patient-specific prognostic outcomes and assess the risk-benefit ratio. Changes in practice are required so that potential candidates are assessed by transplant physicians at an earlier stage, even if this is only to discuss and provide information about the procedure, assess fitness and identify potential donors. JAK inhibitors have become an integral part of the pre-allogeneic SCT pathway for many patients and increasing data on outcome analysis are now available. Overall, the use of these drugs prior to allogeneic SCT appears to be safe, with no adverse effects on engraftment; tapering is appropriate leading up to conditioning therapy and there is emerging data concerning haematologica | 2019; 104(4)

Allogeneic SCT for myelofibrosis

potential reductions in the risks of GvHD and relapse, although these need confirmation. Alertness to potential infectious complications is warranted. How the role of other novel agents will fit into the transplant paradigm

References 1. James C, Ugo V, Le Couédic JP, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera. Nature. 2005;434(7037):1144-1148 2. Kralovics R, Passamonti F, Buser AS, et al. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med. 2005;352(17):1779-1790. 3. Levine RL, Wadleigh M, Cools J, et al. Activating mutation in the tyrosine kinase JAK2 in polycythemia vera, essential thrombocythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell. 2005;7(4),387–97 4. Baxter EJ, Scott LM, Campbell PJ, et al. Cancer Genome Project. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. Lancet. 2005;365 (9464):1054-1061. 5. Harrison C, Kiladjian JJ, Al-Ali HK, et al. JAK inhibition with ruxolitinib versus best available therapy for myelofibrosis. N Engl J Med. 2012;366(9):787–798 6. Verstovsek S, Mesa RA, Gotlib J, et al. A double-blind, placebo-controlled trial of ruxolitinib for myelofibrosis. N Engl J Med. 2012;366(9):799–807. 7. Bose P, Verstovsek S. Developmental therapeutics in myeloproliferative neoplasms. Clin Lymphoma Myeloma Leuk. 2017;17S:S43-S52. 8. Cervantes F, Dupriez B, Pereira A, et al. New prognostic scoring system for primary myelofibrosis based on a study of the International Working Group for Myelofibrosis Research and Treatment. Blood. 2009;113(13),2895-2901. 9. Passamonti F, Cervantes F, Vannucchi AM, et al. A dynamic prognostic model to predict survival in primary myelofibrosis: a study by the IWG-MRT (International Working Group for Myeloproliferative Neoplasms Research and Treatment). Blood. 2010;115(9):1703-1708. 10. Gangat N, Caramazza D, Vaidya R, et al. DIPSS-Plus: a refined Dynamic International Prognostic Scoring System (DIPSS) for primary myelofibrosis that incorporates prognostic information from karyotype, platelet count and transfusion status. J Clin Oncol. 2011;29(4):392-397. 11. Hussein K, Pardanani AD, Van Dyke DL, Hanson CA, Tefferi A. International Prognostic Scoring System-independent cytogenetic risk categorization in primary myelofibrosis. Blood. 2010;115(3):496-499. 12. Kröger NM, Deeg JH, Olavarria E, et al. Indication and management of allogeneic stem cell transplantation in primary myelofibrosis: a consensus process by an EBMT/ELN international working group. Leukemia. 2015;29(11):2126–2133. 13. Kröger N, Giorgino T, Scott BL, et al. Impact of allogeneic stem cell transplantation on survival of patients less than 65 years of age with primary myelofibrosis. Blood. 2015;125(21):3347-3350 14. Grinfeld J, Nangalia, J, Baxter EJ, et al. Personalized prognostic predictions for Patients with myeloproliferative neoplasms

haematologica | 2019; 104(4)

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

remains unknown. Current challenges regarding allogeneic SCT are to reduce the relapse risk and rates of NRM, particularly when using alternative donors, and optimization of graft function.

through integration of comprehensive genomic and clinical information. Blood. 2017;130(Suppl 1):491. Passamonti F, Giorgino T, Mora B, et al. A clinical-molecular prognostic model to predict survival in patients with post polycythemia vera and post essential thrombocythemia myelofibrosis. Leukemia. 2017;31 (12):2726-2731. Scott BL, Gooley T, Linenberger M, et al. International Working Group scores predict post-transplant outcomes in patients with myelofibrosis. Blood. 2010;116(21):3085. Scott BL, Gooley TA, Sorror M, et al. The Dynamic International Prognostic Scoring System for myelofibrosis predicts outcomes after hematopoietic cell transplantation. Blood. 2012;119(11):2657-2664 Gagelmann N, Eikema DJ, Koster L, et al. Comparative performance of prognostic systems in patients with myelofibrosis secondary to PV and ET after allogeneic stem-cell transplantation. EHA Meeting 2018;Abstract S129. Rumi E, Pietra D, Pascutto C, et al. Clinical effect of driver mutations of JAK2, CALR, or MPL in primary myelofibrosis. Blood. 2014;124(7):1062-1069. Tefferi A, Guglielmelli P, Lasho TL, et al. CALR and ASXL1 mutations-based molecular prognostication in primary myelofibrosis: an international study of 570 patients. Leukemia. 2014;28(7):1494-1500. Tefferi A, Lasho TL, Finke C, et al. Type 1 vs type 2 calreticulin mutations in primary myelofibrosis: differences in phenotype and prognostic impact. Leukemia. 2014;28(7): 1568-1570 Guglielmelli P, Lasho TL, Rotunno G, et al. The number of prognostically detrimental mutations and prognosis in primary myelofibrosis: an international study of 797 patients. Leukemia. 2014;28(9):1804-1810. Guglielmelli P, Lasho TL, Rotunno G, et al. MIPSS70: Mutation-enhanced International Prognostic Score System for transplantationage patients wth primary myelofibrosis. J Clin Oncol. 2018;36(4):310-318. Tefferi A, Guglielmelli P, Lasho TL, et al. MIPSS70+ version 2.0: mutation and karyotype-enhanced International Prognostic Scoring System for primary myelofibrosis. J Clin Oncol. 2018;36(17): 1769-1770. Guardiola P, Anderson J.E, Bandini G, et al. Allogeneic stem cell transplantation for agnogenic myeloid metaplasia: a European Group for Blood and Marrow Transplantation, Societe Francaise de Greffe de Moelle, Gruppo Italiano per il Trapianto del Midollo Osseo, and Fred Hutchinson Cancer Research Center collaborative study. Blood. 1999;93(9):2831–2838. Nivison-Smith I, Dodds AJ, Butler J, et al. Allogeneic hematopoietic cell transplantation for chronic myelofibrosis in Australia and New Zealand: older recipients receiving myeloablative conditioning at increased mortality risk. Biol Blood Marrow Transplant. 2012;18(2):302 –308. Abelsson J, Merup M, Birgegard G, et al. The outcome of allo-HSCT for 92 patients with

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

myelofibrosis in the Nordic countries. Bone Marrow Transplant. 2011;47(3):380–386. Samuelson S, Sandmaier B.M, Heslop HE, et al. Allogeneic haematopoietic cell transplantation for myelofibrosis in 30 patients 60-78 years of age. Br J Haematol. 2011;153(1):76– 82. Muffly LS, Kocherginsky M, Stock W, et al. Geriatric assessment to predict survival in older allogeneic hematopoietic cell transplantation recipients. Haematologica. 2014;99(8): 1373-1379 Robin M, Tabrizi R, Mohty M, et al. Allogeneic haematopoietic stem cell transplantation for myelofibrosis: a report of the Société Française de Greffe de Moelle et de Thérapie Cellulaire (SFGM-TC). Br J Haematol. 2011;152(3):331-339. Mesa R.A, Nagorney D.S, Schwager S, Allred J, Tefferi A. Palliative goals, patient selection, and perioperative platelet management: outcomes and lessons from 3 decades of splenectomy for myelofibrosis with myeloid metaplasia at the Mayo Clinic. Cancer. 2006;107(2):361–370. Kroger N, Holler E, Kobbe G, et al. Allogeneic stem cell transplantation after reduced-intensity conditioning in patients with myelofibrosis: a prospective, multicenter study of the Chronic Leukemia Working Party of the European Group for Blood and Marrow Transplantation. Blood. 2009;114 (26):5264–5270 Robin M, Zine M, Chevret S, et al. The impact of splenectomy in myelofibrosis patients before allogeneic hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2017;23(6):958-964. McLornan D, Szydlo R, Robin M, et al. Outcome of myeloablative and reducedintensity conditioned allogeneic haematopoietic stem cell transplantation in myelofibrosis: a retrospective study by the Chronic Malignancies Working Party of the EBMT. Annual EBMT meeting. 2018; OS 5-5 Ballen KK, Shrestha S, Sobocinski KA, et al. Outcome of transplantation for myelofibrosis. Biol Blood Marrow Transplant. 2010;16(3):358–367. Matsubara E, Yamanouchi J, Kitazawa R, et al. Usefulness of low-dose splenic irradiation prior to reduced-intensity conditioning regimen for hematopoietic stem cell transplantation in elderly patients with myelofibrosis. Case Rep Hematol. 2016;2016:8751329. Kalman NS, Mukhopadhyay ND, Roberts CH, et al. Low-dose splenic irradiation prior to hematopoietic cell transplantation in hypersplenic patients with myelofibrosis. Leuk Lymphoma. 2017;58(12):2983-2984. Jaekel N, Behre G, Behning A, et al. Allogeneic hematopoietic cell transplantation for myelofibrosis in patients pretreated with the JAK1 and JAK2 inhibitor ruxolitinib. Bone Marrow Transplant. 2014;49(2): 179184. Stübig T, Alchalby H, Ditschkowski M, et al. JAK inhibition with ruxolitinib as pretreatment for allogeneic stem cell transplantation in primary or post-ET/PV myelofibrosis. Leukemia. 2014;28(8):1736-1738

667

D.P. McLornan et al. 40. Gupta V, Kosiorek HE, Mead A, Klisovic RB, et al. Ruxolitinib therapy followed by reduced-intensity conditioning for hematopoietic cell transplantation for myelofibrosis: myeloproliferative disorders research Consortium 114 study. Biol Blood Marrow Transplant. 2019;25(2):256-264. 41. Robin M, Francois S, Huynh A, et al. Ruxolitinib before allogeneic hematopoietic stem cell transplantation (HSCT) in patients with myelofibrosis: a preliminary descriptive report of the JAK ALLO study, a phase II trial sponsored by Goelams-FIM in collaboration with the SFGMTC. Blood, 2013;122(21):306. 42. Shanavas M, Popat U, Michaelis LC, et al. Outcomes of allogeneic hematopoietic cell transplantation in patients with myelofibrosis with prior exposure to Janus kinase 1/2 inhibitors. Biol Blood Marrow Transplant. 2016;22(3):432-440. 43. Shahnaz Syed Abd Kadir S, Christopeit M, Gerald W, et al. Impact of ruxolitinib pretreatment on outcomes after allogeneic stem cell transplantation in patients with myelofibrosis. Eur J Haematol. 2018;101(3):305-317. 44. Kröger N, Shahnaz Syed Abd Kadir S, Zabelina T, et al. Peritransplantation ruxolitinib prevents acute graft-versus-host disease in patients with myelofibrosis undergoing allogenic stem cell transplantation. Biol Blood Marrow Transplant. 2018;24(10): 2152-2156. 45. Zeiser R, Burchert A, Lengerke C, et al. Ruxolitinib in corticosteroid-refractory graftversus-host disease after allogeneic stem cell transplantation: a multicenter survey. Leukemia. 2015;29(10):2062-2068. 46. Kerbauy DM, Gooley TA, Sale GE, et al. Hematopoietic cell transplantation as curative therapy for idiopathic myelofibrosis, advanced polycythemia vera, and essential thrombocythemia. Biol Blood Marrow Transplant. 2007;13(3):355-365. 47. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant. 2009;15(12):1628–1633. 48. Robin M, Porcher R, Wolschke C, et al. Outcome after transplantation according to reduced-intensity conditioning regimen in patients undergoing Transplantation for Myelofibrosis. Biol Blood Marrow Transplant. 2016;22(7):1206-1211. 49. Rondelli D, Goldberg JD, Isola L, et al. MPDRC 101 prospective study of reduced-intensity allogeneic hematopoietic stem cell transplantation in patients with myelofibrosis. Blood. 2014;124(7):1183-1191. 50. Patriarca F, Bacigalupo A, Sperotto A, et al. Allogeneic hematopoietic stem cell transplantation in myelofibrosis: the 20-year experience of the Gruppo Italiano Trapianto di Midollo Osseo (GITMO). Haematologica. 2008;93(10):1514–1522. 51. Robin M, Chevret S, Koster L, et al. In vivo T cell depletion in patients with myelofibrosis transplanted from an HLA-matched sibling donor: an EBMT study. 44th Annual Meeting of the European Society for Blood and Marrow Transplantation: Physicians Oral Session: Bone Marrow Transplant. 2018 Abstract supplements. 52. Raj K, Eikema DJ, McLornan DP, et al. Family mismatched allogeneic stem cell transplanta-

668

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

tion for myelofibrosis: report from the Chronic Malignancies Working Party of EBMT. Biol Blood Marrow Transplant. 2018 Nov5. [Epub ahead of print] Takagi S, Ota Y, Uchida N, et al. Successful engraftment after reduced intensity umbilical cord blood transplantation for myelofibrosis. Blood. 2010;116(4):649–652. Robin M, Giannotti F, Deconinck E, et al. Unrelated cord blood transplantation for patients with primary or secondary myelofibrosis. Biol Blood Marrow Transplant. 2014;20(11):1841-1846. Tefferi A, Mudireddy M, Mannelli F, et al. Blast phase myeloproliferative neoplasm: Mayo-AGIMM study of 410 patients from two separate cohorts. Leukemia. 2018; 32(5):1200-1210. Alchalby H, Zabelina T, Stübig T, et al. Allogeneic stem cell transplantation for myelofibrosis with leukemic transformation: a study from the Myeloproliferative Neoplasm Subcommittee of the CMWP of the European Group for Blood and Marrow Transplantation. Biol Blood Marrow Transplant. 2014;20(2): 279-281. Cahu X, Chevallier P, Clavert A, et al. AlloSCT for Philadelphia-negative myeloproliferative neoplasms in blast phase: a study from the Societe Française de Greffe de Moelle et de Therapie Cellulaire (SFGM-TC). Bone Marrow Transplant. 2014;49(6):756-760. Ditschkowski M, Elmaagacli AH, Trenschel R, Steckel NK, Koldehoff M, Beelen DW. No influence of V617F mutation in JAK2 on outcome after allogeneic hematopoietic stem cell transplantation (HSCT) for myelofibrosis. Biol Blood Marrow Transplant. 2006;12(12):1350–1351. Alchalby H, Badbaran A, Zabelina T, et al. Impact of JAK2V617F mutation status, allele burden, and clearance after allogeneic stem cell transplantation for myelofibrosis. Blood. 2010;116(18):3572–3581. Panagiota V, Thol F, Markus B, et al. Prognostic effect of calreticulin mutations in patients with myelofibrosis after allogeneic hematopoietic stem cell transplantation. Leukemia. 2014;28(7):1552-1555. Kröger N, Panagiota V, Badbaran A, et al. Impact of molecular genetics on outcome in myelofibrosis patients after allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2017;23(7):1095-1101. Alchalby H, Yunus DR, Zabelina T, Ayuk F, Kröger N. Incidence and risk factors of poor graft function after allogeneic stem cell transplantation for myelofibrosis. Bone Marrow Transplant. 2016;51(9):1223-1227. Hart C, Klatt S, Barop J, et al. Splenic pooling and loss of VCAM-1 causes an engraftment defect in patients with myelofibrosis after allogeneic hematopoietic stem cell transplantation. Haematologica 2016;101 (11):1407-1416. Alchalby H, Badbaran A, Bock O, et al. Screening and monitoring of MPL W515L mutation with real-time PCR in patients with myelofibrosis undergoing allogeneicSCT. Bone Marrow Transplant. 2010;45(9): 1404–1407 Rumi E, Passamonti F, Arcaini L, et al. Molecular remission after allo-SCT in a patient with post-essential thrombo-

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

cythemia myelofibrosis carrying the MPL (W515A) mutation. Bone Marrow Transplant. 2010;45(4):798–800. Wolschke C, Badbaran A, Zabelina T, et al. Impact of molecular residual disease post allografting in myelofibrosis patients. Bone Marrow Transplant. 2017;52(11):1526-1529. Kroger N, Alchalby H, Klyuchnikov E, et al. JAK2-V617F- triggered pre-emptive and salvage adoptive immunotherapy with donorlymphocyte infusion in patients with myelofibrosis after allogeneic stem cell transplantation. Blood. 2009;113(8):1866– 1868. McLornan DP, Szydlo R, Robin M, et al. Outcome of patients with myelofibrosis relapsing after allogeneic stem cell transplant: a retrospective study by the Chronic Malignancies Working Party of EBMT. Br J Haematol. 2018 May 29. [Epub ahead of print] Janson D, Ayuk F. A, Wolschke C, et al. Ruxolitinib for myelofibrosis patients relapsing after allogeneic hematopoietic transplantation. Blood. 2016;128(22):1948. Lasho TL, Jimma T, Finke CM, et al. SRSF2 mutations in primary myelofibrosis: significant clustering with IDH mutations and independent association with inferior overall and leukemia-free survival. Blood. 2012;120(20):4168-4171. Tefferi A, Finke CM, Lasho TL et al. U2AF1 mutations in primary myelofibrosis are strongly associated with anemia and thrombocytopenia despite clustering with JAK2V617F and normal karyotype. Leukemia. 2014;28(2):431-433. Tefferi A, Lasho TL, Finke CM, et al. CALR vs JAK2 vs MPL-mutated or triple-negative myelofibrosis: clinical, cytogenetic and molecular comparisons. Leukemia. 2014;28(7):1472-1477. Lasho TL, Finke CM, Hanson CA, et al. SF3B1 mutations in primary myelofibrosis: clinical, histopathology and genetic correlates among 155 patients. Leukemia. 2012;26(5):1135-1137. Tefferi A, Lasho TL, Finke C, et al. Prognostic significance of ASXL1 mutation types and allele burden in myelofibrosis. Leukemia. 2018;2(3):837-839. Guglielmelli P, Biamonte F, Score J, et al. EZH2 mutational status predicts poor survival in myelofibrosis. Blood. 2011;118(19): 5227-5234. Tefferi A, Jimma T, Sulai NH, et al. IDH mutations in primary myelofibrosis predict leukemic transformation and shortened survival: clinical evidence for leukemogenic collaboration with JAK2V617F. Leukemia. 2012;26(3):475-480. Gupta V, Malone AK, Hari PN, et al. Reduced-intensity hematopoietic cell transplantation for patients with primary myelofibrosis: a cohort analysis from the center for international blood and marrow transplant research. Biol Blood Marrow Transplant. 2014;20(1):89-97. Jain T, Kunze KL, Temkit M, et al. Comparison of reduced intensity conditioning regimens used in patients undergoing hematopoietic stem cell transplantation for myelofibrosis. Bone Marrow Transplant. 2019;54(2):204-211.

haematologica | 2019; 104(4)

ARTICLE

Hematopoiesis

In vitro and in vivo evaluation of possible pro-survival activities of PGE2, EGF, TPO and FLT3L on human hematopoiesis

Ferrata Storti Foundation

Eva-Maria Demmerath,1,* Sheila Bohler,1,2,* Mirjam Kunze3 and Miriam Erlacher1,4

Department of Pediatrics and Adolescent Medicine, Division of Pediatric Hematology and Oncology, University Medical Center Freiburg, Faculty of Medicine, University of Freiburg; 2Faculty of Biology, University of Freiburg; 3Department of Obstetrics and Gynecology, University Medical Center of Freiburg and 4German Cancer Consortium (DKTK), Freiburg and German Cancer Research Center (DKFZ), Heidelberg, Germany 1

*E-MD and SB contributed equally to this work.

Haematologica 2019 Volume 104(4):669-677

ABSTRACT

yelosuppression is a major and frequently dose-limiting side effect of anticancer therapy and is responsible for most treatment-related morbidity and mortality. In addition, repeated cycles of DNA damage and cell death of hematopoietic stem and progenitor cells, followed by compensatory proliferation and selection pressure, lead to genomic instability and pave the way for therapy-related myelodysplastic syndromes and secondary acute myeloid leukemia. Protection of hematopoietic stem and progenitor cells from chemo- and radiotherapy in patients with solid tumors would reduce both immediate complications and long-term sequelae. Epidermal growth factor (EGF) and prostaglandin E2 (PGE2) were reported to prevent chemo- or radiotherapy-induced myelosuppression in mice. We tested both molecules for potentially protective effects on human CD34+ cells in vitro and established a xenograft mouse model to analyze stress resistance and regeneration of human hematopoiesis in vivo. EGF was neither able to protect human stem and progenitor cells in vitro nor to promote hematopoietic regeneration following sublethal irradiation in vivo. PGE2 significantly reduced in vitro apoptotic susceptibility of human CD34+ cells to taxol and etoposide. This could, however, be ascribed to reduced proliferation rather than to a change in apoptosis signaling and BCL-2 protein regulation. Accordingly, 16,16-dimethyl-PGE2 (dmPGE2) did not accelerate regeneration of the human hematopoietic system in vivo. Repeated treatment of sublethally irradiated xenograft mice with known antiapoptotic substances, such as human FLT3L and thrombopoietin (TPO), which suppress transcription of the proapoptotic BCL-2 proteins BIM and BMF, also only marginally promoted human hematopoietic regeneration in vivo.

Introduction Myelosuppression occurs transiently after intensive chemotherapy and radiotherapy, and is characterized by anemia, bleeding tendency and susceptibility to infection. It is a major and frequently dose-limiting side effect of anticancer therapy and is responsible for most treatment-related morbidity and mortality.1 In addition, repeated cycles of DNA damage, cell attrition and subsequent compensatory proliferation of hematopoietic stem and progenitor cells (HSPCs) lead to genomic instability and pave the way for late complications such as therapy-related myelodysplastic syndromes (t-MDS) and secondary acute myeloid leukemia (AML).2 Protection of HSPCs from chemo- and radiotherapy in patients with solid tumors would reduce both immediate complications and long-term sequelae. Numerous endogenous pathways and chemical compounds have been reported to prevent chemo- or radiotherapy-induced myelosuppression in mice, either by haematologica | 2019; 104(4)

Correspondence: MIRIAM ERLACHER miriam.erlacher@uniklinik-freiburg.de Received: February 18, 2018. Accepted: November 14, 2018. Pre-published: November 15, 2018. doi:10.3324/haematol.2018.191569 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/669 ÂŠ2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

669

E.-M. Demmerath et al.

protecting HSPCs from damage and preventing their cell death or by fostering subsequent hematopoietic regeneration.3-10 In contrast, only limited preclinical data on substances protective for the human hematopoietic system in vivo are available.11,12 The aim of this study was to identify substances that protect human HSPCs from irradiation-induced apoptosis in vivo and to delineate their effects on the BCL-2 protein family. BCL-2 proteins are the master regulators of the intrinsic apoptosis pathway and have either pro- or antiapoptotic function. Anti-apoptotic BCL-2 proteins (i.e. BCL-2, BCL-XL, MCL-1 and A1/BFL) protect cells from apoptotic stimuli by binding and inactivating their proapoptotic antagonists. The pro-apoptotic family members can be subdivided into the downstream ‘effector’ proteins, BAK and BAX, and the BH3-only proteins (e.g. BIM, PUMA, BMF, BAD and others) that act upstream as cell stress sensors. Upon activation, BH3-only proteins activate BAX and BAK either directly or indirectly through inhibition of the anti-apoptotic BCL-2 proteins. BAX/BAK activation leads to outer mitochondrial membrane permeabilization, caspase activation and cell death.13 Radiotherapy as well as most conventional chemotherapeutic drugs converge at the level of BCL-2 proteins and engage the intrinsic apoptosis pathway.2 A particularly attractive candidate for our study was the epidermal growth factor (EGF) that was recently described to prevent irradiation-induced apoptosis of murine HSPCs in vivo.10 Mechanistically, EGF receptor was up-regulated on bone marrow HSPCs subjected to irradiation, and binding of EGF resulted in suppression of p53-mediated transcriptional activation of the pro-apoptotic BCL-2 protein PUMA,10 the p53 target responsible for most DNA damage-induced apoptosis in hematopoietic cells.14,15 A second candidate was prostaglandin E2 (PGE2), which was shown to have multiple beneficial effects on both murine and human HSPCs, including increased survival, self-renewal and homing, together resulting in increased long-term repopulation potential.5,16,17 In a mouse model of sublethal total body irradiation (TBI), treatment with the long-acting PGE2 analog, 16,16-dimethyl-PGE2 (dmPGE2), resulted in increased HSPC survival and accelerated hematopoietic regeneration.9 The increase in apoptosis resistance was explained by upregulation of BCL-2 and BCL-XL and reduced expression of BAX.17 Finally, the cytokines FLT3L, stem cell factor (SCF) and thrombopoietin (TPO) have protective effects on murine and human HSPCs and drive their proliferation in vitro.3 Together, these cytokines are frequently used for culture and ex vivo expansion of human CD34+ cells. We have shown earlier that their pro-survival activity can be attributed to reduced transcription of BIM and BMF mRNA.18 None of these molecules have been tested yet for possible protective effects on human hematopoiesis in vivo. We focused on EGF, PGE2, FLT3L and TPO but excluded SCF from our studies since it had been shown earlier to induce extensive proliferation and premature exhaustion in murine HSPCs.19 We analyzed the effects of these substances on human HSPCs subjected to cell stress in vitro and, in addition, developed a xenograft model to analyze stress resistance and regeneration of human hematopoiesis in vivo. Despite promising data obtained in the above-described mouse model, EGF was not able to protect human HSPCs in vitro nor to promote hematopoietic regeneration follow670

ing sublethal irradiation in vivo. PGE2 significantly reduced in vitro apoptotic susceptibility of human HSPCs to taxol and etoposide. This could, however, be ascribed to reduced proliferation rather than to a change in BCL-2 protein regulation. Accordingly, PGE2 did not accelerate regeneration of the human hematopoietic system in vivo. Repeated treatment of sublethally irradiated xenograft mice with the combination of FLT3L and TPO also resulted in only minor beneficial effects during human hematopoietic regeneration.

Methods Cell isolation and culture Human umbilical cord blood was obtained after caesarean birth. Informed consent was obtained from the parents and the study was approved by the local ethics committee. CD34+ cells were enriched by MACS-technology (Miltenyi), cell purity was generally more than 90%. Purified cells were frozen in CryoStor CS10 (Stem Cell Technologies), stored in liquid nitrogen and used at later time points. Thawed cells were cultured in serum-free medium supplemented with 10% ES-FBS (Invitrogen), human TPO (50 ng/mL, Immunotools), FLT3L, SCF, IL3 (100 ng/mL each, Immunotools), human EGF (hEGF) (20 or 200 ng/mL Immunotools), PGE2 (10, 25 or 50 mM, Sigma-Aldrich) and/or cytotoxic drugs (etoposide, taxol, tunicamycin; Sigma-Aldrich). Alternatively, thawed cells were used for xenotransplantation.

Xenotransplantation All experiments were performed according to the guidelines of the German "Tierversuchsgesetz" and approved by the local committee (RP Freiburg/Germany). Rag2-/-γc-/- mice were irradiated at five weeks of age with 3 Gy and 6-8 hours (h) later they were injected intravenously into the retrobulbar venous plexus with 3x105 human CD34+ cells. Four weeks later, animals were irradiated again. Subsequently, xenograft mice were treated once daily intraperitoneally (i.p.) with human EGF (0.5 mg/g body weight), murine EGF (0.5 µg/g), human dmPGE2 (2 mg/g), human FLT3L (40 ng/g), human TPO (40 ng/g), combinations thereof, or respective carrier solutions (Figure 1). At indicated time points, mice were sacrificed for analysis. Alternatively, mice were treated once daily for seven days with etoposide (20 mg/k, i.p.), and the anti-apoptotic substances were given simultaneously.

Proliferation, apoptosis and colony formation assays Cell cycle status and proliferation were determined by double staining for Ki-67 (BioLegend) and DAPI (Sigma-Aldrich) or incubation with CFSE (1 mM; Sigma-Aldrich). Apoptosis was determined by combined staining with 7-AAD and Annexin-V. Specific apoptosis triggered by stress was calculated as follows: (induced apoptosis – spontaneous apoptosis)/(100 – spontaneous apoptosis). For colony forming assays, 150,000 human CD45+ cells isolated from murine bone marrow (BM) were plated for 11 days on a semi-solid medium containing insulin, transferrin, human SCF, IL-3, IL-6, EPO, G-CSF and GM-CSF (SF H4436 MethoCult).

Flow cytometric analysis Single cell suspensions of hematopoietic organs were surfacestained with antibodies conjugated with FITC, PE, APC, PerCP/Cy5.5, PE/Cy7 or biotin. Antibodies for murine markers: anti-CD45 (30-F11). Antibodies for human markers: H13 and 2D1, anti-CD45; AC136, anti-CD34. Biotinylated antibodies were detected using streptavidin-APC. Flow cytometric analysis was performed using a FACS-Fortessa (BD). haematologica | 2019; 104(4)

PGE2, EGF, FLT3L and TPO in human hematopoiesis

Figure 1. Xenograft model for evaluation of radioprotective substances. Cord blood-derived human CD34+ cells were transplanted into sublethally irradiated 5-week old Rag2−/−γc−/− mice. Four weeks later, xenograft mice were again irradiated with 3 Gy in order to subject human hematopoiesis to sublethal stress. Subsequently, mice were treated intraperitoneally (i.p.) once daily with the indicated molecules. Control mice were treated with the respective carrier solution (saline or ethanol). At day 8 after second irradiation, mice were sacrificed for analysis. Single cell suspensions were obtained from bone marrow and spleen. h: hours; hu EGF: human epidermal growth factor; mu EGF: murine epidermal growth factor; hu dmPGE2: human 16,16-dimethyl-PGE2; hu FLT3L: human FLT3L; TPO: human thrombopoietin.

RT-MLPA RNA was isolated with Fast-Spin columns (ZymoResearch). For RT-MLPA (MRC Holland, R011-C1), specific mRNAs were reversely transcribed into cDNA and bound by two oligonucleotides consequently ligated. The generated amplification products of unique length were separated by capillary sequencer (Genescan). Analysis was performed with Sequence Pilot (JSI Medical Systems). The sum of all peak data was set to 100% to normalize for fluctuations between different samples, and single peaks were calculated relative to 100%.

Statistical analysis Statistical analysis was performed using the Mann-WhitneyTest (Graphpad Prism). P<0.05 was considered statistically significant.

Results Epidermal growth factor does not protect human hematopoietic stem cells from DNA damage-induced apoptosis in vitro and in vivo To induce DNA damage-induced apoptosis, CD34+ cells were treated with the topoisomerase inhibitor etoposide (0.5 or 1 mg/mL) for 24 and 48 h. Apoptosis could not be prevented by addition of 20 or 200 ng/mL human EGF, respectively (Figure 2A). To exclude the possibility that EGF contained in medium or serum or produced by the cells themselves was sufficient to reduce apoptotic susceptibility even in the absence of accessory EGF, cells were treated with neutralizing EGF-R-antibodies (cetuximab) for 24 and 48 h. Inhibition of EGF-R-signaling did not induce apoptosis by itself (Figure 2B) nor increase DNA damage-induced apoptosis of human HSPCs (Figure 2C). Since the effects of EGF on murine hematopoiesis were exclusively shown in vivo,10 we performed analogous experiments in a xenograft model using immunodeficient Rag2-/-γ-/- recipient mice. Animals were xenotransplanted at five weeks of age with 3x105 cord blood-derived CD34+ haematologica | 2019; 104(4)

cells. Upon successful human engraftment, mice were irradiated with 3 Gy to mimic myelosuppression occurring during therapeutic irradiation. Subsequently, mice were treated daily with human EGF (0.5 mg/g, i.p.) or saline (100 ml, i.p.) for seven days. Regeneration of human hematopoiesis, as defined by percentage and cell count of human CD45+ cells, was determined in BM and spleen seven days after irradiation (Figure 2D and E and Online Supplementary Figure S1). In addition, we determined the proportion of immature CD34+ cells within all human cells as a surrogate for their regenerative capacity (Figure 2F). While non-irradiated mice showed consistent human engraftment, irradiated mice lost most of their human hematopoietic cells irrespectively of whether they were treated with human EGF or not. To test for a possibly indirect, cell-extrinsic effect of EGF on HSPCs conferred by the murine microenvironment, we performed the same experiment using murine EGF. In addition, we combined treatment with human and murine EGF to test for synergies. None of these treatment regimens resulted in protection of the xenografted human hematopoietic system (Figure 2D and E and Online Supplementary Figure S1). As we observed no protective effects of EGF both in vitro and in vivo, we waived the analysis of BCL-2 protein regulation.

Prostaglandin E2 protects human hematopoietic stem cells short-term from apoptosis but has toxic long-term effects We subjected CD34+ cells for 48 h to various stress stimuli all known to induce intrinsic apoptosis and investigated whether PGE2 was able to reduce their apoptotic susceptibility. Apoptosis induced by the spindle drug taxol or the topoisomerase inhibitor etoposide was significantly reduced in a dose-dependent manner by PGE2. In addition, apoptosis induced by serum and cytokine deprivation was reduced when cells were cultured in the presence of PGE2, albeit not significantly. In contrast, apoptosis induced by the ER stressor tunicamycin could not be pre671

E.-M. Demmerath et al.

vented by PGE2 (Figure 3A). Altogether, the data point towards a dose-dependent protective effect of PGE2 that is, however, restricted to certain forms of cell stress. Unexpectedly, long-term culture of CD34+ cells revealed a toxic effect of PGE2 itself when it was used at high concentrations for longer than two days. Numbers and viabil-

ity of CD34+ cells were strongly reduced when they were cultured in the presence of 50 mM PGE2 for up to eight days (Figure 3B and C). To unravel the paradoxical finding that PGE2 can both protect from and induce apoptosis, we analyzed regulation of BCL-2 family members on mRNA level. After 4 h of PGE2 treatment, CD34+ cells showed

Figure 2. Epidermal growth factor (EGF) does not protect human CD34+ from apoptosis induced by sublethal irradiation. (A) Cord blood-derived human CD34+ cells were treated for 24 and 48 hours (h) with etoposide (1 µg/mL) and different doses of human EGF (hEGF). Apoptosis was measured using AnnexinV/7AAD and specific apoptosis was calculated. Bars represent means±Standard Error of Mean (SEM) of 5 independent experiments. (B) Cord blood-derived human CD34+ cells were cultured for 24 and 48 h in the presence of human EGF and/or cetuximab at indicated concentrations and specific apoptosis was determined. Bars represent means±SEM of 4 independent experiments. (C) Human CD34+ cells were treated with both etoposide and cetuximab, and specific apoptosis was determined 24 h later. Bars represent means±SEM of 3 independent experiments. (D-F) Human CD34+ cells were transplanted into sublethally irradiated Rag2−/−γc−/− mice. Four weeks later, mice were irradiated with 3 Gy or left untreated. Mice that were irradiated received daily injections of human and/or murine EGF. Eight days after irradiation, mice were sacrificed and % human CD45+ cells was determined in bone marrow (D) and spleen (E). In addition, the proportion of CD34+ immature cells was determined within the human cell population (F). Bars represent means±SEM of 4-7 animals from 5 independent experiments. Mann-Whitney test, *P≤0.05; **P≤0.01; ***P≤0.001).

672

haematologica | 2019; 104(4)

PGE2, EGF, FLT3L and TPO in human hematopoiesis cultured in the presence of 50 mM PGE2 (Figure 3B), we suspected a proliferation repression in addition to apoptosis induction. Indeed, more cells were in G0 phase when cells cultured with FLT3L, SCF, TPO and IL3 were treated with PGE2 (50 mM) (Figure 3D and Online Supplementary Figure S3). Consistently, CFSE staining showed 4-5 divisions in four days in most untreated cells but fewer divi-

increased mRNA levels of the anti-apoptotic protein MCL-1 and the pro-apoptotic protein BIM while BMF levels were reduced. mRNA levels returned to normal at 12 h of treatment. Other BCL-2 proteins were not transcriptionally regulated at either time point (Online Supplementary Figure S2A and B). As we observed no HSPC expansion when cells were

Figure 3. Prostaglandin E2 (PGE2) has both protective and toxic effects on human CD34+ cells. (A) Cord blood-derived CD34+ cells were subjected to different cytotoxic agents or serum and cytokine withdrawal. Control cells were treated with serum, FLT3L, stem cell factor (SCF), thrombopoietin (TPO) and IL3. PGE2 was added at indicated concentration. After 48 hour (h), cells were stained with AnnexinV/7AAD and specific apoptosis was determined. Bars represent means±Standard Error of Mean (SEM) of 5-6 from 6 independent experiments. P-values were determined using the Mann-Whitney test. (B and C) CD34+ cells were cultured in serum, FLT3L, SCF, TPO and IL3 plus different concentrations of PGE2. Cell count (B) and viability (C) were determined every other day for eight days. Bars represent means±SEM of 4 from 3 independent experiments. P-values were determined using the Mann-Whitney test. (D) After three days of culture in the presence or absence of PGE2 (50 mM), cell cycle status was determined by Ki67 and DAPI staining and flow cytometric analysis. Bars represent means±SEM of 6 from 5 independent experiments. Mann-Whitney test. (E) PGE2-treated CD34+ cells were cultured in the presence of carboxy-fluorescein diacetate succinimidyl ester (CFSE), and CFSE content representing amount of cell divisions was determined by flow cytometry four days later. Bars represent means ± SEM of n=3 independent experiments.

haematologica | 2019; 104(4)

673

E.-M. Demmerath et al.

sions in PGE2-treated cells (Figure 3E and Online Supplementary Figure S3B). Considering this, the reduced susceptibility of PGE2-treated HSPCs to taxol and etoposide might be rather due to the proliferation repression than to deregulated apoptosis signaling since both drugs primarily target proliferating cells. We hypothesized that PGE2-mediated inhibition of cell proliferation 24 h prior cytotoxic stress would further reduce susceptibility to apoptosis but noted no additional benefit (Online Supplementary Figure S2C).

To test whether the protective in vitro effect of PGE2 can be translated to an in vivo situation, we irradiated recipient mice sublethally four weeks after xenotransplantation. A second group of xenograft mice was subjected once daily to etoposide for one week to better mimic a chemotherapy cycle. Animals from both groups were treated daily with dmPGE2 at a dose that had been shown earlier to protect murine HSPCs from apoptosis induced by sublethal irradiation in vivo.9 Seven days after treatment, animals were sacrificed and analyzed.

Figure 4. Prostaglandin E2 (PGE2) does not promote human hematopoietic regeneration. (A and B) Human CD34+ cells were transplanted into sublethally irradiated Rag2−/−γc−/− mice. Four weeks later, mice were irradiated with 3 Gy or left untreated. Mice that were irradiated received daily injections of dmPGE2. Eight days after irradiation, mice were sacrificed and % human CD45+ cells were determined in bone marrow (BM) (A) and spleen (B). Bars represent means±Standard Error of Mean (SEM) of n=4-7 animals from 5 independent experiments. Mann-Whitney test, *P≤0.05; **P≤0.01; ***P≤0.001. (C-F) Human CD34+ cells were transplanted into sublethally irradiated Rag2−/−γc−/− mice. Four weeks later, mice were treated daily with etoposide alone, or together with dmPGE2. Eight days after treatment start, mice were sacrificed and % human CD45+ cells was determined in BM (C) and spleen (E). Cell counts of human CD45+ cells were determined in BM (D) and spleen (F). Bars represent means±SEM of 2-7 animals from at least 2 independent experiments.

674

haematologica | 2019; 104(4)

PGE2, EGF, FLT3L and TPO in human hematopoiesis

Sublethal irradiation cleared most of human hematopoietic cells and daily dmPGE2 administration did not result in their protection (Figure 4A and B). In contrast, etoposide treatment appeared more toxic for murine hematopoietic cells (Online Supplementary Figure S4A). When etoposide was combined with dmPGE2 treatment, we observed enrichment of human CD45+ cells indicating

haematologica | 2019; 104(4)

some survival advantages over murine cells (Figure 4C and E). However, this was not reflected by increased absolute numbers of human cells (Figure 4D and F). Interestingly, dmPGE2 administration together with total body irradiation or etoposide treatment provoked profuse and nearly fatal diarrhea making such an approach unusable.

Figure 5. Mild beneficial effects of FLT3L and thrombopoietin (TPO) in vivo. (A-D) Human CD34+ cells were transplanted into sublethally irradiated Rag2−/−γc−/− mice. Four weeks later, mice were irradiated with 3 Gy or left untreated. Mice that were irradiated received daily injections of human TPO and/or FLT3L. Eight days after irradiation, mice were sacrificed and % human CD45+ cells were determined in bone marrow (BM) (A) and spleen (C). Human CD45+ cell numbers were determined in BM (B) and spleen (D). Bars represent means±Standard Error of Mean (SEM) of n=4-7 animals from 3 independent experiments. Mann-Whitney test, *P≤0.05; **P≤0.01; ***P≤0.001. (E-H) Treatment with human TPO and FLT3L was given for 14 days and % human CD45+ cells was determined in BM (E) and spleen (G). Cell counts of human CD45+ cells were determined (F and H). Bars represent means±SEM of 5-11 animals from 3 independent experiments. Mann-Whitney test, *P≤0.05; **P≤0.01; ***P≤0.001.

675

E.-M. Demmerath et al.

Minor in vivo effects of the anti-apoptotic cytokines FLT3L and thrombopoietin We next wondered whether substances known to inhibit apoptosis by interfering with the BCL-2 protein family would be able to protect human hematopoiesis from irradiation-induced injury in vivo. We focused on the cytokines FLT3L and TPO that suppress activation of the BH3-only proteins BIM and BMF in both murine and human HSPCs but also have proliferative effects.18 Sublethally irradiated mice were treated daily with FLT3L, TPO or their combination. While treatment of either FLT3L or TPO alone did not have any effects on regeneration of human hematopoiesis (data not shown), combined treatment for seven days resulted in a relative increase of human cells in BM and a mildly increased number of human cells in the spleen (Figure 5A-D and Online Supplementary Figure S5A). Both effects were transient, and no increase in human cell numbers was observed when cytokine combination was administered for 14 days (Figure 5E-H). To investigate the frequency of human HSPCs giving rise to colonies, we isolated 150,000 human CD45+ cells from BM of untreated and treated mice and seeded them into methylcellulose medium. There was no significant difference in colony numbers between mice treated with cytokines and control mice (Online Supplementary Figure S5B). Finally, we started TPO/FLT3L treatment 24 h prior to irradiation, but again did not see any protective effects on human hematopoiesis in vivo (data not shown).

Discussion Although there is a strong clinical need to reduce hematologic side effects in cancer patients treated with chemoor radiotherapy, to date, only few therapeutic options are available.11,12 This is in contrast to the many chemical compounds and endogenous substances that were described to be radioprotective in vitro or in mouse models. A major limitation in translating preclinical findings to routine clinical practice is the lack of suitable model systems. While research on primates is laborious, expensive and not feasible everywhere, xenograft models that allow studies on human cells in vivo can be a good alternative. The first aim of this study was to generate a xenograft mouse model of human hematopoiesis that can be used to analyze hematopoietic regeneration following sublethal stresses. Upon successful engraftment of human hematopoietic cells, recipient mice were subjected to total body irradiation (TBI) or daily etoposide treatment. Immediately after irradiation, or in parallel to etoposide administration, treatment with possibly protective substances was initiated and continued for one week before the human hematopoiesis was analyzed in detail. Although the substances tested in this work did not perform convincingly, we successfully demonstrated that our model can be used to investigate the presumed radioprotective effects of given molecules or chemical compounds on the human hematopoietic system treated with chemoor radiotherapy in vivo. In addition, our model can be used to test novel cytotoxic drugs for their hematotoxicity by substituting TBI with repeated drug treatment. One shortcoming of our model is that the murine microenvironment does not provide optimal support to human hematopoiet676

ic cells. This should, however, be no major issue since radio- and intensive chemotherapy also harm the human hematopoietic niche, thereby imposing an additional layer of stress to HSPCs. A more relevant shortcoming is that irradiation and cytotoxic drugs result in depletion of both human and murine cells, implicating that human cells have to compete against the more dominant murine cells during the stage of subsequent regeneration. Relative increase in human cells within a murine tissue, as observed after TBI and 7-day treatment with TPO/FLT3, or when we applied etoposide together with dmPGE2, might thus indicate that human cells are favored even when absolute human cell numbers do not increase. The substances used here were carefully selected on the basis of published data and own earlier work. However, despite the very promising data obtained in mouse models,9,10 neither EGF nor PGE2 were able to promote hematopoietic regeneration in our model system. In the case of EGF, non-conserved pathways seem to underlie this inconsistency. As for PGE2, both its antiproliferative and toxic effects might contribute to its lack of in vivo efficacy. Anyhow, the severe gastrointestinal side effects provoked by the combination of dmPGE2 and TBI/etoposide would render such a treatment unfeasible. FLT3L and TPO are both known to increase viability and promote proliferation of human HSPCs in vitro. Along that line, we observed some beneficial effects on human hematopoietic regeneration in vivo, even though these were marginal and transient. One could speculate that the maximal beneficial effect on human hematopoietic regeneration can be achieved by the use of substances that foster both survival and proliferation of HSPCs. Certainly, short-term complications such as febrile neutropenia or transfusion-dependent anemia and thrombocytopenia could be reduced by increasing the proliferative capacity of HSPCs following chemoor radiotherapy. Repeated cycles of forced proliferation, however, could result in premature stem cell exhaustion and BM failure in the long term; this has already been shown when mice were repeatedly treated with SCF.19 Forced proliferation of DNA-damaged HSPCs, as caused by chemo- or radiotherapy, together with selection pressure could also increase the risk of genetic instability and clonal evolution eventually leading to t-MDS and secondary AML. Ideally, radioprotective substances should not stimulate proliferation but only inhibit apoptosis of healthy BM cells. Apoptosis resistance reduces therapy-induced myelosuppression, and at the same time lowers the risk of secondary leukemia by prolonging the time available for DNA repair and reducing compensatory proliferation and selection pressure.2,15,20,21 ROS scavengers, for example, are radioprotective by preventing both DNA damage and apoptosis.11 Direct inhibition of the pro-apoptotic effector proteins, BAX and BAK, could also keep hematopoietic cells alive without affecting proliferation. We recently showed that such transient apoptosis resistance can be achieved by short-term overexpression or protein transduction of the anti-apoptotic protein BCL-XL.22 Translation to clinical use will probably be pushed ahead by the development of specific BAX/BAK inhibitors. Similar approaches will be useful to reduce risk of graft failure and shorten the time to full hematopoietic regeneration in the case of autologous or allogeneic hematopoiethaematologica | 2019; 104(4)

PGE2, EGF, FLT3L and TPO in human hematopoiesis

ic stem cell transplantation (HSCT). We have already showed that inhibition of the intrinsic apoptosis pathway in donor cells, even when limited to a few days around transplantation, significantly improves outcome of HSCT, especially when only low donor cell numbers are available.18,22,23 While donor stem cells used for HSCT can be manipulated ex vivo, the major challenge in the field of cancer treatment will be to identify substances that exclusively protect hematopoietic cells without having any beneficial effect on cancer cells. EGF, PGE2 and SCF are known to have proliferative and pro-survival activities on certain types of cancer24-26 and FLT3L has not yet been systematically tested for its effects on solid tumors. As an alterna-

References 1. Wang Y, Probin V, Zhou D. Cancer therapy-induced residual bone marrow injuryMechanisms of induction and implication for therapy. Curr Cancer Ther Rev. 2006;2(3):271-279. 2. Labi V, Erlacher M. How cell death shapes cancer. Cell Death Dis. 2015;6:e1675. 3. Murray LJ, Young JC, Osborne LJ, et al. Thrombopoietin, flt3, and kit ligands together suppress apoptosis of human mobilized CD34+ cells and recruit primitive CD34+ Thy-1+ cells into rapid division. Exp Hematol. 1999;27(6):1019-1028. 4. Varnum-Finney B, Xu L, Brashem-Stein C, et al. Pluripotent, cytokine-dependent, hematopoietic stem cells are immortalized by constitutive Notch1 signaling. Nat Med. 2000;6(11):1278-1281. 5. Hoggatt J, Singh P, Sampath J, Pelus LM. Prostaglandin E2 enhances hematopoietic stem cell homing, survival, and proliferation. Blood. 2009;113(22):5444-5455. 6. Dygai AM, Khmelevskaya ES, Skurikhin EG, et al. Catecholamine regulation of stromal precursors and hemopoietic stem cells in cytostatic myelosuppression. Bull Exp Biol Med. 2012;15(6)2:723-727. 7. Cheng CW, Adams GB, Perin L, et al. Prolonged fasting reduces IGF-1/PKA to promote hematopoietic-stem-cell-based regeneration and reverse immunosuppression. Cell Stem Cell. 2014;14(6):810-823. 8. Burdelya LG, Krivokrysenko VI, Tallant TC, et al. An agonist of toll-like receptor 5 has radioprotective activity in mouse and primate models. Science. 2008; 320(5873):226-230. 9. Porter RL, Georger MA, Bromberg O, et al.

haematologica | 2019; 104(4)

10.

11.

12.

13.

14.

15.

16.

17.

tive, radioprotective substances could be delivered only to healthy BM cells while sparing cancer cells to avoid therapy resistance. Novel approaches in targeted drug delivery will hopefully soon enable such therapies to be developed. Acknowledgments We are grateful to Nora Fischer for excellent technical assistance and to Natalie Krause for animal care. We also would like to thank our colleagues from FOR2036 for insightful discussion. Funding This work was supported by grants from the German Research Foundation (DFG-FOR2036 to ME) and the European Research Council (ERC-StG-2014 â&#x20AC;&#x201C; 638145 ApoptoMDS to ME).

Prostaglandin E2 increases hematopoietic stem cell survival and accelerates hematopoietic recovery after radiation injury. Stem Cells. 2013;31(2):372-383. Doan PL, Himburg HA, Helms K, et al. Epidermal growth factor regulates hematopoietic regeneration after radiation injury. Nat Med. 2013;19(3):295-304. Kamran MZ, Ranjan A, Kaur N, Sur S, Tandon V. Radioprotective Agents: Strategies and Translational Advances. Med Res Rev. 2016;36(3):461-493. Johnke RM, Sattler JA, Allison RR. Radioprotective agents for radiation therapy: future trends. Future Oncol. 2014;10(15):2345-2357. Kollek M, Muller A, Egle A, Erlacher M. Bcl-2 proteins in development, health, and disease of the hematopoietic system. FEBS J. 2016;283(15):2779-2810. Erlacher M, Labi V, Manzl C, et al. Puma cooperates with Bim, the rate-limiting BH3-only protein in cell death during lymphocyte development, in apoptosis induction. J Exp Med. 2006;203(13):2939-2951. Erlacher M, Michalak EM, Coultas L, et al. The BH3-only proteins Puma/bbc3 and Bim are rate-limiting for g-radiation- and glucocorticoid-induced apoptosis of lymphoid cells in vivo. Blood. 2005; 106(13):4131-4138. North TE, Goessling W, Walkley CR, et al. Prostaglandin E2 regulates vertebrate haematopoietic stem cell homeostasis. Nature. 2007;447(7147):1007-1011. Goessling W, Allen RS, Guan X, et al. Prostaglandin E2 enhances human cord blood stem cell xenotransplants and shows long-term safety in preclinical nonhuman primate transplant models. Cell Stem Cell. 2011;8(4):445-458.

18. Labi V, Bertele D, Woess C, et al. Haematopoietic stem cell survival and transplantation efficacy is limited by the BH3-only proteins Bim and Bmf. EMBO Mol Med. 2013;5(1):122-136. 19. Lemoli RM, Gulati SC. Effect of stem cell factor (c-kit ligand), granulocytemacrophage colony stimulating factor and interleukin 3 on hematopoietic progenitors in human long-term bone marrow cultures. Stem Cells. 1993;11(5):435-444. 20. Labi V, Erlacher M, Krumschnabel G, et al. Apoptosis of leukocytes triggered by acute DNA damage promotes lymphoma formation. Genes Dev. 2010;24(15):1602-1607. 21. Michalak EM, Vandenberg CJ, Delbridge AR, et al. Apoptosis-promoted tumorigenesis: gamma-irradiation-induced thymic lymphomagenesis requires Puma-driven leukocyte death. Genes Dev. 2010; 24(15):1608-1613. 22. Kollek M, Voigt G, Molnar C, et al. Transient apoptosis inhibition in donor stem cells improves hematopoietic stem cell transplantation. J Exp Med. 2017; 214(10):2967-2983. 23. Afreen S, Weiss JM, Strahm B, Erlacher M. Cheating death for a better transplant. Stem Cells. 2018 Aug 29. [Epub ahead of print] 24. Mendelsohn J, Baselga J. The EGF receptor family as targets for cancer therapy. Oncogene. 2000;19(56):6550-6565. 25. Huang Q, Li F, Liu X, et al. Caspase 3-mediated stimulation of tumor cell repopulation during cancer radiotherapy. Nat Med. 2011;17(7):860-866. 26. Cardoso HJ, Figueira MI, Socorro S. The stem cell factor (SCF)/c-KIT signalling in testis and prostate cancer. J Cell Commun Signal. 2017;11(4):297-307.

677

ARTICLE Ferrata Storti Foundation

Iron Metabolism & its Disorders

Gastrointestinal iron excretion and reversal of iron excess in a mouse model of inherited iron excess

Courtney J. Mercadante,1 Milankumar Prajapati,1 Jignesh H. Parmar,2 Heather L. Conboy,1 Miriam E. Dash,1 Michael A. Pettiglio,1 Carolina Herrera,1 Julia T. Bu,1 Edward G. Stopa, 1Pedro Mendes2 and Thomas B. Bartnikas1

Department of Pathology and Laboratory Medicine, Brown University, Providence, RI and Center for Quantitative Medicine and Department of Cell Biology, University of Connecticut School of Medicine, Farmington, CT, USA

Haematologica 2019 Volume 104(4):678-689

ABSTRACT

Correspondence: THOMAS BARTNIKAS thomas_bartnikas@brown.edu Received: May 22, 2018. Accepted: November 7, 2018. Pre-published: November 8, 2018.

he current paradigm in the field of mammalian iron biology states that body iron levels are determined by dietary iron absorption, not by iron excretion. Iron absorption is a highly regulated process influenced by iron levels and other factors. Iron excretion is believed to occur at a basal rate irrespective of iron levels and is associated with processes such as turnover of intestinal epithelium, blood loss, and exfoliation of dead skin. Here we explore iron excretion in a mouse model of iron excess due to inherited transferrin deficiency. Iron excess in this model is attributed to impaired regulation of iron absorption leading to excessive dietary iron uptake. Pharmacological correction of transferrin deficiency not only normalized iron absorption rates and halted progression of iron excess but also reversed body iron excess. Transferrin treatment did not alter the half-life of 59Fe in mutant mice. 59Fe-based studies indicated that most iron was excreted via the gastrointestinal tract and suggested that iron-loaded mutant mice had increased rates of iron excretion. Direct measurement of urinary iron levels agreed with 59Fe-based predictions that urinary iron levels were increased in untreated mutant mice. Fecal ferritin levels were also increased in mutant mice relative to wild-type mice. Overall, these data suggest that mice have a significant capacity for iron excretion. We propose that further investigation into iron excretion is warranted in this and other models of perturbed iron homeostasis, as pharmacological targeting of iron excretion may represent a novel means of treatment for diseases of iron excess.

doi:10.3324/haematol.2018.198382 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/678 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

678

Introduction Iron is an essential nutrient but toxic when present in excess. While body iron levels are determined in theory by a balance between absorption and excretion, the current paradigm in the field of iron biology states that the rate of absorption determines iron levels.1–5 Iron absorption is an orchestrated process regulated prominently by hepcidin, a hormone synthesized mainly by the liver which inhibits iron export from duodenal enterocytes and other types of cells. Hepcidin expression is suppressed by anemia, leading to increased absorption of iron, and stimulated by iron excess and inflammation, leading to decreased absorption. Hepcidin deficiency is central to common inherited diseases of iron excess such as hereditary hemochromatosis and β-thalassemia. Hemochromatosis is caused by mutations in genes required for hepcidin expression.6,7 β-thalassemia is caused by β-globin mutations leading to ineffective erythropoiesis, anemia, and hepcidin deficiency.8–10 In contrast to iron absorption, iron excretion is presumed to occur at a basal rate irrespective of iron deficiency or excess.1–5 Excretion is believed to reflect processes such as turnover of intestinal epithelium, blood loss, and exfoliation of dead skin. Renal iron excretion is considered a negligible factor in iron homeostasis. The liver haematologica | 2019; 104(4)

Iron excretion in iron excess

is also believed to contribute minimally to elimination of iron from the body, despite the fact that hepatobiliary excretion is a prominent means of excretion of other metals such as manganese and copper. In this study, we investigated iron excretion in Trfhpx/hpx mice, a model of inherited deficiency of the serum ironbinding protein transferrin.11 These mice develop anemia because transferrin is essential for iron delivery to erythroid precursors. They also develop iron excess because transferrin is essential for hepcidin expression. We and others previously observed that treatment of adult Trfhpx/hpx mice with transferrin for 2 to 3 weeks not only corrected anemia and hepcidin deficiency but also lowered liver iron concentrations.12,13 Here we exploit the latter observation to assess the effect of iron excess on iron excretion and to identify routes of iron excretion using short- and long-term transferrin treatment and radioisotopic studies. Our data suggest that the view that iron levels are dictated solely by absorption needs to be reconsidered. They also suggest that non-gastrointestinal routes of excretion, such as exfoliation of skin, play a minimal role in iron homeostasis.

treatment was continued as before when indicated. Bedding was changed once a week. Some mice were housed with a â&#x20AC;&#x2DC;buddyâ&#x20AC;&#x2122; mouse not administered 59Fe. Every 1 to 2 weeks, body 59Fe levels of all the mice were measured. Buddy mouse 59Fe levels never exceeded background, suggesting that coprophagy was not prominent. Details on the conversion of body 59Fe counts to 59Fe half-lives and excretion rates are given in the Online Supplementary Methods. To identify routes of 59Fe excretion, mice were housed overnight for 16 h in metabolic cages at least three times during the excretion study. Feces and urine were analyzed for 59Fe levels by gamma counting then for iron and ferritin levels using ICPAES and enzyme-linked immunosorbent assay (ELISA) as described in the Online Supplementary Methods.

Mathematical modeling The mathematical modeling of iron levels is described in the Online Supplementary Methods. A manuscript on the model is currently under review and a preprint version of the paper is available.15

Statistical analysis Methods Animals and transferrin treatment Studies were approved by the Animal Care and Use Committee at Brown University. Mice were maintained on LabDiet 5010 containing 270 ppm iron. BALB/cJ Trf+/+ and Trfhpx/hpx mice were generated by crossing Trf+/hpx mice, which were intermittently backcrossed to BALB/cJ mice (Jackson Laboratories). To ensure survival of Trfhpx/hpx mice after weaning, pups were injected intraperitoneally with 3 mg human transferrin (Roche/Sigma) 2 days after birth, then once a week until weaning at 3 weeks of age. For all experiments, mice were aged from weaning to 2 months without transferrin injections, then some were injected intraperitoneally with 3 mg human transferrin three times a week as required for specific experiments.

Non-radioactive sample harvesting and analysis Details on the collection of blood and tissues from mice, transferrin immunoblots, measurement of hemoglobin, hepcidin, and RNA levels, tissue staining, and metal analysis are provided in the Online Supplementary Methods. Body and tissue iron levels were measured by inductively coupled plasma absorption emission spectrometry (ICP-AES) of acid-digested tissues in the Environmental Chemistry Facility at Brown University.

Fe treatments, sample harvesting and analysis

To assess absorption, mice were fasted in metabolic cages (Tecniplast) with access to water for 4 h, then gavaged with 10 mCi 59FeCl3 (Perkin Elmer) and 6 mg FeCl3 in 100 mL 1 M ascorbic acid.14 The mice were then housed in metabolic cages with food and water for 16 h. 59Fe levels were measured in bodies, feces, and urine using a Triathler Gamma Counter and external NaI well-type crystal detector (Hidex). To measure body 59Fe levels, mice were anesthetized with isoflurane, placed nose-first into a 50 mL conical tube positioned vertically in the detector, and radioactivity was counted. Background counts were subtracted from all counts. Percent 59Fe absorption was calculated by expressing the sum of body and urine 59Fe levels as a percent of the sum of body, fecal, and urinary 59Fe levels. To assess excretion, mice from the absorption studies were housed individually in regular cages for 2 months. Transferrin haematologica | 2019; 104(4)

Statistical significance (P<0.05) was calculated by a two-tailed t-test or one- or two-way analysis of variance (ANOVA) with a Holm-Sidak post-hoc test using Sigmaplot. Pearson correlations were also measured using Sigmaplot.

Results Short-term transferrin treatment reduces tissue iron excess in Trfhpx/hpx mice Short-term transferrin treatment of adult Trfhpx/hpx mice decreases liver iron concentrations.12,13 To investigate this phenomenon further, we first determined whether a 2week course of transferrin treatment in 2-month old Trfhpx/hpx mice altered iron concentrations in organs other than the liver. As expected, transferrin treatment increased serum transferrin, blood hemoglobin, and serum hepcidin levels in mutant mice (Figure 1A-C). Treatment also normalized Fam132b RNA levels in the spleen, a site of extramedullary hematopoiesis in Trfhpx/hpx mice, with Fam132b encoding erythroferrone, which is an inhibitor of hepcidin that is expression expressed by erythroid precursors (Figure 1D).16 Transferrin treatment also corrects severe splenomegaly in mutant mice.12 Untreated Trfhpx/hpx mice accumulated iron largely in the liver and pancreas, specifically in hepatic periportal regions and exocrine pancreas, and to a lesser extent in the kidneys, heart, and other tissues (Figure 1E).11 While Trf+/+ mice had stainable iron in the red pulp of the spleen, untreated Trfhpx/hpx mice had splenic iron deficiency and a paucity of stainable iron. We attribute this to the fact that hepcidin also inhibits macrophage iron export - hepcidin deficiency in mutant mice leads to persistent iron export from red pulp macrophages scavenging iron-poor red blood cells. Transferrin treatment decreased iron concentrations and tissue iron staining in the liver, pancreas, and kidneys but not in the heart or duodenum and increased iron concentrations and tissue iron staining in the spleen (Figure 1E and Figure 2). Stainable iron was also detectable in duodenal smooth muscle in untreated and treated Trfhpx/hpx mice but in duodenal enterocytes only in treated mutant mice (Figure 2B,C). Overall, transferrin treatment decreased iron concentrations in multiple organs in Trfhpx/hpx mice. 679

C.J. Mercadante et al.

Long-term transferrin treatment corrects body iron excess in Trfhpx/hpx mice Several scenarios could explain decreased iron concentrations in organs of transferrin-treated Trfhpx/hpx mice. Given the severity of anemia in untreated Trfhpx/hpxmice, the increase in hemoglobin levels in treated mutant mice

should require mobilization of a significant amount of iron from the liver and other organs to the bone marrow. We estimated that 0.30 mg of iron were mobilized from the liver and pancreas during the first 2 weeks of treatment and that the anemia in untreated mutant mice corresponded to a deficit of 0.475 mg of hemoglobin iron

Figure 1. Short-term transferrin treatment reduces tissue iron excess in Trfhpx/hpx mice. (A-D) Trf+/+ (‘+/+’), untreated Trfhpx/hpx (‘hpx/hpx’), and transferrin (TF)-treated Trfhpx/hpx (‘hpx/hpx +TF’) mice were analyzed at 2.5 months of age. Treated mice were injected with TF from 2 to 2.5 months of age. (A) Serum TF levels, measured by immunoblot (top) and Coomassie-stained protein gel (bottom). (B) Hemoglobin levels, measured by complete blood count. (C) Plasma hepcidin levels, measured by ELISA. (D) Splenic RNA level ratios of Fam132b to Actb (β-actin), measured by quantitative polymerase chain reaction and normalized to Trf+/+ levels. (E) Organ iron (Fe) levels (left panels), measured by inductively coupled plasma absorption emission spectrometry, and tissue Fe distribution (right micrographs), assessed by tissue Fe stain. In (B-E), data are represented as the mean ± standard error of mean. Brackets indicate statistical significance (P<0.05) calculated by one-way analysis of variance with a Holm-Sidak post-hoc test. Each value represents data from five mice, with males and females grouped together.

680

haematologica | 2019; 104(4)

Iron excretion in iron excess

(see Online Supplement for details). Based upon these calculations, it is likely that changes in tissue iron levels in the first 2 weeks of transferrin treatment largely represent mobilization of iron to the bone marrow. To explore the effect of transferrin treatment on body iron levels in Trfhpx/hpx mice beyond the first 2 weeks of treatment, we measured body iron content in 1- to 6month old Trf+/+ and Trfhpx/hpx mice and in Trfhpx/hpxmice treated with transferrin from 2 to 6 months of age. The mice were euthanized in order to measure body iron content and no blood was removed prior to euthanasia. After mouse pelts had been removed, gastrointestinal tracts were isolated and cleared of contents. Pelts, cleared gas-

trointestinal tracts, and carcasses were then analyzed for iron levels, which were summed to calculate body iron levels. We focused on body iron levels here as these would not be affected by redistribution between organs. Untreated Trfhpx/hpx mice were smaller than Trf+/+ mice, and treatment increased body sizes of Trfhpx/hpx mice (Figure 3A). Body iron levels (in mg iron) and concentrations (in Âľg iron/g body mass) were greater in untreated Trfhpx/hpx mice than in Trf+/+ mice (Figure 3B,C). Treatment of Trfhpx/hpx mice resulted in no difference in iron levels and a less than two-fold difference in iron concentrations relative to those in Trf+/+ mice by 6 months of age (Figure 3B,C). Most untreated Trfhpx/hpx mice did not survive to 6 months,

Figure 2. Trfhpx/hpx mice accumulate iron in the duodenum. (A-C) Mice from Figure 1 were analyzed for duodenal iron (Fe) levels by inductively coupled plasma absorption emission spectrometry (A) and tissue Fe staining in duodenal smooth muscle (B) and villi (C). In (A), data are represented as the mean Âą standard error of mean. Brackets indicate statistical significance (P<0.05) calculated by oneway analysis of variance with a Holm-Sidak post-hoc test. Each value represents data from five mice, with males and females grouped together. In (C), the arrowhead indicates detectable Fe staining in duodenal enterocytes.

haematologica | 2019; 104(4)

681

C.J. Mercadante et al.

which may reflect chronic effects of iron excess and/or anemia. Overall, given that body iron levels would not be affected by iron redistribution between organs, we hypothesized that the long-term change in body iron levels in transferrin-treated Trfhpx/hpx mice reflected changes in absorption and/or excretion.

Trf+/+ and treated Trfhpx/hpx mice have similar 59 Fe absorption rates Untreated Trfhpx/hpx mice absorb iron excessively.14 If transferrin treatment suppresses absorption rates below excretion rates, body iron levels would decrease without any need for increased excretion rates in Trfhpx/hpx mice. To test this hypothesis, we performed intragastric 59Fe gavage in 2.5-month old Trf+/+, untreated Trfhpx/hpx, and transferrin-treated Trfhpx/hpx mice, then analyzed 59Fe levels 16 h later. From herein we studied males and females separately to detect sex-specific differences. Untreated Trfhpx/hpx mice absorbed more gavaged 59Fe than did Trf+/+ and treated Trfhpx/hpx mice, and Trf+/+ and treated Trfhpx/hpx mice absorbed the same amount of 59Fe (Figure 4A). Similar results were observed when absorption values were normalized to body mass (Figure 4B) or when mice were analyzed 1 h after gavage (data not shown). This indicated that reversal of iron excess in transferrin-treated Trfhpx/hpx mice was not due to ‘hypersuppression’ of iron absorption.

Trf+/+ and treated Trfhpx/hpx mice have similar 59 Fe half-lives We next assessed excretion in Trf+/+, untreated Trfhpx/hpx, and transferrin-treated Trfhpx/hpx mice using 59Fe. Mice from the absorption studies were used, as we rationalized that gavage was the most physiological means of administering 59Fe. We repeatedly measured body 59Fe levels in mice for 2 months from 2.5 to 4.5 months of age (Figure 4C).

We ended the excretion study at 4.5 months as untreated Trfhpx/hpx mice do not consistently survive past this age. Body 59Fe levels, plotted versus time, were fitted to exponential decay curves. Exponential decay equations were then used to calculate two factors: biological 59Fe halflives, expressed in days, and percent body 59Fe excreted per day, referred to here as ‘59Fe excretion rates’. 59Fe halflives and 59Fe excretion rates are inversely proportional to each other. 59Fe half-lives were ~80-120 days in all mice except for untreated male Trfhpx/hpx mice, which had a halflife of ~170 days (Figure 4D). 59Fe excretion rates were ~0.6-0.8% in all mice except for male Trfhpx/hpx mice, which had an excretion rate of ~0.45% (Figure 4E). Notably, 59Fe half-lives and excretion rates did not differ between Trf+/+ and untreated Trfhpx/hpx mice.

Trf mice excrete iron largely via the gastrointestinal tract During the 2-month excretion study, mice were placed repeatedly in metabolic cages for overnight collections of feces and urine. Fecal and urinary 59Fe levels were expressed as a percent of body 59Fe levels at the time of collection, then averaged for each mouse group. Most 59Fe was excreted in feces (Figure 4F). Body 59Fe losses could be accounted for by fecal and urinary 59Fe losses in all mice except transferrin-treated Trfhpx/hpx females (Figure 4G). Overall, these data indicate that Trf mice excreted iron largely via the gastrointestinal tract.

Fe-based analyses predict relative abundance of urinary iron and fecal ferritin in Trf mice

In some of the earliest radioisotope-based studies of iron excretion in mice, Finch and colleagues multiplied body 59Fe excretion rates by body iron levels to estimate the amount of iron excreted per day.17,18 We employed this approach here to further explore iron excretion in Trf

Figure 3. Long-term transferrin treatment corrects body iron excess in Trfhpx/hpx mice. (A-C) Body masses (A) and iron (Fe) levels (B) and concentrations (C), measured in Trf+/+ mice (‘+/+’, orange) and untreated Trfhpx/hpx mice (‘hpx/hpx’, green) from 1 to 6 months of age and in Trfhpx/hpx mice treated with transferrin (TF) from 2 to 6 months of age (‘hpx/hpx +TF’, blue). Top and bottom graphs differ only by markers of significance (P<0.05) assessed by two-way analysis of variance with a HolmSidak post-hoc test. In top panels, for a given age, different letters indicate values that differ significantly. In bottom panels, for a given group, the letter indicates that a value differs significantly from the 2-month old value. In all panels, dashed lines indicate that only two untreated Trfhpx/hpx mice survived to 6 months. Data are represented as mean ± standard error of mean: each value represents data from five mice, with males and females grouped together.

682

haematologica | 2019; 104(4)

Iron excretion in iron excess

mice. We performed this for 2.5- and 4.5-month-old mice, the respective ages at the beginning and end of the excretion study. We first established body iron levels and concentrations in 2.5- and 4.5-month-old mice (Figure 5A,B). For 2.5month-old mice, we measured body iron levels and concentrations in five male and five female mice for each experimental group using the same approach employed for Figure 3. For 4.5-month old mice, we harvested all tissues/compartments from mice at the end of the excretion study and measured iron levels and concentrations (Table 1). Iron levels for 2.5- and 4.5-month old mice were consistent with iron levels shown in Figure 3 where sexes were pooled (Online Supplementary Figure S1). With body

haematologica | 2019; 104(4)

iron levels established in 2.5- and 4.5-month old mice, we next multiplied 59Fe excretion rates (Figure 4E) by body iron levels (Figure 5A) or body iron concentrations (Figure 5B) to estimate iron excretion rates with and without normalization to body mass (Figure 5C,D). At 2.5 months, iron excretion rates were predicted to be increased in all Trfhpx/hpx mice except untreated males when not normalized to body size and in all Trfhpx/hpx mice when normalized to body size. At 4.5 months, excretion rates were predicted to be increased in all Trfhpx/hpx mice irrespective of normalization to body size. The above approach involves several assumptions. The first is that 59Fe is fully equilibrated within each mouse. To explore this, we examined the relative distribution of 59Fe

Figure 4. Trf+/+ and treated Trfhpx/hpx mice have similar 59Fe absorption rates and half-lives and excrete iron largely via the gastrointestinal tract. (A) Percent 59Fe absorbed, measured after 59Fe gavage of male and female 2.5-month old Trf+/+ mice (‘+/+’, orange), untreated Trfhpx/hpx mice (‘hpx/hpx’, green), and Trfhpx/hpx mice treated with transferrin (TF) (‘hpx/hpx +TF’, blue) from 2 to 2.5 months of age. Percent 59Fe absorbed was calculated as the sum of body and urinary 59Fe levels expressed as a percent of body, urine, and feces 59Fe levels. (B) Values from (A) normalized to body size. (C) Representative plots of body 59Fe levels from 2.5 to 4.5 months of age in Trf+/+ mice, untreated Trfhpx/hpx mice, and Trfhpx/hpx mice treated with TF from 2 months of age (“hpx/hpx +TF”, blue). ‘Day 0’ indicates the day after 59Fe gavage. Unfilled circles at day 0 indicate that these points were excluded from lines of best fit. Exponential decay equations, 59Fe half-lives (t1/2), and percent 59Fe excreted per day are included in each graph. (D) 59Fe half-lives, calculated by exponential decay equations from (C). (E) Percent 59Fe excreted per day, or ‘59Fe excretion rates’, calculated by exponential decay equations from (C). (F) Percent body 59Fe excreted per day via feces or urine in mice from 2.5 to 4.5 months of age. Feces and urine were collected by housing each mouse overnight in metabolic cages at least three times during the excretion study. (G) Sums of fecal and urinary 59Fe values from (F) (“urine+feces”) compared to body 59Fe excretion rates from (E) (“body loss”). In all panels except (C), data are represented as mean ± standard error of mean with each value shown representing data from at least five mice and brackets indicating P<0.05. Statistical significance was calculated by one-way analysis of variance with a HolmSidak post-hoc test in (A, B, D, E) and by a two-tailed t-test in (F, G).

683

C.J. Mercadante et al.

versus iron within each experimental group harvested at the end of the excretion study at 4.5 months of age. 59Fe and iron levels correlated significantly in all groups (Online Supplementary Figure S2A). The second assumption is that 59Fe equilibrates similarly in all mice. To explore this, we examined the relative distribution of 59Fe between Trf+/+ and untreated or treated Trfhpx/hpx mice. Relative distributions correlated more strongly between Trf+/+ and treated Trfhpx/hpx mice than between Trf+/+ and untreated Trfhpx/hpx mice (Online Supplementary Figure S2B). This was not unexpected - treated mutant mice were administered 59Fe after 2 weeks of transferrin treatment during which time hemoglobin and hepcidin levels increased significantly (Figure 1B,C). To explicitly test 59Fe-based estimates of excretion, we next used 59Fe levels measured during the excretion study to predict urinary iron levels, then measured and com-

pared actual excreted urinary iron levels to predicted levels. We first focused on urinary iron levels given that they solely reflect excretion. To estimate urinary iron levels, values of percent body 59Fe excreted in urine per day (Figure 4F) were multiplied by body iron levels (Figure 5A). Urinary iron levels were estimated to be increased in untreated Trfhpx/hpx mice compared to Trf+/+ and treated Trfhpx/hpx mice at 2.5 and 4.5 months of age (Figure 5E). Measurement of urinary iron levels agreed with the prediction that untreated mutant mice excreted more iron via urine than wild-type or treated mutant mice (Figure 5E and Online Supplementary Figure S3). Measured urinary iron levels also agreed with 59Fe-based predictions for all treated Trfhpx/hpx mice. However, urinary iron levels were underestimated by 59Fe-based predictions by 5-10 mg iron per day in all untreated mutant mice and by 2 mg iron per day in female untreated mutant mice.

684

Figure 5. 59Fe-based analyses predict relative abundance of urinary iron and fecal ferritin in Trf mice. (A, B) Body iron (Fe) levels (A) and concentrations (B) in Trf+/+ mice (‘+/+’, orange), untreated Trfhpx/hpx mice (‘hpx/hpx’, green), and Trfhpx/hpx mice treated with transferrin (TF) (‘hpx/hpx +TF’, blue) harvested at 2.5 months or 4.5 months of age at the end of the excretion study shown in Figure 4. (C) mg Fe excreted per day, calculated by multiplying values in Figure 4E by values in (A). (D) mg Fe excreted per day normalized to body size, calculated by multiplying values in Figure 4E by values in (B). (E) mg Fe excreted per day in urine. 59Fe-based estimates were calculated by multiplying urinary values in Figure 4F by values in (A). Spectrophotometric measurements (‘spec assay’) were calculated by acid digest and BPS-based assay as described in the Online Supplementary Methods. (F) mg Fe excreted per day in feces, estimated by multiplying fecal values in Figure 4F by values in (A). (G) µg ferritin excreted per day in feces measured by enzyme-linked immunosorbent assay. In all panels, data are represented as mean ± standard error of mean; each value shown represents data from at least five mice. In (A-D), at a given age, different letters indicate P<0.05 between values, calculated by one-way analysis of variance with a HolmSidak post-hoc test; ‘#’ indicates P<0.05 between values from 2.5- and 4.5-month old mice, calculated by a two-tailed t-test. In (EG), brackets indicate P<0.05, calculated by one-way analysis of variance with a HolmSidak post-hoc test. In (E), asterisks indicate P<0.05 between 59Fe-based and spectrophotometry-based values, calculated by a twotailed t-test.

haematologica | 2019; 104(4)

Iron excretion in iron excess

We next used 59Fe measurements to investigate fecal iron excretion. To estimate excreted iron levels in feces, values of percent body 59Fe excreted in feces per day (Figure 4F) were multiplied by body iron levels (Figure

5A). Fecally excreted iron levels were estimated to be increased in all Trfhpx/hpx mice relative to Trf+/+ mice and increased in untreated Trfhpx/hpx mice relative to treated Trfhpx/hpx mice for all mice except 2.5-month old male mice

Table 1. Long-term transferrin treatment reduces or normalizes organ iron content in Trfhpx/hpx mice.

Sex

Tissue

+/+

mg Fe hpx/hpx

hpx/hpx +TF

+/+

mg Fe/g tissue hpx/hpx

hpx/hpx +TF

Male

Liver

164±28

3208±148

1162±127

Pancreas

48±18

1536±173

336±36

Heart

3±1

9±2

5±1

Lungs

29±2

100±8

41±9

Kidneys

22±5

183±19

23±5

Spleen

43±5

88±13

66±6

Stomach S. int. A

13±2 10±1

139±6 25±2

29±4 18±2

S. int. B Cecum Large int. Brain

27±2 7±1 12±1 5±2

97±6 19±2 41±2 47±5

33±2 12±2 15±2 19±3

Pelt Carcass Blood

244±22 958±97

510±25 2130±160

233±22 828±117

106±20 (158±22) 162±59 (39±3) 19±3 (155±23) 158±9 (160±10) 38±7 (74±3) 473±60 (441±53) 63±9 44±6 (41±4) 26±3 32±5 37±2 12±6 (16±2) 44±3 55±5 563±29

3406±102 (2344±171) 7110±717 (1924±64) 58±10 (363±29) 676±91 (389±23) 581±75 (496±32) 111±12 (79±6) 1021±53 167±14 (75±9) 144±7 137±9 237±10 123±13 (66±3) 215±12 259±18 191±10

750±79 (1751±127) 1042±62 (1207±51) 31±3 (318±9) 215±24 (274±11) 54±12 (219±5) 443±26 (259±28) 129±22 87±11 (87±6) 30±2 48±6 48±6 45±6 (37±4) 50±5 53±8 432±35

Liver

233±25

2817±286

1437±125

Pancreas

22±5

1453±151

308±25

Heart

2±0

8±3

5±1

Lungs

26±2

98±11

34±5

Kidneys

30±2

212±17

35±5

Spleen

69±5

93±15

72±6

Stomach S. int. A

15±1 8±2

181±7 24±3

24±2 15±2

S. int. B Cecum Large int. Brain

20±1 5±1 11±1 7±2

108±10 27±4 31±3 43±3

42±8 7±2 14±2 16±2

Pelt Carcass Blood

203±23 506±50

448±23 1913±41

199±6 700±114

196±19 (158±22) 85±11 (39±3) 16±2 (155±23) 162±12 (160±10) 94±7 (74±3) 1029±37 (441±53) 82±8 48±8 (41±4) 22±1 26±3 37±5 16±4 (16±2) 42±5 38±4 471±35

3410±337 (2344±171) 7380±590 (1924±64) 53±18 (363±29) 695±83 (389±23) 719±51 (496±32) 150±15 (79±6) 1019±105 179±18 (75±9) 143±12 144±8 184±12 118±9 (66±3) 208±14 277±14 188±11

1159±64 (1751±127) 1109±68 (1207±51) 37±4 (318±9) 215±21 (274±11) 110±12 (219±5) 470±30 (259±28) 134±18 80±12 (87±6) 40±7 37±9 60±4 40±6 (37±4) 47±1 61±11 460±3

Female

Iron (Fe) levels were measured in 4.5-month old Trf+/+ (“+/+”), Trfhpx/hpx (“hpx/hpx”), and transferrin (TF)-treated Trfhpx/hpx (“hpx/hpx +TF”) mice at the end of the excretion study. TF-treated Trfhpx/hpx mice were treated with TF from 2 to 4.5 months of age. Digestive organs including intestines (‘int’.) were cleared of luminal contents before analysis.‘S. int. A’ refers to the first 4 cm of the small intestine; ‘S. int. B’ refers to the remaining small intestine. Cells with different shading differ significantly (P<0.05) for a specific organ and sex, as calculated by one-way analysis of variance with the Holm-Sidak post-hoc test. Each value represents the mean ± standard error of mean of the data from at least five mice. For reference, tissue Fe levels from 2.5-month old mice shown in Figures 1 and 2 are included in parentheses; note that parenthetical tissue Fe levels represent data from five mice with males and females pooled.

haematologica | 2019; 104(4)

685

C.J. Mercadante et al. A

Figure 6. Mathematical modeling predicts increased excretion rates in treated Trfhpx/hpx mice. (A) The iron (Fe) homeostasis scheme used for mathematical modeling of body Fe levels from Figure 3B. Closed polygons represent liver, spleen, red blood cells (RBC), bone marrow (BM), duodenum (Duo), or ‘rest of body’ (Rest), the last comprising stomach, intestines except the duodenum, integument, muscles, heart, fat, lungs, kidneys, brain, and reproductive organs. The circle represents plasma. Dietary Fe is absorbed into the duodenum (solid dark green arrow) and exported to plasma by ferroportin (solid black arrow). Plasma Fe, in states of Fe homeostasis, exists as Fe-loaded transferrin (FeTF) which is imported into all compartments (solid light green arrows) except RBC or, in states of Fe excess, as non-transferrinbound iron (NTBI) which is imported into the liver and rest of the body (solid blue arrows). Bone marrow Fe is incorporated into RBC and recycled in the spleen (solid red arrows); direct Fe transfer from the bone marrow to the spleen represents recycling of immature erythroid cells in the spleen (solid red arrows). Fe can be exported from compartments by ferroportin (solid black arrows). FeTF stimulates hepcidin expression (dashed black arrow), which inhibits ferroportin-dependent Fe export from compartments (dashed blunt-ended lines). Erythropoietin (EPO) stimulates FeTF import into the bone marrow, transfer of Fe from the bone marrow to RBC, and suppression of hepcidin activity. EPO is suppressed by increased RBC Fe levels. Fe can be excreted from the liver, duodenum, and the rest of the body (solid magenta lines). (B) Results of mathematical modeling. Experimental values from Figure 3B are shown as circles. Modeled values are shown as lines.

(Figure 5F). Measurement of total fecal iron levels indicated no significant difference between any experimental group (Online Supplementary Figure S4). This was expected given that total fecal iron levels are affected by multiple factors beyond excretion, including dietary iron levels and iron absorption. Analysis of 59Fe-labeled compounds in feces was not possible - 59Fe levels were low at the time of collecting the feces and decayed significantly once all mice had been processed in the excretion study. As fecal ferritin levels have been reported to reflect body iron levels19, we next measured fecal ferritin levels by ELISA. Similar to our estimates of fecally excreted iron (Figure 5F), fecal ferritin levels were increased in all Trfhpx/hpx mice relative to Trf+/+ mice and increased in untreated Trfhpx/hpx mice relative to treated Trfhpx/hpx mice (Figure 5G, Online Supplementary Figure S5). We next attempted to assess ferritin iron levels in fecal samples. Using mouse liver lysates, we established that the same iron stain used for histology could be used to detect a species very abundant in Trfhpx/hpx mouse liver that comigrated with ferritin heavy and light chain under native PAGE conditions (Online Supplementary Figure S6). However, native PAGE of fecal lysates failed to reveal stainable iron (data not shown).

reproduce iron levels in 6-month old untreated Trfhpx/hpx mice. As most untreated Trfhpx/hpx mice die before 6 months, iron loss at this age may not reflect physiological excretion but rather cell death secondary to severe iron excess or other long-term adverse effects of transferrin deficiency. When we simulated body iron levels in transferrin-treated Trfhpx/hpx mice, we could reproduce body iron levels up to 4 months but were initially unable to reproduce the decrease in body iron levels from 4 to 6 months. We considered that the decrease in iron levels from 4 to 6 months in transferrin-treated Trfhpx/hpx mice required an increase in excretion from one or more compartments or a decrease in absorption starting at 4 months. The smallest change in absorption or excretion that fitted the data was a four-fold increase in excretion from the duodenum and ‘rest of body’. Given that ‘rest of body’ in our model comprised other gastrointestinal regions including jejunum, ileum, and large intestine, this supports our 59Febased studies indicating that the gastrointestinal tract is the main route of excretion.

Mathematical modeling predicts increased excretion rates in treated Trfhpx/hpx mice

In this study, we exploited our initial observation that transferrin treatment decreases the concentrations of iron in the organs of Trfhpx/hpx mice to explore the basis of iron excretion. For this objective, the Trfhpx/hpx model has some key advantages over other mouse models of common human diseases of iron excess such as hereditary hemochromatosis and β-thalassemia. First, Trfhpx/hpx mice develop more severe iron excess than most other models. We anticipated that the severity of iron excess would significantly increase iron levels in potential routes of excretion such as feces and urine. Second, the primary defect in Trfhpx/hpx mice can be rapidly corrected pharmacological-

We also analyzed body iron levels from Figure 3B using our mathematical model of iron homeostasis, summarized in Figure 6A.20 The goal here was to explore the decrease in body iron content in transferrin-treated Trfhpx/hpx mice from 4 to 6 months (Figure 3B,C) - our excretion study was halted at 4.5 months as not all untreated Trfhpx/hpx mice survived to 6 months. Using the model, we could fit body iron levels from Trf+/+ and untreated Trfhpx/hpx mice with good agreement between experimental and modeled values (Figure 6B). However, we were unable to 686

Discussion

haematologica | 2019; 104(4)

Iron excretion in iron excess ly. In contrast, the primary defect in β-thalassemia mouse models - β-globin mutations - cannot. Red cell transfusions can reverse the anemia in this disease but also introduce a large burden of exogenous iron. As mentioned above, organ iron concentrations decreased in transferrin-treated Trfhpx/hpx mice. A similar phenomenon has been described in mouse models of hereditary hemochromatosis and β-thalassemia. Hepcidin deficiency is a key characteristic of both diseases. Administration of pharmacological agents that induce hepcidin expression or mimic hepcidin activity decreased iron levels in mouse models of these diseases.21– 25 Whether correction of hepcidin deficiency simply prevented worsening of organ iron excess or led to mobilization of iron from these organs and excretion from the body remains to be determined. We propose that the decreased organ iron levels in these models reflects a combination of normalized iron absorption and increased excretion rates. The increased excretion rates we estimated for Trf mice are similar to those previously reported. In two of the earliest studies on iron excretion in mice, injected 55Fe cleared from the body of Swiss mice with a half-life of 140 days, which is equivalent to a loss of 0.5% body 55Fe per day.17,18 We observed that 59Fe cleared from all Trf mice except untreated male Trfhpx/hpx mice with a half-life of 80-120 days and a loss of 0.6-0.8% body 59Fe per day (Figure 4D,E). In the older studies, iron-sufficient Swiss mice excreted 11.5 mg iron/day, while mice with increased body iron secondary to dietary or intravenous iron loading excreted 14-57 mg iron/day. These values were similar to those we predicted for Trf mice based on our 59Fe studies (Figure 5C). We also used rates of body 59Fe loss in urine and feces to estimate the rate at which iron was excreted via urine and feces. Measured urinary iron levels agreed with our 59Febased estimates of urinary iron levels except for untreated mutant mice, in which actual iron levels were much higher than predicted. The reason for this underestimation is not clear, although it may reflect the fact that perturbations in iron homeostasis in untreated Trfhpx/hpx mice are quite severe compared to those in Trf+/+ and treated Trfhpx/hpx mice. We also estimated the levels of fecally excreted iron in all mouse groups. While total fecal iron levels were not informative, the relative abundance of fecal ferritin was matched by the relative abundance of fecally excreted iron. Whether fecal ferritin solely represents a marker of iron excess in Trf mice or plays a mechanistic role in iron excretion remains to be determined. The source of fecal ferritin is also not known at this time. Possible sources include sloughed epithelial cells and biliary excretion. We propose that our study can be used as an initial step in a reconsideration of the physiological basis of iron excretion and the significance of its role in iron homeostasis. While humans and rodents may differ in their rates and routes of iron excretion, the possibility that iron excretion affects body iron levels has implications for treatment of human disease. Notably, a seminal work by Green et al. in 1968 indicated that adult men of Bantu origin, a population with increased iron stores, have increased daily iron losses.26 Development of hepcidin mimetics or agonists is an active area of research and may lead to novel treatments for hereditary hemochromatosis, β-thalassemia, and other diseases of iron excess.27 If body iron levels are regulated largely by absorption, treatment haematologica | 2019; 104(4)

of patients with hepcidin agonists or mimetics will prevent worsening of iron excess but will not reverse it additional treatment modalities such as chelation will be required to clear excess iron from the body. If body iron levels do influence iron excretion, treatment of patients with hepcidin agonists should result in decreased body iron burden - with rates of iron absorption normalized, excess iron will clear from the body through physiological mechanisms. The means by which iron is excreted from the body have not yet been established. Iron excretion is currently attributed to multiple processes including exfoliation of dead skin, blood loss, and turnover of intestinal epithelium (Figure 7). Our data indicating that iron is excreted largely via the gastrointestinal tract suggest that skin exfoliation does not play a prominent role. The possibility that increased blood loss contributes prominently to excretion in Trfhpx/hpx mice is also unlikely given that 59Fe half-lives did not decrease in Trfhpx/hpx mice relative to those in Trf+/+ mice (Figure 4D). The possibility that turnover of intestinal epithelium is a major route of iron excretion is stronger. It is supported by the fact that intestinal epithelium in mammals turns over in less than 1 week.28–30 Trfhpx/hpx mice do load excess iron into gastrointestinal organs (Table 1) but histological iron staining indicates that a considerable fraction of this iron in younger mutant mice resides in smooth muscle, not enterocytes (Figure 2). Pountney et al. previously demonstrated that non-heme iron levels are similar in enterocytes isolated from Trf+/+ mice and treated Trfhpx/hpx mice.31 Based on this, we suggest that the increased duodenal iron levels we measured in Trfhpx/hpx mice largely reflect smooth muscle iron loading (Figure 2). The observation by Pountney et al. that transferrin can be internalized by enterocytes isolated from Trfhpx/hpx mice may also explain stainable iron observed in enterocytes of treated but not untreated Trfhpx/hpx mice - it may represent uptake of diferric transferrin across the basolateral membrane of enterocytes. Overall, a careful investigation of the potential role of epithelial turnover to iron excretion would require a quantitative assessment of multiple factors: enterocyte iron levels, the biochemical form of enterocyte iron, and the rate of epithelial turnover in gastrointestinal organs in multiple models of iron excess and deficiency. Another potential contributor to gastrointestinal iron excretion is hepatobiliary excretion. This process is largely ignored by the current view of mammalian iron biology. The reason for this is not apparent. One study in rats excluded bile as a route of excretion, but this was based on the observation that bile duct ligation did not impair the decrease in body iron levels in rats switched from an iron-rich to an iron-deficient diet.32 The use of bile duct ligation is a concern given that this is an established method for inducing liver cirrhosis.33 Multiple studies, most of which were performed in rats, have shown that iron is readily detectable in bile and that biliary iron levels decrease in conditions of iron deficiency and increase in conditions of iron excess.34–52 Several of these studies involved the use of chelators - our statement that iron is readily detectable in bile refers to the measurement of biliary iron in control animals not exposed to chelators. Overall, a full investigation of the contribution of biliary excretion to systemic iron excretion would require measurement of multiple parameters. While our preliminary analysis indicates that biliary iron levels are increased in 687

C.J. Mercadante et al.

Figure 7. Model of iron absorption and excretion. Dietary non-heme iron (Fe) is absorbed by enterocytes in the small intestine. Enterocytes export Fe into the blood; this process can be inhibited by hepcidin. Fe is then transported to the liver for storage, excretion, or distribution to other organs in the body. Fe can be excreted by the liver into bile and transported into the small intestine, where it can undergo enterohepatic circulation or can be eliminated from the body via the feces. Fe can also be excreted from the body by turnover of epithelial cells lining the intestines or from minor trauma to intestinal epithelium leading to blood loss. Dashed lines indicate minor routes of Fe excretion, which include blood loss, exfoliation of dead skin, and excretion via the urine. For the sake of simplicity, not all organs or pathways of Fe transport are shown, including those that mediate heme Fe absorption.

untreated Trfhpx/hpx mice relative to those in Trf+/+ mice at 2.5 months of age (Online Supplementary Figure S7), the rate at which biliary iron is eliminated from the body is not only influenced by biliary iron levels. Dietary iron deficiency can alter rates of bile synthesis in rats.53 In rats, iron can also undergo enterohepatic circulation, the process by which substances excreted in bile are reabsorbed by the small intestine and transported back to the liver.54 A study of the role of hepatobiliary iron excretion would require measurement of bile synthesis rates and iron levels and rates of enterohepatic circulation in multiple animal models of iron excess and deficiency. Analysis of the biochemical form of biliary iron is also warranted, as this may indicate a potential mechanism for iron excretion. Iancu et

References 1. Coffey R, Ganz T. Iron homeostasis: an anthropocentric perspective. J Biol Chem. 2017;292(31):12727–12734. 2. Dev S, Babitt JL. Overview of iron metabolism in health and disease. Hemodial Int Int Symp Home Hemodial. 2017;21 (Suppl 1)S6–S20. 3. Muckenthaler MU, Rivella S, Hentze MW, Galy B. A red carpet for iron metabolism. Cell. 2017;168(3):344–361. 4. Papanikolaou G, Pantopoulos K. Systemic iron homeostasis and erythropoiesis. IUBMB Life. 2017;69(6):399–413. 5. Wallace DF. The regulation of iron absorption and homeostasis. Clin Biochem Rev. 2016;37(2):51–62. 6. Brissot P, Cavey T, Ropert M, Guggenbuhl P, Loréal O. Genetic hemochromatosis: pathophysiology, diagnostic and therapeutic management. Presse Med. 2017;46(12 Pt 2):e288–e295. 7. Kawabata H. The mechanisms of systemic iron homeostasis and etiology, diagnosis, and treatment of hereditary hemochromatosis. Int J Hematol. 2018;107(1):31–43. 8. Asadov C, Alimirzoeva Z, Mammadova T,

688

10.

11. 12.

13.

14.

al. observed electron-dense material within bile canaliculi similar in appearance to hemosiderin.55 This type of study, along with a study of the contribution of epithelial turnover to iron excretion, would also need to be performed in both male and female subjects given that 59Fe excretion rates were decreased in untreated male but not female Trfhpx/hpx mice (Figure 4E). Acknowledgments The authors would like to thank Joseph Orchardo and David Murray for assistance with metal measurements and Iqbal Hamza and Anatoly Zhitkovich for reviewing the manuscript. This work was supported by NIH grants DK84122 and DK110049 (TBB) and GM080219 (PM).

Aliyeva G, Gafarova S, Mammadov J. βthalassemia intermedia: a comprehensive overview and novel approaches. Int J Hematol. 2018;108(1):5-21. Cappellini MD, Motta I. New therapeutic targets in transfusion-dependent and -independent thalassemia. Hematol Am Soc Hematol Educ Program. 2017;2017(1):278– 283. Gupta R, Musallam KM, Taher AT, Rivella S. Ineffective erythropoiesis: anemia and iron overload. Hematol Oncol Clin North Am. 2018;32(2):213–221. Bu JT, Bartnikas TB. The use of hypotransferrinemic mice in studies of iron biology. Biometals. 2015;28(3):473–480. Bartnikas TB, Andrews NC, Fleming MD. 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. Buys SS, Martin CB, Eldridge M, Kushner JP, Kaplan J. Iron absorption in hypotransferrinemic mice. Blood. 1991;78(12):3288– 3290.

15. Parmar JH, Mendes P. A computational model to understand mouse iron physiology and diseases. bioRxiv. 2018;323899. 16. 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. 17. Stevens AR, White PL, Hegsted DM, Finch CA. Iron excretion in the mouse. J Biol Chem. 1953;203(1):161–165. 18. Chappelle E, Gabrio BW, Stevens AR, Finch CA. Regulation of body iron content through excretion in the mouse. Am J Physiol. 1955;182(2):390–392. 19. Skikne BS, Whittaker P, Cooke A, Cook JD. Ferritin excretion and iron balance in humans. Br J Haematol. 1995;90(3):681– 687. 20. Parmar JH, Davis G, Shevchuk H, Mendes P. Modeling the dynamics of mouse iron body distribution: hepcidin is necessary but not sufficient. BMC Syst Biol. 2017;11(1): 57. 21. Casu C, Oikonomidou PR, Chen H, et al. Minihepcidin peptides as disease modifiers in mice affected by β-thalassemia and polycythemia vera. Blood. 2016;128(2):265– 276.

haematologica | 2019; 104(4)

Iron excretion in iron excess

22. Gelderman MP, Baek JH, Yalamanoglu A, et al. Reversal of hemochromatosis by apotransferrin in non-transfused and transfused Hbbth3/+ (heterozygous b1/b2 globin gene deletion) mice. Haematologica. 2015;100(5):611–622. 23. Guo S, Casu C, Gardenghi S, et al. Reducing TMPRSS6 ameliorates hemochromatosis and β-thalassemia in mice. J Clin Invest. 2013;123(4):1531–1541. 24. Ramos E, Ruchala P, Goodnough JB, et al. Minihepcidins prevent iron overload in a hepcidin-deficient mouse model of severe hemochromatosis. Blood. 2012;120(18): 3829–3836. 25. Schmidt PJ, Toudjarska I, Sendamarai AK, et al. An RNAi therapeutic targeting Tmprss6 decreases iron overload in Hfe(-/-) mice and ameliorates anemia and iron overload in murine β-thalassemia intermedia. Blood. 2013;121(7):1200–1208. 26. Green R, Charlton R, Seftel H, et al. Body iron excretion in man: a collaborative study. Am J Med. 1968;45(3):336–353. 27. Casu C, Nemeth E, Rivella S. Hepcidin agonists as therapeutic tools. Blood. 2018;131(16):1790–1794. 28. Cheng H, Bjerknes M. Cell production in mouse intestinal epithelium measured by stathmokinetic flow cytometry and Coulter particle counting. Anat Rec. 1983;207(3):427–434. 29. Creamer B, Shorter RG, Bamforth J. The turnover and shedding of epithelial cells. I. The turnover in the gastro-intestinal tract. Gut. 1961;2110–118. 30. Williams JM, Duckworth CA, Burkitt MD, Watson AJM, Campbell BJ, Pritchard DM. Epithelial cell shedding and barrier function: a matter of life and death at the small intestinal villus tip. Vet Pathol. 2015;52(3): 445–455. 31. Pountney DJ, Konijn AM, McKie AT, et al. Iron proteins of duodenal enterocytes isolated from mice with genetically and experimentally altered iron metabolism. Br J Haematol 1999;105(4):1066–1073. 32. Oates PS, Jeffrey GP, Basclain KA, Thomas C, Morgan EH. Iron excretion in iron-overloaded rats following the change from an iron-loaded to an iron-deficient diet. J Gastroenterol Hepatol. 2000;15(6):665–674. 33. Marques TG, Chaib E, da Fonseca JH, et al. Review of experimental models for induc-

haematologica | 2019; 104(4)

34.

35.

36.

37.

38.

39.

40.

41.

42.

43. 44.

ing hepatic cirrhosis by bile duct ligation and carbon tetrachloride injection. Acta Cir Bras. 2012;27(8):589–594. Bláha K, Cikrt M, Nerudová J, Ponka HF. Biliary iron excretion in rats following treatment with analogs of pyridoxal isonicotinoyl hydrazone. Blood. 1998;91(11): 4368–4372. Goss JA, Barshes NR, Karpen SJ, Gao F-Q, Wyllie S. Liver ischemia and ischemiareperfusion induces and trafficks the multispecific metal transporter Atp7b to bile duct canaliculi: possible preferential transport of iron into bile. Biol Trace Elem Res. 2008;122(1):26–41. Hultcrantz R, Glaumann H. Studies on the rat liver following iron overload: biochemical studies after iron mobilization. Lab Investig. 1982;46(4):383–392. Drummond GS, Rosenberg DW, Kappas A. Intestinal heme oxygenase inhibition and increased biliary iron excretion by metalloporphyrins. Gastroenterology. 1992;102(4 Pt 1):1170–1175. Schümann K, Schäfer SG, Forth W. Iron absorption and biliary excretion of transferrin in rats. Res Exp Med (Berl.) 1986;186 (3):215–219. Zanninelli G, Choudury R, Loréal O, et al. Novel orally active iron chelators (3hydroxypyridin-4-ones) enhance the biliary excretion of plasma non-transferrinbound iron in rats. J Hepatol 1997;27(1): 176–184. Brissot P, Zanninelli G, Guyader D, Zeind J, Gollan J. Biliary excretion of plasma nontransferrin-bound iron in rats: pathogenetic importance in iron-overload disorders. Am J Physiol. 1994;267(1 Pt 1):G135-142. Dijkstra M, Kuipers F, Smit EP, de Vries JJ, Havinga R, Vonk RJ. Biliary secretion of trace elements and minerals in the rat. Effects of bile flow variation and diurnal rhythms. J Hepatol. 1991;13(1):112–119. Allain P, Leblondel G, Mauras Y. Effect of aluminum and deferoxamine on biliary iron elimination in the rat. Proc Soc Exp Biol Med. 1988;188(4):471–473. Brissot P, Deugnier Y, Guyader D, et al. Iron overload and the biliary route. Adv Exp Med Biol. 1994;356277–283. Hall ED, Symonds HW. The maximum capacity of the bovine liver to excrete manganese in bile, and the effects of a man-

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

ganese load on the rate of excretion of copper, iron and zinc in bile. Br J Nutr. 1981;45(3):605–611. Symonds HW, Mather DL, Hall ED. Surgical procedure for modifying the duodenum in cattle to measure bile flow and the diurnal variation in biliary manganese, iron, copper and zinc excretion. Res Vet Sci. 1982;32(1):6–11. Deol HS, Howell JM, Dorling PR, Symonds HW. The effect of copper and heliotrope on the composition of bile in sheep. Res Vet Sci. 1992;53(3):324–330. Albers CJ, Huizenga JR, Krom RA, Vonk RJ, Gips CH. Composition of human hepatic bile. Ann Clin Biochem. 1985;22 ( Pt 2)129– 132. Cikrt M, Ponka P, Necas E, Neuwirt J. Biliary iron excretion in rats following pyridoxal isonicotinoyl hydrazone. Br J Haematol. 1980;45(2):275–283. Gooneratne SR, Chaplin RK, Trent AM, Christensen DA. Effect of tetrathiomolybdate administration on the excretion of copper, zinc, iron and molybdenum in sheep bile. Br Vet J. 1989;145(1):62–72. Lévy P, Dumont M, Brissot P, et al. Acute infusions of bile salts increase biliary excretion of iron in iron-loaded rats. Gastroenterology. 1991;101(6):1673–1679. LeSage GD, Kost LJ, Barham SS, LaRusso NF. Biliary excretion of iron from hepatocyte lysosomes in the rat. A major excretory pathway in experimental iron overload. J Clin Invest. 1986;77(1):90–97. Hultcrantz R, Angelin B, Björn-Rasmussen E, Ewerth S, Einarsson K. Biliary excretion of iron and ferritin in idiopathic hemochromatosis. Gastroenterology. 1989;96(6): 1539–1545. Prasnicka A, Cermanova J, Hroch M, et al. Iron depletion induces hepatic secretion of biliary lipids and glutathione in rats. Biochim Biophys Acta. 2017;1862(12): 1469–1480. Brissot P, Bolder U, Schteingart CD, Arnaud J, Hofmann AF. Intestinal absorption and enterohepatic cycling of biliary iron originating from plasma non-transferrin-bound iron in rats. Hepatology. 1997;25(6):1457–1461. Iancu TC, Shiloh H, Raja KB, et al. The hypotransferrinaemic mouse: ultrastructural and laser microprobe analysis observations. J Pathol. 1995;177(1):83–94.

689

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):690-699

Red Cell Biology & its Disorders

Hemodynamic provocation with acetazolamide shows impaired cerebrovascular reserve in adults with sickle cell disease Lena Václavů,1 Benoit N. Meynart,1 Henri J.M.M. Mutsaerts,1 Esben Thade Petersen,2 Charles B.L.M. Majoie,1 Ed T. VanBavel,3 John C. Wood,4 Aart J. Nederveen1 and Bart J. Biemond5

Amsterdam UMC, Radiology and Nuclear Medicine, University of Amsterdam, the Netherlands; 2Danish Research Centre for Magnetic Resonance, Centre for Functional and Diagnostic Imaging and Research, Copenhagen University Hospital Hvidovre, Denmark; 3Amsterdam UMC, Biomedical Engineering and Physics, University of Amsterdam, the Netherlands; 4Cardiology and Radiology, Children's Hospital of Los Angeles, CA, USA; 5Amsterdam UMC, Hematology, Internal Medicine, University of Amsterdam, the Netherlands 1

ABSTRACT

Correspondence: LENA VACLAVU l.vaclavu@amc.uva.nl Received: September 5, 2018. Accepted: November 23, 2018. Pre-published: December 6, 2018. doi:10.3324/haematol.2018.206094 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/690 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

690

ickle cell disease is characterized by chronic hemolytic anemia and vascular inflammation, which can diminish the vasodilatory capacity of the small resistance arteries, making them less adept at regulating cerebral blood flow. Autoregulation maintains adequate oxygen delivery, but when vasodilation is maximized, the low arterial oxygen content can lead to ischemia and silent cerebral infarcts. We used magnetic resonance imaging of cerebral blood flow to quantify whole-brain cerebrovascular reserve in 36 adult patients with sickle cell disease (mean age, 31.9±11.3 years) and 11 healthy controls (mean age, 37.4±15.4 years), and we used high-resolution 3D FLAIR magnetic resonance imaging to determine the prevalence of silent cerebral infarcts. Cerebrovascular reserve was calculated as the percentage change in cerebral blood flow after a hemodynamic challenge with acetazolamide. Coregistered lesion maps were used to demonstrate prevalent locations for silent cerebral infarcts. Cerebral blood flow was elevated in patients with sickle cell disease compared to controls (median [interquartile range]: 82.8 [20.1] vs. 51.3 [4.8] mL/100g/min, P<0.001). Cerebral blood flow was inversely associated with age, hemoglobin, and fetal hemoglobin, and correlated positively with bilirubin, and LDH, indicating that cerebral blood flow may reflect surrogates of hemolytic rate. Cerebrovascular reserve in sickle cell disease was decreased by half compared to controls (34.1 [33.4] vs. 69.5 [32.4] %, P<0.001) and was associated with hemoglobin and erythrocyte count indicating anemia-induced hemodynamic adaptations. In total, 29/36 patients (81%) and 5/11 controls (45%) had silent cerebral infarcts (median volume of 0.34 vs. 0.02 mL, P=0.03). Lesions were preferentially located in the borderzone. In conclusion, patients with sickle cell disease have a globally reduced cerebrovascular reserve as determined by arterial spin labeling with acetazolamide and reflects anemia–induced impaired vascular function in sickle cell disease. This study was registered at clinicaltrials.gov identifier 02824406.

Introduction Sickle Cell Disease (SCD) is associated with chronic hemolytic anemia and vascular inflammation,1 with progressive multiorgan damage including nephropathy, pulmonary hypertension, priapism, leg ulcers, and stroke.2 Manifestations of progressive cerebral injury in SCD include overt stroke as well as silent cerebral haematologica | 2019; 104(4)

Impaired CVR in adults with sickle cell disease Table 1. Patient characteristics and baseline measurements. Characteristic Age, years Sex Bodyweight, kg Ethnicity, n (%) South America: Suriname Western Africa: Benin, Congo, Ghana, Guinea, Nigeria, Sierra Leone Caribbean: Dutch Antilles, Jamaica Europe: Turkey Medication & therapy Hydroxyurea, n (%) Chronic blood transfusion therapy, n (%) Cardiovascular risk factors Systolic blood pressure, mmHg Diastolic blood pressure, mmHg MAP, mmHg Heart rate, bpm Nicotine smokers, n (%) Cannabis smokers, n (%) Hematologic characteristics Genotype

Controls

Patients with SCD

n=11 37.4 ± 15.4 6 men, 5 women 76 (14)

n=36 31.9 ± 11.3 23 men, 12 women 70 (18)

0.52 0.46++ <0.01

7 (64) 2 (18) 1 (9) 1 (9)

19 (54) 11 (31) 4 (11) 1 (3)

0.59 0.39 0.83 0.38

13 (37) 3 (9)

133±10 87±7 102±7 72±17 1 (9) 1 (9)

121±10 71±8 87±7 76±11 9 (26) 4 (11)

<0.01 <0.001 <0.001 0.51 0.26++ 0.85++

HbAA (n=9, 82%) HbSS (n=31, 89%) HbAS (n=2, 18%) HbS 0 (n=4, 11%) Hemoglobin, g/dL 13.6±1.3 8.8±1.4↓ <0.001 Reticulocyte count, % 1.3±0.5 8.9±4.2↑ <0.001 Reticulocyte count, #109/L 60±25 261±108↑ <0.001 Bilirubin total, mg/dL 0.6±0.5 3.1±2.0↑ <0.001 ASAT, U/L 39.3±31.1 48.1±16.2↑ <0.05 LDH 190±31 459±165↑ <0.05 ++ 2 c test. *T-test, or Wilcoxon signed rank test was used to test the statistical significance of the difference as appropriate. MAP: mean arterial pressure: [(2 *DiastolicBP) + SystolicBP)]/ 3 mmHg. ↑ above healthy reference / ↓ below healthy reference. LDH: Lactate dehydrogenase.

infarcts (SCI).3,4 Although SCI were once thought to be benign, SCI volume is associated with reduced cognitive performance in children with SCD,5 and a 14-fold increased risk of overt stroke in pediatric SCD.6 SCI risk increases relentlessly with age,7,8 reaching a prevalence of 50% by the age of 30.9 There is currently no treatment for SCIs in adults, although efforts to reduce their incidence by blood transfusions have been made.10 Nevertheless, identifying modifiable risk factors and predictors of these lesions is a focus of current research in adult SCD. Cerebrovascular reserve (CVR) is a measure of the viability of cerebral vessels to respond to a vasoactive stimulus and is often used to study hemodynamic status in neurovascular diseases.11 CVR is defined as the remaining vasodilating capacity of the cerebral arterioles in response to an exogenous stimulus such as CO2 or acetazolamide.12 Recent studies suggest that impaired CVR predicts locations of lesions at one year follow up13 and the risk of stroke in steno-occlusive disease,14,15 providing compelling evidence that this hemodynamic marker can inform future cerebral damage observed on MRI. However, the predictive value of CVR in patients with SCD has not yet been shown. Previous cross-sectional studies have observed reduced CVR in children and adults haematologica | 2019; 104(4)

with SCD without a history of overt stroke,16–20 and we have recently shown dilated cerebral vessels at baseline in SCD.21 Together, these findings suggest that high resting blood flow demands are met by vasodilation. Vasodilation at rest will limit further dilation in times of increased demand, which poses a risk for ischemia. Examples of such risk in SCD include infection, fever, acute anemic events,22 and obstructive sleep apnea.23 CVR was previously associated with anemia,20 and dilatory function of the vessels could additionally be impaired in SCD due to vascular inflammation, low nitric oxide and abnormal endothelial function. Since the brain cannot store its own oxygen, it must maintain constant perfusion, and inadequate oxygen delivery by increases in demand or low hemoglobin may cause ischemia and SCIs due to a lack of vasodilatory reserve. We hypothesized that CVR is lower in patients with SCD compared to healthy controls, and that SCIs are related to low CVR. The objective of this study was to investigate regional CVR measurements in SCD and to investigate the association between CVR and the presence and volume of existing ischemic lesions in SCD. In the current study, we used arterial spin labelling (ASL), a non-contrast perfusion MRI method, to assess whole brain CBF prior to and following 691

L. Václavu° et al.

Figure 1. Lesion maps in adult patients with sickle cell disease and healthy controls. Lesions are scaled to local count of participants with a lesion. The local maximum of 2 lesions were detected in the healthy controls (upper row) and 7 in the SCD cohort (bottom row). Deep white matter, periventricular and border zone regions exhibited the highest probability of lesions. Also note the large posterior infarct in one patient in the bottom row.

cerebral vasodilation with acetazolamide. We compared hemodynamic MRI parameters between adult SCD patients in steady-state and without a history of stroke, with healthy controls.

Methods Participants The local Institutional Review Board at the Academic Medical Center, the Netherlands, approved this study, which was carried out in accordance with the Declaration of Helsinki. Adult patients were recruited by their SCD specialist hematologist from the outpatient clinic, and race-and age-matched controls consisted of their healthy non-SCD friends and family. Inclusion criteria were: informed consent, SCD (HbSS/HbSβ0-thalassemia), age (>18), and steady state (absence of an acute SCD-related event 1 month prior to participation). Exclusion criteria were: contraindications to acetazolamide or MRI, clinical history of overt infarct/hemorrhagic stroke, brain tumor, brain surgery, or serious neurologic event. Participants were asked to refrain from consuming alcohol and caffeinated drinks on the day of the examination.

Order of procedures Participants first underwent a blood pressure measurement and blood draw. Subsequently, an intravenous catheter was placed at the site of cannulation for acetazolamide administration during the MRI scan. Blood pressure was measured before and after the MRI and heart rate was monitored continuously during the MRI. Subjects were asked about side-effects afterwards (Online Supplementary Table S1).

Biological parameters Blood samples were drawn from an antecubital vein directly prior to MRI and assessed using standard laboratory procedures. Genotype was confirmed by high-performance liquid chromatography (HPLC) and DNA analysis. Missing lab data were dealt with by last steady-state observation carried forward. Markers indicating hemolysis were defined as serum levels of bilirubin, reticulocyte count and lactate dehydrogenase (LDH).24 In addition, hemoglobin concentration, MCV, HbF%, HbS%, leukocyte count, platelet count, ASAT, ALAT, creatinine and CRP were determined.

MR imaging We performed 3T MRI (Philips Ingenia) with a 32-channel 692

receive head coil in all participants. For CBF, a pseudo-continuous arterial spin labelling (pCASL) sequence was used with a 2D gradient echo FFE single shot echo-planar imaging (EPI) readout with a TR/TE of 4400/14 ms, FOV 240 x 240 mm, voxel size 3 x 3 x 7 mm, post-label delay 1800 ms, label duration 1800 ms, 19 axial slices, flip angle 90°, SPIR fat suppression, 140 label-control pairs, background suppression, and a total scan duration of 20 min. In addition, we acquired 3D time-of-flight magnetic resonance angiography and a 3D fluid-attenuated inversion recovery (FLAIR) sequence for lesion assessment. After 5 min of pCASL, participants received 16 mg/kg acetazolamide (Diamox®, Mercury Pharmaceuticals Ltd., London, UK) with a maximum of 1400 mg. Acetazolamide was dissolved in 20 mL saline (NaCl 0.9%) and injected intravenously at a flow rate of 0.1 mL/s, and flushed with 10 mL saline. Voxel-wise CVR was calculated by: CVR (%) = (ΔCBF)/CBF-PRE x 100%, where ΔCBF represents the average of the first 5 min (CBF-PRE) of the pCASL CBF timeseries subtracted from the average of the final 5 min. We looked at gray matter (GM) and white matter (WM) CBF and CVR by applying the subject-specific anatomic masks to each subject’s CBF map. CBF quantification was customized to improve the accuracy of CBF by using a dual compartment flow model incorporating T1 of blood, measured directly in each subject in the sagittal sinus.25 We also incorporated a labelling efficiency correction based on velocity measured with phase-contrast MRI, and a correction for the arterial transit time, measured with a separate multiple inversion time sequence. The 3T MRI protocol and image analysis is described in more detail in the Online Supplementary Methods.

Lesions FLAIR images were manually segmented and validated by a neuroradiologist (CBM >20 years of experience), blinded to the medical status of the patient. We quantified voxel-wise prevalence, subject-wise prevalence, total volume, and total number of lesions. Lesions were defined as multiple (>1) signal hyperintensities ≥5 mm in diameter. These lower limits were chosen to maintain external validity with previous studies in adults with SCD.26–28 Lesion diameter was defined as the maximum length along the major axis of a lesion in 3D. Since the contribution of different types of lesions to specific impairments is not known, we also included the following lesions in the total lesion volume calculation: lacunar lesions, defined as round or ovoid subcortical fluid-filled cavities, and cortical infarcts, defined as (fluid-filled) regions of haematologica | 2019; 104(4)

Impaired CVR in adults with sickle cell disease

Figure 2. Dynamic gray matter CBF time-course in response to acetazolamide. Acetazolamide administration elicited a robust response in healthy controls (black solid line with grey standard deviations) and patients with sickle cell disease (SCD)(red solid line with pink standard deviations). Absolute CBF changes in the left plot indicate a higher baseline, smaller absolute increase, and slower time to rise in patients with SCD compared to healthy controls. The relative CBF in the right plot show reduced CVR in patients with SCD. Both the CBF and CVR stabilized 10-15 minutes after injection of acetazolamide.

hyperintense necrotic tissue of variable size and shape located in the cortical tissue. A lesion density map was generated by image registration (described in the Online Supplementary Methods) and lesion contours were overlaid on the CVR images to visualize co-localization.

Statistics Statistical significance was assessed in R 3.4.3 (R Core Team (2017) R Foundation for Statistical Computing, Vienna, Austria) using appropriate tests (parametric for normally distributed and non-parametric for significantly non-normal distributed variables based on a Shapiro-Wilk test) to compare medians, means, or proportions between groups. Scatterplots of correlation analyses show both controls and patients, but exploratory correlation coefficients were computed in the patient group only, using Spearman’s rho (ρ). P-values were adjusted for multiple comparisons using the Benjamini-Hochberg method. Variables that were statistically significant (P<0.05) in univariate analysis, were entered as predictor variables in multivariate analysis using the standard enter method with CVR or lesions as an outcome variable. Lesion volume was used in linear regression, while lesion presence or absence was used in binary logistic regression with CVR as a predictor variable.

Results Demographic and clinical characteristics Thirty-six patients and 11 healthy controls were included in the study (Table 1). Patient and control populations were well matched for age, sex, and ethnicity. HPLC and DNA analysis confirmed that 32 (89%) patients had the HbSS genotype and 4 (11%) had HbSβ0 thalassemia. In the healthy control group, 2 (18%) were sickle cell gene carriers (HbAS). Thirteen (37%) patients with SCD were using hydroxyurea and 3 (9%) were receiving regular (every 3-5 weeks) blood transfusions. For those on transfusions, patients were studied 3-28 days since their last haematologica | 2019; 104(4)

transfusion. Regular blood transfusions were given to one patient for prevention of stroke upon detection of high TCD values when they were still in pediatric care, and the other patients were on transfusions for prevention of frequent hydroxycarbamide refractory vaso-occlusive crises/acute chest syndrome. Patients with SCD had lower blood pressure and body weight compared to healthy controls as well as expected differences in hematologic measurements (Table 1).

Anatomic neuroimaging findings MRA data were of high quality, except for two patients’ scans showing small motion artefacts, which precluded assessment. Incidental findings included the following: 1 patient had bilateral MCA occlusions with moyamoya syndrome and corresponding collaterals with diffuse white matter hyperintensity, 2 patients had a total of three aneurysms (smaller than 3 mm in diameter with a wide base), 2 patients had infundibula at the origin of the ophthalmic artery, 3 patients had tortuous vessels, and no arteriovenous malformations were found. The prevalence of white matter, cortical, periventricular and lacunar lesions in SCD patients was 29/36 (81%) and in healthy controls it was 5/11 (45%), P=0.02 (Figure 1). Two patients had cortical infarct (occipital and frontal lobes) and 5 patients had lacunar (fluid-filled cavity) infarcts. We found no lacunar or cortical infarcts in the healthy controls. Periventricular hyperintensity was observed in both groups. Patients with SCD had a similar number of lesions as compared to healthy volunteers (median [interquartile range] of 6 [19) lesions per patient versus 5 [8.5] lesions per healthy control) but significantly larger lesions with a median lesion volume of 0.34 [1.56] mL compared to healthy controls (0.02 [0.28] mL, P=0.03) (Table 2). The maximum number of lesions per subject in the co-registered lesion count map in patients with SCD was seven, located in the periventricular and borderzone regions (Figure 1). 693

L. VĂĄclavuÂ° et al. Table 2. Neuroimaging findings in healthy controls and patients with sickle cell disease.

MRAa Circle of Willis variantb Hypoplasiac Stenosisd Aneurysms (no. of patients)e Infundibulumf Moyamoya / collateralsg Tortuous/curved vessels FLAIR MRI Lacunar infarcts, n(%) Cortical infarcts, n(%) Periventricular infarcts, n(%) Total lesion count Lesion count per subject, median [IQR] Lesion volume, median mL [IQR] Prevalence of lesions (>1 lesion, >5 mm)

Controls (n=11)

Patients with SCD (n=36)

P++

1 7 0 0 0 0 0

6 12 2 (25-50%, occlusion) 2 (6%) 2 (6%) 1 (3%) 3 (ACA, MCA, PCOM)

0.54 0.07 0.42 0.42 0.42 0.58 0.32

0 0 1(9%) 83 1 [8] 0.02 [0.28] 5 (45%)

5 (14%) 2 (6%) 1(3%) 386 5 [15] 0.34 [1.56] 29 (81%)

0.19 0.42

0.27* 0.03 * 0.02

Chi-square test. * Wilcoxon rank sum test. aNormal MRA: normal defined as full Circle of Willis, excludes anatomic variants and hypoplasia. bVariant MRA: anatomic variant of the circle of Willis such as absence of a PCOM, a fetal variant PCA, non-fusion at origin of vertebro-basilar artery, early branching of a distal artery. cHypoplasia: diameter <1 mm for PCOM, or diameter < 2mm for ACA. dStenosis: (i)<25%; (ii) 25-50%, (iii) 50-75%, (iv) 75-99%, (v) occlusion. eAneurysm: number of patients affected by aneurysms. fInfundibulum: dilatational widening of the origin of a junctional artery. gMoyamoya syndrome: bilateral occlusion of the terminal portion of ICA or proximal MCA, with abnormal vascular networks (collaterals) in the vicinity of the occlusive lesions. MRA: magnetic resonance angiogram; FLAIR MRI: fluid attenuation inversion recovery magnetic resonance imaging; SCD: sickle cell disease; ACA: anterior cerebral artery; MCA: middle cerebral artery; PCOM: posterior communicating artery; IQR: interquartile range. ++

Dynamic cerebral blood flow response to acetazolamide The CBF and CVR time-series plotted in Figure 2 show that maximal dilatation was reached 10-15 minutes after acetazolamide in both patients and controls. All healthy subjects exhibited a robust response to acetazolamide without changes in blood pressure or heart rate (data not shown) measured before and after the scan. The most common side-effects after acetazolamide were dizziness, experienced by 26% of participants, headache in 13%, and also paresthesia in 13%. No side-effects required intervention (Online Supplementary Table S1).

Cerebral hemodynamics in SCD differ from healthy controls The boxplots in Figure 3 show that the 36 patients with SCD had higher gray matter (GM) CBF (median [interquartile range]: 82.8 [20.1] mL/100g/min) at baseline compared to the 11 healthy controls (51.3 [4.8] mL/100g/min, P<0.001). After acetazolamide, median GMCBF increased in patients to 108.3 [25.9] mL/100g/min, and in healthy controls to 85.5 [10.8] mL/100g/min, P<0.001. Patients with SCD had 49% lower median GM CVR (34.1 [33.4] %) compared to controls (69.5 [32.4] %, P<0.001). Median white matter (WM) CBF was higher at baseline in SCD patients (39.6 [10.9] mL/100g/min) compared to healthy controls (26.5 [3.0] mL/100g/min, P<0.001). In WM, there was a significant increase in CBF after acetazolamide in patients with SCD (P<0.001), as well as in healthy controls (P=0.002). Median WMCVR was 41% lower in patients with SCD compared to healthy controls (SCD: 27.1 [16.2]%; controls: 66.1 694

[37.3]%, P<0.001). Figure 4 shows the the co-registered maps indicating the SCD group had higher average CBF and lower average CVR compared to healthy controls. The lowest GM CVR of 5.2 % was found in the one patient who had comorbid moyamoya, and a low hemoglobin concentration of 6.4 g/dL. The small (n=3) group of patients receiving transfusions, and even fewer (n=1) receiving transfusions but no hydroxyurea, precluded a statistical comparison of these subgroups. However, fewer days since last transfusion appeared to be associated with a trend to higher CVR, as shown in the descriptive table in Online Supplementary Table S2. Additionally, there was no difference in GM CVR between patients receiving hydroxyurea and those not receiving hydroxyurea (P=0.89).

Factors associated with cerebral hemodynamics Hematologic parameters (Table 3) were explored for their correlation with CBF and CVR. Resting CBF was inversely associated with age, hemoglobin concentration, erythrocyte count and HbF %, and positively associated with LDH and total bilirubin. After adjustment for multiple comparisons, only hemoglobin concentration remained significantly associated with resting CBF. CVR was associated with baseline CBF, hemoglobin concentration, erythrocyte count, and creatinine, and negatively associated with platelet count. After adjustment for multiple comparisons, none remained significant. In multivariate analysis, only baseline CBF remained significantly associated with CVR (P=0.029).

Cerebral hemodynamics and lesion co-localization We observed regional variation in the group-averaged WM CVR maps as shown in the lower panel of Figure 4. Co-localization of CVR in lesions are shown by the lesion haematologica | 2019; 104(4)

Impaired CVR in adults with sickle cell disease Table 3. Spearman’s correlation coefficients between clinical parameters and GM cerebral blood flow and GM cerebrovascular reserve in adult patients with sickle cell disease.

Parameter Age Baseline GM CBF Hemoglobin concentration ASAT ALAT Leukocytes Platelets HbF% HbS% MCV Creatinine CRP Ferritin Markers of hemolysis Bilirubin LDH Reticulocyte %

GM Cerebral blood flow (CBF) Spearman’s rho (ρ)

GM Cerebrovascular reserve (CVR) Spearman’s rho (ρ)

- 0.36 - 0.61 0.26 - 0.03 0.30 0.30 - 0.41 0.10 0.05 0.07 -0.08 0.11

0.03 <0.001* 0.17 0.85 0.09 0.42 <0.05 0.64 0.80 0.69 0.71 0.57

0.17 - 0.43 0.40 0.01 0.12 - 0.18 - 0.41 0.37 - 0.32 -0.15 0.34 0.02 -0.18

0.33 0.01 0.02 0.94 0.49 0.32 0.03 0.07 0.13 0.38 <0.05 0.92 0.35

0.43 0.45 0.32

0.01 0.01 0.06

-0.23 -0.18 -0.32

0.19 0.33 0.06

*P values that remained significant after Benjamini-Hochberg procedure for multiple-comparison adjustment. GM CBF: granulocyte-macrophage cerebral blood flow; CVR: cerebrovascular reserve; ASAT: aspartate aminotransferase; ALAT: alanine aminotransferase; MCV: mean corpuscular volume; CRP: c-reactive protein; LDH: lactate dehydrogenase;

contour overlay in the bottom row of Figure 4. Binary logistic regression with lesion presence or absence as an outcome variable, showed that GM CVR was not a predictor of lesion prevalence (P=0.28), and neither was WM CVR (P=0.57). The same was true for GM CBF (P=0.18), age (P=0.24), hemoglobin levels (P=0.08), and the other blood markers. In linear regression analysis, CVR was not a significant predictor of lesion volume (P=0.669).

Discussion Chronic inflammation and hemolysis play a key role in the pathologic processes that can diminish the dilatory capacity of small resistance arteries in SCD. In our study, we observed a globally reduced gray matter CVR in patients with SCD without a history of stroke, compared to race-matched healthy controls. We found that CVR was particularly impaired in patients with high baseline CBF, indicating that in these patients, cerebral vasodilation was almost maximal at rest. Indeed, the lowest CVR of 5% was observed in a patient with SCD and comorbid moyamoya syndrome, in whom the highest CBF and diffuse white matter injury was observed. Silent cerebral infarcts (SCIs) were detected in the majority of patients, but these were not related to any of the hemodynamic MRI markers. The association between CBF and elevated lactate dehydrogenase and bilirubin levels suggest that blood flow may be related to higher hemolytic rate. However, this association was not significant after adjustment for multiple comparisons so remains to be investigated in future studies. Acetazolamide was well-tolerated by all participants haematologica | 2019; 104(4)

and did not induce vaso-occlusive crisis in any of the patients, indicating that this test can be performed safely to assess CVR in patients with SCD. A previous study also using intravenous acetazolamide administration in children with SCD to assess CVR with SPECT16 did not report on safety of acetazolamide, so it remains unclear if the authors had the same experiences regarding side-effects. Acetazolamide has the advantage over CO2 inhalation of inducing maximal dilation without metabolic changes, which allows a true assessment of vasodilatory capacity. We observed a plateau in the response to acetazolamide after 10-15 minutes, corresponding to maximal vasodilation. However, there was a difference in the maximal CBF between the groups, clearly showing a reduced vascular reserve capacity. This difference in CVR can be explained by chronically increased resting vessel diameter, as we have shown previously,21 which leaves these patients with little reserve for further vasodilation. Numerous resting ASL studies in children with SCD have shown that chronic anemia leads to high CBF20,29–32 and the high flow requirements in SCD could lead to a loss of autoregulatory capacity if dilatory reserve is being used for perfusion. Our dynamic CBF response supports our hypothesis that adult patients with SCD have severely reduced vasodilatory capacity. Hence, autoregulatory capacity is being used to maintain basal cerebral oxygenation, posing a risk for cerebral ischemia and infarction. The prevalence of SCIs found in our cohort was 81%, which is in line with previous reports on lesions in adults with SCD ranging from 15% to 90%.9,26–28,33 The large aforementioned variation arises from methodologic differences including improvements in imaging technology providing higher sensitivity, differences in age between 695

L. Václavu° et al. A

Figure 3. Differences between patients and controls, and associations among gray matter CBF, gray matter CVR, and hemoglobin. (A) The boxplot shows higher CBF in gray matter (GM) in patients with sickle cell disease (SCD) compared to healthy controls. (B) CVR in GM in patients with SCD was half of that of controls. (C) The scatterplot shows the significant association between CBF in GM and hemoglobin concentration in SCD patients. (D) The scatterplot shows that CVR in GM was significantly associated with hemoglobin levels in SCD. (E) The magnitude of the CVR in GM was significantly associated with resting CBF in GM. CBF: cerebral blood flow; CVR: cerebrovascular reserve; GM: gray matter; SCD: sickle cell disease.

patient cohorts,28 the size and number of the lesions,34 as well as differentiation between silent cerebral infarcts and lacunar infarcts.3,26 This sensitivity to technology was previously demonstrated in a study by our group performed in adult SCD patients,28 in which a 7T imaging field strength with 0.8 mm isotropic resolution was compared to 3T with 1 mm isotropic resolution, and found a lesion prevalence of 50% at 3T and 90% at 7T field strength, respectively. The relatively high prevalence of lesions (45%) identified in the control group seems consistent with having higher detection sensitivity when using improved technology.28 Hence, a consensus on lesion definition and measurement that is harmonized across technologies and sites is needed in order to compare studies from several cohorts in future studies. As demonstrated in the lesion density map in our study, SCI prone areas were primarily located in the deep white matter, watershed and borderzone regions,35 where perfusion is known to be lower, and ischemic risk thought to be higher in children with SCD,36 than in the cerebral cortex. While the pathogenesis of SCIs in SCD is still unclear, some evidence does show that the severity of anemia plays a role in SCIs in children with SCD.37 Even though insufficient cerebral oxygen delivery and the associated ischemic risk is probably due to chronic anemia, recent work demonstrated that additional acute moments of critical hypoperfusion may lead to lesions rather than chronic anemia by itself.7,22 Limited cerebrovascular reserve may 696

be the underlying condition that places patients at increased vulnerability for inadequate oxygen delivery and extraction during acute anemia or superimposed hypoxia. Hence, low CVR itself may not be a sufficient condition to initiate lesion formation, but an additional crucial ‘second hit’, such as acute anemic events or superimposed hypoxia, is probably necessary. We hypothesize that interventions that improve oxygen delivery such as blood transfusion, hydroxyurea or new disease-modifying drugs that reduce hemolysis may improve CVR and thereby reduce risk for cerebral ischemia and infarction. Whether reduced CVR is an independent risk factor for SCIs remains to be demonstrated in a prospective trial. The limitations of this study include potential selection bias of patients with no history of stroke. This may have induced a bias towards less severe patients and thereby also less severe white matter injury burden. Indeed, lesions were mostly small punctate lesions, with a total median lesion volume of 0.34 mL, which is low compared to total brain volume. However, since these lesions were present in the majority of patients, our cohort is likely to be a reasonable representation of the adult SCD population without overt stroke. We included 36 patients with SCD under the premise that we would have enough power to detect differences in CVR between patients and healthy controls based on an a priori sample size calculation. However, given that the correlation between lesion volume and CVR has not been studied previously in SCD, haematologica | 2019; 104(4)

Impaired CVR in adults with sickle cell disease

Figure 4. Axial slices of cohort averaged registered maps of hemodynamic MRI parameters. Upper panel shows CBF at baseline and post acetazolamide in GM for healthy controls and patients with sickle cell disease (SCD), and indicates clearly that CBF is elevated in SCD patients in all brain regions. Middle panel shows cohortaveraged co-registered CVR maps in GM and WM and illustrates that CVR is lower in patients with SCD compared to healthy controls in both grey and white matter regions of interest. CVR was not uniformly distributed and appeared to be higher in posterior compared to anterior regions in healthy controls and appear to be higher in watershed and periventricular regions in patients with SCD, with lowest CVR appearing in the deep white matter interface with grey matter. Bottom panel shows the lesion contours overlaid on white matter CVR in patients with SCD. CBF: cerebral blood flow; CVR: cerebrovascular reserve; GM: gray matter.

haematologica | 2019; 104(4)

697

L. Václavu° et al.

the lack of knowledge of expected values precluded a comprehensive sample size calculation so we were possibly underpowered to detect the hypothesized negative correlation between CVR and lesion volume. Another limitation is the cross-sectional design of our study. If CVR is an early indication of hemodynamic compromise in a certain brain region, then ischemic injury may not occur until oxygen delivery is repeatedly interrupted, which may explain the fact that no association between CVR and SCI was found. Or their treatment has improved their CVR and precluded detection of the association between CVR and SCIs that had formed prior to effective therapy. Interestingly, in a previous study in non-SCD patients severely affected with white matter lesions, a link between low CVR and progression of cerebral lesions was found one year later.13 Additionally, our study may have lacked sensitivity in white matter CVR values. The reason for this is that with ASL, CBF signal in the deep white matter is often below the noise level, which makes small changes in CBF after a CVR challenge even more difficult to detect. Hence, CVR values become less reliable further away from gray matter. Improving CBF signal in white matter can be achieved by acquiring ASL for a longer duration. Improvements in ASL technology and scan acceleration will hopefully make this available for clinical research soon. In conclusion, using ASL MRI in combination with hemodynamic provocation by acetazolamide, we demon-

References 1. Rees DC, Williams TN, Gladwin MT. Sickle-cell disease. Lancet. 2010;376(9757): 2018-2031. 2. Kato GJ, Gladwin MT, Steinberg MH. Deconstructing sickle cell disease: Reappraisal of the role of hemolysis in the development of clinical subphenotypes. Blood Rev. 2007;21(1):37-47. 3. DeBaun MR, Armstrong FD, McKinstry RC, et al. Silent cerebral infarcts: a review on a prevalent and progressive cause of neurologic injury in sickle cell anemia. Blood. 2012;119(20):4587-4597. 4. Prengler M, Pavlakis SG, Prohovnik I, Adams RJ. Sickle cell disease: the neurological complications. Ann Neurol. 2002; 51(5):543-52. 5. van der Land V, Hijmans CT, de Ruiter M, et al. Volume of white matter hyperintensities is an independent predictor of intelligence quotient and processing speed in children with sickle cell disease. Br J Haematol. 2015;168(4):553-556. 6. Miller ST, Macklin EA, Pegelow CH, et al. Silent infarction as a risk factor for overt stroke in children with sickle cell anemia: A report from the Cooperative Study of Sickle Cell Disease. J Pediatr. 2001; 139(3):385-390. 7. Bernaudin F, Verlhac S, Arnaud C, et al. Chronic and acute anemia and extracranial internal carotid stenosis are risk factors for silent cerebral infarcts in sickle cell anemia. Blood. 2015;125(10):1653-1661. 8. DeBaun, MR, & Kirkham, FJ. Central nervous system complications and management in sickle cell disease. Blood, 2016;127(7):829-838.

698

strated that CVR is globally reduced in adult sickle cell patients without a history of stroke. Even in steady state and at rest, patients with SCD utilize half of the cerebral vasodilatory reserve in comparison to control participants to compensate for anemia. Complete depletion of CVR can occur in the presence of additional strain on the vasculature such as in moyamoya syndrome, leading to extreme vulnerability to hypoxia and ischemic events. It remains to be seen whether increasing hemoglobin levels by transfusion, hydroxyurea or new disease-modifying drugs can relieve some of the restrictions imposed on the brain by anemia and whether they also reduce cerebral infarction. Acknowledgments The authors would like to thank all the participants involved in this study, Magdalena J Sokolska and David L Thomas for their expertise and contribution to the simulations on labelling efficiency for the pCASL sequence implementation, Moss Y Zhao and Michael Chappell for their contribution to the arterial transit time calculations, Sandra van den Berg and Raschel van Luijk for their excellent clinical and MR support, Erfan Nur and Charlotte van Tuijn for their help with recruitment and clinical expertise and Jan Petr for assistance with the ASL data analysis. Funding We wish to thank the Dutch fund “Fonds Nuts Ohra” for funding this research (grant no. 1303-055).

9. Kassim AA, Pruthi S, Day M, et al. Silent cerebral infarcts and cerebral aneurysms are prevalent in adults with sickle cell anemia - Letter to the Editor. Blood. 2016; 127(16):2038-2041. 10. DeBaun MR, Gordon M, McKinstry RC, et al. Controlled trial of transfusions for silent cerebral infarcts in sickle cell anemia. N Engl J Med. 2014;371(8):699-710. 11. Juttukonda MR, Donahue MJ. Neuroimaging of vascular reserve in patients with cerebrovascular diseases. Neuroimage. 2017 Oct 12. [Epub ahead of print] 12. Settakis G, Molnár C, Kerényi L, et al. Acetazolamide as a vasodilatory stimulus in cerebrovascular diseases and in conditions affecting the cerebral vasculature. Eur J Neurol. 2003;10(6):609-620. 13. Sam K, Crawley AP, Conklin J, et al. Development of white matter hyperintensity is preceded by reduced cerebrovascular reactivity. Ann Neurol. 2016;80(2):277285. 14. Ogasawara K, Ogawa A, Terasaki K, et al. Use of cerebrovascular reactivity in patients with symptomatic major cerebral artery occlusion to predict 5-year outcome: comparison of xenon-133 and iodine-123IMP single-photon emission computed tomography. J Cereb Blood Flow Metab. 2002;22(9):1142-1148. 15. Silvestrini M, Vernieri F, Pasqualetti P, et al. Impaired cerebral vasoreactivity and risk of stroke in patients with asymptomatic carotid artery stenosis. JAMA. 2000; 283(16):2122-2127. 16. Kedar A, Drane WE, Shaeffer D, Nicole M, Adams C. Measurement of cerebrovascular flow reserve in pediatric patients with sick-

17.

18.

19.

20.

21.

22.

23.

24.

le cell disease. Pediatr Blood Cancer. 2006; 46(2):234-238. Nur E, Kim Y-S, Truijen J, et al. Cerebrovascular reserve capacity is impaired in patients with sickle cell disease. Blood. 2009;114(16):3473-3478. Prohovnik I, Hurlet-Jensen A, Adams R, De Vivo D, Pavlakis SG. Hemodynamic etiology of elevated flow velocity and stroke in sickle-cell disease. J Cereb Blood Flow Metab. 2009;29(4):803-810. Leung J, Duffin J, Fisher JA, Kassner A. MRI-based cerebrovascular reactivity using transfer function analysis reveals temporal group differences between patients with sickle cell disease and healthy controls. NeuroImage Clin. 2016;12:624-630. Kosinski PD, Croal PL, Leung J, et al. The severity of anaemia depletes cerebrovascular dilatory reserve in children with sickle cell disease: a quantitative magnetic resonance imaging study. Br J Haematol. 2017;1 76(2):280-287. Václavů L, Baldew ZA, Gevers S, et al. Intracranial 4D flow magnetic resonance imaging reveals altered haemodynamics in sickle cell disease. Br J Haematol. 2018; 180(3):432-442. Dowling M, Quinn C, Plumb P, et al. Acute silent cerebral infarction occurs during acute anemic events in children with and without sickle cell disease. Blood. 2012; 120(19):3891-3897. Kirkham FJ, Hewes DKM, Prengler M, et al. Nocturnal hypoxaemia and central-nervous-system events in sickle-cell disease. Lancet. 2001;357(9269):1656-1659. Kato GJ, Steinberg MH, Gladwin MT, et al. Intravascular hemolysis and the pathophysiology of sickle cell disease. J Clin

haematologica | 2019; 104(4)

Impaired CVR in adults with sickle cell disease

Invest. 2017;127(3):750-760. 25. Václavů L, van der Land V, Heijtel DF, et al. In vivo T1 of blood measurements in children with sickle cell disease improve cerebral blood flow quantification from arterial spin-labeling MRI. Am J Neuroradiol. 2016; 37(9):1727-1732. 26. Vichinsky EP, Neumayr LD, Gold JI, et al. Neuropsychological dysfunction and neuroimaging abnormalities in neurologically intact adults with sickle cell anemia. JAMA. 2010;303(18):1823-1831. 27. Calvet D, Tuilier T, Mélé N, et al. Low fetal hemoglobin percentage is associated with silent brain lesions in adults with homozygous sickle cell disease. Blood Adv. 2017; 1(26):2503-2509. 28. van Der Land V, Zwanenburg JJM, Fijnvandraat K, et al. Cerebral lesions on 7 Tesla MRI in patients with sickle cell anemia. Cerebrovasc. Dis. 2015;39(3–4):181-189. 29. Gevers S, Nederveen AJ, Fijnvandraat K, et

haematologica | 2019; 104(4)

30.

31.

32.

33.

al. Arterial spin labeling measurement of cerebral perfusion in children with sickle cell disease. J Magn Reson Imaging. 2012; 35(4):779-787. Kawadler JM, Hales PW, Barker S, et al. Cerebral perfusion characteristics show differences in younger versus older children with sickle cell anaemia: Results from a multiple-inflow-time arterial spin labelling study. NMR Biomed. 2018;31(6):1-11. Guilliams KP, Fields ME, Ragan DK, et al. Red cell exchange transfusions lower cerebral blood flow and oxygen extraction fraction in pediatric sickle cell anemia. Blood. 2017;131(9):1012-1021. Croal P, Leung J, Kosinski P, et al. Assessment of cerebral blood flow with magnetic resonance imaging in children with sickle cell disease: A quantitative comparison with transcranial Doppler ultrasonography. Brain Behav. 2017; 7(11): e00811. Solomou E, Kraniotis P, Kourakli A, Petsas

34.

35.

36.

37.

T. Extent of silent cerebral infarcts in adult sickle-cell disease patients on magnetic resonance imaging: Is there a correlation with the clinical severity of disease? Hematol. Rep. 2013;5(1):8-12. Liem RI, Liu J, Gordon MO, et al. Reproducibility of detecting silent cerebral infarcts in pediatric sickle cell anemia. J Child Neurol. 2014;29(12):1685-1691. Mangla R, Kolar B, Almast J, Ekholm SE. Border zone infarcts: pathophysiologic and imaging characteristics. Radiographics. 2011;31(5):1201-1214. Ford AL, Ragan DK, Fellah S, et al. Silent infarcts in sickle cell anemia occur in the borderzone region and are associated with low cerebral blood flow. Blood. 2018; 132(16):1714-1723. Kwiatkowski JL, Zimmerman RA, Pollock AN, et al. Silent infarcts in young children with sickle cell disease. Br J Haematol. 2009;146(3):300-305.

699

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):700-709

Myelodysplastic Syndromes

Azacitidine with or without lenalidomide in higher risk myelodysplastic syndrome & low blast acute myeloid leukemia

Melita Kenealy,1,2 Mark Hertzberg,3 Warwick Benson,4 Kerry Taylor,5 Ilona Cunningham,6,7 Will Stevenson,8 Devendra Hiwase,9,10,11 Richard Eek,12 Daniela Zantomio,13 Steve Jong,14 Meaghan Wall,15,16,17 Piers Blombery,18,19 Tracey Gerber,20 Marlyse Debrincat,20,21,22 Diana Zannino20 and John F. Seymour18,23

Cabrini Health, Melbourne; 2Monash University, Melbourne; 3Prince of Wales Hospital, Randwick, Sydney; 4Westmead Hospital, Sydney; 5Icon Cancer Care, Brisbane; 6Concord Hospital University of Sydney; 7University of Sydney; 8Royal North Shore Hospital, St Leonards; 9Haematology Department, Royal Adelaide Hospital; 10School of Medicine, Univeristy of Adelaide; 11Cancer Theme, South Australian Health and Medical Research (SAHMRI), Adelaide; 12Border Medical Oncology, Albury; 13Austin Health, Melbourne; 14 Andrew Love Cancer Centre, University Hospital, Geelong; 15Victorian Cancer Cytogenetics Service, St Vincent’s Hospital, Fitzroy, Victoria; 16Department of Medicine, St Vincent’s Hospital, University of Melbourne, Fitzroy, Victoria; 17St Vincent’s Institute of Medical Research, Fitzroy, Victoria; 18Peter MacCallum Cancer Centre, Melbourne; 19 Sir Peter MacCallum Department of Oncology, University of Melbourne; 20Australasian Leukaemia and Lymphoma Group, Richmond; 21Systems Biology and Personalised Medicine Division, The Walter and Eliza Hall Institute of Medical Research, Melbourne; 22 Department of Medical Biology, University of Melbourne and 23University of Melbourne, Australia 1

Correspondence: MELITA KENEALY melita.kenealy@thebloodunit.com.au Received: July 10, 2018. Accepted: November 23, 2018. Pre-published: December 13, 2018. doi:10.3324/haematol.2018.201152 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/700 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

700

ABSTRACT

tandard treatment for higher risk myelodysplastic syndromes, chronic myelomonocytic leukemia and low blast acute myeloid leukemia is azacitidine. In single arm studies, adding lenalidomide had been suggested to improve outcomes. The ALLG MDS4 phase II trial randomized such patients to standard azacitidine or combination azacitidine (75mg/m2/d days 1 to 5) with lenalidomide (10mg days 1-21 of 28-day cycle from cycle 3) to assess clinical benefit (alive without progressive disease) at 12 months. A total of 160 patients were enrolled; median age 70.7 years (range 42.5-87.2), 31.3% female with 14% chronic myelomonocytic leukemia, 12% acute myeloid leukemia and 74% myelodysplastic syndromes. Adverse events were similar in both arms. There was excellent delivery of protocol therapy (median azacitidine cycles 11 both arms) with few dose reductions, delays or early cessations. At median follow up 33.1 months (range 0.7-59.5), the rate of clinical benefit at 12 months was 65% azacitidine arm and 54% lenalidomide+azacitidine arm (P=0.2). There was no difference in clinical benefit between each arm according to WHO diagnostic subgroup or IPSS-R. Overall response rate was 57% in azacitidine arm and 69% in lenalidomide+azacitidine (P=0.14). There was no difference in progression- free or overall survival between the arms (each P>0.12). Although the combination of lenalidomide and azacitidine was tolerable, there was no improvement in clinical benefit, response rates or overall survival in higher risk myelodysplastic syndrome, chronic myelomonocytic leukemia or low blast acute myeloid leukemia patients compared to treatment with azacitidine alone. This trial was registered at www.anzctr.org.au as ACTRN12610000271000. haematologica | 2019; 104(4)

AZA or LEN plus AZA in MDS, CMML and low blast AML

Introduction The myelodysplastic syndromes (MDS) and chronic myelomonocytic leukemia (CMML), an MDS/myeloproliferative neoplasm overlap syndrome, are a group of clonal bone marrow disorders characterized by active but ineffective and clonal hematopoiesis accompanied by morphological dysplasia and variable cytopenias. Cytogenetic abnormalities and/or recurrent somatic mutations are present in the majority of cases.1,2 The prognosis is variable with 30% of patients transforming to acute myeloid leukemia (AML).2,3 AML with a “low blast” count of 2029% has a similar prognosis to MDS with blasts of 1019%.4 The most widely used tool for stratifying clinical risk in MDS is the IPSS score.4,5 Azacitidine is approved and available for use in subsets of intermediate- to high-risk MDS. It is a nucleoside analogue that has direct cytotoxicity and gives rise to DNA hypomethylation through interference with DNA methyltransferase.6 Clinical responses are manifest by an improvement in hematologic parameters and quality of life in a broad population of MDS patients including those with lower-risk disease but significant cytopenias.7,8 Overall survival is prolonged in those with higher-risk disease.9 There is also an established role for azacitidine in low-blast count AML and elderly AML with >30% BM blasts.10,11 Azacitidine is an established standard of care in these patients, but even so the disease does not respond in many patients and survival remains suboptimal. Ball et al. reviewed a number of studies that combined hypomethylating agents (azacitidine and decitabine) with a number of different medication classes including small molecules, immunomodulators and monoclonal antibodies, but found a lack of survival advantage in these combinations compared to HMA monotherapy.12 Emerging data suggests molecular profiles may influence response to azacitidine.13 Lenalidomide is a thalidomide analogue and is both more potent and tolerable relative to thalidomide.14,15 Its efficacy in MDS is most pronounced in patients with 5qMDS (low risk MDS).16 Targeted degradation of CK1a (encoded by the retained allele of CSNK1A1 at 5q32 in cells with 5q-) achieves cytogenetic remission and transfusion independence in the majority of patients.15 Clinically relevant responses are also seen in lower-risk disease without 5q-.17,18,19 In MDS without 5q-, the primary mechanism of disease control with lenalidomide appears to be immunomodulation.14 Defective or reduced immune interaction between host and tumor contributes to the pathogenesis of MDS. Lenalidomide overcomes this by reducing pro-inflammatory cytokines, upregulation of T- and NK-cell activity and inhibition of angiogenic activity. These effects prevent apoptosis of healthy stem cells, improve erythropoiesis and direct immune responses against abnormal hematopoietic clones.14 The combination of a demethylating agent and an immunomodulatory drug has been explored in phase-I and -II studies in MDS, CMML and low blast AML in an attempt to improve outcomes. The ALLG MDS3 trial of azacitidine and thalidomide20 showed promising response rates, and a phase-II study by Sekeres et al.21 including the combination of azacitidine and lenalidomide in higherrisk MDS (blasts ≥ 5% or IPSS ≥1.5) or CMML resulted in an overall response rate of 49% compared to 38% azacitidine alone (P=0.14), with the subgroup of CMML patients haematologica | 2019; 104(4)

on combination therapy achieving an improved ORR compared to aza alone (68% vs. 28%, P=0.02). Other groups have gone on to review the safety and efficacy of this combination in similar disease groups; elderly AML patients and high risk MDS and AML with ≤30% blasts.22,23 Narayan et al. demonstrated a modest 25% response rate in elderly patients with previously treated AML and high-risk MDS. In these responders, the overall survival was 9.6 months compared to 4 months for non responders.22 We conducted an open-label, multicentre randomized phase-II study across 30 sites in Australia to assess the efficacy of azacitidine in combination with lenalidomide compared to standard azacitidine alone in the treatment of higher-risk MDS, CMML and low blast AML.

Methods Study design and treatment ALLG MDS4 was an open-label, multi-centre study conducted across 30 Australian sites. The study was registered at anzctr.org.au ACTRN12610000271000, was reviewed and approved by the Human Research Ethics Committees of each centre and conducted according to the Declaration of Helsinki. All patients provided written informed consent prior to participation. The primary objective was to demonstrate improved efficacy with the combination compared to azacitidine alone. Secondary objectives were to describe response rates, response duration, overall survival, tolerability and changes in quality of life, and to explore biomarkers of response and mechanism of action of azacitidine and lenalidomide. Patients were stratified according to IPSS (low-Int1 or Int2high),5 by centre and by disease category (MDS, AML or CMML),24 and randomized 1:1 to either azacitidine alone at standard dosing of 75mg/m2/d x 7 days (on a 5-2-2 interrupted schedule25) each 28 day cycle subcutaneously, or to the combination azacitidine plus lenalidomide. Patients on the combination arm received azacitidine alone at the above dose and schedule for the first 2 cycles, then commenced lenalidomide 10mg/d from day 1 of cycle 3 with a reduction in azacitidine dose with the combination to 75mg/m2/d for 5 consecutive days per cycle as per phase 1 data available at the time.26 The rationale for this was to limit the expected myelotoxicity typically seen in the first 2 cycles of treatment with azacitidine and so to improve the deliverability of combination treatment. Lenalidomide was continued only until completion of C12 due to limited data on longer-term combination toxicity. Azacitidine as a single agent was continued after the primary endpoint assessment at 12 months, until disease progression or unacceptable toxicity. Patients were followed for transformation to AML and survival until the last registered patient had been followed for a minimum 2 years after completion of the first 12 months of treatment.

Patient population Patients were eligible with a diagnosis of non-proliferative CMML, AML with blasts <30% or MDS by WHO criteria;24 those with refractory cytopenia with unilineage dysplasia (RCUD) and refractory anemia with ringed sideroblasts (RARS) had to have at least one clinically significant cytopenia as defined in the protocol (refer Online Supplementary Appendix), consistent with early studies of azacitidine in a broader group of patients with MDS.7 Patients were 18 years or older and could have either de novo or secondary disease. They must have received no prior chemotherapy for MDS or AML except low dose cytarabine or hydroxyurea, 701

M. Kenealy et al.

and no prior demethylating agent or immunomodulatory drug. They were to have a performance status of ECOG 0-2 and adequate organ function (Online Supplementary Appendix). GCSF was only used for short term management of severe neutropenic infections with no response assessment performed within 21 days of use. Patients on a stable dose of EPO prior to study entry were allowed to continue unchanged while on study.

Statistical plan Analyses were carried out using the SAS (Statistical Analysis System, Version 9.3, SAS Institute, North Carolina, USA) software and graphs were produced in R version 3.2.3 software (R Foundation for Statistical Computing, Vienna, Austria, http://www.R-project.org). All comparisons were by intention to treat. The close-out date for this analysis was 12th March 2016. A sample size of 160 patients (80 per arm) would provide 90% power, assuming a two-sided type I error of 5%, to detect an improvement of at least 25% in clinical benefit at 12 months, where the expected rate of clinical benefit at 12 months in the control arm of 50%, given a median time to progressive disease, relapse or death in the AZA001 study of 14.1 months,9 and an expected rate of clinical benefit in the combination arm being 75%.

Toxicity Adverse event rates are based on the worst grade reported during study treatment for those patients who commenced treatment. Fisher’s exact test was used to compare adverse event rates between the randomized arms. All events and grades are based on the CTCAE v4.0 unless otherwise specified. Emerging grade 3+ haematologic toxicity applied to patients who did not have a haematologic toxicity at baseline but developed whilst on treatment. Results are based on absolute value from the screening averaged haematology counts. A grade 3+ toxicity for neutrophil and platelet data was defined as a reduction of more than 50% from baseline but for haemoglobin Grade 3+ was defined as Hb <80g/L.

Efficacy All patients who were randomized (intention to treat group) were considered in efficacy analysis. 2006 IWG criteria were used for all responses.27 The primary endpoint was “clinical benefit at 12 months”, defined as the patient being alive and progression/relapse free at 12 months (+/- 1 month) post commencement of treatment, and so included those patients with stable disease at 12 months as achieving clinical benefit. Best response was determined using all assessments performed at the commencement of each cycle from C3 until treatment discontinuation, with bone marrow biopsies performed after C2, C4, C8 and C12. The overall response rate (ORR) included all patients achieving improvement (HI), marrow CR, PR and CR as best response. Univariable logistic regression models were used to assess the impact of the following pre-defined variables on response (marrow CR or better): treatment arm, IPSS-R, IPSS, cytogenetic risk group, WHO diagnosis (MDS vs. AML vs. CMML). Progression-free survival (PFS) was measured from the commencement of treatment to disease progression or death from any cause. Overall survival (OS) was measured from the commencement of treatment to death from any cause. OS and PFS duration was censored at the study close-out date for patients who were still being followed up without having experienced the relevant event by the close-out date, or at the date of last contact for patients who were lost to follow up before the study close-out date. The Kaplan-Meier (product-limit) method was used to estimate PFS and OS and median follow-up time (using the censoring 702

distribution). The logrank test was used to compare survival between the treatment arms and IPSS-R subgroups.

Quality of life (QoL) The EORTC QLQ C30 was utilized to describe differences in QoL parameters. These analyses were performed on five functional scales (physical, role, emotional, social and cognitive), three symptom scales (fatigue, nausea & vomiting and pain) and a global health status/QoL scale and six single items. Regression methods accounting for repeated measures (i.e., generalized estimating equations (GEE) with an exchangeable correlation structure) were used to estimate the difference between the treatment arms adjusted for baseline QoL score and weeks on trial. Differences between arms are expressed such that positive differences favour the LEN+AZA arm and negative differences favour the AZA arm.

Molecular and Biomarkers Next generation sequencing (NGS). Sequence analysis of targeted regions within 26 genes involved in myeloid malignancy (Peter MacCallum Cancer Centre Myeloid Amplicon Panel (v5.4)) was performed in duplicate using Access Array methodology (Fluidigm, South San Francisco, CA, USA) to prepare ampliconbased, indexed libraries that were sequenced to a depth of ∼1000 reads per amplicon on a MiSeq instrument using v2 chemistry (Illumina, San Diego, CA, USA). Alignment, variant calling and annotation were performed using a custom pipeline. Variants were evaluated using multiple functional and quality filters to identify likely pathogenic variants. SNP-Array testing. DNA (200 ng) was hybridized to CytoSNP-12 BeadChip arrays (Illumina, San Diego, CA) according to the manufacturer’s instructions. Analysis was performed using Karyostudio v1.4 software (Illumina). Karyotypes were reported according to the International System for Cytogenetic Nomenclature (ISCN 2013). Exploratory analyses include changes in promoter DNA methylation during cycle 1 and at additional time points on treatment, immunophenotyping of MDS population, T-cell subsets, NK-cell function and cytokine profile as predictors of response to treatment and will be reported separately.

Results Baseline demographics and disease features (Table 1) One hundred and sixty patients with a median age of 70.7 years (range 42.5-87.2) were enrolled on study between August 2010 and August 2012; 159 received study drug. The median time from diagnosis to treatment was 1.0 year (0.0-13.2). Twenty-two patients (14%) had CMML, 19 (12%) AML and the remaining 74% MDS were mostly RCMD or RAEB 1/2 subtypes. Overall IPSS was Low-Int 1 in 61%; by IPSS-R Very Low/Low/Intermed in 63% with no difference in prognostic or cytogenetic subgroups between arms. Fiftyseven percent of patients were transfusion dependent at study entry. The 5q- cytogenetic abnormality was present as an isolated abnormality in only 3 patients ( IPSS Int-1 in 2 patients, Int-2 in 1).

Molecular characteristics at baseline Targeted amplicon sequencing and SNP-A testing was successfully performed in 66 cases. Targeted amplicon sequencing detected pathogenic mutations in one or more genes in 94% (62/66) of patients and SNP-array detected haematologica | 2019; 104(4)

AZA or LEN plus AZA in MDS, CMML and low blast AML

abnormalities in 50% (33/66) of patients consistent with published literature.28,29 Of the 4 patients without abnormalities detectable by NGS, 1 patient had monosomy 7 detectable on SNP-A resulting in 95.4% (63/66) of cases having a detectable molecular aberration. The average

number of SNP-A abnormalities per case was 7 (range 037). Identification of additional SNP-A abnormalities upstaged cytogenetic risk in 24% (16/65) cases (G-banded karyotype not available in one case). SNP-A and mutational data are summarized in Table 2.

Table 1. Clinical and hematologic characteristics of 160 patients by assigned treatment cohort.

Baseline characteristics Age years, median (range) Male Female ECOG 0 1 2 Diagnosis WHO RCUD (RA, RN, RT) RARS RCMD RAEB-1 RAEB-2 MDS-U MDS isolated del5qCMML AML IPSS risk group Low Int-1 Int-2 High IPSS-R risk group Very Low Low Intermediate High Very High missing Cytogenetics (IPSS-R) Very Good Good Intermed Poor Very poor Carrying 5qmissing Past therapy EPO GCSF Low dose cytarabine Baseline cytopenias Hb (<100g/L) Neutrophils (<1.5x109/L) Platelet (<100x109/L)

AZA (n=80)

LEN+AZA (n=80)

Total (n=160)

69.1 (42.5-85.9) 52 (65) 28 (35) 42 (53) 36 (45) 2 (3)

71.4 (44.1-87.2) 58 (73) 22 (28) 41 (51) 33 (41) 6 (8)

70.7 (42.5-87.2) 110 (68.8) 50 (31.3) 83 (51.9) 69 (43.1) 8 (5.0)

1 (1) 6 (8) 24 (30) 11 (14) 16 (20) 0 (0) 2 (3) 12 (15) 8 (10)

0 (0) 3 (4) 28 (35) 11 (14) 15 (19) 1 (1) 1 (1) 10 (13) 11 (14)

1 (0.6) 9 (5.6) 52 (32.5) 22 (13.8) 31 (19.4) 1 (0.6) 3 (1.9) 22 (13.8) 19 (11.9)

12 (15) 37 (46) 22 (28) 9 (11)

10 (13) 38 (48) 17 (21) 15 (19)

22 (13.8) 75 (46.9) 39 (24.4) 24 (15.0)

2 (3) 24 (32) 23 (31) 16 (21) 10 (13) 5

3 (4) 18 (24) 25 (33) 12 (16) 17 (23) 5

5 (3.3) 42 (28.0) 48 (32.0) 28 (18.7) 27 (18.0 10

1 (1) 55 (73) 11 (15) 3 (4) 5 (7) 10 (13) 5

2 (3) 51 (68) 11 (15) 2 (3) 9 (12) 13 (17) 5

3 (2.0) 106 (70.7) 22 (14.7) 5 (3.3) 14 (9.3) 23 (15.3) 10

3 (4) 2 (3) 0 (0)

0 (0) 2 (3) 1 (1)

3 (1.9) 4 (2.5) 1 (0.6)

57 (71) 47 (59) 42 (53)

53 (66) 37 (46) 48 (60)

110 (68.8) 84 (52.5) 90 (56.3)

Hb: hemoglobin; ECOG: Eastern Cooperative Oncology Group; Int: intermediate; MDS-(U): myelodysplastic syndrome-(unclassifiable); RA: refractory anemia, RAEB: refractory anemia with excess blasts; RCMD: refractory cytopenia with multilineage dysplasia; RARS: refractory anemia with ringed sideroblasts; RCUD: refractory cytopenia with unilineage dysplasia; RN: refractory neutropenia; RT: refractory thrombocytopenia; WHO: World Health Organisation. Median (range) reported and N (%) unless otherwise specified.

haematologica | 2019; 104(4)

703

M. Kenealy et al.

Most CMML cases (5/7, 71%) had a normal SNP-A karyotype. Rates of SNP-A karyotypic complexity and del(5q) were highest in RAEB-2 (7/16, 43%) and AML (2/4, 50%). Both AML cases with an abnormal karyotype also had deletion of 17p. There were no significant differences between the AZA and AZA+LEN groups in terms of high-risk molecular profile, SNP-A complexity or cytogenetic risk group. Clinical benefit was most frequent in the

normal SNP-A (29/39, 82%) and IPPS-Rsnp good cytogenetic risk group (28/39, 82%) cases.

Treatment With a close out date of 12th March 2016, median follow up was 33.1 months (range 0.7-59.5). There was excellent drug delivery, with the median number of azacitidine cycles per patient administered of 11 in both arms with

Figure 1. Rates of Grade 3+ Anemia (Hb less than 80g/L) baseline and on treatment, rates of Grade 3+ neutropenia (reduction neutrophils to less than 50% baseline) and rates of Grade 3+ thrombocytopenia (reduction in platelets to less than 50% baseline).

704

haematologica | 2019; 104(4)

AZA or LEN plus AZA in MDS, CMML and low blast AML

only 2.6% cycles dose- reduced. For those on combination treatment the median duration of lenalidomide treatment was 9 cycles (range 1-12) with only 2.8% lenalidomide cycles dose- reduced. Early discontinuation of azacitidine was mainly due to investigator/patient decision, relapse/progressive disease, or death. Early discontinuation of lenalidomide was mainly due to toxicity (11 patients) or investigator/patient decision (9 patients). Six patients were treated with lenalidomide after completing the protocol-specified 12 months of study therapy; 2 in AZA arm and 4 in LEN+AZA. This was initiated by individual investigators, and continued for up to 2 years post study therapy. The extended lenalidomide treatment was associated with grade 3 diarrhea in one patient and resulted in no improvement in response in these patients.

Safety Non hematologic toxicity. Rates of all adverse events grade 3 or higher according to system and treatment arm are summarised in Online Supplementary Table S1 with no differences observed. The most common non-hematologic toxicity was infection; the overall number of infectious episodes grade 3 or worse was 132 in 42.8% patients. There was no difference between the arms for overall rates of infection with sepsis being the most common infection type. The difference between the severity of sepsis between the two treatment arms was significant with greater severity in the combination arm; sepsis Grade 4+ was seen in 11 patients in the combination arm compared to 2 patients in azacitidine alone arm (Table 3). There were 17 deaths due to

Table 2. Molecular characteristics of cohort with baseline samples.

AZA (n=35)

LEN-AZA (n=31)

TOTAL (n=66)

3 14

5 10

8 24

6 6 4 0 2 4

6 3 4 6* 4 2

12 9 8 6 6 4

3 22

5 21

8 43

Targeted Amplicon Sequencing TP53mut TET2mut High-risk molecular profile (TP53mut and/or ASXL1mut and/or RUNX1mut and/or EZH2mut ) low-risk molecular profile (SF3B1mut only) SNP-Array 5qMonosomy 7/7q20qTrisomy 8 17p7q CN LOH Combined molecular profile TP53 abnormality (TP53mut and/or 17p-) High-risk molecular profile (TP53mut , ASXL1mut , RUNX1mut , EZH2mut , IPSS-RsnpP, IPSS-RsnpVP)

mut: mutated; LOH: loss of heterozygosity; IPSS-RsnpP or VP: IPSS-R SNP-A poor cytogenetic or very poor cytogenetic risk.

Table 3. Non-hematologic toxicity; infections, grade 3 and above.

Infection type

AZA N. patients

Any Infection GI/Abdo Renal/Urologic Respiratory Sepsis overall Sepsis grade 3 Sepsis grade 4 Sepsis grade 5 Skin/Mucosal/Eye All other infections

haematologica | 2019; 104(4)

34 (43%) 6 (8%) 0 (0%) 9 (11%) 18 (23%) 16 2 0 9 (11%) 5 (6%)

LEN+AZA No. episodes

N patients

No. episodes

62 7 0 10 29

34 (43%) 7 (9%) 2 (3%) 14 (18%) 19 (24%) 8 10 1 9 (11%) 6 (8%)

71 10 2 18 29

11 5

P=0.02 9 6

705

M. Kenealy et al.

infection (7 in AZA, 10 in LEN+AZA). For full listing of cause of death according to treatment arm see Online Supplementary Table S2. The only other non hematologic toxicity seen at rates >5% was raised GGT in 15 patients with no significant difference between the arms (AZA n=4 LEN+AZA n=11; P=0.1). Hematologic toxicity. Comparing data from cycle 3 to cycle 12 (as LEN was introduced from cycle 3 onwards), there was no association between treatment arm and cycle for any of the hematologic grade 3+ toxicity rates was observed. For Hb <80g/L and for neutrophils and platelets >50% reduction from baseline count, there was no difference between treatment arms but a statistical significant difference across the cycles for both arms. See Figure 1. Emerging Grade 3+ hematologic toxicity. For those patients who did not have a grade 3+ toxicity at baseline as defined by CT CAE V4.0 (N), there was a non significant trend to greater treatment emergent neutropenia (78% vs. 68%) and thrombocytopenia (63% vs. 50%) in the combination arm (Online Supplementary Table S3).

Efficacy Summary of efficacy endpoints is provided in Table 4. The primary endpoint of rate of clinical benefit at 12 months (alive with stable disease or better) in the AZA arm was 65%, and 54% in the LEN+AZA arm (Fishers Exact test, P=0.2). There was no difference in rate of clinical benefit between each treatment arm according to WHO diagnostic subgroup (MDS, AML or CMML) or according to IPSS-R. There was no difference in clinical benefit across disease subtype within either the LEN+AZA or the AZA treated groups. The overall response rate (best response of HI, PR, marrow CR or CR) with AZA was 57% and 69% in LEN+AZA (P=0.14). CR was achieved in 17 patients (22%) on AZA and 20 patients (25%) on LEN+AZA. There was no difference in type of HI across the 2 arms. The median time to best response for those achieving a HI or better was not different between treatment arms; 5.5 months (range 1.8-11.7) AZA and 4.8 months (range 1.8-12.4) LEN+AZA. Median time to first response (of HI or better) was 2.8 months (range 1.6-9.2), with no difference between the arms (P=0.13). There were no significant associations found for these variables with respect to response â&#x20AC;&#x201C; either clinical benefit at 12 months as defined by primary endpoint, or for overall response rate of best response HI or better. Using univariable subgroup logistic regression models of treatment effect on primary endpoint response (clinical benefit stable disease or better at 12 months), there were no significant associations for age, sex, WHO diagnosis, IPSS, IPSS-R or cytogenetic risk group (Online Supplementary Figure S1).

Cytogenetic response Fifty-nine patients had a karyotypic abnormality detected at baseline, 28 in the AZA arm and 31 in the LEN+AZA arm. A total of 29% (8) of patients on the AZA arm had a cytogenetic response. Half (4) of those achieved a complete cytogenetic response while the other half (4) achieved a partial response with a â&#x2030;Ľ50% reduction in the chromosomal abnormality. A total of 39% (12) patients in the LEN+AZA arm had a cytogenetic response, 11 of them 706

Figure 2. Time to relapse, progression-free and overall survival between treatment cohorts, and overall survival according to risk. A. Kaplan-Meier curves of time to relapse after achieving CR/PR, or disease progression between both treatment cohorts. B. Kaplan-Meier curves of progression-free survival (PFS). C. Overall survival according to assigned treatment cohort. D. Overall survival according to IPSS-R; very low/low versus intermed/high/very high risk.

haematologica | 2019; 104(4)

AZA or LEN plus AZA in MDS, CMML and low blast AML

achieving a complete cytogenetic response whilst 1 achieved a partial response..

Time to disease relapse or progression (Figure 2A) No association with treatment arm was found for time to relapse after CR/PR or PD. Median time to progression to AML (MDS and CMML patients by WHO criteria) or death (all patients) from any cause was 37.2 months in the AZA arm and 28.8 months in the LEN+AZA arm.

Progression free-survival (PFS) (Figure 2B) PFS was measured from first day of treatment to date of first confirmed disease progression or death from any cause. Median PFS time on AZA was 31.2 months (95% CI 25.0-37.4) and on LEN+AZA was 19.8 months (95% CI 14.7-29.2) with no difference between the arms observed (Figure 2B, logrank P=0.12).

Overall survival (Figures 2C and 2D) The median follow-up time (estimated with the inverse Kaplan-Meier method) was 47.2 months (range 0.7-59.5); median survival time on AZA was 38.8 months (95%CI 35.8-52.6) compared to 29.2 months on LEN+AZA (95%CI 19.8-35.1) (logrank P=0.2). Forty-one patients on AZA had died compared to 49 on LEN+AZA (Table 8). Cause of death was mostly due to disease progression and infections with 5 overall due to hemorrhage and 9 other/unknown. There was a significant difference in median overall survival in IPSS-R Very Low/Low-risk and Intermediate/High/Very High-risk groups (Figure 2D, logrank P<0.001).

Quality of Life (Figure 3) Completion rates for the EORTC QLQ-C30 at baseline/screening was 96%, at C4D22 83%, C8D22 82% and C12D22 or at primary endpoint visit was 84%. The only effect of treatment on QoL scores during study was a higher rate of diarrhea in LEN+AZA arm.

Discussion This randomized phase II study aimed to find out whether outcomes were improved for patients with higher risk MDS, CMML and low blast AML by adding lenalidomide treatment to the established regimen of azacitidine. These two agents have shown synergistic activity in vitro, with promising early results of the combination treatment from smaller single arm clinical trials. In this study, there was excellent duration and delivery of treatment in both arms due to strong recognition of the value of prolonged therapy, particularly with a clinical benefit endpoint at 12 months. Despite this and the good tolerability of the combination, there was no improvement in response rates, clinical benefit or survival. As in the study by Sekeres et al.,21 we showed a trend towards improved responses without translation to improved clinical benefit or survival, though this study was not adequately powered to show a difference in overall survival. The lack of clinical benefit was not due to an excess of toxicity in the combination arm. No subgroup in this study, including those with CMML, and in contrast to the recent report by Sekeres et al.,21 had improved responses with the addition of lenalidomide to

Figure 3. Quality of life differences on EORTC QLQ C30 questionnaire between treatment cohorts using GEE regression model

haematologica | 2019; 104(4)

707

M. Kenealy et al. Table 4. Efficacy: clinical benefit at 12 months, overall response rate (ORR) & best response achieved by assigned treatment cohort; those who received treatment.

Clinical benefit at 12 months (SD or better) MDS AML CMML IPSS-R Very Low/Low IPSS-R Intermed/High/Very High Overall response rate (Best response) MDS AML CMML IPSS-R Very Low/Low IPSS-R Intermed/High/Very High Best response achieved CR PR Marrow CR Marrow CR+HI HI only SD PD Death prior to C3 first response assessment Missing data/not evaluable

AZA n=79 n (%[Exact 95% CI])

LEN+AZA n=80 n (%[Exact 95% CI])

52 (65% [54-75]) 38 (63% [50-75]) 5 (62% [24-91]) 9 (75% [43-95]) P=0.7 22 (85% [65-96]) 28 (57% [42-71]) 45 (57% [45-68]) 36 (60% [47 - 72]) 3 (43% [10-82]) 6 (50% [21-79]) 15 (58% [37-77]) 28 (58% [43-72])

43 (54% [42-65]) 34 (58% [44-70]) 4 (36% [11-69]) 5 (50% [19-81]) P=0.4 12 (57% [34-78]) 28 (52% [38-66]) 55 (69% [57-79]) 41 (69% [56 - 81]) 6 (55% [23-83]) 8 (80% [44-97]) 14 (67% [43-85]) 37 (69% [54-80])

0.2 0.6 0.4 0.4

17 (22%) 0 2 (2%) 8 (10%) 18 (23%) 22 (28%) 3 (4%) 6 (8%) 3 (4%)

20 (25%) 2 (2%) 5 (6%) 5 (6%) 23 (29%) 15 (19%) 4 (5%) 4 (5%) 2 (2%)

0.052 0.7 0.14 0.3 >0.99 0.2 0.6 0.3

CR: complete response; HI: hematologic improvement; PD: progressive disease; PR: partial response; SD: stable disease.

azacitidine. Overall, there was very good durability of responses and good survival. Unsurprisingly, those with lower-risk disease according to established prognostic scores lived longer. Our eligibility included patients with potentially lowerrisk disease subtypes in contrast to other recent clinical trials such as AZA0019 and SWOG S111721 which defined eligibility based on prognostic score. This inclusion was based on earlier data showing similar response rates across all IPSS groups7 and an acknowledgement that a proportion of those with apparent lower-risk disease have outcomes more in keeping with those with higher prognostic scores. Despite this, and though it is an indirect comparison of populations, our cohort risk compares similarly to the SWOG S1117 cohort with respect to proportion of patients with Very Low/Low IPSS-R; 31.3% Sekeres cohort compared to our ALLG MDS4 32% (AZA) and 38% (AZA + LEN). It is possible that the dose and scheduling of treatment in this protocol may have impacted on responses. Phase 1 data by Sekeres26 supported the decision to reduce the number of days of azacitidine dosing to five when combining with lenalidomide, in order to reduce the risk of treatment limiting toxicity in the first two cycles. Given the lack of excessive toxicity in our combination arm and the high median number of azacitidine cycles (11 cycles in our cohort compared to SWOG 1117 median 23-25 weeks treatment) we did achieve the implementation of this 708

treatment combination on a broad multi-centre setting. However, we have not shown that full azacitidine dosing of seven days per cycle in combination with lenalidomide is feasible. In addition, the concurrent as opposed to consecutive administration of the two agents on this protocol may have reduced overall efficacy. The dose of lenalidomide selected for this study was based on a Phase II study in MDS,21 however, subsequent studies have utilised higher doses of lenalidomide in combination â&#x20AC;&#x201C; mostly sequential - in AML30 and high-risk MDS31 which may improve efficacy. The option of using lenalidomide prior to the introduction of azacitidine could be considered as an extrapolation of the findings by Zeidan et al. who demonstrated enhanced erythroid improvement in low -risk (non-5q deletion) MDS.31 Finally, consideration could be given to commencing both agents from C1 rather than delaying the introduction of lenalidomide until C3 in an attempt to improve efficacy, although early progressions or deaths in our study were uncommon with 10 deaths or disease progression within the first 2 cycles of treatment (5 in each arm). There was no central review of pathology in this study. Responses were provided by the site investigators and only reviewed centrally if there were discrepancies or questions. IWG criteria for response was adopted, though its application in patients experiencing both disease and treatment related cytopenias is complex, and the application and consistency across many sites was haematologica | 2019; 104(4)

AZA or LEN plus AZA in MDS, CMML and low blast AML

difficult to ensure. A more robust, refined IWG criteria is awaited and would make this more consistent in future studies. We have shown the feasibility on a broad scale of the combination of lenalidomide and azacitidine in patients with higher risk MDS, CMML and low blast count AML, but the lack of improvement in responses and clinical benefit do not support the utilization of this combination in higher-risk MDS in clinical practice. Other combinations and novel agents are needed to improve the outcomes for this large and vulnerable group of patients, who at this stage have limited therapeutic options. Other

References 1. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization Classification of Myeloid neoplasms and Acute leukemia. Blood. 2016;127(20):2391-2405. 2. Gangat N, Patnaik MM, & Tefferi A. Myelodysplastic syndromes: Contemporary Review and how we treat. Am J Hematol. 2016;91(1):76-89. 3. Shukron O, Vainstein V, KĂźndgen A, Germing U, & Agur Z. Analyzing transformation of myelodysplastic syndrome to secondary acute myeloid leukemia using a large patient database. Am J Hematol. 2012;87(9):853-860. 4. Greenberg PL, Tuechler H, Schanz J, et al. Revised International Prognostic Scoring System for myelodysplastic syndromes. Blood. 2012;120(12):2454-2465. 5. Greenberg P, Cox C, LeBeau, et al. International Scoring System for evaluating prognosis in myelodysplastic syndromes. Blood. 1997;89(6):2079-2088. 6. Christman JK. 5-Azacytidine and 5-aza-2â&#x20AC;˛deoxycytidine as inhibitors of DNA methylation: mechanistic studies and their implications for cancer therapy. Oncogene. 2012;21(35):5483-5495. 7. Silverman LR, Demakos E. Peterson B, et al. Randomized controlled trial of azacitidine in patients with the myelodysplastic yndrome: a study of the Cancer and Leukemia Group B. J Clin Oncol. 2002;20(10):2429-2440. 8. Kornblith AB, Herndon JE, Silverman LR, et al. Impact of azacytidine on the quality of life of patients with myelodysplastic syndrome treated in a randomized phase III trial: a cancer and leukemia group B study. J Clin Oncol. 2002;20(10):2441-2452. 9. Fenaux P, Mufti GJ, Hellstrom E, et al. Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, phase III study. Lancet Oncol. 2009;10(3):223-232. 10. Dombret H, Seymour JF, Butrym A, et al. International phase 3 study of Azacitidine vs conventional care regimens in older patients with newly diagnosed AML with > 30% blasts. Blood. 2015;126(3):291-299. 11. Fenaux P, Mufti,GJ, Hellstrom E, et al. Azacitidine prolongs overall survival compared with conventional care regimens in elderly patients with low bone marrow

haematologica | 2019; 104(4)

12.

13.

14. 15.

16.

17.

18.

19.

20.

21.

groups are currently exploring novel combination studies with azacitidine such as enasidinib and venetoclax.

Acknowledgments The authors wish to acknowledge the Australasian Leukaemia & Lymphoma Group for conduct of the study, all site investigators and study personnel. We also wish to thank our patients and families.

Funding This study was supported by Celgene with grant funding for study conduct, Snowdome Foundation and the Victorian Epigenetic Group for funding support for molecular studies.

blast count acute myeloid leukemia. J Clin Oncol. 2010;28(4):562-569. Ball B, Zeidan A, Gore SD & Prebet T. Hypomethylating agent combination strategies in myelodysplastic syndromes: hopes and shortcomings. Leuk Lymphoma. 2017;58(5):1022-1036. Rami SK, Sallman D, Ali M, et al. Outcome of myelodysplastic syndrome patients with TP53 mutation treated with hypomethylating agents. Blood. 2017;130(Suppl 1):1687. Kotla V, Goel S, Nischal S, et al. Mechanism of action of lenalidomide in hematological malignancies. J Hematol Oncol. 2009;2:36 List A, Kurtin S, Roe DJ, et al. Efficacy of lenalidomide in myelodysplastic syndromes. N Engl J Med. 2005;352(6):549557. Fenaux P, Giagounidis A, Selleslag D, et al. A randomized phase 3 study of lenalidomide versus placebo in RBC transfusiondependent patients with low-/intermediate-1-risk myelodysplastic syndromes with del5q. Blood. 2011; 118(14):3765-3776. 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. Santini V, Almeida A, Giagounidis A, et al. Randomized phase III study of lenalidomide versus placebo in RBC transfusiondependent patients with lower-risk nondel(5q) myelodysplastic syndromes and ineligible for or refractory to erythropoiesis-stimulating agents. J Clin Oncol. 2016;34(25):2988-2996. Sibon D, Cannas G, Baracco F. et al. Lenalidomide in lower-risk myelodysplastic syndromes with karyotypes other than deletion 5q and refractory to erythropoiesis-stimulating agents. Br J Haematol. 2012;156(5):619-625. Kenealy M, Patton N, Filshie R, et al. Results of a phase II study of thalidomide and azacitidine in patients with clinically advanced myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML) and low blast count acute myeloid leukemia (AML). Leuk Lymphoma. 2017;58(2):298-307. Sekeres MA, Othus M, List AF, et al. Randomized phase II study of azacitidine alone or in combination with lenalidomide or with vorinostat in higher-risk myelodysplastic syndromes and chronic

22.

23.

24.

25.

26.

27.

28. 29.

30.

31.

myelomonocytic leukemia: North American Intergroup Study SWOG S1117. J Clin Oncol. 2017;35(24):2745-2753. Narayan R, Garcia JS, Percival MM, et al. Sequential azacitidine plus lenalidomide in previously treated elderly patients with acute myeloid leukemia and higher risk myelodysplastic syndrome. Leuk Lymphoma. 2016;57(3):609-615. DiNardo CD, Daver N, Jabbour, et al. Sequential azacitidine and lenalidomide in patients with high-risk myelodysplastic syndromes and acute myeloid leukemia: a single arm, phase 1/2 study. Lancet Haematol. 2015;2(1):e12-e20. Swerdlow SH, Campo E, Harris NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues 4th Edition. International Agency for Research on Cancer (2008). Lyons RM, Cosgriff TM, Modi SS, et al. Hematologic response to three alternative dosing schedules of azacitidine in patients with myelodysplastic syndromes. J Clin Oncol. 2009;27(11):1850-1856. Sekeres MA, List AF, Cuthbertson D, et al. Phase I combination trial of lenalidomide and azacitidine in patients with higher-risk myelodysplastic syndromes. J Clin Oncol. 2010;28(13):2253-2258. Cheson BD, Greenberg PL, Bennett J, et al. Clinical application and proposal for modification of the International Working Group (IWG) response criteria in myelodysplasia. Blood. 2006;108(2):419425. Bejar R. Clinical and genetic predictors of prognosis in myelodysplastic syndromes. Haematologica. 2014;99(6):956-964. Kulasekararaj AG, Mohamedali AM, & Mufti GJ. Recent advances in understanding the molecular pathogenesis of myelodysplastic syndromes. Br J Haematol. 2013;162(5):587-605. Medeiros BC, McCaul K, Kambhampati S, et al. Randomized study of continuous high dose lenalidomide, sequential azacitidine and lenalidomide, or azacitidine in persons 65 years and over with newly diagnosed acute myeloid leukemia. Haematologica. 2018;103(1):101-106. Zeidan AM, Al Ali NH, Padron E, et al. Lenalidomide treatment for lower risk nondeletion 5q myelodysplastic syndromes patients yields higher response rates when used before Azacitidine. Clin Lymphoma Myeloma Leuk. 2015;15(11):705-710.

709

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):710-716

Myeloproliferative Neoplasms

Beatrice Drexler,1 Jakob R. Passweg,1 Alexandar Tzankov,2 Martin Bigler,3 Alexandre PA Theocharides,4 Nathan Cantoni,5 Peter Keller,6 Georg Stussi,7 Axel Ruefer,8 Rudolf Benz,9 Geneviève Favre,10 Pontus Lundberg,1 Ronny Nienhold,11 Andrea Fuhrer,3 Christine Biaggi,3 Markus G. Manz,4 Mario Bargetzi,5 Simon Mendez-Ferrer,12 and Radek C. Skoda;11 on behalf of the Swiss Group for Clinical Cancer Research (SAKK)

Division of Hematology, University Hospital Basel and University of Basel, Switzerland; Institute of Pathology, University Hospital Basel and University of Basel, Switzerland; 3 Swiss Group for Clinical Cancer Research, Bern, Switzerland; 4Hematology and Oncology, University Hospital Zurich and University of Zurich, Switzerland; 5Oncology, Hematology & Transfusion Medicine, Kantonsspital Aarau AG, Switzerland; 6University Clinic of Hematology and Central Hematology Laboratory, University Hospital Bern, Switzerland; 7Clinic of Hematology, Oncology Institute of Southern Switzerland, Bellinzona, Switzerland; 8Departement Medizin, Luzerner Kantonsspital, Switzerland; 9 Kantonsspital Muensterlingen, Switzerland; 10Cantonal Hospital Liestal, Switzerland; 11 Department of Biomedicine, Experimental Hematology, University Hospital Basel and University of Basel, Switzerland and 12Wellcome Trust-Medical Research Council Cambridge Stem Cell Institute, Department of Haematology, University of Cambridge, and National Health Service Blood and Transplant, Cambridge, UK 1 2

ABSTRACT

Correspondence: RADEK SKODA radek.skoda@unibas.ch JAKOB PASSWEG jakob.passweg@usb.ch Received: June 20, 2018. Accepted: November 8, 2018. Pre-published: November 8, 2018. doi:10.3324/haematol.2018.200014 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/710 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

710

he β-3 sympathomimetic agonist BRL37344 restored nestin-positive cells within the stem cell niche, and thereby normalized blood counts and improved myelofibrosis in a mouse model of JAK2V617F-positive myeloproliferative neoplasms. We therefore tested the effectiveness of mirabegron, a β-3 sympathomimetic agonist, in a phase II trial including 39 JAK2-V617F-positive patients with myeloproliferative neoplasms and a mutant allele burden more than 20%. Treatment consisted of mirabegron 50 mg daily for 24 weeks. The primary end point was reduction of JAK2-V617F allele burden of 50% or over, but this was not reached in any of the patients. One patient achieved a 25% reduction in JAK2-V617F allele burden by 24 weeks. A small subgroup of patients showed hematologic improvement. As a side study, bone marrow biopsies were evaluated in 20 patients. We found an increase in the nestin+ cells from a median of 1.09 (interquartile range 0.38-3.27)/mm2 to 3.95 (interquartile range 1.98-8.79)/mm2 (P<0.0001) and a slight decrease of reticulin fibrosis from a median grade of 1.0 (interquartile range 0-3) to 0.5 (interquartile range 0-2) (P=0.01) between start and end of mirabegron treatment. Despite the fact that the primary end point of reducing JAK2-V617F allele burden was not reached, the observed effects on nestin+ mesenchymal stem cells and reticulin fibrosis is encouraging, and shows that mirabegron can modify the microenvironment where the JAK2-mutant stem cells are maintained. (Registered at clinicaltrials.gov identifier: 02311569.)

Introduction Myeloproliferative neoplasms (MPN) are thought to be initiated and maintained from a mutated hematopoietic stem cell (HSC).1 An acquired mutation in JAK2 (JAK2-V617F) is present in the majority of MPN patients.2-5 The interplay between haematologica | 2019; 104(4)

Mirabegron in MPN

the MPN HSCs and the stem cell niche is being increasingly recognized as crucial for the biology of the disease. Nestin-positive mesenchymal stem cells (nestin+ MSCs) within the bone marrow (BM) niche are innervated by sympathetic nerve fibers and are important in regulating normal HSCs.6,7 These nestin+ MSCs are strongly reduced in BM from patients with MPN.8 In a mouse model of MPN expressing human JAK2-V617F, this effect was found to be caused by early glial and sympathetic nerve damage and subsequent apoptosis of nestin+ MSCs triggered by the mutant hematopoietic cells. In vivo depletion of nestin+ cells accelerated MPN progression. Conversely, MPN phenotype could be reversed by compensating for the sympathetic neuropathy by systemic administration of a β-3-sympathomimetic agonist. Mice with JAK2V617F-driven MPN treated with the β-3-sympathomimetic drug BRL37344 not only restored nestin+ MSCs numbers, but also showed correction of thrombocytosis, neutrophilia, and BM fibrosis, and efficiently reduced mutant hematopoietic progenitor numbers in BM and peripheral blood (PB).8 Treatment with BRL37344 also corrected the damage inflicted by the MPN clone on the stem cell niche and led to an increase in nestin+ cells.8 Thus, β-3 sympathomimetic agonists represent a promising novel therapeutic approach to MPN by targeting the stem cell niche rather than the MPN clone itself. haematologica | 2019; 104(4)

Figure 1. Changes in JAK2-V617F allele burden during the treatment period. (A) Waterfall plot representing the percent changes in JAK2-V617F allele burden at 24 weeks. (B) Time course of JAK2-V617F allele burden in patients treated with mirabegron. The values measured before treatment (week 0), and at week 12 and week 24 of mirabegron treatment are shown for each individual patient, connected by dashed lines. Filled triangles, circles and squares are used for essential thrombocythemia (ET), polycythemia vera (PV) and primary myelofibrosis (PMF) patients, respectively. Within the group of myelofibrosis (MF) patients, post-ET and post-PV myelofibrosis are indicated by open triangles and circles, respectively.

Recently, mirabegron, a β3-adrenoceptor agonist, was approved in North America, Europe, Japan and Australia for the treatment of an overactive bladder.9 Here, we report the results of a phase II study that tested the efficacy of mirabegron in patients with JAK2-V617F-positive MPN.

Methods Study population Overall, 39 patients with MPN, including 7 patients with essential thrombocythemia (ET) (18%), 21 with polycythemia vera (PV) (54%), and 11 with myelofibrosis (MF) [28%; of whom 5 were primary myelofibrosis (PMF), 3 post-ET MF and 3 post-PV MF] have been accrued in 10 institutions across Switzerland between May 2015 and February 2016. The patients fulfilled the 2008 World Health Organization (WHO) diagnostic criteria for MPN.10 All patients were JAK2-V617F-positive with a mutant allele burden at study entry more than 20% in granulocyte DNA. The trial was planned and conducted in accordance with the Declaration of Helsinki, the Guidelines for Good Clinical Practice (GCP) issued by the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use and the requirements of the respective national regulatory authorities. The local ethics committees of all participating centers have given approval to the trial and written informed consent was obtained 711

B. Drexler et al. Table 1. Patients’ and disease characteristics Gender Female Male Age at registration [years] Disease ET PV MF PMF Post-ET MF Post-PV MF WHO status 0 1 Blood counts Hemoglobin [g/L] Neutrophils [x109/L] Platelets [x109/L] White blood cells [x109/L] Burden of mutated alleles [%] Organomegaly Liver palpable Spleen palpable Liver longitudinal diameter (ultrasound) [cm] Spleen longitudinal diameter (ultrasound) [cm] Clinical signs of diseases other than MPN Medical history prior to inclusion in study Venous complications Deep vein thrombosis Pulmonary embolism Splanchnic veins Retinal vein Unknown/Missing Arterial complications Cerebral Extremity Cardiac Raynaud's phenomenon Erythromelalgia Unknown/Missing Hemorrhagic complications Gastrointestinal Mucocutaneous Intraocular Unknown/Missing Previous therapies Cytoreductive Alkylating agents Hydroxyurea Pipobroman Thioguanin 712

Value N=39 12 (31%) 27 (69%) 62 (53–72) 7 (18%) 21 (54%) 11 (28%) 5 (13%) 3 (8%) 3 (8%) 32 (82%) 7 (18%) 134 (127–143) 6 (4–8) 392 (310–564) 8 (6–12) 52 (33–73) 4 (11%) 11 (29%) 15 (13–16) 14 (12–18) 28 (72%) 39 (100%) 30 (77%) 3 (8%) 0 1 (3%) 0 26 (67%) 36 (92%) 7 (18%) 2 (5%) 4 (10%) 0 0 24 (62%) 28 (72%) 1 (3%) 3 (8%) 0 24 (62%) 39 (100%) 28 (72%) 0 23 (59%) 0 0

Anagrelide Antiaggregation Anticoagulation Interferon Phlebotomy

1 (3%) 33 (87%) 7 (19%) 2 (5%) 21 (55%)

MF: myelofibrosis; PMF: primary myelofibrosis; ET: essential thrombocythemia; PV: polycythemia vera; WHO: World Health Organization; MPN: myeloproliferative neoplasms. Data are presented as number (N) of patients (%) or median (interquartile range).

from all patients prior to enrollment. Details of the inclusion and exclusion criteria are specified in the Online Supplementary Appendix.

Study design and treatment We performed a multicenter, prospective, single-arm, singlestage and open phase II trial (SAKK 33/14; clinicaltrials.gov identifier: 02311569) with the β-3-sympathomimetic agonist mirabegron (Betmiga®). Before the study began, the drug had already been approved in the US, EU and Switzerland for the treatment of patients with an overactive bladder with a maximal recommended dose of 50 mg daily. The trial consisted of mirabegron treatment for at least 24 weeks with an initial dose of 25 mg daily during the first week followed upon good tolerance by 50 mg mirabegron daily during the remaining treatment period. The following treatments were not allowed during the trial treatment phase: other anticancer treatments, drugs known to influence JAK2-V617F allele level (e.g. interferon-α), ruxolitinib, or investigational treatments. Established cytoreductive treatment for MPN (e.g. hydroxyurea, pipobroman, or thioguanin) could be continued as previously prescribed. For further details on the study design see the Online Supplementary Methods.

Primary end point The primary end point was defined as reduction in the JAK2-V617F allele burden of 50% or more at 24 weeks after registration. Secondary end points and response criteria are described in the Online Supplementary Methods).11,12

Molecular analyses The JAK2-V617F allele burden was determined on DNA from purified granulocytes isolated from PB sampled in EDTA-containing tubes. The allele-specific PCR of JAK2 genotyping was performed as previously described.13 The JAK2-V617F allele burden was validated by retesting. Capture-based next-generation sequencing with a panel of 94 genes to detect somatic mutations in granulocyte DNA was performed in patients who consented to this subproject on a voluntary basis. For details see the Online Supplementary Methods.14

Assessment of myelofibrosis and nestin+ mesenchymal stem cells Patients who entered the study could also participate on a voluntary basis in a subproject with the goal to test whether mirabegron can restore the nestin+ niche and may have a beneficial effect on BM morphology and the degree of myelofibrosis. BM trephine biopsies were performed at study entry and at week 24. Reticulin and collagen fibrosis was evaluated following established criteria.15-18

Statistical analysis Statistical methods are defined in the Online Supplementary Methods. haematologica | 2019; 104(4)

Mirabegron in MPN

Results Patients and study treatment The characteristics of the 39 MPN patients enrolled in the study are summarized in Table 1. None of the patients was newly diagnosed. The median time between MPN diagnosis and trial registration was 3.6 years (range 1.6-8.6 years). Prior to inclusion, 30 patients (77%) had received cytoreductive therapy and 21 (55%) were treated by phlebotomy. Treatment with mirabegron was completed as per protocol in 32 out of 39 patients (82%). In 2 patients (5%), treatment was stopped due to toxicity, in 2 patients (5%) due to patients' preference, and in one patient (3%) due to breast cancer diagnosis (Table 2). Treatment deviation was described in 16 patients (41%) and was due to patient’s decision (n=6; 15%), doctor’s decision (n=1; 3%), toxicity (n=2; 5%) or other reasons (n=12; 31%). Thirtysix patients (92%) received concomitant medication (Table 2).

Mutational profiles In 33 out of 39 patients (84%) who consented to this subproject, granulocyte DNA was sequenced at study entry using a next-generation sequencing (NGS) panel of 94 genes (Table 3). In 10 out of 33 patients (30%), additional somatic mutations were detected (Table 3) and in 3 of these patients (9%) two concomitant mutations were present (TET2 and DNMT3A, PIAS2 and TYK2, TP53 and PRPF40B). The presence or absence of additional mutations was not associated with clinical or laboratory parameters.

Response None of the patients reached the primary end point of a 50% or more reduction of JAK2-V617F allele burden at 24 weeks (Figure 1 and Online Supplementary Appendix). The median percent change from baseline to week 24 was an

Table 2. Treatment.

Value N=39 Total dose of mirabegron [mg] Total treatment duration [weeks] Total dose per week [mg] Main reason for stopping treatment Treatment was completed as per protocol Unacceptable toxicity Other* Missing Patients receiving any concomitant medication Concomitant medication Hydroxyurea Thioguanin Pipobroman Anagrelide Phlebotomy Other cytoreductive drugs Other treatment

9275 (8875–9775) 27.0 (25.9–28.3) 345 (344–345) 32 (82%) 2 (5%) 3 (8%) 2 (5%) 36 (92%) 20 (51%) 0 0 0 11 (28%) 3 (8%) 35 (90%)

Data are presented as number of (N) of patients (%) or median (interquartile range). *The reasons categorized as “Other” are patients’ wish and a breast cancer diagnosis.

haematologica | 2019; 104(4)

increase of 6.1% [interquartile range (IQR) 3.2-13.8%]. One patient reached the secondary end point with a reduction of JAK2-V617F allele burden of 25% or more after 24 weeks. Hematologic response according to European LeukemiaNet (ELN) and International Working Group Myeloproliferative Neoplasms Research and Treatment (IWG-MRT) criteria was observed in 5 out of 21 patients with PV (24%): 2 of them had a complete response (CR) (10%) and 3 a partial response (PR) (14%) (Table 4). Two of 7 ET patients (29%) showed a PR, i.e. reduction in platelet count. One patient with MF became transfusion-independent (9%), all other MF patients (n=10) showed no response (91%). There was no difference in spleen size between baseline and 24 weeks by ultrasound [median spleen longitudinal diameter 15 cm (IQR 13-21) vs. 16 cm (IQR 14-19)]. All parameters measured are listed in Online Supplementary Table S1.

Adverse events Overall, 33 patients (85%) had at least one adverse event, 3 (8%) of them with Common Terminology Criteria for Adverse Events (CTCAE) 4.0 grade 3, 12 (31%) with worst grade 2, 18 (46%) with worst grade 1, and no patient with grade 4 or 5 event. Five adverse events Table 3. Additional somatic mutations in myeloproliferative neoplasm patients.

Gene

Mutation

TET2

M695fs, K1125E, E1339D, T1393I, Y1345C V468M R687Q E282G E45G S533delinsWS A597T H179R V673L

DNMT3A GSN JAK2 NFE PIAS2 PRPF40B TP53 TYK2

Patients (n=33) N % 5

1 1 1 1 1 1 1 1

3 3 3 3 3 3 3 3

Table 4. Response to mirabegron therapy.

Outcome Percent change of allele burden at 24 weeks Allele burden reduction of 50% at 24 weeks Allele burden reduction of 25% at 24 weeks Overall hematologic response in PV, n = 21 CR PR No response Overall hematologic response in ET, n = 7 PR No response Overall hematologic response in MF, n = 11 Improvement of anemia No response

Value N=39 6.1 (3.2–13.8) 0 1 (3%) 2 (10%) 3 (14%) 16 (76%) 2 (29%) 5 (71%) 1 (9%) 10 (91%)

Data are presented as number (N/n) of patients (%) or median (IQR). PV: polycythemia vera; CR: complete response; PR: partial response; ET: essential thrombocythemia; MF: myelofibrosis.

713

B. Drexler et al.

were of grade 3, including gastrointestinal disorders (nausea, vomiting), nervous system disorders (headache, paresthesia) and secondary malignancy (one case of breast cancer). The latter was reported as a serious adverse event and assessed as unrelated to the trial treatment. Nine adverse events were considered to be possibly related to the trial treatment by the investigators (nausea, vomiting, headache, paresthesia, insomnia, pruritus, prostatic obstruction, mucositis, vestibular disorder). No death was observed. The observed adverse events reflect the known profile of mirabegron.

Disease-related symptoms During the trial, 20 patients (51%) suffered from at least one disease-related symptom (DRS). Considering the highest CTCAE 4.0 grade DRS per patient, only one (3%) DRS was grade 3 (abdominal distension), 6 (15%) were grade 2, and 13 (33%) grade 1. Most DRS were gastrointestinal (abdominal distension, early satiety), general (fatigue, fever), microvascular (erythromelalgia, acroparesthesia, digital ischemia), headache, and pruritus.

Figure 2. Reticulin fibrosis and nestin positive (nestin+) cells before and after treatment with mirabegron. (A) Bone marrow histology of a patient before (week 0) and at the end (week 24) of treatment with mirabegron. (Top) Reticulin fibers are stained black by silver impregnation (Gรถmรถri). (Bottom) Immunohistochemical staining with a monoclonal antibody against human nestin protein. Note decrease of reticulin fibrosis and increase of nestin+ cells (brown staining) after 24 weeks of treatment. Magnification: 200x. (B) Single patient evolutional curves of the grade of reticulin fibrosis (left) and nestin+ mesenchymal cells/mm2 (right) at study inclusion and after 24 weeks of mirabegron. n: number; PMF: primary myelofibrosis; ET: essential thrombocythemia; PV: polycythemia vera; MF: myelofibrosis.

714

haematologica | 2019; 104(4)

Mirabegron in MPN

Bone marrow histology Bone marrow biopsies before and after mirabegron treatment were obtained in 20 patients of the 39 patients who consented to this subproject (51%). These included 9 PV, 4 ET, 4 PMF, 2 post-ET, and 1 post-PV MF patients. The biopsies were evaluated in a blinded fashion. A slight decrease in reticulin fiber content from a median grade of 1.0 (IQR 0-3) to 0.5 (IQR 0-2) (P=0.01) and an increase in the nestin+ MSCs cells from a median of 1.09/mm2 (IQR 0.38-3.27) to 3.95/mm2 (IQR 1.98-8.79) (P<0.0001) were observed (Figure 2). The mean change in the nestin+ cells from baseline to week 24 was 3.52/mm2 [95% confidence interval (CI): 1.65-5.39]. We found no correlation between reticulin fibrosis or nestin+ cell content with time from diagnosis to study inclusion, blood counts, splenomegaly, and JAK2-V617F allele burden. The decrease in reticulin fibrosis was limited to patients without hydroxyurea treatment (-0.85/mm2 without hydroxyurea vs. 0.0/m2 with hydroxyurea; P=0.042). No statistically significant differences in CD34+ cell numbers were noted on paired samples before and after 24 weeks of mirabegron. Quantitative assessment of megakaryocyte numbers showed no differences between baseline to week 24 (median 25.5/mm2, IQR 16.75-34.25 vs. 22/mm2, IQR 14.38-29.63; P=0.371), but a trend towards reduction in megakaryocyte cluster formation and decrease in numbers of large megakaryocytes with staghorn-like morphology was noted in some patients.

Discussion Mirabegron was safe and well tolerated in patients with JAK2-mutated MPNs. However, the primary end point of reducing the JAK2-V617F allele burden was not reached (Figure 1). A slight overall hematologic improvement was seen in a subset of patients, but was not considered clinically relevant (Table 4). In a JAK2-V617F-driven mouse model of MPN, treatment with the β-3-sympathomimetic agonist BRL37344 lowered platelet and neutrophil counts, and decreased mutant hematopoietic progenitor numbers and spleen size.8 However, we did not observe effects on blood counts, spleen size or CD34+ cells in our phase II study. Species differences in the β3adrenergic signaling and responsiveness of β3-adrenergic receptors towards different agonists between human and mouse could contribute to the observed discrepancies. Mirabegron is selective for the human β3-adrenergic receptor and was less effective in mice,8 whereas BRL37344 shows higher affinity for the murine β3-adrenergic receptor. Nevertheless, some of the effects observed in the preclinical JAK2-V617F mouse model treated with BRL37344, i.e. increase in nestin+ bone marrow MSCs and

References 1. Mead AJ, Mullally A. Myeloproliferative neoplasm stem cells. Blood. 2017; 129(12):1607-1616. 2. James C, Ugo V, Le Couedic JP, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes poly-

haematologica | 2019; 104(4)

decrease in myelofibrosis, were also seen in our mirabegron study: BM biopsies performed in a subset of 20 patients revealed a significant increase in the nestin+ MSCs and a decrease in reticulin fibrosis (Figure 2). Although the beneficial effect of mirabegron on reticulin fibrosis was moderate, the duration of treatment was also rather short (24 weeks), as it was mainly designed to assess the primary end point of reduction of allele burden. The question of whether a higher dose of mirabegron might have been more effective is difficult to answer. Although doses of 100 mg daily have been tested in earlier clinical studies, no clear dose-dependent effect has been observed, while cardiovascular symptoms and a prolongation of the QT interval were noted.9,19 The fact that nestin+ cells showed a robust increase at 24 weeks of treatment indicates that mirabegron at 50 mg daily had one of the expected biological effects that had previously been described in mouse experiments. Surprisingly, the effect on reticulin fibrosis was limited to patients who did not receive hydroxyurea treatment (41% of patients). The mechanism of how hydroxyurea interfered with the effect of mirabegron on reticulin fibrosis is currently unknown. Previous reports suggest that hydroxyurea alone can reduce reticulin fibrosis in some MPN patients.20,21 Selecting patients who have not previously received hydroxyurea, a longer trial duration and higher dosage of mirabegron will be considered for future studies in MPN. Despite the fact that the primary end point of reducing JAK2-V617F allele burden was not reached in this trial, the observed effects on nestin+ MSCs and reticulin fibrosis is encouraging and shows that a β-3-sympathomimetic agonist can modify the microenvironment where the JAK2-mutant stem cells are maintained. These results generate an interest in evaluating β-3-sympathomimetic agonists specifically in patients with myelofibrosis not pretreated with hydroxyurea, and possibly in combination with other substances. Acknowledgments The authors thank the patients for participating in the study and the local data managers for collecting patient data. Funding This investigator-initiated trial was supported by grants from the Rising Tide Foundation, Gateway for Cancer Research, Swiss Cancer League (KFS-3655-02-2015), and the Swiss National Science Foundation (KFS-3539-08-2014) (31003A_166613) to RCS, ERC-2014-CoG-648765 grant to SMF, and the Swiss Cancer League to JRP. The SAKK organization is supported by the Swiss State Secretary for Education, Research and Innovation, Swiss Cancer Research Foundation and the Swiss Cancer League. The study drug mirabegron was provided free of charge by Astellas Pharma AG Switzerland.

cythaemia vera. Nature. 2005;434(7037): 1144-1148. 3. Kralovics R, Passamonti F, Buser AS, et al. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med. 2005;352(17):1779-1790. 4. Levine RL, Wadleigh M, Cools J, et al. Activating mutation in the tyrosine kinase JAK2 in polycythemia vera, essential

thrombocythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell. 2005; 7(4):387-397. 5. Baxter EJ, Scott LM, Campbell PJ, et al. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. Lancet. 2005;365(9464):1054-1061. 6. Mendez-Ferrer S, Lucas D, Battista M, Frenette PS. Haematopoietic stem cell

715

B. Drexler et al.

10.

11.

716

release is regulated by circadian oscillations. Nature. 2008;452(7186):442-447. Mendez-Ferrer S, Michurina TV, Ferraro F, et al. Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature. 2010;466(7308):829-834. Arranz L, Sanchez-Aguilera A, MartinPerez D, et al. Neuropathy of haematopoietic stem cell niche is essential for myeloproliferative neoplasms. Nature. 2014; 512(7512):78-81. Khullar V, Amarenco G, Angulo JC, et al. Efficacy and tolerability of mirabegron, a beta(3)-adrenoceptor agonist, in patients with overactive bladder: results from a randomised European-Australian phase 3 trial. Eur Urol. 2013;63(2):283-295. Vardiman JW, Thiele J, Arber DA, et al. The 2008 revision of the WHO classification of myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009;114(5):937-951. Barosi G, Mesa R, Finazzi G, et al. Revised response criteria for polycythemia vera and essential thrombocythemia: an ELN and IWG-MRT consensus project. Blood.

2013;121(23):4778-4781. 12. Tefferi A, Cervantes F, Mesa R, et al. Revised response criteria for myelofibrosis: International Working GroupMyeloproliferative Neoplasms Research and Treatment (IWG-MRT) and European LeukemiaNet (ELN) consensus report. Blood. 2013;122(8):1395-1398. 13. Kralovics R, Teo SS, Li S, et al. Acquisition of the V617F mutation of JAK2 is a late genetic event in a subset of patients with myeloproliferative disorders. Blood. 2006;108(4):1377-1380. 14. Lundberg P, Karow A, Nienhold R, et al. Clonal evolution and clinical correlates of somatic mutations in myeloproliferative neoplasms. Blood. 2014;123(14):2220-2228. 15. Thiele J, Kvasnicka HM, Facchetti F, Franco V, van der Walt J, Orazi A. European consensus on grading bone marrow fibrosis and assessment of cellularity. Haematologica. 2005;90(8):1128-1132. 16. 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. 17. Thiele J, Kvasnicka HM, Orazi A, et al. Myeloprolifierative Neoplasms. In: Swerdlow SH, Campo E, Harris NL, et al., eds. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Lyon: International Agency for Research on Cancer (IARC); 2017:585. 18. Kvasnicka HM, Beham-Schmid C, Bob R, et al. Problems and pitfalls in grading of bone marrow fibrosis, collagen deposition and osteosclerosis - a consensus-based study. Histopathology. 2016;68(6):905-915. 19. Nitti VW, Auerbach S, Martin N, Calhoun A, Lee M, Herschorn S. Results of a randomized phase III trial of mirabegron in patients with overactive bladder. J Urol. 2013;189(4):1388-1395. 20. Lofvenberg E, Wahlin A, Roos G, Ost A. Reversal of myelofibrosis by hydroxyurea. Eur J Haematol. 1990;44(1):33-38. 21. Harrison CN, Campbell PJ, Buck G, et al. Hydroxyurea compared with anagrelide in high-risk essential thrombocythemia. N Engl J Med. 2005;353(1):33-45.

haematologica | 2019; 104(4)

ARTICLE

Chronic Myeloid Leukemia

A new BCR-ABL1 Drosophila model as a powerful tool to elucidate the pathogenesis and progression of chronic myeloid leukemia

Ferrata Storti Foundation

Roberto Bernardoni,1,2,§,# Giorgia Giordani,1,3,4,§ Elisabetta Signorino,3,§ Sara Monticelli,1 Francesca Messa,1 Monica Pradotto,3 Valentina Rosso,3 Enrico Bracco,5 Angela Giangrande,6 Giovanni Perini,1,2* Giuseppe Saglio3,* and Daniela Cilloni3,*#

1 Department of Pharmacy and Biotechnology (FABIT), University of Bologna, Italy; 2Health Sciences and Technology - Interdepartmental Center for Industrial Research (HST-ICIR), University of Bologna, Ozzano Emilia, Italy; 3Department of Clinical and Biological Sciences, University of Turin, Italy; 4Present address: Department of Biological Sciences, School of Applied Sciences, University of Huddersfield, Queensgate, UK; 5Department of Oncology, University of Turin, Italy and 6Institut de Génétique et de Biologie Moléculaire et Cellulaire, CNRS/INSERM/ULP 67404 Illkirch, France

Haematologica 2019 Volume 104(4):717-728

These authors share first authorship.

These authors share last authorship.

ABSTRACT

he oncoprotein BCR-ABL1 triggers chronic myeloid leukemia. It is clear that the disease relies on constitutive BCR-ABL1 kinase activity, but not all the interactors and regulators of the oncoprotein are known. We describe and validate a Drosophila leukemia model based on inducible human BCR-ABL1 expression controlled by tissue-specific promoters. The model was conceived to be a versatile tool for performing genetic screens. BCR-ABL1 expression in the developing eye interferes with ommatidia differentiation and expression in the hematopoietic precursors increases the number of circulating blood cells. We show that BCR-ABL1 interferes with the pathway of endogenous dAbl with which it shares the target protein Ena. Loss of function of ena or Dab, an upstream regulator of dAbl, respectively suppresses or enhances both the BCR-ABL1-dependent phenotypes. Importantly, in patients with leukemia decreased human Dab1 and Dab2 expression correlates with more severe disease and Dab1 expression reduces the proliferation of leukemia cells. Globally, these observations validate our Drosophila model, which promises to be an excellent system for performing unbiased genetic screens aimed at identifying new BCR-ABL1 interactors and regulators in order to better elucidate the mechanism of leukemia onset and progression.

Introduction Chronic myeloid leukemia (CML) is a clonal myeloproliferative disorder associated with a reciprocal translocation between chromosomes 9 and 22. This process leads to the fusion of the Abelson (ABL1) tyrosine kinase gene with the breakpoint cluster region (BCR) sequences generating a fusion gene encoding the constitutively active protein tyrosine kinase BCR-ABL1. Due to its high frequency in CML patients (95%), the translocation is considered the cytogenetic hallmark of this disease.1,2 Although BCR-ABL1 is one of the most studied oncogenic proteins, some molecular mechanisms leading to cellular transformation are still partially unknown. In particular, positive or negative regulators of BCR-ABL1 have not been completely identified. The fruitfly, Drosophila melanogaster, represents a powerful tool for genome-wide genetic analysis and screens, given the functional conservation and sequence homology between human and Drosophila genes. Genome-wide approaches may allow identification of genetic pathways that contribute to disease onset and/or progression without a priori knowledge of the gene function.3 The high degree of conservation between human and haematologica | 2019; 104(4)

Correspondence: DANIELA CILLONI daniela.cilloni@unito.it ROBERTO BERNARDONI roberto.bernardoni@unibo.it Received: May 21, 2018. Accepted: November 8, 2018. Pre-published: November 8, 2018. doi:10.3324/haematol.2018.198267 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/717 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

717

R. Bernardoni et al.

Drosophila Abl (dAbl) proteins and the existence of Drosophila homologs for many proteins that interact functionally with BCR-ABL1 in mammals strongly support the idea that dAbl and presumably BCR-ABL1 signal transduction pathways could be highly conserved from fly to human. The dAbl gene is expressed at high levels in differentiating neurons and plays an important role in central nervous system, eye and epithelia development, mainly regulating cytoskeleton remodeling.4-6 Interestingly, Forgerty and colleagues demonstrated that the neural expression of a chimeric BCR-ABL protein carrying the human BCR fused to dAbl is able to rescue the dAbl mutant phenotype, suggesting that the chimeric BCR-ABL protein can effectively compensate for lack of dAbl.7 To further identify genes and pathways involved in the onset and progression of CML, we developed and validated a genetic model based on transgenic flies that drive inducible human BCR-ABL1 expression under the control of tissue- and stage-specific promoters, providing both an excellent and powerful model to identify novel functional interactors.

Methods Generation of BCR-ABL1 transgenic flies The BCR-ABL1 coding sequence was amplified by polymerase chain reactions and cloned into the P-element expression vector pKS69. BCR-ABL1 kinase-dead (BCR-ABL1KD) was obtained through site-directed mutagenesis (Online Supplementary Data). Plasmids were prepared using QiafilterTM Plasmid Maxi Kit (Qiagen, Venlo, the Netherlands) and injected in Drosophila embryos (The BestGene, Inc, Chino Hills, CA, USA).

Drosophila stocks Fly stocks were obtained from Bloomington Drosophila Stock Center (Department of Biology, Indiana University, Bloomington, IN, USA). RNA interference (RNAi) lines were obtained from the Vienna Drosophila RNAi Center (Vienna, Austria). domelessGal4 and STATDN flies were kindly provided by A. Giangrande (IGBMC, Illkirch, France) (Online Supplementary Data).

(DSHB), incubated with a Cy3-conjugated anti-rat secondary antibody (Jackson Immunoresearch, Newmarket, UK) and exposed to HOECHST (Sigma-Aldrich Corp., St. Louis, MO, USA) before mounting in Fluormount-G (Electron Microscopy Sciences, Hatfield, PA, USA) (Online Supplementary Data).

Primary cells The protocol was approved by the local ethics committee (approval n. 212/2015). White blood cells (105) were obtained from peripheral blood. Immunofluorescence was performed as previously described8. Mouse anti-Dab1 and anti-Dab2 primary antibodies (sc-271136 and sc-136963, Santa Cruz BIotechnology) and anti-mouse Alexa Fluor 568 secondary antibody (Molecular Probes-Invitrogen, ThermoFisher Scientific, Waltham, MA, USA) were used (Online Supplementary Data).

Genetic analysis Eye Flies carrying gmrGal4 or sevGal4 driver constructs were crossed to the UAS-BCR-ABL1 transgenic lines. To analyze the phenotype, flies from a recombinant line carrying both gmrGal4 and UASBCR-ABL1 on the third chromosome (gmrGal4,UAS-BCR-ABL1 4M/TM3) were crossed to lines carrying single gene mutations, deficiencies or RNAi constructs. Fifteen to 30 F1 flies from three independent crosses were classified into three phenotypic classes described in the Results section.

Melanotic nodules domelessGal4-driven BCR-ABL1 expression was controlled with the TARGET system9,10 (Online Supplementary Data). We performed conditional expression in the medullary zone of the lymph gland starting at different stages during larvae development by moving the animals from 18°C to 29°C. Analysis of the melanotic nodule phenotype and temperature shift experiments were performed as previously described.11

RNA extraction and quantitative analysis RNA was extracted using standard procedures. Expression levels of Dab1 and Dab2 were evaluated by real-time polymerase chain reaction using specific on-demand kits (Hs00245445_m1 for ABL1, Hs00221518_m1 for Dab1, Hs00184598_m1 for Dab2, Applied Biosystems, ThermoFisher Scientific) according to published methods.12

Immunoblotting Adult heads were dissected and homogenized in a protein extraction buffer. For cell lines, 107 cells were lysed in RIPA buffer. The following primary antibodies were used: c-Abl (sc23), Dab1 (sc-271136), p-Tyr (sc-7020), GAPDH (sc-137179) (Santa Cruz Biotechnology, Santa Cruz, CA, USA), α-tubulin (CP06; Oncogene Research Products, Merck KGaA, Darmstadt, Germany) mouse monoclonal antibodies, BCR (sc-20707) rabbit polyclonal antibody (Santa Cruz Biotechnology) and mouse 5G2 anti-Enabled supernatant (Developmental Studies Hybridoma Bank - DSHB, University of Iowa, IA, USA). For immunoprecipitation, 1 mg of total protein extract was incubated with anti-Enabled supernatant and subsequently with protein A sepharose (Amersham Bioscience, GE Healthcare, Waukesha, WI, USA) (Online Supplementary Data).

Fluorescent Immunolabeling Fly eye primordium Eye imaginal discs were dissected from third instar larvae, fixed in 4% paraformaldehyde, permeabilized with 0.3% Triton X-100, labeled with the rat anti-Elav 7E8A10 supernatant 718

Results Expression of human BCR-ABL1 affects eye cell differentiation The aim of this work was to set up a CML Drosophila model based on the expression of a completely human BCR-ABL1 fusion protein. Available Drosophila genetic tools allow expression of proteins of interest in developing eye cells, often inducing viable and visible phenotypic traits that can be used as a bait in genetic screening. The Drosophila eye differentiates during the third instar larva (L3) from the eye imaginal disc, a monolayer epithelium that is accessible to dissection. We generated several stable transgenic fly lines to express BCR-ABL1 protein using the yeast Gal4/UAS (Upstream Activating Sequence) transcriptional regulation system controlled by a gene promoter active in specific tissues and stages (Gal4 drivers).13 BCR-ABL1 expression was first triggered with the sevenlessGal4 (sevGal4) construct that drives haematologica | 2019; 104(4)

New Drosophila model for chronic myeloid leukemia

high levels of expression in some but not all photoreceptors,14 producing a mild rough eye similar to the one observed by Fogerty7 (Figure 1A-E). This suggests that BCR-ABL1 interferes with eye development as described for the human/fly chimera. To drive BCR-ABL1 expression in more eye cells, we used the glass multimer reporterGal4 (gmrGal4) driver, active in all cells committed to differentiation and located posteriorly to the morphogenetic furrow,15 the cell indentation crossing the eye pri-

mordium from posterior to anterior (Figure 1N,O). BCRABL1 expression in these cells produced a severe “glazed” eye phenotype (Figure 1F-J, Online Supplementary Figure S1A,B,H,I). The regular structure of the eye was almost completely lost: ommatidia, the functional units of the eye, failed to differentiate and were no longer distinguishable. The eye was smaller, bar-shaped and misplaced extra sensory bristles appeared in the dorsal region (Figure 1H-J). Western blot

haematologica | 2019; 104(4)

Figure 1. BCR-ABL1 expression in the developing eye cells affects photoreceptor differentiation. (A-E) Adult eyes expressing EGFP (A) or BCR-ABL1 in four independent transgenic lines, 1M (B), 3M (C), 4M (D), or 7M (E), in a subset of differentiating photoreceptor cells under the control of the sevenlessGal4 driver construct. (C-E) High levels of BCR-ABL1 induce a “rough” eye phenotype due to impairment of cell differentiation. (F-J) Adult eyes expressing EGFP (F) or BCR-ABL1 (G-J) in all differentiating eye cells under the control of the gmrGal4 driver construct. (H-J) BCR-ABL1 expressed at high level in all differentiating eye cells profoundly disrupts ommatidia development inducing a “glazed” phenotype, depigmented area and the appearance of extra bristles (black arrows). (K-M) Quantification of BCRABL1 expression (K,L) and tyrosine-phosphorylation (K,M) in protein extracts from adult heads of flies expressing either EGFP (lane 1) or BCR-ABL1 in independent transgenic fly lines (lanes 2-5). The protein extracts were probed with antibodies raised against BCR, phosphorylated tyrosine residues (p-Tyr) or α–tubulin as the loading control. (N) Schematic of the eye-antenna imaginal disc from a late third instar larva; the positions of the eye and antenna primordia and of the morphogenetic furrows are indicated. The eye imaginal disc area posterior to the morphogenetic furrow, made of cells committed to terminal differentiation, is indicated in green. The thin black square indicates the region of interest shown in panels O-T. (O,P) Eye imaginal disc from wildtype late third instar larvae expressing EGFP under the control of the gmrGal4 driver in cells posterior to the morphogenetic furrow and expressing the pan-neuronal marker Elav in cells committed to terminal differentiation. (Q-T) Elav expression in eye imaginal discs from third instar larvae of the four independent transgenic lines that express BCR-ABL1 under the control of the gmrGal4 driver construct. BCR-ABL1 expression reduces the number of differentiated photoreceptors as indicated by the decrease of Elav-expressing cells.

719

R. Bernardoni et al.

analysis demonstrated that the severity of the phenotype correlated with the amount and phosphorylation of BCR-ABL1 protein (Figure 1K-M): indeed the low level of BCR-ABL1 expression observed in line 1M (Figure 1K-M) resulted in a very mild phenotype (Figure 1G). To better understand the origin of the phenotype, we analyzed the expression of the pan-neuronal and eye photoreceptor marker Elav16 in eye imaginal discs expressing BCRABL1. The typical Elav+ photoreceptor clusters (Figure 1P) were reduced in number and altered in BCR-ABL1expressing flies and this correlated with the described defects of the eye’s ordered structure (Figure 1P-T). To assess whether the phenotype depends on BCR-ABL1 kinase activity, we generated transgenic flies to express a kinase-dead mutant BCR-ABL1. gmrGal4-driven expression of the mutant protein did not affect eye development, indicating that the BCR-ABL1 phenotype requires the enzymatic activity of the oncoprotein (Online Supplementary Figure S1A-C,H).

Expression of human BCR-ABL1 interferes with eye development by altering dAbl signaling To better understand the consequences of BCR-ABL1 overexpression in the eye, we investigated whether the human oncoprotein could activate the endogenous pathway regulated by the Drosophila Abl kinase (dAbl). To quantify the phenotype we classified BCR-ABL1 eyes (line 4M) into three phenotypic classes. Class 0 represents the most frequent “glazed” phenotype. Class +1 is less severe: the eye is bigger and more prominent, and some ommatidia can be observed. Class -1 is more severe, being characterized by a less differentiated eye with evident lack of pigmentation in the most posterior region (Figure 2A). Interestingly, phenotype expressivity did not change comparing gmrGal4,UAS-BCR-ABL1 4M animals with gmrGal4,UAS-BCR-ABL1 4M;UAS-EGFP (Figure 2B, Online Supplementary Figure S1H,I) indicating that a single gmrGal4 copy does not express a Gal4 limiting amount that could be titrated by increasing the number of UAS sequences. Since overexpression of dAbl (UAS-Abl) induces a very mild rough eye phenotype (Online Supplementary Figure S1A,B,G), we investigated whether it could enhance the BCR-ABL1 phenotype. We observed a worsening of the phenotype: all of the eyes belonged to class -1, showing smaller eyes and more evident loss of pigmentation (Figure 2B, Online Supplementary Figure S1G,H,N). We then investigated whether dAbl loss of function (LOF) could suppress the BCR-ABL1 phenotype. gmrGal4,UAS-BCR-ABL1 4M animals heterozygous for a dAbl hypomorphic recessive lethal allele (Abl1/+) showed a very mild phenotypic suppression but were not statistically different from controls (Figure 2B, Online Supplementary Figure S1E,H,L). However, dAbl downregulation through RNAi (AblRNAi) or expression of a dominant negative kinasedefective dAbl (UAS-AblK417N) induced a significant suppression of the BCR-ABL1 phenotype (Figure 2B, Online Supplementary Figure S1D,F,H,I,K,M). Interestingly, we observed that animals expressing either Abl-RNAi or UAS-Abl or UAS-AblK417N showed a similar mild disorganization of the ommatidia (Online Supplementary Figure S1A,B,D,F,G) suggesting that the pathway activated by dAbl is indeed implicated in eye development. Furthermore, the genetic interactions between BCRABL1 expression and dAbl loss or gain of function sug720

gest that dAbl, dAblK417N and overexpressed BCR-ABL1 could compete for common binding targets. To confirm that BCR-ABL1 overexpression affects eye development by altering dAbl signaling cascade, we analyzed whether BCR-ABL1 could functionally interact with components of the dAbl pathway. In detail, we focused on four genes whose LOF mutations genetically interact with a dAbl mutant phenotype. Mutations of prospero (pros), a transcription factor that regulates neuronal differentiation17, failed axon connections (fax), implicated both in neurogenesis and axonogenesis18 and Disabled (Dab) that regulates cellular localization of dAbl19, enhance the mutant dAbl phenotype. Moreover, enabled (ena) gene mutations suppress a dAbl mutant phenotype.20 Interestingly, we found that either a deletion or a mutant allele of pros (Figure 2C, Online Supplementary Figure S2A-C) and fax (Figure 2D, Online Supplementary Figure S2A,D,E) was able to enhance the BCR-ABL1 phenotype. Moreover, although the insertional DabEY10190 allele did not change the BCRABL1 phenotype significantly, a deletion uncovering the Dab locus enhanced it (Figure 2E, Online Supplementary Figure S2A,F,G), confirming that BCR-ABL1 expression alters eye development likely by interacting with components of the dAbl pathway.

BCR-ABL1 expression increases phosphorylation of the dAbl substrate Ena A genetic screen had previously identified an ena LOF allele as a suppressor of the recessive lethality due to dAbl LOF mutations.20 Ena is a cytoskeletal regulator that facilitates actin polymerization.21 Its cellular localization depends on dAbl5,20,22 and it is phosphorylated by both human and Drosophila Abl.23,24 Heterozygosis of a LOF ena allele or of an ena deletion suppressed the BCR-ABL1 phenotype (Figure 2F, Online Supplementary Figure S2A,J,K). ena silencing with two independent constructs (ena-RNAi), induced a size increase and strong decrease of depigmented tissue in eyes expressing BCR-ABL1 (Figure 3A, Online Supplementary Figure S2A,L,M). Consistently, the analysis of Elav expression highlighted a more correct organization of photoreceptor clusters (Figure 3B-E). Furthermore, we looked at tyrosine-phosphorylation of the endogenous Ena. Flies expressing BCR-ABL1 showed increased levels of Ena tyrosinephosphorylation (Figure 3F,H) even after Ena immunoprecipitation (Figure 3G,H) suggesting that Ena might be phosphorylated by BCR-ABL1. Taken together our data indicate that alteration of several components of the dAbl pathway could be important for the mechanism by which BCR-ABL1 overexpression affects eye development, likely phosphorylating conserved targets in fly eye cells.

A component of the BCR-ABL1-activated pathway in human leukemia modulates the eye phenotype in Drosophila To further assess the effectiveness of the model, we investigated whether a Drosophila homolog of a gene known to be involved in BCR-ABL1 signaling in human leukemia was also able to modulate the BCR-ABL1 phenotype. Signal transducer and activator of transcription 5 (STAT5) is a transcription factor activated in response to cytokines and its role in malignant transformation is well established.25 Several studies showed that BCR-ABL1 induces phosphorylation and constitutive activation of haematologica | 2019; 104(4)

New Drosophila model for chronic myeloid leukemia

haematologica | 2019; 104(4)

Figure 2. BCR-ABL1 expression affects ommatidia development by altering the dAbl signaling pathway. (A) Adult eyes showing the phenotypic classes used to quantify the severity of the variable phenotype due to BCR-ABL1 expression in eye cells committed to terminal differentiation. Posterior is on the left. Class 0 (green) corresponds to the average phenotype shown by gmrGal4, UAS-BCR-ABL1 4M flies; the ommatidia are almost totally absent, the eye depigmented region is very small and the eye appears flatter than that of wild-type eyes. Class -1 (pale blue) corresponds to more severe phenotypes: the eyes are even flatter than class 0 eyes and the depigmented area is enlarged including a dorso-ventral sector in the most posterior region of the eye. Class +1 (pink) corresponds to less severe phenotypes: a few ommatidia are visible, the eyes are more bulging and the depigmented area is absent indicating better eye cell differentiation. (B-F) Adult eyes from flies of the indicated genotypes were classified to evaluate the frequency of the three phenotypic classes. (B) A piled histogram chart showing the frequencies of the different phenotypic classes in flies expressing BCR-ABL1 (gmrGal4,4M/+), co-expressing BCR-ABL1 and the EGFP (UASEGFP/+;gmrGal4,4M/+), expressing BCR-ABL1 but having a partial loss of the endogenous Abl gene through the mutation heterozygous Abl1 (Abl1/+;gmrGal4,4M/+), RNAi targeting Abl (Abl-RNAi/+;gmrGal4,4M/+), expression of a dominant negative dAbl mutant (UAS-AblK417N;gmrGal4,4M/+) or overexpression of the wild-type Abl protein (UAS-Abl;gmrGal4,4M/+). (C-F) Piled histogram charts showing the frequencies of the three phenotypic classes in flies expressing BCR-ABL1 (gmrGal4,4M) and heterozygous for a loss of function allele or for a deletion of genes that behave as genetic modifiers of the embryonic lethality due to Abl LOF: prospero (C), failed axon connection (D), Disabled (E) and enabled (F). The statistical comparisons were conducted using a Mann-Whitney test (*P<0.05, **P<0.01, ***P<0.001).

721

R. Bernardoni et al.

Figure 3. Ena loss of function suppresses the eye phenotype due to BCR-ABL1 expression which increases phosphorylation of the dAbl target Ena. (A) Piled histogram chart showing the frequencies of the three phenotypic classes in flies co-expressing BCR-ABL1 (gmrGal4,4M) and EGFP (UAS-EGFP/+;gmrGal4,4M/+), or one of two independent ena-RNAi lines (VDRC#43056 and VDRC#106484). (B,C) Eye imaginal discs from wild-type late third instar larvae expressing EGFP under the control of the gmrGal4 driver in cells posterior to the morphogenetic furrow (B) and expressing the pan-neuronal marker Elav in cells committed to terminal differentiation (C). (D,E) Elav expression in eye imaginal discs from late third instar larvae expressing BCR-ABL1 (D) or larvae co-expressing BCR-ABL1 and ena-RNAi (E) under the control of the gmrGal4 driver construct. BCR-ABL1 expression reduces the number of differentiated photoreceptors, as indicated by a decrease of Elav-expressing cells, and ena downregulation suppresses this phenotypic trait. (F-H) Quantification of Ena expression and tyrosine-phosphorylation in protein extracts (F) or Enaimmunoprecipitated proteins (G) from the heads of adult flies expressing either EGFP (lane 1) or BCR-ABL1 (lane 2). Independent loads of equal amount of protein extracts or Ena-immunoprecipitated proteins were probed with antibodies raised against phosphorylated tyrosine residues (p-Tyr), Ena or Îąâ&#x20AC;&#x201C;tubulin as loading control. (H) Average signal intensity from replica of the experiment shown in (F) and (G). Ena immunoprecipitation and probing for tyrosine-phosphorylation confirmed the increase of Ena tyrosine-phosphorylation in animals expressing BCR-ABL1. The statistical comparisons in (A) were conducted using the Mann-Whitney test (*P<0.05, **P<0.01, ***P<0.001).

STAT5, hindering apoptosis in leukemic cells.26 The JAK/STAT pathway is required during Drosophila eye morphogenesis and larval hematopoiesis.27,28 Interestingly, loss of STAT92E function (STAT92E06346), the fly counterpart of STAT5, induced strong suppression of the BCR-ABL1 phenotype (Figure 4, Online Supplementary Figure S3A,B). Flies coexpressing a STAT dominant negative allele (STATDN) and BCR-ABL1 showed an even weaker phenotype (Figure 4A, Online Supplementary Figure S3A,C) confirming that STAT is involved in the BCR-ABL1-activated pathway in the Drosophila eye.

The human homologs of Disabled, Dab1 and Dab2, are altered in patients with chronic myeloid leukemia To better explore the efficacy of the model we analyzed the Disabled gene that encodes for an adaptor protein acting downstream of many receptor tyrosine kinases.17,29 In the embryo Dab LOF disrupts the intracellular localization of dAbl and consequently that of phosphorylated Ena and F-Actin accumulation.30 In the fly eye we observed an enhancement of the BCR-ABL1 phenotype in animals heterozygous for a Dab deletion. Thus, we further reduced Dab function by gene silencing. 722

Figure 4. A component of the BCR-ABL1-activated pathway in human leukemia modulates the eye phenotype in Drosophila. Piled histogram chart showing the frequencies of the three phenotypic classes in flies expressing BCR-ABL1 (gmrGal4,4M) and heterozygous for a loss of function STAT92E06346 allele or overexpressing a dominant negative allele STATDN. Reduction of the function of STAT, a gene encoding the homolog of the STAT5 protein involved in the BCR-ABL1-activated pathway in human leukemia cells, suppresses the BCR-ABL1-dependent phenotype in the fly eye. The statistical comparisons were conducted using the Mann-Whitney test (*P<0.05, **P<0.01, ***P<0.001).

haematologica | 2019; 104(4)

New Drosophila model for chronic myeloid leukemia

Interestingly, two independent RNAi lines worsened the BCR-ABL1 phenotype more than the Dab deletion did (Figures 2E and 5A, Online Supplementary Figure S2A,F-I): most of the eyes were smaller and showed depigmented scar-like tissue (Online Supplementary Figure S2A,F,H,I). Consistently, alterations of the ommatidia clusters, revealed by Elav expression, worsened compared to those of the control (Figure 5B-E). To establish whether Dab might have a role in CML we analyzed the two human counterparts of Disabled, Dab1 and Dab2 in human primary cells. Dab1 is a large, common fragile site gene and the Dab1 protein acts as a signal transducer that interacts with many receptor tyrosine kinase pathways.31 Dab2 encodes for an adaptor protein implicated in growth factor signaling, endocytosis, cell adhesion, hematopoietic cell differentiation and cell signaling of various receptor tyrosine kinases.32 The expression of both genes is often decreased in many human solid cancers, suggesting their possible role in oncogenesis.31,33 Interestingly, quantitative real-time polymerase chain reaction analysis revealed a significant downregulation of both genes in CML patients at diagnosis compared to controls in peripheral blood or bone marrow samples (Figure 6A,G). Analysis of bone marrow samples from CML patients during molecular remission showed increased levels of expression of both Dab1 (Figure 6B) and Dab2 (Figure 6H) with respect to the levels in treatment-resistant patients. Moreover, immunofluorescence assays demonstrated a significant down-modulation of both proteins in peripheral blood samples at diagnosis compared to the levels in controls or patients in molecular remission (Figure 6C,D,I,J). Finally, transfection experiments in K562 cells using a plasmid carrying the whole Dab1 coding sequence demonstrated that reactivation of Dab1 expression reduced cell proliferation (Figure 6E,F).

BCR-ABL1 expression impairs Drosophila blood cell homeostasis To further confirm the efficacy of the model, we inves-

tigated the effects of BCR-ABL1 expression in the lymph gland, the hematopoietic organ of the larva. The lymph gland begins to develop in the embryo34 and grows up from multipotent progenitor cells (prohemocytes) that proliferate and enter a quiescent phase during the second instar (L2). During the third instar (L3) some prohemocytes start to proliferate again and differentiate. The lymph gland breaks apart at the beginning of metamorphosis releasing differentiated blood cells (hemocytes) into the hemolymph, the Drosophila blood.35,36 During the L3, three functional regions can be distinguished in the lymph gland:37 the medullary zone, populated by prohemocytes; the posterior signaling center that regulates the exit of prohemocytes from quiescence; and the cortical zone, made up of differentiating hemocytes.38,39 The lymph gland can break up prematurely in late-L3 if the number of differentiating hemocyte increases. As a reaction to excessive hematopoiesis, the hemocytes aggregate and a spontaneous process of melanization takes place inducing the formation of melanotic nodules.11,36,40,41 Constitutive BCR-ABL1 expression under the control of the domelessGal4 (domeGal4) driver, active in the medullary zone of the lymph gland,11,42 is lethal (data not shown). To overcome this problem, we repressed expression of BCR-ABL1 by co-expressing a heat-sensitive mutant of the Gal4 repressor Gal80 (tubGal80TS) until larvae reached the desired instar (TARGET system).10 While BCR-ABL1 expression from the first instar (L1) induced lethality (data not shown), expression from the L2 allowed larvae to survive and to develop melanotic nodules at L3 (Figure 7A,B). This suggests that BCR-ABL1 expression in the medullary zone precursors might induce an increase of circulating hemocytes (Figure 7C). When compared to controls (Figure 7A), 45% of domeGal4,BCR-ABL1 3M,tubGal80TS larvae showed two to three small melanotic nodules (Figure 7B,C). This correlates with an increased number of circulating hemocytes in hemolymph preparations (Figure 7D). BCRABL1 expression starting from the early L3 did not show

Figure 5. Dab downregulation enhances the eye phenotype due to BCR-ABL1. (A) Piled histogram chart showing the frequencies of the three phenotypic classes in flies co-expressing BCR-ABL1 (gmrGal4,4M) and EGFP (UAS-EGFP/+;gmrGal4,4M/+), or one of two independent Dab-RNAi constructs (VDRC#13005, VDRC#14008). (B,C) Eye imaginal discs from wild-type late third instar larvae expressing EGFP under the control of the gmrGal4 driver in cells posterior to the morphogenetic furrow and expressing the pan-neuronal marker Elav in cells committed to terminal differentiation. (D,E) Elav expression in eye imaginal discs from late third instar larvae expressing BCR-ABL1 (D) or larvae co-expressing BCR-ABL1 and Dab-RNAi (E) under the control of the gmrGal4 driver construct. BCR-ABL1 expression reduces the number of differentiated photoreceptors, as indicated by a decrease of Elav-expressing cells, and Dab downregulation enhances this phenotypic trait. The statistical comparisons were conducted using a Mann-Whitney test (*P<0.05, **P<0.01, ***P<0.001).

haematologica | 2019; 104(4)

723

R. Bernardoni et al.

any significant phenotype (Figure 7C), indicating that only when BCR-ABL1 is expressed when prohemocytes enter quiescence is it able to increase hematopoiesis. Consistently, constitutive expression of the kinase-dead mutant BCR-ABL1KD did not induce any significant phenotype (Figure 7C). Since dAbl, like Dab and ena, is expressed in the lymph gland,43 we assessed whether decreased dAbl function is able to rescue the phenotype. We co-expressed BCR-ABL1 and Abl-RNAi, and observed a significant

724

decrease of the phenotype penetrance (Figure 7E). We then investigated whether Dab or ena downregulation interacts genetically with BCR-ABL1 expression during hematopoiesis as well. Dab-RNAi in the medullary zone starting from L2 was able to enhance the melanotic nodule phenotype, inducing a significant increase of the penetrance (Figure 7F). Consistently, larvae co-expressing Dab (UAS-Dab) and BCR-ABL1 in the medullary zone starting from L2 showed phenotypic suppression (Figure 7F).

Figure 6. Altered pattern of expression of the human Disabled homologs, Dab1 and Dab2, in patients with chronic myeloid leukemia. (A) Downregulation of Dab1 RNA expression in patients with chronic myeloid leukemia (CML) compared to the expression in healthy donors (CTRL). In particular we found a 1 log reduction of Dab1 expression in both peripheral blood (PB) (P<0.01) and bone marrow (BM) (P<0.01) (median values 2-Î&#x201D;Î&#x201D;ct: 0.02 versus 0.3 in PB and 0.008 versus 0.04 in BM). (B) Expression pattern of Dab1 in CML patients during molecular remission (MR) compared to that in treatment-resistant patients. (C) Immunofluorescence staining of Dab1 protein (red) in PB samples of healthy donors, CML patients at diagnosis and CML patients during MR. Nuclei are stained in blue. (D) Quantification of Dab1 protein expression in the immunofluorescence assay. (E) A 3H-thymidine proliferation assay showing a 20% reduction of cell proliferation in K562 cells transfected with Dab1 plasmid compared to control. (F) Western blot of protein extracts from K562 cells transfected with an empty vector (lane 1) and transfected with a Dab1 expression vector (lane 2), showing detectable expression of Dab1 only in K562 cells transfected with the Dab1 vector. Independent loads of equal amounts of protein extract were probed with antibodies raised against BCR, Dab1 and GAPDH as a loading control. (G) Down-regulation of Dab2 RNA expression in CML patients compared to the expression in healthy donors. In particular Dab2 expression was found to be statistically decreased (P<0.0001 and P<0.0001 in PB and BM, respectively) with median values of 0.12 versus 2.8 and 0.12 versus 0.7 in PB and BM, respectively. (H) Pattern of expression of Dab2 in CML patients during MR compared to that in treatmentresistant patients. (I) Immunofluorescence staining of Dab2 protein (red) in PB samples of healthy donors, CML patients at diagnosis and CML patients during MR. Nuclei are stained in blue. (J) Quantification of Dab2 protein expression in an immunofluorescence assay. The statistical comparisons were conducted using a Student t test (*P<0.05, **P<0.01, ****P<0.0001). Bars indicate the standard error.

haematologica | 2019; 104(4)

New Drosophila model for chronic myeloid leukemia

Moreover, ena-RNAi weakly suppressed the BCR-ABL1 phenotype (Figure 7G), decreasing the phenotype penetrance. As a control, we did not observe any phenotype due to Dab or ena downregulation or Dab overexpression in prohemocytes (Figure 7F,G).

Discussion In order to identify candidate genes and pathways involved in the onset and progression of CML we developed and validated a CML genetic model based on trans-

haematologica | 2019; 104(4)

genic Drosophila expressing BCR-ABL1. In order to build and characterize a human functional model that could be sensitive to pharmacological inhibition and suitable for studying the effects of BCR-ABL1 mutations identified in patients with CML, we chose to express a completely human p210-BCR-ABL1 protein, in contrast what has been done previously.7 The expression of the oncoprotein in all eye cells committed to differentiation as photoreceptors or accessory cells (gmrGal4 driver) induces a strong phenotype characterized by altered differentiation of the ommatidia cells.44 The lack of phenotype in flies expressing a BCR-ABL1 kinase-dead mutant sup-

Figure 7. BCR-ABL1 expression in the hematopoietic precursor cells of the lymph gland impairs Drosophila blood cell homeostasis increasing the number of circulating blood cells. (A) A w1118 mid-L3 instar larva used as the wild-type control. (B) A mid-L3 larva conditionally expressing BCR-ABL1 in the hematopoietic precursors of the lymph gland medullary zone under the control of the domelessGal4 driver construct (dome:GFP/+;BCR-ABL1_3M,tub80TS/+). BCRABL1 expression was induced in stage L2 or early-L3 larvae by exposing the animals to 29Â°C during the indicated larval instars to disrupt the ability of the temperature-sensitive Gal80 mutant to inhibit Gal4 transactivation activity. The black arrows in (B) point to melanotic nodules. Anterior is on the left. (C) Penetrance of the melanotic nodule phenotype in mid-L3 control larvae expressing GFP under the control of the domelessGal4 driver (domeGFP), in larvae constitutively expressing a kinase-dead BCR-ABL1 mutant protein (dome:GFP/+;BCR-ABL1 KD/+) and in larvae in which BCR-ABL1 (dome:GFP/+;BCR- ABL1_3M,tub80T S/+) expression was induced starting from the L2 (L2) or from the early-L3 (eL3) instars. (D) Evaluation of the average number of hemocytes per field after bleeding of dome:GFP/+;BCRABL1_3M,tub80TS/+ and dome:GFP/+ larvae. BCR-ABL1 expression induces the appearance of melanotic nodules and this correlates with an increase of circulating hemocytes. (E) Penetrance of the melanotic nodule phenotype in mid-L3 control larvae (dome:GFP), in larvae expressing Abl-RNAi (dome/Abl-RNAi), in larvae in which BCR-ABL1 alone (dome:GFP/+;BCRABL1_3M,tub80TS/+) or together with Abl-RNAi (dome/Abl-RNAi;BCR-ABL1_3M,tub80TS/+) is expressed from the L2 instar. (F) Penetrance of the melanotic nodule phenotype in mid-L3 control larvae (dome:GFP), in larvae expressing DabRNAi (dome/+;Dab-RNAi/+), in larvae conditionally expressing the Dab protein (dome:GFP/+;tub80TS/+;UAS-Dab/+), and in larvae in which BCR-ABL1 alone (dome:GFP/+;BCRABL1_3M,tub80TS/+) or together with either Dab-RNAi (dome/+;BCRABL1_3M,tub80TS/+;Dab-RNAi/+) or UAS-Dab (dome/+;BCR-ABL1_3M,tub80TS/+;UAS-Dab/+) is expressed from the L2 instar. (G) Penetrance of the melanotic nodule phenotype in mid-L3 control larvae (dome:GFP), in larvae expressing ena-RNAi (dome/+;ena-RNAi/+), and in larvae in which BCR-ABL1 alone (dome:GFP/+;BCRABL1_3M,tub80TS/+) or together with ena-RNAi (dome/+;BCR-ABL1_3M,tub80TS/ena-RNAi) is expressed from the L2 instar. The average phenotype penetrance is calculated from three independent experiments, each involving 15-95 larvae. The statistical comparisons were conducted using a Student t test (*P<0.05, **P<0.01, ***P<0.001, ns=not significant). Bars indicate the standard error.

725

R. Bernardoni et al.

ports the role of kinase activity in the eye phenotype. Moreover, BCR-ABL1 expression and phosphorylation levels correlate with the severity of the phenotype. Consistently, BCR-ABL1 expression under the control of gmrGal4 induces a decrease of photoreceptors expressing Elav in eye imaginal discs and this correlates with the disruption of the adult eye. Interestingly, partial loss of dAbl function also slightly reduces the number of eye cells expressing Elav at L3, and to a much greater extent at later stages of development. This suggests that dAbl is implicated in the maintenance of neuronal commitment45,46 and confirms that loss or gain of function of dAbl/BCR-ABL1 can alter eye cell development.7 We have shown that human BCR-ABL1 interacts and interferes with the dAbl signaling pathway. Animals expressing BCR-ABL1 and heterozygous for the recessive Abl1 allele or coexpressing either Abl-RNAi or a kinase-dead dominant negative Abl (AblK417N) showed a weaker phenotype, suggesting that BCR-ABL1 and dAbl proteins most likely share binding sites and/or targets of the kinase activity. Consistently, co-expression of human BCRABL1 and dAbl synergizes and the phenotype becomes more severe. Notably, dAbl overexpression per se induces a weak â&#x20AC;&#x153;roughâ&#x20AC;? eye phenotype but the differentiation program is not severely disrupted. We cannot exclude that this is due to a level of dAbl expression below a critic threshold but it could also suggest that excessive dAbl might be still, at least partially, negatively regulated. This possible negative regulation seems to be overcome by BCR-ABL1 since all animals co-expressing dAbl and BCRABL1 showed a severe class -1 phenotype. Consistently, LOF or downregulation of genes known to interact genetically with dAbl LOF mutations interact in the same way with BCR-ABL1 expression. Namely, pros and fax alleles or deletions enhance the phenotype and this is consistent with their roles in neuronogenesis and neuronal differentiation. Moreover, ena LOF suppresses and Dab LOF enhances the dAbl LOF phenotype19,20 and we observed that both ena and Dab LOF and downregulation through RNAi also modify the BCR-ABL1 phenotype in the same way. Ena belongs to the ENA/VASP protein family involved in regulation of the actin cytoskeleton.47,48 dAbl regulates Ena by modulating its localization, most likely through its phosphorylation. It is known that both dAbl and the human/Drosophila BCR-ABL chimera phosphorylate Ena7 in vitro and we established that human BCR-ABL1 expression in the eye also increases Ena phosphorylation. This conservation of phosphorylation targets significantly increases the reliability of our model for identifying relevant BCR-ABL1 functional interactors. In this view the observation that decreased Ena function suppresses phenotypes due to both dAbl mutations24 and BCR-ABL1 expression suggests that both phenotypes can be due to Ena mislocalization and consequently actin cytoskeleton alterations can be suppressed if Ena expression decreases. In Drosophila, Abl and Dab are often co-expressed and the phenotype due to Dab mutations mimics the dAbl phenotype. Epistasis experiments have shown that Dab functions upstream of both dAbl and Ena, controlling their localization and thus the actin cytoskeleton, and Dab LOF does indeed enhance the phenotype due to dAbl mutations.30 Interestingly, Dab deletion or downregulation has the same effect on the BCR-ABL1 phenotype. These findings could be explained if Dab is able to regulate, at least partially, BCR-ABL1 localization. This interaction might 726

mitigate more severe BCR-ABL1-dependent effects when Dab is expressed at a physiological level but not if Dab is downregulated or its gene dosage is halved. Furthermore, our study showed that Dab human homologs are less expressed in both peripheral blood and bone marrow of CML patients at diagnosis compared to their expression in controls and are re-expressed in patients during molecular remission. Moreover, Dab1 expression in transfected K562 cells significantly decreases cell proliferation, confirming that Dab activity might alleviate the pathogenic effects of BCR-ABL1. We then assessed whether our model could help to fish-out homologs of leukemia-relevant genes in an ongoing dosage-sensitive genetic screen of the whole Drosophila genome. To this aim we considered STAT5, a transcription factor phosphorylated and activated by BCR-ABL1. Interestingly, LOF conditions of STAT92E, encoding the fly homolog of various human STAT, led to suppression of the BCR-ABL1 phenotype. In order to discover a tissue that could be a reliable second read-out for identifying BCR-ABL1 interactors relevant for hematopoiesis and leukemia, we moved to the larval hematopoietic organ, the lymph gland. We conditionally expressed human BCR-ABL1 in the lymph gland medullary zone where quiescent prohemocytes reside. Only BCR-ABL1 expression during L2 induces the appearance of melanotic nodules, which correlates with an increase of circulating hemocytes. This phenotype can be suppressed by dAbl downregulation, confirming that dAbl is expressed in the lymph gland medullary zone43 where it contributes to BCR-ABL1 pathway activation and to induction of the hematopoietic phenotype. It is worth noting that both Dab and ena interact functionally with BCR-ABL1 during hematopoiesis. In fact, while Dab downregulation enhances the melanotic nodule phenotype and Dab overexpression suppresses it, ena downregulation decreases the penetrance of this phenotype, confirming that ena and Dab are also expressed in the lymph gland medullary zone43 and modulate BCR-ABL1 activity. This phenotype is visible if BCR-ABL1 is expressed from the L2, when prohemocytes become quiescent, but not if it is expressed from the early L3, when the quiescent prohemocytes are still present in the medullary zone of the lymph gland. We are tempted to speculate that the dAbl pathway, activated by BCR-ABL1, could be involved in the mechanisms that regulate entry of prohemocytes into the quiescent state rather than maintenance of this state. This seems consistent with the observation that the lymph glands in mid-L3 larvae expressing BCR-ABL1 from L2 are very small compared to those in controls and do not show any clear partition (Giordani and Bernardoni, unpublished data). This suggests that, upon BCR-ABL1 expression, most of the prohemocytes could undertake the differentiation pathway and leave the lymph gland prematurely without becoming quiescent. We did not test all pathways interacting with BCR-ABL1, for example the Tyr-receptor/Ras pathway, which is known to compete with BCR-ABL1 for binding with the Grb2/Drk proteins1 and is likely involved in the eye phenotype since the Sevenless Tyr-receptor has an established role in eye differentiation.49,50 Nevertheless, we present here a new and efficient CML model based on Drosophila transgenic for human BCR-ABL1. This model could be a powerful tool for identifying new genes and pathways involved in the pathogenesis and progression of CML. haematologica | 2019; 104(4)

New Drosophila model for chronic myeloid leukemia

Acknowledgments We thank L. Giardino, V.A. Baldassarro, C. Mangano, and L. CalzĂ (Fondazione IRET, Ozzano dellâ&#x20AC;&#x2122;Emilia-Bologna, Italy) for assistance with the confocal microscopy analysis; D. Manzoni and M. Voltattorni for excellent technical help; M. Capovilla and the Trans-FlyER, Startup Company, Ferrara, Italy for generating the BCR-ABL1 kinase-dead transgenic lines. Drosophila lines were

References 1. Chereda B, Melo JV. Natural course and biology of CML. Ann Hematol. 2015;94 (Suppl 2):S107-121. 2. Quintas-Cardama A, Cortes J. Molecular biology of bcr-abl1-positive chronic myeloid leukemia. Blood. 2009;113(8):1619-1630. 3. Gonzalez C. Drosophila melanogaster: a model and a tool to investigate malignancy and identify new therapeutics. Nat Rev Cancer. 2013;13(3):172-183. 4. Bashaw GJ, Kidd T, Murray D, Pawson T, Goodman CS. Repulsive axon guidance: Abelson and Enabled play opposing roles downstream of the roundabout receptor. Cell. 2000;101(7):703-715. 5. Grevengoed EE, Loureiro JJ, Jesse TL, Peifer M. Abelson kinase regulates epithelial morphogenesis in Drosophila. J Cell Biol. 2001;155(7):1185-1198. 6. Liebl EC, Forsthoefel DJ, Franco LS, et al. Dosage-sensitive, reciprocal genetic interactions between the Abl tyrosine kinase and the putative GEF trio reveal trio's role in axon pathfinding. Neuron. 2000;26(1):107118. 7. Fogerty FJ, Juang JL, Petersen J, Clark MJ, Hoffmann FM, Mosher DF. Dominant effects of the bcr-abl oncogene on Drosophila morphogenesis. Oncogene. 1999;18(1):219-232. 8. Cilloni D, Messa F, Arruga F, et al. The NFkappaB pathway blockade by the IKK inhibitor PS1145 can overcome imatinib resistance. Leukemia. 2006;20(1):61-67. 9. McGuire SE, Mao Z, Davis RL. Spatiotemporal gene expression targeting with the TARGET and gene-switch systems in Drosophila. Sci STKE. 2004;2004 (220):pl6. 10. McGuire SE, Roman G, Davis RL. Gene expression systems in Drosophila: a synthesis of time and space. Trends Genet. 2004;20(8):384-391. 11. Giordani G, Barraco M, Giangrande A, et al. The human Smoothened inhibitor PF04449913 induces exit from quiescence and loss of multipotent Drosophila hematopoietic progenitor cells. Oncotarget. 2016;7(34):55313-55327. 12. Cilloni D, Carturan S, Bracco E, et al. Aberrant activation of ROS1 represents a new molecular defect in chronic myelomonocytic leukemia. Leuk Res. 2013;37(5):520-530. 13. Brand AH, Perrimon N. Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development. 1993;118(2):401-415. 14. Basler K, Yen D, Tomlinson A, Hafen E. Reprogramming cell fate in the developing Drosophila retina: transformation of R7 cells by ectopic expression of rough. Genes Dev. 1990;4(5):728-739.

haematologica | 2019; 104(4)

obtained from the Bloomington Drosophila Stock Center-BDSC (NIH P40OD018537) and primary antibodies from the Developmental Studies Hybridoma Bank (created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA, USA). We thank the Italian Association for Research on Cancer (AIRC) for funding D. Cilloni (IG10005) and G. Perini (IG11400, IG15182) and for supporting this work.

15. Freeman M. Reiterative use of the EGF receptor triggers differentiation of all cell types in the Drosophila eye. Cell. 1996;87(4):651-660. 16. Robinow S, White K. The locus elav of Drosophila melanogaster is expressed in neurons at all developmental stages. Dev Biol. 1988;126(2):294-303. 17. Gertler FB, Hill KK, Clark MJ, Hoffmann FM. Dosage-sensitive modifiers of Drosophila abl tyrosine kinase function: prospero, a regulator of axonal outgrowth, and disabled, a novel tyrosine kinase substrate. Genes Dev. 1993;7(3):441-453. 18. Hill KK, Bedian V, Juang JL, Hoffmann FM. Genetic interactions between the Drosophila Abelson (Abl) tyrosine kinase and failed axon connections (fax), a novel protein in axon bundles. Genetics. 1995;141(2):595-606. 19. Gertler FB, Bennett RL, Clark MJ, Hoffmann FM. Drosophila abl tyrosine kinase in embryonic CNS axons: a role in axonogenesis is revealed through dosage-sensitive interactions with disabled. Cell. 1989;58(1):103-113. 20. Gertler FB, Doctor JS, Hoffmann FM. Genetic suppression of mutations in the Drosophila abl proto-oncogene homolog. Science. 1990;248(4957):857-860. 21. Krause M, Bear JE, Loureiro JJ, Gertler FB. The Ena/VASP enigma. J Cell Sci. 2002;115(Pt 24):4721-4726. 22. Grevengoed EE, Fox DT, Gates J, Peifer M. Balancing different types of actin polymerization at distinct sites: roles for Abelson kinase and Enabled. J Cell Biol. 2003;163(6):1267-1279. 23. Comer AR, Ahern-Djamali SM, Juang JL, Jackson PD, Hoffmann FM. Phosphorylation of Enabled by the Drosophila Abelson tyrosine kinase regulates the in vivo function and protein-protein interactions of Enabled. Mol Cell Biol. 1998;18(1):152-160. 24. Gertler FB, Comer AR, Juang JL, et al. enabled, a dosage-sensitive suppressor of mutations in the Drosophila Abl tyrosine kinase, encodes an Abl substrate with SH3 domain-binding properties. Genes Dev. 1995;9(5):521-533. 25. Nieborowska-Skorska M, Wasik MA, Slupianek A, et al. Signal transducer and activator of transcription (STAT)5 activation by BCR/ABL is dependent on intact Src homology (SH)3 and SH2 domains of BCR/ABL and is required for leukemogenesis. J Exp Med. 1999;189(8):1229-1242. 26. de Groot RP, Raaijmakers JA, Lammers JW, Jove R, Koenderman L. STAT5 activation by BCR-Abl contributes to transformation of K562 leukemia cells. Blood. 1999;94(3):11081112. 27. Bina S, Wright VM, Fisher KH, Milo M, Zeidler MP. Transcriptional targets of Drosophila JAK/STAT pathway signalling as

28. 29.

30.

31.

32.

33.

34.

35.

36. 37.

38.

39.

40.

41. 42.

43.

effectors of haematopoietic tumour formation. EMBO Rep. 2010;11(3):201-207. Zeidler MP, Bach EA, Perrimon N. The roles of the Drosophila JAK/STAT pathway. Oncogene. 2000;19(21):2598-2606. Le N, Simon MA. Disabled is a putative adaptor protein that functions during signaling by the sevenless receptor tyrosine kinase. Mol Cell Biol. 1998;18(8):4844-4854. Song JK, Kannan R, Merdes G, Singh J, Mlodzik M, Giniger E. Disabled is a bona fide component of the Abl signaling network. Development. 2010;137(21):37193727. McAvoy S, Zhu Y, Perez DS, James CD, Smith DI. Disabled-1 is a large common fragile site gene, inactivated in multiple cancers. Genes Chromosomes Cancer. 2008;47(2):165-174. Finkielstein CV, Capelluto DG. Disabled-2: A modular scaffold protein with multifaceted functions in signaling. Bioessays. 2016;38 (Suppl 1):S45-55. Zhang Z, Chen Y, Tang J, Xie X. Frequent loss expression of dab2 and promotor hypermethylation in human cancers: a metaanalysis and systematic review. Pak J Med Sci. 2014;30(2):432-437. Holz A, Bossinger B, Strasser T, Janning W, Klapper R. The two origins of hemocytes in Drosophila. Development. 2003;130(20): 4955-4962. Evans CJ, Hartenstein V, Banerjee U. Thicker than blood: conserved mechanisms in Drosophila and vertebrate hematopoiesis. Dev Cell. 2003;5(5):673-690. Lanot R, Zachary D, Holder F, Meister M. Postembryonic hematopoiesis in Drosophila. Dev Biol. 2001;230(2):243-257. Jung SH, Evans CJ, Uemura C, Banerjee U. The Drosophila lymph gland as a developmental model of hematopoiesis. Development. 2005;132(11):2521-2533. Krzemien J, Dubois L, Makki R, Meister M, Vincent A, Crozatier M. Control of blood cell homeostasis in Drosophila larvae by the posterior signalling centre. Nature. 2007;446 (7133):325-328. Lebestky T, Jung SH, Banerjee U. A Serrateexpressing signaling center controls Drosophila hematopoiesis. Genes Dev. 2003;17(3):348-353. Luo H, Hanratty WP, Dearolf CR. An amino acid substitution in the Drosophila hopTuml Jak kinase causes leukemia-like hematopoietic defects. EMBO J. 1995;14(7):1412-1420. Minakhina S, Steward R. Melanotic mutants in Drosophila: pathways and phenotypes. Genetics. 2006;174(1):253-263. Bourbon HM, Gonzy-Treboul G, Peronnet F, et al. A P-insertion screen identifying novel X-linked essential genes in Drosophila. Mech Dev. 2002;110(1-2):71-83. Irving P, Ubeda JM, Doucet D, et al. New insights into Drosophila larval haemocyte

727

R. Bernardoni et al. functions through genome-wide analysis. Cell Microbiol. 2005;7(3):335-350. 44. Ellis MC, O'Neill EM, Rubin GM. Expression of Drosophila glass protein and evidence for negative regulation of its activity in non-neuronal cells by another DNAbinding protein. Development. 1993;119(3): 855-865. 45. Bennett RL, Hoffmann FM. Increased levels of the Drosophila Abelson tyrosine kinase in nerves and muscles: subcellular localization and mutant phenotypes imply a role in cellcell interactions. Development. 1992;116(4):

728

953-966. 46. Xiong W, Morillo SA, Rebay I. The Abelson tyrosine kinase regulates Notch endocytosis and signaling to maintain neuronal cell fate in Drosophila photoreceptors. Development. 2013;140(1):176-184. 47. Barzik M, Kotova TI, Higgs HN, et al. Ena/VASP proteins enhance actin polymerization in the presence of barbed end capping proteins. J Biol Chem. 2005;280(31): 28653-28662. 48. Bear JE, Svitkina TM, Krause M, et al. Antagonism between Ena/VASP proteins

and actin filament capping regulates fibroblast motility. Cell. 2002;109(4):509-521. 49. Olivier JP, Raabe T, Henkemeyer M, et al. A Drosophila SH2-SH3 adaptor protein implicated in coupling the sevenless tyrosine kinase to an activator of Ras guanine nucleotide exchange, Sos. Cell. 1993;73 (1):179-191. 50. Simon MA, Dodson GS, Rubin GM. An SH3-SH2-SH3 protein is required for p21Ras1 activation and binds to sevenless and Sos proteins in vitro. Cell. 1993;73(1): 169-177.

haematologica | 2019; 104(4)

ARTICLE

Myeloid Neoplasms

Blastic plasmacytoid dendritic cell neoplasm: genomics mark epigenetic dysregulation as a primary therapeutic target Maria Rosaria Sapienza,1* Francesco Abate,2,3* Federica Melle,4 Stefania Orecchioni,5 Fabio Fuligni,6 Maryam Etebari,1 Valentina Tabanelli,4 Maria Antonella Laginestra,1 Alessandro Pileri,7,8 Giovanna Motta,4 Maura Rossi,1 Claudio Agostinelli,1 Elena Sabattini,1 Nicola Pimpinelli,8 Mauro Truni,9 Brunangelo Falini,10 Lorenzo Cerroni,11 Giovanna Talarico,5 Rossana Piccioni,12 Stefano Amente,13 Valentina Indio,14 Giuseppe Tarantino,14 Francesco Brundu,2 Marco Paulli,15 Emilio Berti,16 Fabio Facchetti,17 Gaetano Ivan Dellino,12,18 Francesco Bertolini,5 Claudio Tripodo,19* Raul Rabadan2,3* and Stefano A. Pileri4ǂ*

Hematopathology Unit, Department of Experimental, Diagnostic, and Specialty Medicine, S. Orsola-Malpighi Hospital, University of Bologna, Italy; 2Department of Systems Biology, Columbia University College of Physicians and Surgeons, New York, NY, USA; 3Department of Biomedical Informatics, Columbia University College of Physicians and Surgeons, New York, NY, USA; 4Division of Haematopathology, European Institute of Oncology, Milan, Italy; 5Laboratory of Hematology-Oncology, European Institute of Oncology, Milan, Italy; 6Department of Genetics and Genome Biology, The Hospital for Sick Children, Toronto, ON, Canada: 7Dermatology Unit, Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna, Italy; 8Division of Dermatology, Department of Surgery and Translational Medicine, University of Florence, Italy; 9Pathological Anatomy Histology & Cytogenetics, Niguarda Cancer Center, Niguarda-Ca' Granda Hospital, Milan, Italy; 10Institute of Hematology and Center for Hemato-Oncology Research (CREO), University and Hospital of Perugia, Italy; 11 Universitätsklinik für Dermatologie und Venerologie, LKH-Universitatsklinikum Graz, Austria; 12Department of Experimental Oncology, European Institute of Oncology, Milan, Italy; 13Department of Molecular Medicine and Medical Biotechnologies, University of Naples ‘Federico II’, Italy; 14"Giorgio Prodi" Cancer Research Center, University of Bologna, Italy; 15Unit of Anatomic Pathology, Department of Molecular Medicine, University of Pavia and Fondazione IRCCS San Matteo Policlinic, Pavia, Italy; 16 Department of Dermatology, Fondazione IRCCS Ca' Granda - Ospedale Maggiore Policlinic and Milan University, Milan, Italy; 17Pathology Section, Department of Molecular and Translational Medicine, University of Brescia, Italy; 18Department of Oncology and Hemato-Oncology, University of Milan, Italy and 19Tumor Immunology Unit, Department of Health Science, Human Pathology Section, University of Palermo School of Medicine, Italy

Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):729-737

Correspondence: MARIA ROSARIA SAPIENZA mariarosaria.sapienza@gmail.com

*MRS, FA, CT, RR and SAP contributed equally to this work. ǂAlma Mater Professor, Bologna University

ABSTRACT

lastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare and aggressive hematologic malignancy for which there is still no effective therapy. In order to identify genetic alterations useful for a new treatment design, we used whole-exome sequencing to analyze 14 BPDCN patients and the patient-derived CAL-1 cell line. The functional enrichment analysis of mutational data reported the epigenetic regulatory program to be the most significantly undermined (P<0.0001). In particular, twenty-five epigenetic modifiers were found mutated (e.g. ASXL1, TET2, SUZ12, ARID1A, PHF2, CHD8); ASXL1 was the most frequently affected (28.6% of cases). To evaluate the impact of the identified epigenetic mutations at the gene-expression and Histone H3 lysine 27 trimethylation/acetylation levels, we performed additional RNA and pathology tissue-chromatin immunoprecipitation sequencing experiments. The patients displayed enrichment in gene signatures regulated by methylation and modifiable by decitabine administration, shared common H3K27-acetylated regions, and had a set of cell-cycle genes aberrantly up-regulated and marked by promoter acetylation. Collectively, the integration of sequencing data showed the potential of a therapy based on epigenetic agents. Through the adoption of a preclinical BPDCN mouse model, established by CAL-1 cell line xenografting, we demonstrated the efficacy of the combination of the epigenetic drugs 5’-azacytidine and decitabine in controlling disease progression in vivo. haematologica | 2019; 104(4)

Received: July 16, 2018. Accepted: October 30, 2018. Pre-published: October 31, 2018. doi:10.3324/haematol.2018.202093 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/729 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

729

M.R. Sapienza et al.

Introduction Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare malignancy derived from precursors of plasmacytoid dendritic cells.1-4 It has no clear racial or ethnic predisposition and more often affects elderly males (male/female ratio 3.3:1; mean/median age at diagnosis: 61-67 years). BPDCN patients usually respond to firstline chemotherapy, but despite this they almost invariably relapse and display a dismal prognosis with a median overall survival (OS) ranging from 10 to 19 months.2 No standardized therapeutic approach has so far been established for BPDCN, even if hematopoietic stem cell transplantation has been shown to achieve remission in selected patients.5-6 Therefore, the development of effective treatments still represents an unmet need.7 The pathobiology of BPDCN is poorly understood and the number of reports exploring its molecular features is still limited.8-21 Recent advances in the understanding of the BPDCN molecular landscape have paved the way for novel treatment approaches based on the inhibition of the BCL2 protein,22 the activation of the cholesterol efflux,23 the repression of the Bromodomain-containing protein 4 (BRD4),24 and binding to the interleukin-3 receptor (IL3R).25 All these potential therapeutic options (which are worthy of further evaluation) have mainly emerged from the analysis of the BPDCN transcriptome or from its antigenic repertoire. The genomic landscape of BPDCN has not been well investigated, and only a few studies have explored the mutational events occurring in BPDCN, mainly through targeted sequencing approaches.14,16,19,20 Unfortunately, these have not offered any novel prospects of treatment options. In this study, we performed whole-exome sequencing (WES) of 14 BPDCN samples and of the BPDCN-derived CAL-1 cell line to look for specific BPDCN genetic vulnerabilities that may support the design of new therapeutic strategies. The WES mutational findings were complemented by copy number variant (CNV) analysis, RNA and pathology tissue-chromatin immunoprecipitation (PATChIP) sequencing results. The integration of data allowed us to identify a successful combinatorial therapy based on epigenetic drugs able to control disease progression in a rapidly progressive BPDCN xenograft model.

Whole-exome sequencing analysis We performed paired-end sequencing of matched tumor/normal DNA samples (9 cases), tumor only DNA samples (5 cases), and the CAL-1 cell line (Online Supplementary Table S3) using the TruSeq Exome Kit and Nextera Rapid Capture Exome kit (Illumina). Further details are available in the Online Supplementary Appendix.

Sanger sequencing We used Sanger sequencing to analyze two candidate nonsense somatic mutations of SUZ12 and ASXL1 occurring in 2 patients, respectively, as described in the Online Supplementary Appendix.

Targeted sequencing We performed MiSeq targeted sequencing (Illumina) of the 14 BPDCN tumor patients, 7 normal matched saliva samples and the CAL-1 cell line, already analyzed by WES. More bioinformatics details are provided in the Online Supplementary Appendix and Online Supplementary Tables S4 and S5.

RNA sequencing analysis Five BPDCN cases studied by WES and targeted sequencing had sufficient material for RNA extraction and sequencing; these samples represented the RNA sequencing (RNAseq) discovery set. We also collected an additional 4 BPDCN cryopreserved cutaneous biopsies, sufficient only for RNA sequencing experiments, used as an RNA-seq extension set. RNA of 4 normal plasmacytoid dendritic cell (pDCs) samples was purchased from AllCells (Alameda, CA, US) and used for comparison. For details, see Online Supplementary Table S6 and the Online Supplementary Appendix.

Pathology tissue-chromatin immunoprecipitation sequencing The BPDCN_25 and BPDCN_37 patients were provided with one skin biopsy: half was cryopreserved and used for WES, targeted and RNA sequencing analysis, and the other half was fixed in formalin, included in paraffin and used for pathology tissue-chromatin immunoprecipitation (PATChIP) sequencing analysis. PAT-ChIP experiments were performed as in Fanelli et al.26 Further details are available in the Online Supplementary Appendix.

CAL-1 cell line Methods Blastic plasmacytoid dendritic cell neoplasm samples We collected 14 BPDCN cryopreserved cutaneous biopsies at diagnosis, 9 matched saliva samples and the BPDCN patient-derived cell line, CAL-1. The pathological cases were evaluated as previously described17 and diagnosed by experienced hematopathologists (CA, EB, FF, LC, MP, ES, CT, MT, and SAP) according to World Health Organization diagnostic criteria.2 Informed consent was obtained from each patient in accordance with the Ethical Review Board of the Department of Experimental, Diagnostic, and Specialty Medicine of the University of Bologna, Italy, and the Declaration of Helsinki. DNA was extracted as reported in the Online Supplementary Appendix. The main clinical, immunohistochemical and cytogenetic features of the BPDCN patients are shown in Online Supplementary Tables S1 and S2. 730

CAL-1, a BPDCN cell line27 was cultured as reported previously.18 The CAL-1 gene expression profile of a previous study was used17 (http://www.ncbi.nlm.nih.gov/geo/query/ acc.cgi?acc=GSE62014).

Mouse model and in vivo treatments Experiments were carried out on 6-8-week old nonobese diabetic severe combined immunodeficient NOD/SCID interleukin-2 receptor g (IL-2Rg)â&#x20AC;&#x201C;null (NSG) mice, as previously reported.13 All animal experiments were carried out in accordance with the Italian laws in force (Legislative Decree 26/14 and subsequent amendments) and institutional guidelines. All in vivo studies were ratified by the Italian Ministry of Health. For induction of BPDCN in mice, 5000 CAL-1 cells were injected intravenously (i.v.) through the lateral tail vein in non-irradiated mice. Engrafted mice were then treated with bortezomib, 5â&#x20AC;&#x2122;-azacytidine, decitabine and romidepsin, as detailed in the Online Supplementary Appendix. haematologica | 2019; 104(4)

Epigenetic matrix and targeted therapy of BPDCN

Results Whole-exome sequencing reveals the epigenetic program dysregulation as the main theme of the blastic plasmacytoid dendritic cell neoplasm mutational landscape We collected 14 BPDCN cases with a mean age of 56 years at diagnosis (range 9-89 years), a male-to-female

ratio of 10:4, and the classical BPDCN presentation (Online Supplementary Tables S1 and S2).1 The enrolled patients underwent different treatment regimens and 78.5% (11 out of 14) died of the disease 6.3-76 months after the diagnosis or were lost at follow up. Most patients who underwent autologous and/or allogeneic hematopoietic stem cell transplantation experienced a prolonged survival. We performed WES on 14 BPDCN cases, and on the

Figure 1. The genomic characterization of blastic plasmacytoid dendritic cell neoplasm (BPDCN). (A) Circos plot graphical representation of the functional analysis performed on 54 genes recurrently mutated and/or affected by nonsense and frameshift single nucleotide variants (SNVs) in BPDCN whole-exome sequencing (WES) samples. The four biological processes most significantly enriched are reported in the counterclockwise order from the highest to the lowest P-value: the gamma-aminobutyric acid (GABA) secretion (in violet), the Rac signaling (in red), the hematopoietic stem cell homeostasis (in light blue) and the epigenetic process (in green). The genes are colored according to their belonging to one or more of the biological processes represented. Genes not involved are in gray. (B) Overview of the TET2 and ASXL1 mutations identified. Structure of ASXL1 protein with C-terminal plant homeodomain catalytic region and structure of TET2 protein showing the catalytic core region: the cysteinerich (Cys) and double-stranded Î˛-helix (DSBH) domains. Empty circles: somatic SNVs. + : frameshift SNV; *nonsense SNV. (C) Heatmap representation of SNVs in BPDCN WES samples and its distribution among selected pathways commonly mutated in myeloid disorders. The SNVs, the affected genes and the related pathways are reported in rows, while, the BPDCN samples are in columns.

haematologica | 2019; 104(4)

731

M.R. Sapienza et al. A

Figure 2. The transcriptome and H3K27 trymethylation/acetylation profiling of blastic plasmacytoid dendritic cell neoplasm (BPDCN). (A) Unsupervised hierarchical clustering performed on 5 BPDCN samples and 4 plasmacytoid dendritic cell (pDCs) samples according to the expression level of the RNA sequencing data. In the heat-map each row represents a gene and each column a sample. The color scale exemplifies the relative expression level of a gene across all samples: (red) represented genes with an expression level above the mean; (blue) the genes with an expression level lower than the mean. Tumors (BPDCNs) and controls (pDCs) cluster in two distinct groups. (B) Gene Set Enrichment Analysis (GSEA) plot illustrating the enrichment of the KDM5B and PRMT5 gene signatures in BPDCN patients reported in literature34-36 as well as the enrichment of a set of genes, described by Missiaglia et al.37 as responsive to hypomethylating treatment, namely decitabine. Normalized enrichment score (NES) â&#x2030;Ľ 2; false discovery rate (FDR) q-value false discovery rate â&#x2030;¤0.0001. (C) Visualization of anti-H3K27ac and anti-H3K27me3 normalized pathology tissue-chromatin immunoprecipitation (PAT-ChIP) sequencing profiles in the UCSC Genome Browser showing genomic regions from patient BPDCN_25 and BPDCN_37. (Red boxes) Exemplificative regions displaying a similar level of anti-H3K27ac in both patients. (Black solid rectangles) Genes in correspondence of the anti-H3K27ac peaks. (D) The cases BPDCN_25 and BPDCN_37 share common H3k27ac regions. (E) Histogram representation of the top 10 significant biological processes emerged by Gene Ontology (GO) analysis of 86 up-regulated genes marked by H3K27ac in their promoters. GO categories are shown in x-axis and the fold enrichment values of observed versus expected genes are reported in the y-axis (FDR q-value <0.001).

732

haematologica | 2019; 104(4)

Epigenetic matrix and targeted therapy of BPDCN

Figure 3. The efficacy of epigenetic agents in a preclinical blastic plasmacytoid dendritic cell neoplasm (BPDCN) mouse model. (A) Pharmacodynamic assessment of the percentage of human CD56+CD38+ cells in the peripheral blood (PB), bone marrow (BM) of the femur and spine, spleen, and liver of a representative vehicletreated BPDCN mouse model, 39 days after CAL-1 injection. The cytofluorometric assays shows the tumor dissemination in all the tissues analyzed. (B) Hematoxylin & eosin (H&E) staining of BM and spleen samples collected in a representative vehicle-treated NSG mouse 39 days after CAL-1 injection (H&E; x400; Olympus DP2SAL). The histological assay shows a marked dissemination of blast elements. The immunohistochemistry detection of the CD303 (BDCA-2) antigen, in the murine BM, indicates the presence of specific BPDCN blasts cells (Immmuno-alkaline phosphatase; Gill’s hematoxylin nuclear counterstaining; x400; Olympus DP2-SAL). These results further confirmed the effective engraftment of CAL-1 cell line. (C) Graphical representation of the treatment schedules observed in a BPDCN mouse model. Each treatment is represented by a single color or by a combination of colors and was administered for four weeks as follows: 5’-azacytidine 5 mg/kg 5 doses at 2-day intervals (green), decitabine 2.5 mg/kg 3 doses at 2-day intervals (light brown), romidepsin 0.5 mg/kg every day (violet), bortezomib 0.5 mg/kg two times weekly (fuchsia). The same doses were also administered in various combinations. (D) Kaplan-Meier curves comparing overall survival of BPDCN mice models respectively treated with vehicle or the above reported treatments. Each treatment is summarized by a box colored as described above. *Indicates that the combination of decitabine and 5’-azacytidine was the most effective in prolonging mice survival. Curves were compared by log-rank test, n=5 mice/treatment arm. (E) Pharmacodynamic assessment of spleen size in 4 representative NSG mice CAL-1 injected after 39 days of treatment with vehicle (mouse Control), Decitabine (mouse Deci), 5’-azacytidine (mouse Aza), and 5’-azacytidine plus decitabine (mouse Deci+Aza) according to the dosing schedule reported above. ns: not significant.

haematologica | 2019; 104(4)

733

M.R. Sapienza et al.

BPDCN patient-derived CAL-1 cell line and identified 1302 non-synonymous single nucleotide variants (SNVs) represented predominantly by missense SNVs (n=1251), nonsense SNVs (n=47), and frameshift insertions/deletions (n=7). To verify the robustness of the WES analysis, 2 randomly-selected variants occurring in the ASXL1 and SUZ12 genes, respectively, were validated by Sanger sequencing (Online Supplementary Figure S1). To extend the validation to a higher number of samples and mutations, a targeted sequencing approach was adopted: 21 SNVs were interrogated by MiSeq Illumina technology in the same BPDCN cases analyzed by WES and a concordance of 95.2% (20 out of 21 SNVs) was achieved, underlining a high consistency of data (Online Supplementary Table S7). The 1302 non-synonymous SNVs detected by WES affected 1166 genes, all but 7 known to be related with pathological conditions and reported as mutated in the Catalogue of Somatic Mutations in Cancer (COSMIC_v66). To identify the biological processes that were most altered by the mutational events, we performed a functional enrichment analysis of the 9 genes that were recurrently mutated (≥ 3 samples) and the 45 genes impacted by deleterious (nonsense or frameshift) SNVs. Among the top 10 significantly enriched biological processes, the epigenetic program was the most represented (P=0.0001), followed by hematopoietic stem cell homeostasis, Rac signaling and gamma-aminobutyric acid (GABA) secretion (Figure 1A and Online Supplementary Table S8). The ASXL1 gene proved to be the most frequently mutated (28%, 4 out of 14 samples), followed by TET2 (21%, 3 out of 14 samples); both genes displayed mainly nonsense or frameshift SNVs located within or upstream of the catalytic domain of the proteins, potentially leading to their functional disruption (Figure 1B and Online Supplementary Table S9). We consulted the Gene Ontology database28 to identify among the 1166 BPDCN mutated genes those implicated in the epigenetic regulation. We found 25 mutated epigenetic modifier genes controlling chromatin accessibility (ARID1a, CHD8, SMARCA1), DNA methylation (TET2, IDH2), or histone post-transcriptional modifications [methylation (ASXL1, SUZ12, MLL family), demethylation (KDM4D), acetylation (EP300, EP400), ubiquitination (PHC1, PHC2), dephosphorylation (EYA2), and exchange (SRCAP)]. Of note, 12 out of 14 BPDCN samples (86%) harbored at least one of the 25 epigenetic regulator genes mutated, and specifically 8 out of 14 patients (57.14%) presented a deleterious lesion (nonsense/frameshift SNV) (Online Supplementary Table S10). Many SNVs clustered in the histone methylation pathway, specifically in genes belonging to the Polycomb-repressive complex 2 (ASXL1, ASXL3, SUZ12) and in histone methyltransferases (ASHL1, SETMAR, MLL), possibly compromising the integrity of the methylation program. Besides genetic lesions targeting epigenetic regulators, we also detected mutations potentially affecting molecular programs commonly deregulated in myeloid malignancies, such as RAS signaling29 (hot-spot SNVs on KRAS or NRAS, alternatively), DNA repair/cycle progression30 (SNVs on BRCA1, ATM, ATR, and RAD52), Wnt-signaling31 (SNVs on WNT3, WNT7B, WNT10 and BCL9L), cell growth32 (SNVs on RUNX2, MAPK1), and splicing machinery33 (an SNV on ZRSR2) (Figure 1C). Whole-exome sequencing data were also used for 734

cytogenetic CNV analysis, which highlighted extensive losses along the chromosome 9 and the associated deletion of the tumor suppressor CDKN2A gene in 8 out of 14 BPDCN samples (57%) (Online Supplementary Figure S2), as already reported in the literature.12,15,20 In addition, CNV analysis showed that deletions affected six of the nine genes recurrently mutated; deletions were always mutually exclusive with mutations (Online Supplementary Figure S3). However, no significant correlation was found between genetic lesions and the clinical data.

Blastic plasmacytoid dendritic cell neoplasm transcriptome profiling confirms the dysregulation of epigenetic programs Genetic lesions in key epigenetic modifier genes and in related regulatory networks can induce profound perturbations in the transcriptional homeostasis of the cell. To further substantiate the impact of mutations affecting the chromatin remodeling pathway in BPDCN, we performed RNA sequencing of 5 BPDCNs, considered as the discovery set, already studied by WES and MiSeq targeted sequencing. We compared the patients' transcriptomes with those of 4 normal plasmacytoid dendritic cell (pDC) samples isolated from the peripheral blood of healthy individuals and used as controls. BPDCN tumor samples and pDCs segregated separately according to their gene expression profiles (Figure 2A). Two thousand and thirty-four genes (2034) were significantly deregulated among patients, and approximately half of them were up-regulated (46%) in the BPDCN setting. Gene set enrichment analysis (GSEA) reported the significant deregulation of two genetic signatures involved in the methylation process, driven by the KDM5B34 histone demethylase and PRMT535 methyltransferase-associated genes, respectively. Of note, GSEA also detected the significant enrichment of a set of genes associated with the response to a DNA demethylating agent,36 namely decitabine (Figure 2B). The GSEA results [normalized enrichment score (NES) ≥2; false discovery rate (FDR) qvalue ≤0.0001] were then validated in an extension set of 4 BPDCN samples and in a CAL-1 cell line (Online Supplementary Figures S4 and S5).

Genome-wide ChIP-sequencing substantiates epigenetic dysregulation of cell cycle program in blastic plasmacytoid dendritic cell neoplasms To investigate if the transcriptional deregulation of BPDCNs could be linked to specific epigenetic features, we analyzed the histone acetylation/methylation profiles of 2 selected BPDCN patients (BPDCN_25 and BPDCN_37). The trimethylation at lysine 27 of histone 3 (H3K27me3) is closely associated with inactive gene promoters, while its acetylation (H3K27ac) closely correlates with gene activation, the two epigenetic modifications being mutually exclusive. Given this, we analyzed the genome-wide distribution of trimethylation and acetylation profiles of H3K27 in BPDCN cases. The analysis of PAT-ChIP sequencing data demonstrated that the 2 patients converged on the same pattern of histone acetylation, sharing as much as 43.6% of the acetylated promoters (Figure 2C and D). PAT-ChIP sequencing results were then integrated into the RNA sequencing data leading to the identification of a signature of 86 genes marked by promoter acetylation and significantly overexpressed in the BPDCN RNA sequencing sets. Gene haematologica | 2019; 104(4)

Epigenetic matrix and targeted therapy of BPDCN

Ontology analysis of the 86 selected genes highlighted the enrichment in biological processes involved in cell cycle progression (FDR q-value <0.001) (Figure 2E and Online Supplementary Table S11).

In vivo blastic plasmacytoid dendritic cell neoplasm modeling demonstrates combined epigenetic therapy as effective in controlling disease progression The integration of results obtained from WES, RNA sequencing and PAT-ChIP-sequencing experiments clearly pointed to a fundamental role for epigenetic dysregulation in BPDCN and allowed us to hypothesize that this malignancy could display susceptibility to drugs active on the epigenetic regulation. Following the demonstration that the CAL-1 cell line, like primary BPDCN samples, had mutations clustering in chromatin remodeling pathway (Figure 1C) and enrichment in the same epigenetic programs (Online Supplementary Figure S5), we developed an in vivo CAL-1 xenograft BPDCN-like model to explore the effects of treatments targeting the acetylation, methylation, and also the NF-kB pathways, according to previous results.17,18 To this end, we focused on four different FDA-approved compounds: 5’-azacytidine, decitabine, romidepsin and bortezomib. NSG mice intravenously injected with 5x103 CAL-1 cells rapidly developed a systemic BPDCN-like progressive disease, which was defined by the flow cytometry identification of human CD56+CD38+ malignant cells in the peripheral blood, bone marrow, spleen and liver, as evaluated at 39 days after injection (Figure 3A). The pathological infiltration by malignant BPDCN cells in the mouse model was also confirmed at the same time point by the histopathological analysis of the bone marrow and spleen samples, which showed the presence of atypical cells with blastic morphology and expressing the human CD303/BDCA2 pDC marker (Figure 3B). Xenografted mice were divided into 11 treatment groups (n=110 mice) one day after CAL-1 injection and treated with either saline or with the hypo-methylating agents 5’-azacitidine or decitabine, the proteasome inhibitor bortezomib, and the histone deacetylase inhibitor romidepsin, used as single agents or in combination, according to the treatment schedule summarized in Figure 3C. The administration of 5’-azacytidine and decitabine used as single agents significantly prolonged OS of the mice when compared with saline (median survival 43.6 days vs. 32 days, P<0.01 for 5’-azacytidine; median survival 44.7 days vs. 32 days, P<0.05 for decitabine) while neither bortezomib nor romidepsin alone showed beneficial effects on disease outcome. When the same agents were associated in combined treatment experiments, three different combinations were seen to significantly prolong mouse survival: i) the association of romidepsin and decitabine (median survival 42.8 days vs. 32 days, P<0.05); ii) the combination of romidepsin, decitabine, and 5’-azacytidine (median survival 41.8 days vs. 32 days, P<0.01); and iii) the association of decitabine and 5’-azacytidine (median survival 52.8 days vs. 32 days, P<0.01), which achieved the best result in terms of survival (Figure 3D). Consistently, 5’azacytdine and decitabine administered alone reduced the CAL-1-induced splenomegaly as evaluated at day 39 post injection and their combination proved to be even more effective (Figure 3E). haematologica | 2019; 104(4)

Discussion Blastic plasmacytoid dendritic cell neoplasm is a rare myeloid malignancy with dismal prognosis and no standard therapy. In the present study, we performed WES on the largest series of BPDCNs that, to the best of our knowledge, has so far been reported in the literature. Thanks to the integration of WES with RNA and PATChIP sequencing, we provide new insights into BPDCN pathobiology by highlighting the dysregulation of the epigenetic program as a hallmark of the disease and suggest possible novel therapeutic interventions. We found BPDCN patients extensively affected by mutations of genes involved in the epigenetic regulation: 25 epigenetic modifiers were mutated in almost all BPDCN patients (13 out of 14) and the CAL-1 cell line. In more than half of the patients (8 out of 14), the mutations heralded damaging functional alterations (Figure 1C). Some of the mutated genes have already been reported in previous studies (e.g. ASXL1, RAS, ATM, ARID1A, and IDH2), although, at times, at different rates than in our series (see ASXL1 and TET2, which were found to be mutated in 28.6% and 21.4% of our samples vs. 32% and 36% of those of Menezes et al.19). In this respect, it should be remembered that the aim of our study was not only to extensively explore the mutational landscape of BPDCN, but also to possibly translate molecular notions into a preclinical approach. In any case, thanks to the employment of a WES approach, which did not limit our investigation to a priori-selected genes, we recognized additional mutated epigenetic factors that have never been described before but which are potentially relevant in the context of BPDCN, like PHF2 histone demethylase, that enhances the TP53-tumor suppressor activity,37 and the CHD8 Chromodomain helicase DNA-binding protein-8, that promotes the E2F-dependent transcription and cell cycle progression.38 Besides the epigenetic pathway, we also detected mutations affecting programs common to other myeloid malignancies, such as DNA repair process,30 Wnt/β-catenin signaling,31 and the differentiation pathway.32 Importantly, the functional enrichment analysis of WES data showed that among all genes/pathways explored the epigenetic program was the most deregulated (P<0.0001). To evaluate the impact of the identified epigenetic mutations at gene expression level, we analyzed the transcriptome of samples studied by WES. Among up-regulated genes, GSEA revealed the significant enrichment of two methylation pathways, driven respectively by the KDM5B histone-demethylase34 and by the PRMT5 arginine methyltransferase-5;35 these two epigenetic modifiers are reported to be over-expressed in several cancer types and also represent promising therapeutic targets.39 Blockade of the PRMT5 activity reduces cell survival in chronic myelogenous leukemia40 and inhibition of KDM5B demethylation correlates with cell growth arrest in hepatocellular carcinoma and breast cancers.41,42 We also identified the overexpression of one gene signature36 specifically responsive to the administration of the hypomethylating agent decitabine; a molecular finding bearing important therapeutic implications (FDR q=1.85E5). To gain a functional insight into the epigenetic landscape of BPDCN samples, we performed PAT-ChIP sequencing of H3K27-acetylation/trymethylation signals of 2 BPDCN patients. The trimethylation of H3K27 marks 735

M.R. Sapienza et al.

inactive gene promoters and enhancers, while its acetylation correlates with gene activation.43 PAT-ChIP sequencing data showed that the 2 patients converged on the same epigenetic pattern sharing approximately half of the identified H3K27-acetylated promoters. Interestingly, the common acetylated regions comprised 10 superenhancers (SE) bound by the Bromodomain-containing protein 4 (BRD4), as described by Ceribelli et al. in a recent work on BPDCN (data not shown).24 The integration of PAT-ChIP and the RNA sequencing data highlighted a set of 86 genes involved in the cell-cycle progression aberrantly over-expressed and marked by H3K27-promoter acetylation. This finding suggests that the cell-cycle deregulation could be driven by H3K27acetylation signals, a hypothesis meriting future ad hoc studies that could help to clarify the mechanism of proliferation of this largely obscure disease. The rarity of the disease (with an incidence of 0.000045%) and its extremely aggressive behavior (OS 1019 months) limits the number of available patients included in biological and/or clinical studies. For these reasons, not surprisingly, BPDCN is still an orphan tumor lacking a standardized and effective therapeutic approach. In the last few years, new molecular studies have opened the way to innovative target therapies (e.g. bortezomib,17,18 venetoclax,22 BET-inhibitors,24 SL-40125) being used in clinical trials. Some of these are showing promising results, although still concerns remain regarding their safety. Of note, all the treatments proposed are mainly the result of investigation into the RNA transcriptome, while the DNA features of BPDCN patients have barely been evaluated. We therefore decided to tackle this yet incurable disease by designing the first therapeutic strategy modeled on the DNA mutational status of BPDCN patients, analyzed by WES. The WES mutational findings enhanced by the RNA and PAT-ChIP sequencing results clearly evidenced the prominent role of the epigenetic program dysregulation among BPDCN patients and guided our therapeutic approach towards the use of epigenetic agents. In particular, we tested in vivo the efficacy of US Food and Drug Administration-approved epigenetic drugs which could be considered for potential repositioning in clinical

References 1. Chaperot L, Bendriss N, Manches O, et al. Identification of a leukemic counterpart of the plasmacytoid dendritic cells. Blood. 2001;97(10):3210-3217. 2. Swerdlow SH, Campo E, Hazzis NL, et al. Facchetti F, Jones D, Petrella T. Blastic plasmacytoid dendritic cell neoplasm. In: Swerdlow SH, Campo E, Hazzis NL, et al., eds. WHO Classification of Tumors of Haematopoietic and Lymphoid Tissues. Lyon: IARC Press; 2008:145-147. 3. Facchetti F, Cigognetti M, Fisogni S, Rossi G, Lonardi S, Vermi W. Neoplasms derived from plasmacytoid dendritic cells. Mod Pathol. 2016;29(2):98-111. 4. Garnache-Ottou F, Feuillard J, Ferrand C, et al. Extended diagnostic criteria for plasmacytoid dendritic cell leukaemia. Br J Haematol. 2009;145(5):624-636. 5. Pagano L, Valentini CG, Pulsoni A, et al.

736

7. 8.

trials: two hypomethylating agents such as decitabine and 5’-azacytidine, and the histone deacetylase inhibitor romidepsin. We hypothesized that these drugs could impact on tumor progression because: i) BPDCN patients displayed potential sensitivity to hypomethylating agents, particularly to decitabine, as detected by GSEA analysis; ii) both 5’-azacytidine and decitabine are currently used for the treatment of myelodysplastic syndromes,44,45 which are myeloid neoplasms sharing many epigenetic mutated genes with BPDCN; iii) preclinical studies on several malignancies demonstrated that the action of decitabine is synergized by romidepsin.46 In the light of this, our experimental design focused on epigenetic drugs with a largescale activity, aiming to explore whether we might induce cell death by perturbation of the malignant epigenetic programme. In addition to the epigenetic drugs, we also verified the efficacy of bortezomib, a proteasome inhibitor, which had previously been shown to significantly induce in vitro and in vivo BPDCN cell death.17,18 Our experiments revealed that the treatment with 5’-azacytidine in combination with decitabine significantly inhibits disease progression and extends survival (P<0.01) in a preclinical mouse model. In the past, two reports experimented the use of 5’-azacytidine in elderly BPDCN patients, though this therapeutic choice was not yet sustained by a molecular rationale.47,48 Here we demonstrate that 5’-azacytidine is more effective in tumor eradication when combined with decitabine. Further studies are ongoing to elucidate the synergistic mechanisms between the two drugs. In conclusion, we have identified the deregulation of the epigenetic program as a genetic hallmark of BPDCN and suggest a novel therapeutic approach based on the combination of two hypomethylating agents, 5’-azacytidine and decitabine, to be tested in future clinical trials. Funding The present work was supported by the AIRC grants IG 15762 and 5x1000 10007 “Genetics-driven targeted management of lymphoid malignancies” and the Grant “Innovative approaches to the diagnosis and pharmacogenetic-based therapies of primary hepatic tumours, peripheral B and T-cell lymphomas and lymphoblastic leukaemias” Strategic Programme 2010-2012 Regione Emilia Romagna - Università (all to SAP).

Blastic plasmacytoid dendritic cell neoplasm with leukemic presentation: an Italian multicenter study. Haematologica. 2013;98(2): 239-246. Roos-Weil D, Dietrich S, Boumendil A, et al. Stem cell transplantation can provide durable disease control in blastic plasmacytoid dendritic cell neoplasm: a retrospective study from the European Group for Blood and Marrow Transplantation. Blood. 2013;121(3):440-446. Pemmaraju N. Blastic plasmacytoid dendritic cell neoplasm. Clin Adv Hematol Oncol. 2016;14(4):220-222. Petrella T, Dalac S, Maynadie M, et al. CD4+ CD56+ cutaneous neoplasms: a distinct hematological entity? Groupe Francais d'Etude des Lymphomes Cutanes (GFELC). Am J Surg Pathol. 1999;23(2):137-146. Leroux D, Mugneret F, Callanan M, et al. CD4(+), CD56(+) DC2 acute leukemia is characterized by recurrent clonal chromosomal changes affecting 6 major targets: a

10.

11.

12.

13.

study of 21 cases by the Groupe Francais de Cytogenetique Hematologique. Blood. 2002;99(11):4154-4159. Reichard KK, Burks EJ, Foucar MK, et al. CD4(+) CD56(+) lineage-negative malignancies are rare tumors of plasmacytoid dendritic cells. Am J Surg Pathol. 2005;29(10):12741283. Dijkman R, van Doorn R, Szuhai K, Willemze R, Vermeer MH, Tensen CP. Gene-expression profiling and array-based CGH classify CD4+CD56+ hematodermic neoplasm and cutaneous myelomonocytic leukemia as distinct disease entities. Blood. 2007;109(4):1720-1727. Wiesner T, Obenauf AC, Cota C, Fried I, Speicher MR, Cerroni L. Alterations of the cell-cycle inhibitors p27(KIP1) and p16(INK4a) are frequent in blastic plasmacytoid dendritic cell neoplasms. J Invest Dermatol. 2010;130(4):1152-1157. Agliano A, Martin-Padura I, Marighetti P, et al. Therapeutic effect of lenalidomide in a

haematologica | 2019; 104(4)

Epigenetic matrix and targeted therapy of BPDCN

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

novel xenograft mouse model of human blastic NK cell lymphoma/blastic plasmacytoid dendritic cell neoplasm. Clin Cancer Res. 2011;17(19):6163-6173. Jardin F, Ruminy P, Parmentier F, et al. TET2 and TP53 mutations are frequently observed in blastic plasmacytoid dendritic cell neoplasm. Br J Haematol. 2011;153(3):413-416. Lucioni M, Novara F, Fiandrino G, et al. Twenty-one cases of blastic plasmacytoid dendritic cell neoplasm: focus on biallelic locus 9p21.3 deletion. Blood. 2011;118(17):4591-4594. Alayed K, Patel KP, Konoplev S, et al. TET2 mutations, myelodysplastic features, and a distinct immunoprofile characterize blastic plasmacytoid dendritic cell neoplasm in the bone marrow. Am J Hematol. 2013;88(12):1055-1061. Sapienza MR, Fuligni F, Agostinelli C, et al. Molecular profiling of blastic plasmacytoid dendritic cell neoplasm reveals a unique pattern and suggests selective sensitivity to NFkB pathway inhibition. Leukemia. 2014;28(8):1606-1616. Philippe L, Ceroi A, Bole-Richard E, et al. Bortezomib as a new therapeutic approach for blastic plasmacytoid dendritic cell neoplasm. Haematologica. 2017;102(11):18611868. Menezes J, Acquadro F, Wiseman M, et al. Exome sequencing reveals novel and recurrent mutations with clinical impact in blastic plasmacytoid dendritic cell neoplasm. Leukemia. 2014;28(4):823-829. Stenzinger A, Endris V, Pfarr N, et al. Targeted ultra-deep sequencing reveals recurrent and mutually exclusive mutations of cancer genes in blastic plasmacytoid dendritic cell neoplasm. Oncotarget. 2014;5(15): 6404-6413. Emadali A, Hoghoughi N, Duley S, et al. Haploinsufficiency for NR3C1, the gene encoding the glucocorticoid receptor, in blastic plasmacytoid dendritic cell neoplasms. Blood. 2016;127(24):3040-3053. Montero J, Stephansky J, Cai T, et al. Blastic Plasmacytoid Dendritic Cell Neoplasm Is Dependent on BCL2 and Sensitive to Venetoclax. Cancer Discov. 2017;7(2):156164. Ceroi A, Masson D, Roggy A, et al. LXR agonist treatment of blastic plasmacytoid dendritic cell neoplasm restores cholesterol efflux and triggers apoptosis. Blood. 2016;128(23):2694-2707. Ceribelli M, Hou ZE, Kelly PN, et al. A Druggable TCF4- and BRD4-Dependent Transcriptional Network Sustains Malignancy in Blastic Plasmacytoid

haematologica | 2019; 104(4)

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Dendritic Cell Neoplasm. Cancer Cell. 2016;30(5):764-778. Angelot-Delettre F, Roggy A, Frankel AE, et al. In vivo and in vitro sensitivity of blastic plasmacytoid dendritic cell neoplasm to SL401, an interleukin-3 receptor targeted biologic agent. Haematologica. 2015;100(2): 223-230. Fanelli M, Amatori S, Barozzi I, et al. Pathology tissue-chromatin immunoprecipitation, coupled with high-throughput sequencing, allows the epigenetic profiling of patient samples. Proc Natl Acad Sci U S A. 2010;107(50):21535-21540. Maeda T, Murata K, Fukushima T, et al. A novel plasmacytoid dendritic cell line, CAL1, established from a patient with blastic natural killer cell lymphoma. Int J Hematol. 2005;81(2):148-154. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25-29. Ward AF, Braun BS, Shannon KM. Targeting oncogenic Ras signaling in hematologic malignancies. Blood. 2012; 120(17):33973406. Guarini A, Marinelli M, Tavolaro S, et al. ATM gene alterations in chronic lymphocytic leukemia patients induce a distinct gene expression profile and predict disease progression. Haematologica. 2012;97(1):47-55. Simon M, Grandage VL, Linch DC, Khwaja A. Constitutive activation of the Wnt/betacatenin signalling pathway in acute myeloid leukaemia. Oncogene. 2005; 24(14):24102420. Kuo YH, Zaidi SK, Gornostaeva S, Komori T, Stein GS, Castilla LH. Runx2 induces acute myeloid leukemia in cooperation with Cbfbeta-SMMHC in mice. Blood. 2009;113(14):3323-3332. Togami K, Madan V, Li J, et al. Blastic Plasmacytoid Dendritic Cell Neoplasm (BPDCN) Harbors Frequent Splicesosome Mutations That Cause Aberrant RNA Splicing Affecting Genes Critical in pDC Differentiation and Function. Blood. 2016; 128(22):738. Scibetta AG, Santangelo S, Coleman J, et al. Functional analysis of the transcription repressor PLU-1/JARID1B. Mol Cell Biol. 2007;27(20):7220-7235. Pal S, Vishwanath SN, Erdjument-Bromage H, Tempst P, Sif S. Human SWI/SNF-associated PRMT5 methylates histone H3 arginine 8 and negatively regulates expression of ST7 and NM23 tumor suppressor genes. Mol Cell Biol. 2004;24(21):9630-9645. Missiaglia E, Donadelli M, Palmieri M,

37.

38.

39.

40.

41.

42.

43. 44.

45.

46.

47.

48.

Crnogorac-Jurcevic T, Scarpa A, Lemoine NR. Growth delay of human pancreatic cancer cells by methylase inhibitor 5-aza-2'deoxycytidine treatment is associated with activation of the interferon signalling pathway. Oncogene. 2005;24(1):199-211. Lee KH, Park JW, Sung HS, et al. PHF2 histone demethylase acts as a tumor suppressor in association with p53 in cancer. Oncogene. 2015;34(22):2897-2909. Subtil-Rodriguez A, Vazquez-Chavez E, Ceballos-Chavez M, et al. The chromatin remodeller CHD8 is required for E2F-dependent transcription activation of S-phase genes. Nucleic Acids Res. 2014;42(4):21852196. Rotili D, Mai A. Targeting Histone Demethylases: A New Avenue for the Fight against Cancer. Genes Cancer. 2011;2(6): 663-679. Jin Y, Zhou J, Xu F, et al. Targeting methyltransferase PRMT5 eliminates leukemia stem cells in chronic myelogenous leukemia. J Clin Invest. 2016;126(10):3961-3980. Yamane K, Tateishi K, Klose RJ, et al. PLU-1 is an H3K4 demethylase involved in transcriptional repression and breast cancer cell proliferation. Mol Cell. 2007;25(6):801-812. Tang B, Qi G, Tang F, et al. JARID1B promotes metastasis and epithelial-mesenchymal transition via PTEN/AKT signaling in hepatocellular carcinoma cells. Oncotarget. 2015;6(14):12723-12739. Zhang T, Cooper S, Brockdorff N. The interplay of histone modifications - writers that read. EMBO Rep. 2015;16(11):1467-1481. Quintas-Cardama A, Santos FP, GarciaManero G. Therapy with azanucleosides for myelodysplastic syndromes. Nat Rev Clin Oncol. 2010;7(8):433-444. Jabbour E, Short NJ, Montalban-Bravo G, et al. Randomized phase 2 study of low-dose decitabine vs low-dose azacitidine in lowerrisk MDS and MDS/MPN. Blood. 2017;130(13):1514-1522. Kalac M, Scotto L, Marchi E, et al. HDAC inhibitors and decitabine are highly synergistic and associated with unique geneexpression and epigenetic profiles in models of DLBCL. Blood. 2011;118(20):55065516. Laribi K, Denizon N, Ghnaya H, et al. Blastic plasmacytoid dendritic cell neoplasm: the first report of two cases treated by 5-azacytidine. Eur J Haematol. 2014;93(1):81-85. Khwaja R, Daly A, Wong M, Mahe E, Cerquozzi S, Owen C. Azacitidine in the treatment of blastic plasmacytoid dendritic cell neoplasm: a report of 3 cases. Leuk Lymphoma. 2016;57(11):2720-2722.

737

ARTICLE Ferrata Storti Foundation

Acute Lymphoblastic Leukemia

Autophagy inhibition as a potential future targeted therapy for ETV6-RUNX1-driven B-cell precursor acute lymphoblastic leukemia Roel Polak,1 Marc B. Bierings,2,3 Cindy S. van der Leije,4 Mathijs A. Sanders,4 Onno Roovers,4 João R. M. Marchante,1 Judith M. Boer,1 Jan J. Cornelissen,4 Rob Pieters,3 Monique L. den Boer1,3* and Miranda Buitenhuis4* Department of Pediatric Oncology, Erasmus MC - Sophia Children’s Hospital, Rotterdam; 2Department of Pediatric Oncology, University Medical Center Utrecht; 3 Princess Máxima Center for Pediatric Oncology, Utrecht and 4Department of Hematology, Erasmus Medical Center, Rotterdam, the Netherlands 1

Haematologica 2019 Volume 104(4):738-748

*MdB and MB contributed equally to this work.

ABSTRACT

Correspondence: MONIQUE L. DEN BOER m.l.denboer@prinsesmaximacentrum.nl Received: March 18, 2018. Accepted: October 30, 2018. Pre-published: October 31, 2018. doi:10.3324/haematol.2018.193631 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/738 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

738

ranslocation t(12;21), resulting in the ETV6-RUNX1 (or TELAML1) fusion protein, is present in 25% of pediatric patients with B-cell precursor acute lymphoblastic leukemia and is considered a first hit in leukemogenesis. A targeted therapy approach is not available for children with this subtype of leukemia. To identify the molecular mechanisms underlying ETV6-RUNX1-driven leukemia, we performed gene expression profiling of healthy hematopoietic progenitors in which we ectopically expressed ETV6-RUNX1. We reveal an ETV6-RUNX1-driven transcriptional network that induces proliferation, survival and cellular homeostasis. In addition, Vps34, an important regulator of autophagy, was found to be induced by ETV6-RUNX1 and up-regulated in ETV6-RUNX1-positive leukemic patient cells. We show that induction of Vps34 was transcriptionally regulated by ETV6RUNX1 and correlated with high levels of autophagy. Knockdown of Vps34 in ETV6-RUNX1-positive cell lines severely reduced proliferation and survival. Inhibition of autophagy by hydroxychloroquine, a well-tolerated autophagy inhibitor, reduced cell viability in both ETV6RUNX1-positive cell lines and primary acute lymphoblastic leukemia samples, and selectively sensitized primary ETV6-RUNX1-positive leukemia samples to L asparaginase. These findings reveal a causal relationship between ETV6-RUNX1 and autophagy, and provide pre-clinical evidence for the efficacy of autophagy inhibitors in ETV6-RUNX1driven leukemia. Introduction Acute lymphoblastic leukemia (ALL) is the most common pediatric malignancy. During the last decades, the overall survival rates of pediatric ALL have improved significantly.1 This is primarily due to optimization of conventional chemotherapeutic drug regimens combined with risk-directed therapies.1 However, to date, still 20% of pediatric ALL cases relapse because of resistance to therapy.2 In addition, long-term treatment-induced side effects remain considerable.3 New treatment regimens increasingly aim to target specific intrinsic characteristics of leukemia. This approach has, for example, led to the successful development of BCR-ABL1 inhibitors.4 Regrettably, such a targeted approach is not available for the majority of children suffering from leukemia. Translocation t(12;21)(p13;q22), resulting in the ETV6-RUNX1 fusion protein (also known as TEL-AML1), is present in 25% of pediatric patients with B-cell precursor acute lymphoblastic leukemia (BCP-ALL) and is therefore the most common fusion protein in childhood cancer.5 The t(12;21)(p13;q22) rearrangement fuses the 5' non-DNA binding region of the ETS family transcription factor ETV6 (TEL) to almost the entire RUNX1 (AML1) locus.5,6 Despite the favorable prognosis haematologica | 2019; 104(4)

Autophagy drives ETV6-RUNX1-positive leukemia

associated with this cytogenetic type of BCP-ALL,7 resistance to chemotherapeutic drugs and relapse occur in approximately 10% of these patients.7-9 The ETV6-RUNX1 fusion protein induces a silent preleukemic clone that requires additional genetic hits for the transition to leukemia.10-12 Although these pre-leukemic ETV6-RUNX1-positive hematopoietic stem cells (HSCs) still possess self-renewal properties and are capable of contributing to hematopoiesis, they fail to outcompete normal HSCs.11,12 In ETV6-RUNX1-positive leukemia, this early genetic lesion is followed by a number of ‘driver’ copy number alterations, including loss of ETV6 and alterations directed to genes regulating normal B-cell differentiation.13 These alterations are acquired independently without pref-

erential order, thereby generating a dynamic clonal architecture.13 This genetic variation implies that targeted therapy in ETV6-RUNX1-driven ALL should preferably be directed to targets that are present in all subclones, i.e. those being deregulated by the ETV6-RUNX1 fusion protein itself. This concept is further supported by the observation that ETV6-RUNX1-positive cell lines are highly dependent on the expression of the fusion protein for their survival.14,15 Previous reports revealed that enhanced levels of STAT3, heat-shock proteins, survivin, has-mir-125b-2, the erythropoietin receptor, cytoskeleton regulatory genes, and the PI3K/PKB/mTOR pathway, as well as aberrant regulation of the TGFβ pathway, are important for ETV6RUNX1-positive BCP-ALL.15-20 However, the molecular

Figure 1. Vps34 is recurrently up-regulated in ETV6-RUNX1-positive B-cell precursor acute lymphoblastic leukemia (BCP-ALL) and is driven by the ETV6-RUNX1 fusion protein. (A) 2log expression levels of the gene probe set mapped to Vps34 were analyzed in a cohort of 132 pediatric ALL patients published by Yeoh et al.26 Gene expression of ETV6-RUNX1-positive patients (green bar) was compared to gene expression of all other B-ALL patients (excluding T-ALL): ***False Discovery Rate (FDR)-adjusted P=3.45*10-15. (B) In addition, 2log expression levels of the gene probe set mapped to Vps34 were analyzed in a cohort of 653 pediatric ALL patients published by Van der Veer et al.30 Gene expression of ETV6-RUNX1-positive patients (green bar) was compared to gene expression of all other B-ALL patients (excluding T-ALL). Gray dashed line represents mean expression of all patients: ***FDR-adjusted P=7.24*10-39. (C) CB-CD34+ cells were transduced with ETV6RUNX1-IRES-eGFP or with control EV-IRES-eGFP after which eGFP+ cells were sorted and Vps34 mRNA levels were determined by Q-PCR and normalized to hypoxanthine-guanine phosphoribosyltransferase (HPRT). Gray bars represent the mean of 5 biological replicates. Gene expression of Vps34 was compared between ETV6RUNX1+ CB-CD34+ cells and EV-control CB-CD34+ cells (n=5; P=0.03). (D) ETV6-RUNX1-positive cells were transfected with siRNAs directed to the ETV6-RUNX1 breakpoint or scrambled control siRNAs. Vps34 mRNA levels were determined in ETV6-RUNX1+ cells (REH (n=2) and ETV6-RUNX1 transduced CB-CD34+ cells (n=2) by QPCR, normalized to HPRT, and compared to the average expression of cells transfected with scrambled control siRNAs (n=3; P≤0.001). Vps34 protein expression was quantified in REH cells by western blot and compared to protein expression of REH cells transfected with scrambled control siRNAs (n=2; P≤0.05 and P≤0.01 for siRNA#1 and siRNA#5, respectively). Bars represent mean. Error bars represent Standard Error of Mean. *P≤0.05, **P≤0.01, ***P≤0.001. See also Online Supplementary Figures S1-S5.

haematologica | 2019; 104(4)

739

R. Polak et al. A

Figure 2. ETV6-RUNX1 and ETV6-RUNX1 target genes enhance Vps34 promoter activity. (A) The UCSC genome browser (GRCh37/hg19) was used to analyze the transcription factors that bind to the Vps34 promoter. Several transcription factors known to play an important role in regulation of hematopoiesis were found to be interacting with this promoter region. Using publically available ChIP-seq data, the interaction of the transcription factors RUNX1, GATA1, GATA2, EGR1 and HEY1 with the Vps34 promoter is shown. (B) Umbilical cord blood-derived CD34+ cells were retrovirally transduced to ectopically express the ETV6-RUNX1 fusion protein. Gene expression analysis was performed 20 and 40 hours (hr) after transduction. For flow chart; see left panel. The gene expression levels of GATA1, GATA2, EGR1 and HEY1 are shown. P-values represent the differences between ETV6-RUNX1 positive (green bars) compared to ETV6-RUNX1-negative (white bars) CB-CD34+ cells (FDR-adjusted). (C) (Top) Schematic representation of the -1383 to +58 region of the Vps34 promoter cloned upstream of the Gaussia luciferase gene. Bar graphs: HEK293T cells were co-transfected with a Vps34 promoter construct, RUNX1 and/or CBFβ (left), increasing concentrations of GATA1 and GATA2 (middle) or HEY1 and EGR1 (right) expression constructs. Luciferase activity was determined 48 hr after transfection. eGFP expression was quantified in each sample by flow cytometric analysis and used to normalize luciferase activity. Data were depicted as fold induction compared to empty vector controls. Error bars represent Standard Error of Mean. *P≤0.05, **P≤0.01, ***P≤0.001. See also Online Supplementary Figure S2A-C.

740

haematologica | 2019; 104(4)

Autophagy drives ETV6-RUNX1-positive leukemia

network underlying the persistence and maintenance of ETV6-RUNX1 BCP-ALL remains to be elucidated. In the present study, we address the role of autophagy in ETV6-RUNX1-driven leukemia. Autophagy is a cellular recycling system in which unwanted or damaged cellular components are degraded and recycled. The core autophagy-regulating complex includes Vps34, Beclin-1, and Vps15.21,22 Although autophagy can sustain cell survival during stress conditions, it can also result in cell death because of progressive cellular consumption.23 Whether autophagy plays an initiating or suppressive role in cancer is a question of debate and most likely depends on the (onco)genetic context of cells.24,25 This potential dual role of autophagy in cancer highlights the importance of studies on the context-specific role and the functional importance of autophagy in neoplastic processes before the start of autophagy-based therapeutic interventions. We show here that ETV6-RUNX1 targets the autophagy process, which in turn affects sensitivity to LAsparaginase, a key enzyme used in the treatment of ALL

that affects the asparagine (and to a lesser extent glutamine) levels in cells.

Methods Transduction and gene expression profiling of primary cells CD34-positive hematopoietic progenitor cells (CB-CD34+ cells) were derived from human cord blood and transduced with retrovirus expressing ETV6-RUNX1 and eGFP. DAPI-CD34+ GFP+ CBCD34+ cells were sorted using a BD ARIA II sorter. After sorting, cells were lysed and RNA was extracted and subsequently linearly amplified. Bone marrow aspirates were obtained from children with newly diagnosed BCP-ALL prior to treatment. Leukemic blasts were collected and processed as previously described. Affymetrix GeneChip HG-U133-Plus-2.0 microarrays were used for all samples. Microarray data of CB-CD34+ cells are available in the ArrayExpress database under accession number E

Figure 3. Autophagy levels are high in ETV6-RUNX1-positive B-cell precursor acute lymphoblastic leukemia (BCP-ALL) and regulated by ETV6-RUNX1 and Vps34. (A) Western blot analysis was performed to determine the expression levels Vps34, p62 (sequestosome 1), and LC3B in ALL cell lines. PIK3R4 was used as a loading control. (B) Quantification of protein levels of Vps34 and p62 measured by reverse phase protein array (RPPA) in 30 ETV6-RUNX1-positive primary BCP-ALL patient samples (ETV6-RUNX1+), and 29 B-Other primary BCP-ALL patient samples (BO). Data are means±Standard Error of Mean. *P≤0.05, **P≤0.01. (C) Representative confocal images showing LC3B-positive vesicles in ETV6-RUNX1-positive BCP-ALL cell line REH. (Left) Overlay of LC3B expression and DAPI staining (nuclear staining). (Middle) Only LC3B expression. (Right) 3D representations after deconvolution of the 488nm signal representing the LC3B expression. For the quantification of number and volume of LC3B-positive vesicles, we excluded cells with atypical nuclei. (Top) Control conditions after transfection with scrambled siRNAs. Bottom 6 panels represent conditions after transfection with siRNAs against ETV6-RUNX1 or Vps34. (D and E) Quantification of the number of LC3B-positive vesicles after 3D deconvolution of images (D), and quantification of the number of LC3B-positive vesicles multiplied by the volume of these vesicles after 3D deconvolution of images (E) (n=3, *P≤0.05,***P≤0.001). See also Online Supplementary Figure S6.

haematologica | 2019; 104(4)

741

R. Polak et al.

MTAB-3466. Microarray data of BCP-ALL blasts are available in the Gene Expression Omnibus database. Informed consent was provided according to the Declaration of Helsinki. Use of left-over materials for research purposes was approved by the Institutional Review Board (IRB) of the University Medical Center Rotterdam; IRB approval file number MEC 2004-203.

Functional assays For protein quantification, both western blot analysis and reverse phase protein arrays (RPPA) were used. Vps34 promoter activity was studied using a reporter construct consisting of a 1.4 kB region of the Vps34 promoter upstream of the Gaussia luciferase gene. Cell viability was quantified using MTT cytotoxicity assays or flow cytometry-based Annexin V - Propidium Iodide assays. Silencing of genes was achieved using transfec-

tion of specific siRNAs or a lentiviral knockdown approach using the pLKO.1 Mission vector containing a puromycin selection marker. Autophagy levels (number and volume of LC3Bpositive vesicles) were quantified using confocal scanning microscopy (Leica SP5).

Statistical analysis Statistical analysis of microarray data of paired CB-CD34+ cells was performed using a linear mixed model. Microarray data of primary ALL samples were analyzed using LIMMA. Functional analysis of differential gene expression was performed using QIAGEN’s Ingenuity Pathway Analysis. Both the Student t-test and the Student paired t-test were used when applicable. Bar graphs represent the mean of biological replicates. Error bars represent standard error of mean (SEM). Further details of the methods used are available in the Online Supplementary Appendix.

Figure 4. Vps34 is essential for the survival of ETV6-RUNX1-positive leukemic cells. (A) Schematic representation of the known transcript variants of Vps34 (Ensemble Genome Browser; ENSG00000078142) and the shVps34#1 and shVps34#2 recognition sites. (B) ETV6-RUNX1-positive (REH) BCP-ALL cells were lentivirally transduced with scrambled shRNA control, or two distinct shRNA constructs to silence Vps34 expression. Western blot analysis was performed with an antibody against Vps34 or β-actin to visualize the knockdown of Vps34 in REH cells. A representative experiment is shown in which increasing concentrations of virus were used to emphasize the specificity of Vps34 knockdown. (C) Data were quantified and are depicted as the relative Vps34 expression in comparison to the expression in cells transduced with scrambled (non-silencing) controls. (D) ETV6-RUNX1-positive (REH and REHS1) and ETV6-RUNX1-negative (NALM6) BCP-ALL cells were lentivirally transduced with scrambled shRNA control (NSC) or two distinct Vps34 shRNA constructs. NI: non-infected cells. Cells were cultured for 18 days. To determine the effect on proliferation, cell counts were performed every 2-3 days. Representative graphs are shown (n=3). (E and F) ETV6-RUNX1-positive (REH) BCP-ALL cells were lentivirally transduced with scrambled shRNA control (NSC) or two distinct shRNA constructs to silence Vps34 expression. After seven days of culture, flow cytometrical analysis was performed to determine the effect of Vps34 knockdown on survival and cell-cycle progression. (E) The percentage of viable (AnnexinV-positive, Propidium-Iodide negative), actively cycling cells was determined using DyeCycle. Data are shown as the percentage of cells in S, G2 M phase (n=2; *P≤0.05). Error bars represent Standard Error of Mean (SEM). (F) The percentages of early apoptotic (AnnexinV-positive, Propidium-Iodide negative) and late apoptotic (Propidium-Iodide positive) cells were determined seven days after transduction (n=2, **P≤0.01, ***P≤0.001). Error bars represent SEM. See also Online Supplementary Figure S7.

742

haematologica | 2019; 104(4)

Autophagy drives ETV6-RUNX1-positive leukemia

Results

Supplementary Figure S1D and E and Online Supplementary Figure S2A and B). For the gene expression analysis of primary ALL samples, we used data from cohorts in which ETV6 RUNX1positive BCP-ALL patients were included26-30 and data from an ETV6-RUNX1 knockdown study performed in a leukemic cell line31 (Online Supplementary Figure S3). Pathway analysis on the top 500 differentially expressed genes in the largest patient cohort (Erasmus MC cohort, 654 ALL patients including 172 ETV6-RUNX1-positive BCP-ALL patients and 401 ETV6vRUNX1-negative BCP ALL patients30) revealed a pro-survival and pro-proliferative signature in ETV6 RUNX1-positive patient cells, similar to the phenotype predicted in CB-CD34+ cells ectopically expressing ETV6-RUNX1 (Online Supplementary Table S1 and Online Supplementary Figure S2C). The class III PI3 kinase Vps34 (PIK3C3) was found to be recurrently up-regulated in ETV6-RUNX1-positive ALL patient cells (2.7 fold-higher expression in Erasmus MC

Vps34 is up-regulated in ETV6-RUNX1-positive BCP-ALL and is induced by the ETV6-RUNX1 fusion protein To identify novel ETV6-RUNX1 target genes, we compared gene expression profiles of primary ALL samples with those of ETV6-RUNX1-transduced umbilical cord blood-derived, healthy CD34-positive hematopoietic progenitors (CB-CD34+) (Online Supplementary Figure S1A). Gene expression analysis revealed that 196 genes were differentially expressed in ETV6-RUNX1 transduced CB CD34+ cells in comparison to empty vector controls (2-40 fold; P≤0.05) (Online Supplementary Figure S1B). Ingenuity Pathway Analysis on these 196 genes predicted an interacting gene network of 36 genes (Online Supplementary Figure S1C). Analysis of gene ontology (GO) functional categories indicated that ETV6-RUNX1 induced a signature associated with pro-survival and pro-proliferative gene expression and cellular homeostasis (Online

Figure 5. ETV6-RUNX1-positive acute lymphoblastic leukemia (ALL) cells are relatively sensitive to treatment with hydroxychloroquine (HCQ). (A) ETV6-RUNX1-positive (REH and REHS1) and ETV6-RUNX1-negative (NALM6) BCP-ALL cells were cultured in absence or presence of HCQ (6.25 mg/mL) for 48 hours. Western blot analysis was performed using an antibody against LC3B to determine the effectiveness of HCQ treatment. LC3B-I represents the cytosolic form of LC3B, while LC3BII represents the autophagosome membrane-bound form of LC3B (see also Figure 7). HCQ treatment leads to the accumulation of autophagosomes by blocking the fusion of autophagosomes to lysosomes and hence to an increase in LC3B-II in cells. (B) ETV6-RUNX1-positive BCP-ALL, ETV6-RUNX1-negative BCP-ALL, and T-ALL cell lines were cultured in absence or presence of HCQ (20 mg/mL) for four days. An MTT assay was performed to determine the effect of HCQ treatment on the viability of the cells. Data are depicted as the percentage of viable cells compared to untreated control. Error bars represent Standard Error of Mean (SEM) (n=3). (C) ETV6-RUNX1-positive BCP-ALL, ETV6-RUNX1-negative BCP-ALL, and T-ALL cell lines were cultured in absence or presence of increasing concentrations of HCQ for four days. An MTT assay was performed to determine the effect of HCQ treatment on the viability of the cells. A representative experiment is shown in which data are depicted as the percentage of viable cells. (D) Primary ETV6-RUNX1-positive BCP-ALL cells were cultured in absence or presence of 5 or 10 mg/mL HCQ for five days. Flow cytometric analysis was performed to determine the percentage of non-apoptotic (Annexin-V negative, Propidium-Iodide negative, CD19-positive) cells. Representative FACS plots are shown (n=5). (E) Co-culture experiments were performed with primary ETV6-RUNX1-positive BCP-ALL cells and mesenchymal stromal cells (MSCs). Cells were cultured in absence or presence of increasing concentrations of HCQ for five days. Flow cytometrical analysis was performed to determine the percentage of non-apoptotic (Annexin V-negative, Propidium Iodide-negative, CD19-positive) cells. Data are depicted as the relative reduction in survival compared to untreated cells (n=6 for HCQ 5 mg/mL and HCQ 10 mg/mL; n=4 for HCQ 20 mg/mL). Error bars represent SEM. *P≤0.05, **P≤0.01. See also Online Supplementary Figure S8.

haematologica | 2019; 104(4)

743

R. Polak et al.

cohort, FDR-adjusted P=7.24E-39) (Figure 1A and B and Online Supplementary Figure S3). To determine the direct effect of the ETV6-RUNX1 fusion protein on Vps34 expression, ETV6-RUNX1 was ectopically expressed in CB-CD34+ cells. Forty hours after transduction, Vps34 mRNA expression was significantly up-regulated by 1.3 fold (P=0.03) (Figure 1D), suggesting a causal relationship between ETV6-RUNX1 and Vps34. Reciprocal experiments were performed in ETV6-RUNX1-positive cells (REH cell line and ETV6-RUNX1 transduced CB-CD34+ cells) using siRNAs directed to the ETV6-RUNX1 breakpoint. Although the ETV6-RUNX1 mRNA levels could only be reduced by 30-35% (P≤0.05) (Online Supplementary Figure S5B), this reduction was sufficient to significantly reduce the levels of Vps34 mRNA and protein both by approximately 40% (P≤0.05) (Figure 1D).

ETV6-RUNX1 and ETV6-RUNX1 target genes enhance Vps34 promoter activity The upregulation of Vps34 expression in ETV6-RUNX1positive BCP-ALL patients and ETV6-RUNX1 transduced CB-CD34+ cells, suggests that the Vps34 promoter is activated directly or indirectly by ETV6-RUNX1. Analysis of the Vps34 promoter, using publically available ChIP-seq data, revealed that transcription factors known to play an important role in regulation of hematopoiesis, including

GATA1, GATA2, EGR1 and HEY1, can interact with the Vps34 promoter (Figure 2A). Four of these transcription factors, namely GATA1, GATA2, EGR1 and HEY1, were also found to be up-regulated in ETV6-RUNX1-transcells (Figure 2B and Online duced CB-CD34+ Supplementary Figure S1C and Online Supplementary Figure S4). The mRNA expression levels of these four genes were modestly increased in ETV6-RUNX1-transduced CBCD34+ cells 20 hours after transduction and significantly up-regulated after 40 hours: GATA1, GATA2, EGR1, and HEY1 were up-regulated 3.8-fold (P=0.046), 2.2-fold (P=0.019), 5.0-fold (P=0.004), and 24.9-fold (P=0.015), respectively (Figure 2B). To investigate the role of ETV6-RUNX1 and the transcription factors GATA1, GATA2, EGR1, and HEY1 in regulation of the activity of the Vps34 promoter, luciferase reporter assays were performed using a reporter construct consisting of the Gaussia luciferase gene downstream of a 1.4 kB region (-1383 to +58) of the Vps34 promoter (Figure 2C). Although RUNX1 expression alone was sufficient to induce Vps34 promoter activity (2.4-fold compared to control, P≤0.01), co-expression of its co-factor CBFβ further enhanced Vps34 promoter activity (4.5-fold compared to control, P≤0.05). Similarly, expression of ETV6RUNX1 was sufficient to induce Vps34 promoter activity (4.2-fold compared to control, P≤0.05), and luciferase

Figure 6. Autophagy inhibition sensitizes ETV6-RUNX1-positive acute lymphoblastic leukemia (ALL) cells to L-Asparaginase. (A) Primary ETV6-RUNX1-positive BCPALL cells were cultured in absence or presence of IC 50 concentrations of L-Asparaginase and increasing concentrations of hydroxychloroquine (HCQ). Flow cytometric analysis was performed to determine the percentage of non-apoptotic (Annexin V-negative, Propidium Iodide-negative, CD19-positive) cells (for gating strategy see also Online Supplementary Figure S7C). The survival of primary leukemic blasts in the presence of L-Asparaginase was compared to their survival in the absence of L-Asparaginase. White bars represent the relative survival in the absence of HCQ. Gray bars represent the relative survival in the presence of HCQ: 5 mg/mL HCQ (light gray) and 10 mg/mL HCQ (gray). (B) Averages of data presented in (A), representing sensitization of primary leukemic blasts by HCQ to L-Asparaginase (n=5 for ETV6-RUNX1-positive; n=3 for ETV6-RUNX1-negative primary patient cells; *P≤0.05, **P≤0.01, *** P≤0.001). Error bars represent Standard Error of Mean (SEM). (C and D) Co-culture experiments were performed with primary ETV6-RUNX1-positive BCP-ALL cells and mesenchymal stem cells (MSC). Cells were cultured in the presence or the absence of L-Asparaginase and increasing concentrations of HCQ. Flow cytometric analysis was performed to determine the percentage of nonapoptotic (Annexin V-negative, Propidium-Iodide-negative, CD19-positive) cells. First, the survival of primary leukemic blasts in the presence of L-Asparaginase was compared to their survival in absence of L-Asparaginase. Next, data are depicted as fold reduction compared to HCQ-untreated controls (n=5 for conditions in the absence of MSCs, n=7 for conditions in the presence of MSCs for (C), n=3 for conditions in the absence of MSCs, n=4 for conditions in the presence of MSCs for (D). *P≤0.05, **P≤0.01, ***P≤0.001. Error bars represent SEM. See also Online Supplementary Figure S9.

744

haematologica | 2019; 104(4)

Autophagy drives ETV6-RUNX1-positive leukemia

expression was further enhanced by co-expression of CBFβ (13-fold compared to control, P≤0.05). These results demonstrate that, although both RUNX1 and ETV6RUNX1 function as transcriptional activators of Vps34, ETV6-RUNX1 induces Vps34 promoter activity more efficiently (Figure 2C). In addition, these results demonstrate that CBFβ acts as a co-activator for ETV6-RUNX1 in inducing Vps34 promoter activity. Additional luciferase reporter assays revealed that the ETV6-RUNX1 target genes HEY1, EGR1 and GATA1 induce Vps34 promoter activity in a dose-dependent manner. HEY1, EGR1, and GATA1 induced Vps34 promoter activity up to 13-fold (P≤0.05), 5.0-fold (P≤0.001), and 10-fold (P≤0.05), respectively (Figure 2C). GATA2 expression did not induce luciferase expression, which was in concordance with the absence of a GATA2 DNA binding domain in the -1338/+58 promoter region used for the reported assays (Figure 2A and C). Together, these results demonstrate that the Vps34 promoter is not only positively regulated by the ETV6-RUNX1 fusion protein itself, but also by its target genes HEY1, EGR1 and GATA1.

Autophagy levels are high in ETV6-RUNX1-positive BCP-ALL and regulated by ETV6-RUNX1 and Vps34 Since Vps34 is a key player in autophagy regulation, we hypothesized that ETV6-RUNX1-mediated upregulation of Vps34 induces autophagy in BCP-ALL cells. To investi-

gate this, autophagy levels were determined in ALL cell lines and primary BCP-ALL samples by western blot analysis and RPPA. Western blot analysis in a panel of ALL cell lines revealed that Vps34 protein levels are the highest in the ETV6-RUNX1-positive cell line REH (Figure 3A). In comparison to ETV6-RUNX1-negative ALL cell lines, lower levels of p62 (SQSTM1) and LC3B, both specifically degraded by autophagy, were observed in REH cells. These results suggest high levels of autophagy in REH cells (Figure 3A). RPPA on samples from a large cohort of newly diagnosed BCP-ALL patients revealed that the level of Vps34 was 9.6-fold higher in BCP-ALL cells compared to healthy bone marrow-derived mononuclear cells (59 BCP-ALL patients vs. 10 healthy controls; P≤0.001) (Online Supplementary Figure S6C). Vps34 levels were also significantly higher in ETV6-RUNX1-positive patient cells in comparison to ETV6-RUNX1-negative BCP ALL (B-Other) patient cells (1.2-fold, P≤0.05) (Figure 3B). In line with this, lower p62 protein levels were observed in ETV6-RUNX1positive in comparison to ETV6-RUNX1-negative BCPALL patient cells (1.4-fold, P≤0.01) (Figure 3B). These RPPA data were confirmed by western blot analysis in a smaller set of patients (n=11 patients) (Online Supplementary Figure S6A and B). Next, an siRNA approach was used to investigate whether ETV6-RUNX1 regulates autophagy. To quantify the level of autophagy, the number and volume of LC3B-

Figure 7. Proposed model for the induction of the Vps34-autophagy pathway in ETV6 RUNX1-positive B-cell precursor acute lymphoblastic leukemia (BCP-ALL) cells. The ETV6-RUNX1 fusion protein can transcriptionally induce the expression of various transcription factors, including GATA1, GATA2, HEY1, and EGR1. ETV6RUNX1 and its co-factor CBFβ, together with GATA1, HEY1 and EGR1 can activate the Vps34 promoter, resulting in enhanced Vps34 expression in ETV6-RUNX1-positive leukemic cells. Vps34, in turn, can initiate autophagy by forming a core autophagy-regulating complex with Beclin 1 and Vps15. This complex plays an important role in: 1) the early initiation (in complex with Atg14L); and 2) the vesicle elongation phase (together with UVRAG) of autophagosome formation. Induction of autophagy allows ETV6-RUNX1-positive cells to maintain homeostasis by degrading and recycling damaged proteins and organelles. In addition, activation of the autophagy program in ETV6-RUNX1-positive cells results in enhanced proliferation, survival and drug resistance. Inhibition of autophagy in ETV6-RUNX1-positive cells, by treatment with hydroxychloroquine (HCQ) or knockdown of Vps34, is sufficient to reduce proliferation and survival of leukemic blasts and to induce sensitization to L-Asparaginase.

haematologica | 2019; 104(4)

745

R. Polak et al.

positive vesicles were calculated per cell with a 3D object analyzer using Huygens Professional software (Online Supplementary Figure S6D). Knockdown of ETV6-RUNX1 resulted in a 45% reduction in the number of LC3B positive vesicles and 50% reduction in the volume of LC3Bpositive vesicles per cell (P≤0.001 and P≤0.05, respectively) (Figure 3C-E). In addition, knockdown of Vps34 reduced the number of LC3B-positive vesicles and the volume of LC3B-positive vesicles with 79% (P≤0.001) and 84% (P≤0.05), respectively (Figure 3C-E). These results suggest that autophagy levels are high in ETV6-RUNX1 positive BCP-ALL cells. In addition, both ETV6-RUNX1 and Vps34 are important for maintaining high levels of autophagy in ETV6-RUNX1-positive leukemic cells.

Vps34 is essential for the survival of ETV6-RUNX1-positive leukemic cells To determine the functional role of Vps34 in ETV6RUNX1-positive leukemic cells, lentiviral knockdown studies were performed in two ETV6-RUNX1-positive cell lines (REH and REHS1). For these studies, four independent short hairpin RNAs were used. Short hairpins shVps34#1 and shVps34#2 were directed against the main Vps34 transcript variants, whereas shVps34#3 and shVps34#4 were directed against the two full length transcript variants only (Figure 4A and Online Supplementary Figure S7C). Vps34 knockdown of at least 80% was achieved with all four individual shRNAs (Figure 4B and C and Online Supplementary Figure S7D). Knockdown of only the full-length Vps34 transcript variants with shVps34#3 and shVps34#4 significantly reduced proliferation of the ETV6-RUNX1-positive cell lines REH and REHS1 (P≤0.01) (Figure 4A and Online Supplementary Figure S7E and F). Knockdown of all main Vps34 transcript variants with shVps34#1 and shVps34#2 resulted in a complete growth arrest of these ETV6-RUNX1-positive cell lines (P≤0.001) (Figure 4D and Online Supplementary Figure S7A). In contrast, knockdown of either all main Vps34 transcript variants or the full-length Vps34 transcripts in ETV6-RUNX1negative NALM6 cells decreased proliferation to a significantly lesser extent (P≤0.001) (Figure 4D and Online Supplementary Figure S7A, E and F). To investigate whether the observed growth arrest in ETV6-RUNX1-positive cells was due to a block in cell cycle progression or enhanced apoptosis, flow cytometric analysis was performed using DyeCycle and Annexin V. Cell cycle analysis revealed that knockdown of Vps34 modestly reduces the percentage of cycling ETV6-RUNX1 positive cells (Figure 4E and Online Supplementary Figure S7B and G). In contrast, a remarkable reduction in survival was observed upon Vps34 knockdown. While shRNA mediated knockdown of the full length Vps34 transcript variants in ETV6-RUNX1 positive cells (shRNA#3 and shRNA#4) already resulted in 40-50% apoptotic cells (Online Supplementary Figure S7H), targeting of all main Vps34 transcript variants resulted in even higher levels of apoptosis (80-90%, P<0.01) (Figure 4F). In conclusion, knockdown of Vps34 completely arrests cell growth of ETV6-RUNX1-positive cells by modestly reducing cell cycle progression and strongly inducing apoptosis. Although Vps34 knockdown also affects cell cycle progression and survival in ETV6-RUNX1-negative cells (data not shown), these cells were still able to proliferate. 746

ETV6-RUNX1-positive ALL cells are relatively sensitive to hydroxychloroquine To date, no agents are clinically available that specifically inhibit Vps34 activity. However, the efficacy of autophagy inhibitors is currently being examined in clinical cancer treatment trials. To investigate the effect of autophagy inhibition on ETV6-RUNX1-positive leukemia, we exposed leukemic cells to hydroxychloroquine (HCQ). This agent has favorable pharmacological properties and has been safely used for decades in the treatment of malaria and rheumatoid arthritis.32 HCQ accumulates within and de-acidifies the lysosome, resulting in increased LC3B-II levels, which here is indicative of impaired autophagy (Figure 5A). MTT assays were performed to determine the effect of HCQ on cell viability of ETV6RUNX1-positive and ETV6-RUNX1-negative ALL cell lines (Figure 5B and C and Online Supplementary Figure S8A and B). While treatment with 20 mg/mL HCQ resulted in 82% and 95% reduced cell viability of ETV6-RUNX1-positive BCP-ALL cell lines (REH and REHS1), the viability of ETV6-RUNX1-negative cell lines was reduced to a lesser extent (NALM6: 43%, TOM1: 50%, Loucy: 40%, Jurkat: 0%) (Figure 5B). The IC50 of HCQ was significantly lower in ETV6-RUNX1-positive ALL cell lines compared to ETV6-RUNX1-negative cell lines (P≤0.001) (Online Supplementary Figure S8A and B). In addition, the effect of HCQ on the survival of primary BCP-ALL samples was determined. Primary BCP-ALL samples were cultured for five days upon which flow cytometric analysis was performed (see Online Supplementary Figure S8C for flow cytometric gating strategy). Survival of primary ETV6-RUNX1-positive BCP-ALL samples was significantly reduced after treatment with 10 mg/mL HCQ (26%, P≤0.05) (Figure 5D and E). In contrast, this treatment did not affect cell viability of primary ETV6-RUNX1-negative BCP-ALL samples (Online Supplementary Figure S8D). The survival of primary ETV6-RUNX1-positive ALL samples was even further reduced after treatment with 20 mg/mL HCQ (79%, P≤0.01) (Figure 5E). Co-culture of these ALL samples in the presence of primary bone marrow-derived mesenchymal stromal cells (MSCs) significantly rescued the HCQ-mediated induction of apoptosis in primary ETV6-RUNX1-positive ALL samples (P≤0.05) (Figure 5E). These results demonstrate that although ETV6-RUNX1positive BCP-ALL cells are relatively sensitive to HCQ treatment, this sensitivity is abrogated by primary MSCs.

Autophagy inhibition sensitizes ETV6-RUNX1-positive acute lymphoblastic leukemia cells to L-Asparaginase As our results indicate that inhibition of autophagy reduces survival of ETV6-RUNX1-positive ALL samples, we investigated whether HCQ-mediated inhibition of autophagy could sensitize primary BCP-ALL samples to commonly used chemotherapeutics. To investigate the potential of HCQ treatment in sensitization to chemotherapeutics, the percentage of apoptotic cells was determined by flow cytometry after five days of culture either in absence or presence of IC50 values of the chemotherapeutic drug. Our results indicate that HCQ treatment (in clinically relevant concentrations33,34) selectively sensitizes ETV6 RUNX1-positive leukemic cells to L-Asparaginase treatment (Figure 6A and Online Supplementary Figure S9A). Treatment of primary ETV6 RUNX1-positive ALL samples with 5 mg/mL or 10 mg/mL HCQ resulted in a 48% haematologica | 2019; 104(4)

Autophagy drives ETV6-RUNX1-positive leukemia

and 71% reduction in cell survival during L-Asparaginase exposure, respectively (P≤0.01 and P≤0.001) (Figure 6B), while ETV6-RUNX1-negative BCP-ALL cells were not sensitized to L-Asparaginase (Figure 6B and Online Supplementary Figure S9D). Co-culture of primary BCPALL samples with primary MSCs significantly reversed the HCQ-mediated sensitization to L-Asparaginase (P≤0.01) (Figure 6C). To investigate whether these primary samples could still be sensitized, similar experiments were performed with a higher dose of HCQ (20 mg/mL). Inhibition of autophagy with 20 mg/mL HCQ was indeed sufficient to significantly sensitize primary ETV6-RUNX1positive BCP-ALL samples to L-Asparaginase both in absence or presence of primary MSCs (80% reduced survival, P≤0.05 and P≤0.01, respectively) (Figure 6D and Online Supplementary Figure S9B and C). In contrast to L-Asparaginase, HCQ did not significantly induce apoptosis of primary BCP-ALL samples upon treatment with prednisolone or 6-mercaptopurine (Online Supplementary Figure S9E and F). These data show that HCQ-mediated inhibition of autophagy results in sensitization of ETV6-RUNX1-positive BCP-ALL cells, but not ETV6-RUNX1-negative BCPALL cells, to L-Asparaginase.

Discussion In this study, we show that the ETV6-RUNX1 fusion gene induces a transcriptional network regulating preleukemic features in hematopoietic progenitors. We show that this network facilitates the induction of autophagy by up-regulating Vps34 expression in ETV6-RUNX1-positive BCP-ALL (Figure 7). In addition, our data show for the first time that inhibition of autophagy is a promising strategy for sensitization of ETV6-RUNX1-positive BCP-ALL cells to the important anti-leukemic agent L-Asparaginase. The importance of the ETV6-RUNX1 fusion protein for modulation of proliferation, survival and cell cycle distribution has already been shown in cell lines14,17,31 and mouse models.10 Similarly, expression of the ETV6RUNX1 fusion gene induces survival properties in human cord blood-derived progenitors transplanted in NOD/SCID mice or co-cultured in the presence of murine MS-5 stromal cells.11,20 However, to date, the downstream effectors of this pro-survival and pro-proliferative phenotype have not been elucidated. In this study, we uncovered the transcriptional network regulating these phenotypes in a human progenitor population by analyzing the gene expression profile after ectopic expression of ETV6RUNX1. This approach allowed us to examine early effects of ETV6-RUNX1 expression in human hematopoietic progenitors. These data, therefore, provide a comprehensive and functional list of ETV6-RUNX1 target genes (Online Supplementary Table S2). In addition to a pro-survival and pro-proliferative phenotype, genes involved in cytoskeleton rearrangements and cellular homeostasis were found to be regulated by the ETV6-RUNX1 fusion protein (Online Supplementary Figure S2). Importantly, our results reveal that autophagy is induced in ETV6-RUNX1positive cells because of transcriptional activation of Vps34, a member of the core (macro)autophagy-regulating complex (Figures 1-3).22 In the present study, we show that the ETV6-RUNX1 fusion gene can directly up-regulate the level of autophagy haematologica | 2019; 104(4)

in leukemic cells in absence of cellular stress. Our results demonstrate that these enhanced levels of autophagy are important to maintain proliferation and survival of ETV6RUNX1-positive leukemic cells (Figures 4-6). Knockdown of Vps34 and inhibition of autophagy with HCQ reduced the proliferation and survival of ETV6 RUNX1-positive BCP-ALL cells, confirming the importance of induced autophagy in these cells (Figures 4 and 5). These results are in line with a recently published study showing sensitization of REH cells to L-Asparaginase during chloroquine treatment in a xenograft model.35 Importantly, we found that primary ETV6-RUNX1-negative BCP-ALL samples were not affected by autophagy inhibition. Autophagy might play an important role in protecting leukemic cells during chemotherapeutic treatment with nutrient-modulating drugs like L-Asparaginase that actively inhibits protein biosynthesis by asparagine depletion, which leads to nutritional deprivation and effective killing of leukemic cells.36 Here, we show that autophagy selectively protects ETV6-RUNX1-positive leukemic cells against L-Asparaginase treatment, whereas this effect is absent in ETV6-RUNX1-negative leukemic cells (Figure 6). This is in contrast with the study of Takashi et al.,35 highlighting the importance of the use of primary human leukemic cells in studies investigating targeted therapy. HCQ-mediated inhibition of autophagy did not sensitize cells to prednisolone or 6-mercaptopurine, two other chemotherapeutics often used in treatment of BCP-ALL. These results highlight the importance of the cellular and molecular context in which autophagy inhibition is embedded and show that caution is warranted before the general introduction of autophagy inhibitors in the treatment of leukemia. The leukemic microenvironment or niche has been shown to protect leukemic cells from elimination by immune responses and chemotherapeutic agents and facilitates the development of drug resistance to classic and targeted chemotherapy.37 Here, we show that MSCs can abrogate the effects of autophagy inhibition in ETV6RUNX1-positive BCP-ALL cells. This highlights the crucial role of the leukemic niche in induction of resistance to chemotherapy, including autophagy inhibition. However, MSC-induced resistance of ETV6-RUNX1-positive cells could still be overcome when adequate concentrations of the autophagy inhibitor HCQ were used (Figure 6). The efficacy of autophagy inhibitors during cancer treatment is currently being examined in clinical trials (reviewed, for example, by White25). Initial results indicate that HCQ treatment is safe and tolerated at high concentrations and might be effective in a subset of patients (reviewed, for example, by Vogl et al.38). In addition, autophagy independent “off-target” effects of chloroquines, resulting in enhanced response to chemotherapy have been reported.39 This strengthens the rationale to use HCQ in clinical practice. However, more specific and potent autophagy inhibitors are currently being developed and preclinical studies with these novel inhibitors (e.g. Lys05) show promising results.40 In addition, the recent determination of the crystal structure of Vps3441 will enable the development of Vps34 inhibitors for clinical use in the near future.42,43 Our observation that the ETV6RUNX1 fusion protein induces Vps34 expression and subsequently autophagy, strongly indicates that Vps34/autophagy inhibitors should be considered in future protocols of ETV6-RUNX1-positive BCP-ALL. 747

R. Polak et al.

Acknowledgments We thank all members of the research laboratory Pediatric Oncology of the Erasmus MC for their help in processing leukemic and mesenchymal stromal cell samples, in particular E. Bindels and B. de Rooij for scientific input and critical dis-

References 1. Pui CH, Evans WE. Drug therapy Treatment of acute lymphoblastic leukemia. N Engl J Med. 2006;354(2):166178. 2. Pui CH, Carroll WL, Meshinchi S, Arceci RJ. Biology, risk stratification, and therapy of pediatric acute leukemias: an update. J Clin Oncol. 2011;29(5):551-565. 3. Robison LL. Late effects of acute lymphoblastic leukemia therapy in patients diagnosed at 0-20 years of age. Hematology Am Soc Hematol Educ Program. 2011;2011:238-242. 4. Goldman JM, Melo JV. Chronic myeloid leukemia--advances in biology and new approaches to treatment. N Engl J Med. 2003;349(15):1451-1464. 5. Golub TR, Barker GF, Bohlander SK, et al. Fusion of the TEL gene on 12p13 to the AML1 gene on 21q22 in acute lymphoblastic leukemia. Proc Natl Acad Sci U S A. 1995;92(11):4917-4921. 6. Zelent A, Greaves M, Enver T. Role of the TEL-AML1 fusion gene in the molecular pathogenesis of childhood acute lymphoblastic leukaemia. Oncogene. 2004;23(24):4275-4283. 7. Loh ML, Goldwasser MA, Silverman LB, et al. Prospective analysis of TEL/AML1-positive patients treated on Dana-Farber Cancer Institute Consortium Protocol 9501. Blood. 2006;107(11):4508-4513. 8. Stams WA, Beverloo HB, den Boer ML, et al. Incidence of additional genetic changes in the TEL and AML1 genes in DCOG and COALL-treated t(12;21)-positive pediatric ALL, and their relation with drug sensitivity and clinical outcome. Leukemia. 2006;20(3):410-416. 9. van Delft FW, Horsley S, Colman S, et al. Clonal origins of relapse in ETV6-RUNX1 acute lymphoblastic leukemia. Blood. 2011; 117(23):6247-6254. 10. Tsuzuki S, Seto M, Greaves M, Enver T. Modeling first-hit functions of the t(12;21) TEL-AML1 translocation in mice. Proc Natl Acad Sci U S A. 2004;101(22):8443-8448. 11. Hong D, Gupta R, Ancliff P, et al. Initiating and cancer-propagating cells in TEL-AML1associated childhood leukemia. Science. 2008;319(5861):336-339. 12. Schindler JW, Van Buren D, Foudi A, et al. TEL-AML1 corrupts hematopoietic stem cells to persist in the bone marrow and initiate leukemia. Cell Stem Cell. 2009; 5(1):43-53. 13. Anderson K, Lutz C, van Delft FW, et al. Genetic variegation of clonal architecture and propagating cells in leukaemia. Nature. 2011;469(7330):356-361. 14. Fuka G, Kantner HP, Grausenburger R, et al. Silencing of ETV6/RUNX1 abrogates PI3K/AKT/mTOR signaling and impairs reconstitution of leukemia in xenografts. Leukemia. 2012;26(5):927-933. 15. Diakos C, Krapf G, Gerner C, et al. RNAimediated silencing of TEL/AML1 reveals a heat-shock protein- and survivin-depen-

748

16.

17.

18.

19.

20.

21.

22.

23.

24. 25. 26.

27.

28.

29.

30.

cussions; The Erasmus Optical Imaging Centre for providing support of CLSM; The Department of Hematology of the Erasmus MC for providing the use of CLSM and Flow Cytometers; The Vlietland Ziekenhuis for collecting and providing cord blood.

dent mechanism for survival. Blood. 2007;109(6):2607-2610. Mangolini M, de Boer J, WalfVorderwulbecke V, Pieters R, den Boer ML, Williams O. STAT3 mediates oncogenic addiction to TEL-AML1 in t(12;21) acute lymphoblastic leukemia. Blood. 2013; 122(4):542-549. Torrano V, Procter J, Cardus P, Greaves M, Ford AM. ETV6-RUNX1 promotes survival of early B lineage progenitor cells via a dysregulated erythropoietin receptor. Blood. 2011;118(18):4910-4918. Gefen N, Binder V, Zaliova M, et al. Hsamir-125b-2 is highly expressed in childhood ETV6/RUNX1 (TEL/AML1) leukemias and confers survival advantage to growth inhibitory signals independent of p53. Leukemia. 2010;24(1):89-96. Palmi C, Fazio G, Savino AM, et al. Cytoskeletal Regulatory Gene Expression and Migratory Properties of B Cell Progenitors are Affected by the ETV6RUNX1 Rearrangement. Mol Cancer Res. 2014;12(12):1796-1806. Ford AM, Palmi C, Bueno C, et al. The TELAML1 leukemia fusion gene dysregulates the TGF-beta pathway in early B lineage progenitor cells. J Clin Invest. 2009; 119(4):826-836. Jaber N, Dou Z, Chen JS, et al. Class III PI3K Vps34 plays an essential role in autophagy and in heart and liver function. Proc Natl Acad Sci U S A. 2012;109(6):2003-2008. Funderburk SF, Wang QJ, Yue Z. The Beclin 1-VPS34 complex--at the crossroads of autophagy and beyond. Trends Cell Biol. 2010;20(6):355-362. Levine B, Klionsky DJ. Development by self-digestion: molecular mechanisms and biological functions of autophagy. Dev Cell. 2004;6(4):463-477. Levine B, Kroemer G. Autophagy in the pathogenesis of disease. Cell. 2008; 132(1):27-42. White E. Deconvoluting the contextdependent role for autophagy in cancer. Nat Rev Cancer. 2012;12(6):401-410. Yeoh EJ, Ross ME, Shurtleff SA, et al. Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling. Cancer Cell. 2002;1(2):133-143. Gandemer V, Rio AG, de Tayrac M, et al. Five distinct biological processes and 14 differentially expressed genes characterize TEL/AML1-positive leukemia. BMC Genomics. 2007;8:385. Andersson A, Olofsson T, Lindgren D, et al. Molecular signatures in childhood acute leukemia and their correlations to expression patterns in normal hematopoietic subpopulations. Proc Natl Acad Sci U S A. 2005;102(52):19069-19074. Fine BM, Stanulla M, Schrappe M, et al. Gene expression patterns associated with recurrent chromosomal translocations in acute lymphoblastic leukemia. Blood. 2004; 103(3):1043-1049. van der Veer A, Waanders E, Pieters R, et al.

31.

32. 33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

Independent prognostic value of BCRABL1-like signature and IKZF1 deletion, but not high CRLF2 expression, in children with B-cell precursor ALL. Blood. 2013; 122(15):2622-2629. Fuka G, Kauer M, Kofler R, Haas OA, Panzer-Grumayer R. The leukemia-specific fusion gene ETV6/RUNX1 perturbs distinct key biological functions primarily by gene repression. PLoS One. 2011;6(10):e26348. Mackenzie AH. Antimalarial drugs for rheumatoid arthritis. Am J Med. 1983;75(6A):48-58. Munster T, Gibbs JP, Shen D, Baethge BA, Botstein GR, Caldwell J, et al. Hydroxychloroquine concentrationresponse relationships in patients with rheumatoid arthritis. Arthritis Rheum. 2002;46(6):1460-1469. Rangwala R, Leone R, Chang YC, et al. Phase I trial of hydroxychloroquine with dose-intense temozolomide in patients with advanced solid tumors and melanoma. Autophagy. 2014;10(8):13691379. Takahashi H, Inoue J, Sakaguchi K, Takagi M, Mizutani S, Inazawa J. Autophagy is required for cell survival under L-asparaginase-induced metabolic stress in acute lymphoblastic leukemia cells. Oncogene. 2017; 36(30):4267-4276. Pieters R, Hunger SP, Boos J, et al. Lasparaginase treatment in acute lymphoblastic leukemia: a focus on Erwinia asparaginase. Cancer. 2011;117(2):238-249. McMillin DW, Negri JM, Mitsiades CS. The role of tumour-stromal interactions in modifying drug response: challenges and opportunities. Nat Rev Drug Discov. 2013; 12(3):217-228. Vogl DT, Stadtmauer EA, Tan KS, et al. Combined autophagy and proteasome inhibition: A phase 1 trial of hydroxychloroquine and bortezomib in patients with relapsed/refractory myeloma. Autophagy. 2014;10(8):1380-1390. Maes H, Kuchnio A, Peric A, et al. Tumor Vessel Normalization by Chloroquine Independent of Autophagy. Cancer Cell. 2014;26(2):190-206. McAfee Q, Zhang Z, Samanta A, et al. Autophagy inhibitor Lys05 has single-agent antitumor activity and reproduces the phenotype of a genetic autophagy deficiency. Proc Natl Acad Sci U S A. 2012; 109(21):8253-8258. Miller S, Tavshanjian B, Oleksy A, et al. Shaping development of autophagy inhibitors with the structure of the lipid kinase Vps34. Science. 2010;327(5973): 1638-1642. Ronan B, Flamand O, Vescovi L, et al. A highly potent and selective Vps34 inhibitor alters vesicle trafficking and autophagy. Nat Chem Biol. 2014;10(12):1013-1019. Dowdle WE, Nyfeler B, Nagel J, et al. Selective VPS34 inhibitor blocks autophagy and uncovers a role for NCOA4 in ferritin degradation and iron homeostasis in vivo. Nat Cell Biol. 2014;16(11):1069-1079.

haematologica | 2019; 104(4)

ARTICLE

Acute Lymphoblastic Leukemia

CD123 expression patterns and selective targeting with a CD123-targeted antibody-drug conjugate (IMGN632) in acute lymphoblastic leukemia

Evgeniya Angelova,1 Charlene Audette2, Yelena Kovtun,2 Naval Daver,3 Sa A. Wang,1 Sherry Pierce,3 Sergej N. Konoplev,1 Haitham Khogeer,1 Jeffrey L. Jorgensen,1 Marina Konopleva,3 Patrick A. Zweidler-McKay,2 L. Jeffrey Medeiros,1 Hagop M. Kantarjian,3 Elias J. Jabbour3 and Joseph D. Khoury1

Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, TX; 2ImmunoGen, Inc., Waltham, MA and 3Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA 1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):749-755

ABSTRACT

he potential of CD123-targeted therapies in acute lymphoblastic leukemia/lymphoma remains largely unexplored. We examined CD123 expression levels in a large cohort of patients with acute lymphoblastic leukemia/lymphoma and assessed the in vitro impact of IMGN632, a conjugate of CD123-binding antibody with a novel DNAalkylating payload. CD123 expression on leukemic blasts was surveyed using multicolor/multiparameter flow cytometry. The in vitro effect of IMGN632 was evaluated on B acute lymphoblastic leukemia/lymphoma cell lines and primary B acute lymphoblastic leukemia/lymphoma blasts. The study cohort (n=213) included 183 patients with B acute lymphoblastic leukemia/lymphoma and 30 with T acute lymphoblastic leukemia/lymphoma. CD123 expression was more prevalent in B acute lymphoblastic leukemia/lymphoma than in T acute lymphoblastic leukemia/lymphoma (164/183, 89.6% versus 13/30, 43.3%; P<0.0001), and within B acute lymphoblastic leukemia/lymphoma CD123 expression was more prevalent in Philadelphia chromosome-positive patients than in Philadelphia chromosome-negative patients (96.6% versus 86.3%; P=0.033). In T acute lymphoblastic leukemia/lymphoma, 12/13 (92.3%) patients with CD123-positive blasts had either early T precursor (ETP) or early nonETP immunophenotype. IMGN632 was highly cytotoxic to B acute lymphoblastic leukemia/lymphoma cell lines, with half maximal inhibitory concentrations (IC50) between 0.6 and 20 pM. In five of eight patientsâ&#x20AC;&#x2122; samples, low pico-molar concentrations of IMGN632 eliminated more than 90% of the B acute lymphoblastic leukemia/lymphoma blast population, sparing normal lymphocytes. In conclusion, CD123 expression is prevalent across acute lymphoblastic leukemia/lymphoma subtypes, and the CD123-targeted antibody-drug conjugate IMGN632 demonstrates promising selective activity in preclinical models of B acute lymphoblastic leukemia/lymphoma.

Introduction Acute lymphoblastic leukemia (ALL) is an aggressive hematologic neoplasm arising from immature lymphoid precursors. In adults, 75% of ALL cases develop from B-cell precursors (B-ALL) and the remainder from T-cell precursors (TALL). B-ALL has been long known for its genetic heterogeneity and includes a subset of cases that harbors the BCR-ABL1 fusion located on the derivative chromosome 22 (Philadelphia chromosome) resulting from t(9;22)(q34.1;q11.2). Philadelphia chromosome (Ph)-positive (Ph+) B-ALL patients and patients with haematologica | 2019; 104(4)

Correspondence: JOSEPH D. KHOURY jkhoury@mdanderson.org Received: August 23, 2018. Accepted: October 22, 2018. Pre-published: October 25, 2018. doi:10.3324/haematol.2018.205252 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/749 ÂŠ2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

749

E. Angelova et al.

Ph-like molecular and cytogenetic signatures are currently treated on frontline protocols with tyrosine kinase inhibitors, which have dramatically improved the outcome of this previously poor prognostic group.1-4 Outcomes of patients with T-ALL are generally inferior to those of their B-ALL counterparts, particularly in adults, and the molecular heterogeneity of T-ALL has only recently been uncovered using high-throughput molecular methods. Early T-cell precursor ALL (ETPALL) is a subset of T-ALL that was identified recently and found to include a sizeable proportion of patients with poor outcomes.5,6 In contrast to children, only 3040% of adults with ALL achieve long-term remission, and survival drops substantially in patients over 60 years of age.7,8 Despite advances in frontline treatment of adult ALL, the prognosis of patients who fail frontline and first salvage therapy is extremely poor9,10 and justifies the need to explore new therapeutic modalities. CD123, the interleukin-3 (IL-3) receptor α-chain, is the primary low-affinity subunit of the IL-3 receptor and promotes high-affinity binding to IL-3 when co-expressed with the β-subunit. IL-3 is mainly produced by T-lymphocytes; it regulates the production of hematopoietic cells by stimulating cell cycle progression, differentiation, and inhibition of apoptosis. Early studies suggested that IL-3 plays a critical role in leukemogenesis through enabling leukemic cells to escape programmed cell death and grow autonomously.11 CD123 was previously reported to be expressed at a low level or to be absent on normal hematopoietic stem cells, but it is expressed at various levels in hematologic malignancies, including hairy cell leukemia,12 acute myeloid leukemia,13 blastic plasmacytoid dendritic cell neoplasm,14-16 and systemic mastocytosis.17 Differential overexpression of CD123 by neoplastic cells and their normal precursors has positioned this cell surface receptor as an attractive target of therapy. The potential of CD123-targeted therapies in ALL remains largely unexplored. There are some data on CD123 expression in B-ALL, but only limited data for TALL.13,15,18 In this report, we present a comprehensive single-institution survey of CD123 expression in adult ALL and assess the correlation between CD123 expression and clinicopathological factors and outcomes. We also describe the in vitro impact of IMGN632, a conjugate of CD123-targeting antibody with a novel DNA-alkylating payload, in ALL cell lines and patients’ samples.

approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center (MDACC) and conducted in accordance with the Declaration of Helsinki.

Antibodies and reagents For the flow cytometry analysis (details below), we used an allophycocyanin-conjugated anti-CD123 (IL-3 receptor α chain) antibody (clone 7G3; BD Pharmingen, BD Biosciences) according to the manufacturer’s recommendation. The humanized anti-CD123 antibody G4723A and a humanized non-targeting control antibody of the same IgG1 isotype and identical Fc sequences, were generated at ImmunoGen. IMGN632 and the control DGN549 antibody drug conjugate were produced via conjugation of DGN549 to the G4723A and the non-targeting IgG1 antibodies, respectively, as described previously.19

Multicolor/multiparameter flow cytometry analysis The blast gate was defined on the basis of CD45dim expression and side-scatter characteristics and quantified as a percentage of total gated events. For analysis of CD123 expression, measurements included mean fluorescence intensity (MFI) on leukemic blasts (adjusted for background fluorescence using negative internal controls) and relative mean fluorescence intensity (RFI) ratio (leukemic blasts versus non-leukemic events). In patients’ samples, positive CD123 expression (CD123+) was defined as expression in ≥20% of leukemic blasts using MFI by comparison to background fluorescence and fluorescence on nonleukemic gated events, respectively. Additional details are provided in the Online Supplement.

Cytogenetics and molecular diagnostics Conventional cytogenetics, fluorescence in situ hybridization, polymerase chain reaction-based molecular diagnostics, and next-generation sequencing-based mutation profiling were performed on bone marrow aspirate specimens as described previously.20-22

Cell lines B-ALL cell lines CRF-SB and JM-1 (from American Type Culture Collection) and KOPN-8, SEM, 380, TOM-1, and SD-1 (from Deutsche Sammlung von Mikroorganismen und Zellkulturen) were procured between 2000 and 2015; they were characterized by the respective vendors using DNA profiling. Cytotoxicity was assessed using either a water-soluble tetrazolium salt 8 (WST-8)-based cell viability assay (Dojindo Molecular Technologies) as described previously23 or the alamarBlue Cell Viability Reagent (Invitrogen). Further details are given in the Online Supplement.

Methods Study group A total of 213 consecutive patients (183 with B-ALL, 30 with T-ALL) were identified and included in the study group. B-ALL patients were further subdivided into Ph+ (121/124 treatmentnaïve) and Ph-negative (Ph–) (51/59 treatment-naïve) subsets based on cytogenetic, fluorescence in situ hybridization, and/or molecular detection of t(9;22)(q34.1;q11.2)/BCR-ABL1. In the TALL group, 19 patients were treatment-naïve and 11 presented with relapsed/refractory disease after prior treatment. T-ALL patients were subdivided into immunophenotypic subsets based on the expression of CD1a and sCD3: the subsets were early (CD1a−, sCD3−), thymic (CD1a+, sCD3−), or mature (CD1a−, sCD3+) T-ALL. Patients with ETP-ALL were defined as described previously.6 Additional details regarding the study group are provided in the Online Supplement. This study was 750

In vitro evaluation of primary B-cell acute lymphoblastic leukemia samples Bone marrow mononuclear cells from 11 newly diagnosed and 10 relapsed/refractory B-ALL patients were obtained from MDACC or ConversantBio. The number of CD123 antibodybinding sites per cell (ABC) was quantified by the BD QuantiBRITE™ Fluorescence Quantitation Kit (BD Biosciences) using G4723A conjugated to R-phycoerythrin at a 1:1 ratio, as already described.24 Cell proliferation for samples treated with IMGN632 was assessed using the CellTiter-Glo® (Promega) or Cell Trace™ Violet stain (Invitrogen) techniques. Additional details are provided in the Online Supplement.

Statistical analysis The statistical analysis methodology is described in the Online Supplement. haematologica | 2019; 104(4)

Targeted CD123 therapy in ALL

Results

CD123 expression by multicolor/multiparameter flow cytometry in patients’ samples

Patients’ characteristics

The median number of blasts detected by multicolor/multiparameter flow cytometry across the entire study group was 76.5% (range, 7.2-98.0%) and did not differ between the three groups (P=0.872). When the entire group was considered, the median percentage of CD45dim blasts expressing CD123 was 64.7% (range, 0.399.9%), and the median RFI was 11.0 (range, 0.0-420.5). CD123 positivity (>20% blasts) was seen in 177 (83%) of all ALL patients. CD123+ blasts were positive for CD34 in 58/59 (98.3%) Ph+ B-ALL, in 95/124 (76.6%) Ph– B-ALL and in 13/30 (43.3%) T-ALL patients. We identified a strong correlation between CD123 and CD34 expression on leukemic blasts across the entire study group (r=0.483; P<0.0001).

The study group included a total of 213 patients, 131 men and 82 women, with a median age of 40.5 years (range, 1.3-88.3 years) at diagnosis; 25 (11.7%) patients were under 18 years of age (22 with B-ALL and 3 with TALL). For the purposes of this study, patients were divided into three clinically relevant groups: BCR-ABL1-positive (Ph+) B-ALL, BCR-ABL1-negative (Ph–) B-ALL, and TALL. Most Ph+ patients (42/59; 71.2%) expressed the e1a2 BCR-ABL1 fusion transcript, followed by the b2a2 and b3a2 types. Detailed comparisons of the clinical and laboratory features of the CD123+ and CD123– patients in the three groups are provided in Table 1.

Table 1. Clinical and laboratory features of patients in the study group (n=213) categorized by CD123 expression status.

B-ALL, BCR-ABL1-positive (n=59) CD123+ CD123– P-value Patients, n (%) 57 (96.6) 2 (3.4) Age in years, median (range) 54.3 (16.2-80.2)51.2 (37.1-65.3) Sex, n (%) Male 31 (96.9) 1 (3.1) Female 26 (96.3) 1 (3.7) CNS involvement, n (%) 7 (100) 0 (0) Allogeneic SCT, n Peripheral blood parameters White blood count (x109/L), 20 (0.7-381.6) 7 (1.8-12.2) median (range) Hemoglobin concentration 9.7 (7.1-15.8) 10.4 (8.2-12.6) (g/dL), median (range) Platelet count (x109/L), median (range) 44 (3-257) 190 (120-260) Bone marrow parameters BM blast percentage, 82 (15-98) 69 (66-72) median (range) CD123 MFI ratio, 22.4 (1.7 – 203.2) 1.25 (1.0-1.5) median (range) Immunophenotypic groups* Early T precursor NA NA Early NA NA Thymic NA NA Mature NA NA Cytogenetic groups, n (%) Hypodiploid NA NA Hyperdiploid NA NA Flow cytometry CD45 blast gate, % (range) 77.1 (7.2-96.1)73.5 (64.5-82.5) CD123+ blasts, % (range) 91.5 (31.6-99.7)13.8 (10.7-16.9) CD123 RFI, median 21.3 1.3 Treatment/response groups Hyper-CVAD, n 40 1 CR/CRp, n 53 2 Positive MRD, n (%) NA NA Dead, n (%) 15 (26.3) 1 (50) Leukemia-free survival in months, median (range) 16 (0-54) 15 (7-22) Follow up duration in months, median (range) 29 (1-74) 22 (8-36)

NA 0.802

B-ALL, BCR-ABL1-negative (n=124) CD123+ CD123– P-value 107 (86.3) 17 (13.7) 38.1 (1.3-79.9) 56.0 (10.3-88.3)

NA 0.229

CD123+

T-ALL (n=30) CD123– P-value

13 (43.3) 17 (56.7) 28.1 (9.8-65.3) 30.9 (13.8-58.1)

0.710 NA 0.775

61 (57.0) 46 (43.0) 9 (8.4) 15

10 (58.8) 7 (41.2) 1 (5.9) 1

0.553 NA 0.589

0.224

6.1 (0.2-612.3)

5.9 (1.0-33.2)

0.435

9.4 (1.0-137.3) 10.9 (1.8-309.2)

0.983

0.917

9.3 (4.2-13.8)

9.4 (6.3-13.2)

0.802

10.7 (7.0-14.9) 11.7 (6.9-16.0)

0.615

0.040

42 (4-334)

38 (13-210)

0.802

44 (13-326)

58 (6-332)

0.950

0.093

84 (21-99)

90 (62-97)

0.025

84 (24-96)

75 (4-94)

0.315

16.9 (0.1-420)

1.1 (0.3-3.4)

6.3 (1.9-101)

1.0 (0-6.0)

NA NA NA NA

5 (38.5) 7 (53.8) 0 (0) 1 (7.7)

1 (5.9) 3 (17.6) 12 (70.6) 1 (5.9)

0.001

NA NA

53/99 (53.5) 12/99 (12.1)

11/17 (64.7) 2/17 (11.8)

0.655

6/11 (54.5) 1/11 (9.1)

14/16 (87.5) 2/16 (12.5)

0.033

0.933 0.017 0.021

74.6 (11.7-97.4) 84.0 (15.3-94.1) 82.4 (20.6-99.9) 7.5 (0.8-19.5) 16.9 1.1

0.307 <0.0001 <0.0001

13 (100) 0 (0) 1/11 (9.1) 6

15 (88.2) 2 (11.8) 5 (29.4) 6

NA 1.000 0.313 NA 0.214

86.3 (10.6-98.0) 80.7 (17.3-98.0) 0.660 66.0 (22.7-93.5) 1.0 (0.3-19.7) <0.0001 6.3 1.0 <0.0001

NA NA NA 0.472

49 76 25/73 (34.2) 35 (32.7)

8 13 3/13 (23.1) 5 (29.4)

NA NA 0.328 0.513

8 6 5/6 (83.3) 6 (46.2)

8 10 4/9 (44.4) 7 (41.2)

NA NA 0.168 0.538

0.840

11 (0-53)

20 (3-42)

0.186

6 (0-13)

10 (2-34)

0.135

0.675

22 (1-99)

22 (1-54)

0.876

14 (1-50)

21 (1-44)

0.304

ALL: acute lymphoblastic leukemia/lymphoma; BM: bone marrow; CNS: central nervous system; MFI: mean fluorescence intensity; RFI: relative fluorescence intensity; Hyper-CVAD: hyperfractionated cyclophosphamide, vincristine, doxorubicin and dexamethasone; CR/CRp: clinical remission/without platelet recovery; MRD: minimal residual disease (by flow standardized multiparameter flow cytometry analysis). NA: not-applicable, SCT: stem cell transplantation; P-values refer to comparisons of CD123+ and CD123– patients within a given clinical subgroup.

haematologica | 2019; 104(4)

751

E. Angelova et al.

The frequency of CD123 expression was higher in BALL than in T-ALL (164/183, 89.6% versus 13/30, 43.3%; P<0.0001), and within B-ALL the frequency of CD123 expression was higher in Ph+ versus Ph– B-ALL patients (96.6% versus 86.3%; P=0.033). (Figure 1A) The intensity of CD123 expression across groups of patients showed similar patterns. Namely, the median CD123 RFI was higher in B-ALL compared to T-ALL (14.45 versus 2.35; P<0.001), and within B-ALL it tended to be higher in Ph+ compared with Ph– patients, although the difference did not attain statistical significance (20.85 versus 11.8; P=0.08) (Figure 1B). We analyzed our findings using various CD123 MFI-based cutoffs of 20%, 30% 50%, as well as RFI ≥10. All conclusions remained valid at cutoff levels of 20% and 30% as well as RFI ≥10 but not at the 50% cutoff level (data not shown). In the T-ALL group (40% thymic; 33% early non-ETP; 20% ETP; 7% mature), the median CD123+ blast proportion was 17.95% (range, 0.5-93%) with a median RFI of 2.35 (range, 0-101). We identified a correlation (P=0.001) between T-ALL immunophenotypic subgroups and CD123 expression status; namely, 12/13 (92.3%) patients with CD123+ T-ALL had either ETP or early non-ETP immunophenotype.

age were excluded. Notably, the impact of CD123 expression on relapse-free survival correlated with the intensity of CD123 expression in adult T-ALL using RFI ≥10 vs. <10 as a cutoff (P=0.015) (Online Supplementary Figure S2). For patients with B-ALL, CD123 expression showed no correlation with relapse-free survival (Online Supplementary Figure S3). We performed multivariate analysis using Cox regression modeling (Wald backward stepwise method) to evaluate the impact of CD123 expression status on relapse-free survival alongside other prognostic variables that included: minimal residual disease status at complete remission, central nervous system involvement, and age (patients <18 years excluded). Analyses were conducted in each of the clinical subgroups separately. In the Ph+ B-ALL group only minimal residual disease status at complete remission was independently associated with relapse-free survival (P=0.019), and in the Ph– B-ALL group only central nervous system involvement was independently associated with relapse-free survival (P=0.005). In the T-ALL group, none of the variables was associated with leukemia-free survival.

Correlation between CD123 expression and clinical parameters

IMGN632 is a conjugate of the CD123-binding antibody G4723A with a novel DNA-alkylating IGN payload, DGN549. The numbers of binding sites for the antibody component of IMGN632 were quantified on leukemic blasts and on lymphocytes from 21 patients with newly diagnosed or relapsed/refractory B-ALL (Online Supplementary Figure S4). As described above, the blast population was identified based on the CD45dim/SSClow/med/CD19+/CD34+ profile, while the lymphocyte population was identified based on CD45bright/SSClow characteristics. Of the 21 samples analyzed, 20 expressed the CD123 antigen, with a median number of ABC of 1,085 on leukemic blasts and 57 on lymphocytes. There was no appreciable difference in the number of ABC between blasts from newly diagnosed and relapsed/ refractory patients. The cytotoxicity of IMGN632 was assessed in a panel of

CD123 expression status did not correlate with minimal residual disease status at the end of induction. The median follow-up duration for the entire group was 23 months (range, 1-79 months) and was comparable for patients with CD123+ and CD123– ALL. There was no difference in Performance Status between CD123+ and CD123– ALL patients in any of the clinical subgroups (data not shown). At last follow up, 69 (32.4%) patients were dead. There was no difference in overall survival between patients with CD123+ and CD123– ALL within any of the subgroups. However, for T-ALL patients, the relapse-free survival rate was more favorable in those with CD123– disease compared with those with CD123+ disease (P=0.03) (Online Supplementary Figure S1). Similar results were obtained when patients under 18 years of

In vitro anti-leukemia activity of IMGN632 in acute lymphoblastic leukemia

Figure 1. Expression of CD123 on leukemic blasts in B and T acute lymphoblastic leukemia/lymphoma by multiparameter flow cytometry immunophenotyping. (A) Percentage of leukemic blasts with CD123 expression based on mean fluorescence intensity relative to background in each of the groups of patients. CD123 expression was significantly different between B-acute lymphoblastic leukemia/lymphoma (ALL) and T-ALL (P<0.0001; Mann-Whitney U test). In addition, CD123 expression was significantly different between Philadelphia chromosome (Ph)-positive and Ph-negative patients (P=0.033; Mann-Whitney U test). (Lines: median, 25th-75th percentiles). (B) Relative fluorescence intensity (RFI) on leukemic blasts compared to non-leukemic gated events (lines: mean ± standard deviation). Ph: Philadelphia chromosome and/or BCR/ABL1 fusion status.

752

haematologica | 2019; 104(4)

Targeted CD123 therapy in ALL

B-ALL cell lines that expressed CD123 at levels similar to those of blasts from B-ALL patients (ABC values: 4573,741) and encompassed genetic variants associated with a poor prognosis (Table 2). IMGN632 was highly cytotoxic to B-ALL cell lines with half maximal inhibitory concentrations (IC50) between 0.6 and 20 pM. The conjugate cytotoxicity was CD123-specific, as dosing with a non-targeting conjugate (Ab-DGN549) prepared with a similar linker and payload reduced the potency more than 1000-fold. IMGN632 potency was evaluated in primary samples from eight patients with B-ALL. The blast cells in each of the untreated samples proliferated in culture, as manifested by a dramatic decrease in the cell trace violet fluorescent signal, while only a small shift in fluorescence (likely due to nonspecific dissociation of the dye) was observed in the lymphocyte population, suggesting lack of proliferation (Figure 2A). IMGN632 eliminated more than 90% of the BALL blast population in five of eight samples, including those from newly diagnosed as well as relapsed/refractory patients, at low pico-molar concentrations (Figures 2B,C). Normal lymphocytes were not affected by IMGN632 at 100-fold higher concentrations.

Discussion In this study, we demonstrate pervasive CD123 expression in a large cohort of ALL patients. In B-ALL, previously reported to be often CD123+,13,15,18 our findings demonstrate that CD123 expression translates into susceptibility to targeting CD123 via the antibody-drug conjugate IMGN632 in vitro and ex vivo. Coupled with previously published data on the differential overexpression of CD123 on the surface of acute myeloid leukemia blasts, but not on normal hematopoietic stem cells,25 these findings provide a plausible basis for the exploration of anti-CD123 therapy in B-ALL patients in whom frontline treatment fails. Furthermore, in T-ALL our findings suggest that CD123 expression might be associated with an early precursor immunophenotype, including ETP-ALL. Since the latter is a high-risk subset, targeting CD123 in these patients might be of potential benefit if the association is confirmed in a larger group of ETP-ALL patients. Previously, the CD123-targeting conjugate IMGN632 was reported to have high potency in pre-clinical models of acute myeloid leukemia.24 In this study we demonstrated the efficacy of IMGN632 in pre-clinical B-ALL models, suggesting that IMGN632 is a promising anti-

leukemia agent in CD123-expressing ALL. In addition to its promising anti-leukemia activity, CD123 offers an alternative to CD19-targeting therapeutic approaches to circumvent CD19 antigen-loss relapses and/or elimination of CD19– stem cells.26 B-ALL is a heterogeneous disease that includes nine genetic subgroups recognized in the World Health Organization classification.27 Data on correlations between CD123 expression and B-ALL subgroups are limited. Djokic et al. showed that B-ALL cases with a hyperdiploid karyotype have significantly higher rates of CD123 expression compared to other subgroups such as BCR/ABL1, MLL, ETV6/RUNX1 and diploid karyotype.18 However, their study included mostly pediatric patients, and the BCR/ABL1 subgroup formed a small proportion of their cohort (2/81 pediatric patients; 3/13 adult patients). In this study, we identified a significant correlation between CD123 expression and the presence of the BCR/ABL1 fusion, with a higher percentage of CD123+ events and a higher intensity in comparison to the Ph– B-ALL and T-ALL groups. The association between the BCR/ABL1 fusion and CD123 has also been identified in chronic myeloid leukemia, in which targeted CD123 inhibition appears to deplete chronic myeloid leukemia progenitor and stem cells.28,29 The findings in this study suggest a close correlation between BCR/ABL1 and CD123, this time in the context of B-ALL. Whether the correlation between Ph– B-ALL and CD123 is related to an underlying Ph-like biology remains to be determined and is being pursued in an ongoing separate study. There are limited and somewhat controversial data on CD123 expression in T-ALL. Several prior studies have suggested that CD123 is not or only scarcely expressed in T-ALL, although most of these studies had relatively small sample sizes.13,15,18 Two other studies reported expression of CD123 by T-ALL blasts.30,31 In a study by Lhermitte et al. CD123 expression was detected in 16% of T-ALL; notably, CD123 positivity was reported to be less frequent in children than in adults (9% versus 23%).31 Another study of a large cohort of T-ALL patients by Du et al. found a slightly higher frequency of CD123 expression and noted an age-dependent difference, further suggesting a lower frequency of CD123 expression in pediatric patients (27% versus 42%).30 A crucial difference precluding direct comparison of the results of those studies with the current study is lack of data on CD123 intensity (RFI) in the earlier studies. ETP-ALL comprises a unique subset of T-ALL previously belonging to the subset of early T-ALL. ETP ALL

Table 2. In vitro cytotoxicity of IMGN632 in B-acute lymphoblastic leukemia/lymphoma cell lines.

Cell line

Genetic variant

IMGN632 (IC50 pM)

Ab-DGN549 (IC50 pM)

CCRF-SB JM1 KOPN-8 SEM 380 TOM-1 SD-1

EBNA-positive

20 10 6 3 2 2 0.6

20,000 20,000 13,000 20,000 7,000 20,000 3,000

t(11;19)(q23;p13)/KMT2A-MLLT1 t(4;11)/KMT2A-AFF1 t(8;14)/MYC-IGH and t(14;18)/IGH-BCL2 EBNA-negative t(9;22)(q34.1;q11.2)/BCR-ABL1 fusion gene (e1-a2) t(9;22)(q34.1;q11.2)/BCR-ABL1 fusion gene (e1-a2)

IC50: half maximal inhibitory concentration; pM: pico-males.

haematologica | 2019; 104(4)

753

E. Angelova et al.

has been reported in 10-13% of childhood T-ALL and in 5-17% of adults.5,6 Jain et al. reported significantly poorer outcomes in patients with ETP-ALL, with lower complete remission and 5-year overall survival rates, compared with other T-ALL patients.6 In the present study, patients with ETP-ALL represented 20% of the T-ALL cohort. We postulate that this high rate of ETP-ALL is likely due, in part, to higher detection rates and to a referral bias to our institution in view of the higher relapse rates of patients with this subtype. Only limited data, using different stratification approaches, are available from other studies on the correlation between CD123 expression and immunophenotype in T-ALL. Lhermitte et al. stratified patients with T-ALL on the basis of a cytoplasmic T-cell receptor β-negative phenotype and patterns of TRD, TRG and TRB configurations,31 as opposed to commonly used multicolor/multiparame-

ter flow cytometry immunophenotyping criteria according to the World Health Organization classification. These authors found that CD123 expression was more common in “immature” T-ALL, being detectable in 68% of adults and 36% of children; in contrast, they identified CD123 expression in 5% of “non-immature” T-ALL cases. Du et al. classified T-ALL patients into “early Tprecursor” (CD7+), “T-precursor” (CD2+ and/or CD5+ and/or CD8+), and “mature” (CD3+),30 according to the European Group for the Immunological Characterization of Leukemias (EGIL) criteria. In their study, 83% of cases with an “early T-precursor” immunophenotype were positive for CD123. In keeping with those previous observations, our ETP-ALL group – as well as early-nonETP – showed the highest frequency of CD123 expression compared with the thymic and mature immunophenotypes.

Figure 2. In vitro potency of IMGN632 on leukemic blasts cells and lymphocytes from patients with B-acute lymphoblastic leukemia/lymphoma. (A) The leukemic blast population, but not lymphocyte population, proliferates in culture. (B) IMGN632 is cytotoxic to leukemic blasts from five of eight patients with B-acute lymphoblastic leukemia/lymphoma (B-ALL), while sparing lymphocytes. (C) IMGN632 eliminates blasts, but not lymphocytes in a specimen (sample H) from a patient with relapsed-refractory B-ALL.

754

haematologica | 2019; 104(4)

Targeted CD123 therapy in ALL

Although a higher number of CD123+ leukemic stem cells was shown to have a negative impact on leukemiafree and overall survival in patients with acute myeloid leukemia,25,32 information on the prognostic impact of CD123 expression in ALL has been limited and unclear. In a large study of pediatric ALL, high-level CD123 expression was most common in patients with a hyperdiploid karyotype, a favorable prognostic marker.18 We did not identify such a correlation in this study group of mostly adult patients. In T-ALL patients, CD123 expression was not found to be prognostic of outcome after first induction treatment.30 In our study group, we did not identify a cor-

References 1. Roberts KG, Gu Z, Payne-Turner D, et al. High frequency and poor outcome of Philadelphia chromosome-like acute lymphoblastic leukemia in adults. J Clin Oncol. 2017;35(4):394-401. 2. Jain N, Roberts KG, Jabbour E, et al. Ph-like acute lymphoblastic leukemia: a high-risk subtype in adults. Blood. 2017;129(5):572581. 3. Roberts KG, Li Y, Payne-Turner D, et al. Targetable kinase-activating lesions in Phlike acute lymphoblastic leukemia. N Engl J Med. 2014;371(11):1005-1015. 4. El Fakih R, Jabbour E, Ravandi F, et al. Current paradigms in the management of Philadelphia chromosome positive acute lymphoblastic leukemia in adults. Am J Hematol. 2018;93(2):286-295. 5. Coustan-Smith E, Mullighan CG, Onciu M, et al. Early T-cell precursor leukaemia: a subtype of very high-risk acute lymphoblastic leukaemia. Lancet Oncol. 2009;10(2):147156. 6. Jain N, Lamb AV, O'Brien S, et al. Early T-cell precursor acute lymphoblastic leukemia/lymphoma (ETP-ALL/LBL) in adolescents and adults: a high-risk subtype. Blood. 2016;127(15):1863-1869. 7. Sive JI, Buck G, Fielding A, et al. Outcomes in older adults with acute lymphoblastic leukaemia (ALL): results from the international MRC UKALL XII/ECOG2993 trial. Br J Haematol. 2012;157(4):463-471. 8. Jabbour E, O'Brien S, Konopleva M, Kantarjian H. New insights into the pathophysiology and therapy of adult acute lymphoblastic leukemia. Cancer. 2015;121(15): 2517-2528. 9. O'Brien S, Thomas D, Ravandi F, et al. Outcome of adults with acute lymphocytic leukemia after second salvage therapy. Cancer. 2008;113(11):3186-3191. 10. Gokbuget N, Stanze D, Beck J, et al. Outcome of relapsed adult lymphoblastic leukemia depends on response to salvage chemotherapy, prognostic factors, and performance of stem cell transplantation. Blood. 2012;120(10):2032-2041. 11. Lotem J, Cragoe EJ, Jr., Sachs L. Rescue from programmed cell death in leukemic and normal myeloid cells. Blood. 1991;78(4):953960.

haematologica | 2019; 104(4)

relation between CD123 prevalence and leukemia-free survival or overall survival in B-ALL, but we did observe a significant correlation between the intensity of CD123 expression and leukemia-free survival. We acknowledge that this finding in a limited group of patients warrants further validation in a larger cohort. In conclusion, data from this study show that CD123 is widely expressed in B-ALL and to a lesser degree in TALL, and they confirm its potential utility as a therapeutic target. IMGN632, an antibody-drug conjugate which targets CD123, has promising preclinical activity against B-ALL.

12. Del Giudice I, Matutes E, Morilla R, et al. The diagnostic value of CD123 in B-cell disorders with hairy or villous lymphocytes. Haematologica. 2004;89(3):303-308. 13. Munoz L, Nomdedeu JF, Lopez O, et al. Interleukin-3 receptor alpha chain (CD123) is widely expressed in hematologic malignancies. Haematologica. 2001;86(12):12611269. 14. Angelot-Delettre F, Roggy A, Frankel AE, et al. In vivo and in vitro sensitivity of blastic plasmacytoid dendritic cell neoplasm to SL401, an interleukin-3 receptor targeted biologic agent. Haematologica. 2015;100(2): 223-230. 15. Testa U, Pelosi E, Frankel A. CD 123 is a membrane biomarker and a therapeutic target in hematologic malignancies. Biomark Res. 2014;2(1):4. 16. Alayed K, Patel KP, Konoplev S, et al. TET2 mutations, myelodysplastic features, and a distinct immunoprofile characterize blastic plasmacytoid dendritic cell neoplasm in the bone marrow. Am J Hematol. 2013;88(12): 1055-1061. 17. Pardanani A, Reichard KK, Zblewski D, et al. CD123 immunostaining patterns in systemic mastocytosis: differential expression in disease subgroups and potential prognostic value. Leukemia. 2016;30(4):914-918. 18. Djokic M, Bjorklund E, Blennow E, Mazur J, Soderhall S, Porwit A. Overexpression of CD123 correlates with the hyperdiploid genotype in acute lymphoblastic leukemia. Haematologica. 2009;94(7):1016-1019. 19. Miller ML, Fishkin NE, Li W, et al. A new class of antibody-drug conjugates with potent DNA alkylating activity. Mol Cancer Ther. 2016;15(8):1870-1878. 20. Zhang L, Singh RR, Patel KP, et al. BRAF kinase domain mutations are present in a subset of chronic myelomonocytic leukemia with wild-type RAS. Am J Hematol. 2014;89(5):499-504. 21. Khoury JD, Sen F, Abruzzo LV, Hayes K, Glassman A, Medeiros LJ. Cytogenetic findings in blastoid mantle cell lymphoma. Hum Pathol. 2003;34(10):1022-1029. 22. Salem A, Loghavi S, Tang G, et al. Myeloid neoplasms with concurrent BCR-ABL1 and CBFB rearrangements: a series of 10 cases of a clinically aggressive neoplasm. Am J Hematol. 2017;92(6):520-528. 23. Kovtun YV, Audette CA, Mayo MF, et al.

24.

25.

26.

27.

28.

29.

30.

31.

32.

Antibody-maytansinoid conjugates designed to bypass multidrug resistance. Cancer Res. 2010;70(6):2528-2537. Kovtun Y, Jones GE, Adams S, et al. A CD123-targeting antibody-drug conjugate, IMGN632, designed to eradicate AML while sparing normal bone marrow cells. Blood Adv. 2018;2(8):848-858. Testa U, Riccioni R, Militi S, et al. Elevated expression of IL-3Ralpha in acute myelogenous leukemia is associated with enhanced blast proliferation, increased cellularity, and poor prognosis. Blood. 2002;100(8):29802988. Ruella M, Barrett DM, Kenderian SS, et al. Dual CD19 and CD123 targeting prevents antigen-loss relapses after CD19-directed immunotherapies. J Clin Invest. 2016;126 (10):3814-3826. Borowitz MJ, Chan JKC, Downing JR, Le Beau MM, Arber DA, Bene MC. Precursor lymphoid neoplasms. . In: Swerdlow SH, Campo E, Harris NL, et al., eds. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues Lyon: IARC, 2017:200-213. Nievergall E, Ramshaw HS, Yong AS, et al. Monoclonal antibody targeting of IL-3 receptor alpha with CSL362 effectively depletes CML progenitor and stem cells. Blood. 2014;123(8):1218-1228. Frolova O, Benito J, Brooks C, et al. SL-401 and SL-501, targeted therapeutics directed at the interleukin-3 receptor, inhibit the growth of leukaemic cells and stem cells in advanced phase chronic myeloid leukaemia. Br J Haematol. 2014;166(6):862-874. Du W, Li J, Liu W, et al. Interleukin-3 receptor alpha chain (CD123) is preferentially expressed in immature T-ALL and may not associate with outcomes of chemotherapy. Tumour Biol. 2016;37(3):3817-3821. Lhermitte L, de Labarthe A, Dupret C, et al. Most immature T-ALLs express Ra-IL3 (CD123): possible target for DT-IL3 therapy. Leukemia. 2006;20(10):1908-1910. 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.

755

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):756-765

Acute Lymphoblastic Leukemia

Jeanette Greiner,1 Martin Schrappe,2 Alexander Claviez,2 Martin Zimmermann,3 Charlotte Niemeyer, 4Reinhard Kolb,5 Wolfgang Eberl,6 Frank Berthold,7 Eva Bergsträsser,8 Astrid Gnekow,9 Elisabeth Lassay,10 Peter Vorwerk,11 Melchior Lauten,12 Axel Sauerbrey,13 Johannes Rischewski,14 Andreas Beilken,3 Günter Henze,15 Wolfgang Korte16* and Anja Möricke2* for the THROMBOTECT Study Investigators†

Children’s Hospital of Eastern Switzerland, Hematology and Oncology Department, St. Gallen, Switzerland; 2Department of Pediatrics, Christian-Albrechts-University Kiel and University Medical Center Schleswig-Holstein, Kiel, Germany; 3Department of Pediatric Hematology and Oncology, Hannover Medical School, Germany; 4Department of Pediatrics and Adolescent Medicine, Division of Pediatric Hematology and Oncology, Medical Center - Faculty of Medicine, University of Freiburg, Germany; 5Department of Pediatrics, Zentrum für Kinder- und Jugendmedizin, Klinikum Oldenburg GmbH, Germany; 6Institute for Clinical Transfusion Medicine and Children’s Hospital, Klinikum Braunschweig GmbH, Germany; 7 Department of Pediatric Hematology and Oncology, Children's Hospital, University of Cologne, Germany; 8Department of Pediatric Oncology, University Children's Hospital, Zurich, Switzerland; 9Hospital for Children and Adolescents, Klinikum Augsburg, Germany; 10 Division of Pediatric Hematology, Oncology and Stem Cell Transplantation, Medical Faculty, RWTH Aachen University, Germany; 11Pediatric Oncology, Otto von Guericke University Children’s Hospital, Magdeburg, Germany; 12University Hospital SchleswigHolstein, Department of Pediatrics, University of Lübeck, Germany; 13HELIOS Children's Hospital GmbH, Erfurt, Germany; 14Department of Oncology/Hematology, Children's Hospital, Cantonal Hospital Lucerne, Switzerland; 15Department of Pediatric Oncology/Hematology, Charité Universitätsmedizin Berlin, Germany and 16Center for Laboratory Medicine and Hemostasis and Hemophilia Center, St. Gallen, Switzerland 1

Correspondence: JEANETTE GREINER jeanette.greiner@kispisg.ch Received: March 29, 2018. Accepted: September 27, 2018. Pre-published: September 27, 2018. doi:10.3324/haematol.2018.194175 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/756 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

756

*WK and AM share last authorship

†A complete list of the THROMBOTECT study investigators is provided in the Online Supplementary Appendix

ABSTRACT

hromboembolism is a serious complication of induction therapy for childhood acute lymphoblastic leukemia. We prospectively compared the efficacy and safety of antithrombotic interventions in the consecutive leukemia trials ALL-BFM 2000 and AIEOP-BFM ALL 2009. Patients with newly diagnosed acute lymphoblastic leukemia (n=949, age 1 to 18 years) were randomized to receive low-dose unfractionated heparin, prophylactic low molecular weight heparin (enoxaparin) or activity-adapted antithrombin throughout induction therapy. The primary objective of the study was to determine whether enoxaparin or antithrombin reduces the incidence of thromboembolism as compared to unfractionated heparin. The principal safety outcome was hemorrhage; leukemia outcome was a secondary endpoint. Thromboembolism occurred in 42 patients (4.4%). Patients assigned to unfractionated heparin had a higher risk of thromboembolism (8.0%) compared with those randomized to enoxaparin (3.5%; P=0.011) or antithrombin (1.9%; P<0.001). The proportion of patients who refused antithrombotic treatment as allocated was 3% in the unfractionated heparin or antithrombin arms, and 33% in the enoxaparin arm. Major hemorrhage occurred in eight patients (no differences between the groups). The 5-year event-free survival was 80.9±2.2% among patients haematologica | 2019; 104(4)

Thromboembolism and thromboprophylaxis in pediatric ALL

assigned to antithrombin compared to 85.9±2.0% in the unfractionated heparin group (P=0.06), and 86.2±2.0% in the enoxaparin group (P=0.10). In conclusion, prophylactic use of antithrombin or enoxaparin significantly reduced thromboembolism. Despite the considerable number of patients rejecting the assigned treatment with subcutaneous injections, the result remains unambiguous. Thromboprophylaxis - for the present time primarily with enoxaparin - can be recommended for children and adolescents with acute lymphoblastic leukemia during induction therapy. Whether and how antithrombin may affect leukemia outcome remains to be determined.

Introduction Thromboembolism is a serious complication of glucocorticoid and E. coli asparaginase-containing induction therapy for childhood acute lymphoblastic leukemia (ALL). Reported incidences vary between 1% and 37%, depending on the study design and definition of thrombosis, as well as diagnostic, supportive and therapeutic methods.1-6 Acquired antithrombin deficiency as a result of asparaginase-induced asparagine depletion is considered to be a crucial mechanism for the development of thromboembolism during ALL induction therapy. The presence of a central venous catheter (CVC) seems to be an additional – at least local – risk factor for thromboembolism as a significant proportion of thromboembolic events during ALL treatment is related to an indwelling CVC. Furthermore, the risk of thromboembolism has been shown to be associated with CVC location and insertion technique.1,5,7-12 Published data also provide good evidence that adolescent age is an important risk factor for thromboembolism whereas the additional impact of inherited thrombophilia in the context of childhood ALL treatment is controversial.5,13-16 Sufficiently powered randomized trials on thromboprophylaxis in children during ALL induction therapy have not been available,16-23 and evidence for the benefit of specific thromboprophylactic measures has therefore been lacking so far. In the absence of valid medical standards of care regarding thromboprophylaxis and the use of a CVC during ALL induction, various different approaches existed in the pediatric cancer centers in Switzerland and Germany in the early 2000s, each based on individual experiences and institutional standards. This unsatisfactory situation gave the impetus to initiate the THROMBOTECT trial, a prospective randomized study to evaluate the efficacy and safety of antithrombotic prophylaxis in children treated for ALL. As drug administration through an indwelling CVC provides significant gain in comfort for the patients and increases the safety of therapy with tissue-toxic agents, the THROMBOTECT study was initially designed to include patients with implanted CVC from the initiation of the induction phase and was only later on also opened for patients without CVC. Two mechanisms of action to prevent thromboembolism were utilized in the two interventional arms of the trial: inhibition of thrombin through inactivation of coagulation factor X by treatment with the low molecular weight heparin enoxaparin (ClexaneTM) and replacement of antithrombin by the plasma-derived antithrombin preparation KyberninTM to compensate for asparaginase-related aquired antithrombin deficiency. Being aware of the published data of NowakGöttl et al., which reported an almost 50% incidence of haematologica | 2019; 104(4)

thromboembolism among ALL patients with a prothrombotic defect, and considering the additional risk factor of an indwelling CVC, a control arm without any intervention appeared difficult to justify.15 The third arm therefore included continuous infusion of low-dose unfractionated heparin (UFH) while the CVC was in use, with the aim of preventing local clot formation at the tip of the catheter, thereby preventing thrombotic occlusion of the indwelling CVC without causing relevant systemic anticoagulatory effects.7,24-27 Low-dose UFH was, therefore, considered the control arm. The current report presents the clinical results of the THROMBOTECT study with respect to the incidence of symptomatic thromboembolism and hemorrhage as primary efficacy and safety outcomes as well as the secondary safety outcome of leukemia-related survival.

Methods Study design THROMBOTECT was an open-label, prospective, randomized, multicenter study to evaluate two different preventive antithrombotic measures during induction chemotherapy in children with ALL treated according to ALL-BFM 2000 (NCT 00430118) and AIEOP-BFM-ALL 2009 treatment protocols (NCT 01117441). THROMBOTECT was an add-on study to the ALLBFM protocols and was approved by the leading ethics committees of the Medical School Hannover, Germany, and St. Gallen, Switzerland, and by the local ethics committees of each participating site. Written informed consent to participation in the study was obtained from guardians and/or patients before randomization. The detailed study protocol is available in the Online Supplementary Material.

Patients’ eligibility Patients were eligible if treated on the ALL-BFM 2000 or AIEOP-BFM ALL 2009 protocol,28-30 if they had a CVC inserted by day 8 of induction and if the CVC remained in place until at least day 33. The choice of the CVC and decisions regarding its maintenance were made by the treating physicians according to institutional guidelines. In August 2004, the protocol was amended to allow participation of patients without a CVC. Exclusion criteria were known hemorrhagic disorders unrelated to leukemia, active gastrointestinal ulcer, previous cerebrovascular accident and/or known hypersensitivity to heparin.

Randomization and study treatment After written informed consent had been given, randomization was performed by day 8 in a 1:1:1 ratio using permuted blocks of six patients and stratification by country and the glucocorticoid preparation (dexamethasone or prednisone) administered during induction.29 Randomization was performed cen757

J. Greiner et al.

trally by the ALL-BFM study coordination center using computer-generated random number lists. This ensured that the participating centers had no access to the allocation sequence. The assigned arm was submitted to the center by fax. Patients were randomly assigned to receive one of the two experimental thromboprophylactic treatments with either the low molecular weight heparin enoxaparin or with activityadjusted antithrombin or to the control arm, i.e., low-dose UFH. Thromboprophylaxis was started on day 8 and ended on day 33 of induction chemotherapy (Online Supplementary Figure S1). The observation period covered the induction and consolidation phases (Online Supplementary Figure S2) up to and including protocol day 64. Patients in the enoxaparin group received ClexaneTM at a dose of 80-100 IU/kg body weight once daily subcutaneously31-34 with a target anti-Xa level not exceeding 0.4 U/L, measured 4 h after the third or fourth injection. On days with lumbar puncture or other invasive procedures, enoxaparin was postponed until at least 4 h after the procedure. In the case of thrombocytopenia <30 x 109/L, platelet tranfusion was required or enoxaparin had to be withheld until platelet regeneration. In the antithrombin group, antithrombin activity was measured every 3 days prior to each asparaginase administration. If antithrombin activity was below the lower limit of normal of 80%, the plasma-derived antithrombin preparation KyberninTM was substituted calculating the dose according to the formula [antithrombintarget 100% – antithrombinactual] x kg body weight targeting at 100% AT activity. Patients assigned to the control arm received UFH at a dose of 2 IU/kg body weight/h as long as an infusion drip was running to prevent local thrombotic occlusion of the indwelling CVC.24 Treatment with coagulation factors or anticoagulants beyond the interventions intended per protocol was not allowed unless clinically indicated. Management of thromboembolism was at the discretion of the treating physician.

Outcome measures The diagnosis of thromboembolism was based on clinical suspicion and had to be confirmed by one or more suitable imaging methods within a routine diagnostic work-up (Online Supplementary Table S1). No systematic provision was made for blinding the attending physicians or radiologists to the randomization arm. Intermittent dysfunction of the CVC by a clot at the tip of the catheter was not considered a thrombotic event as long as CVC patency was restored. The principal safety outcome was absence of bleeding complications during the study period. The definition of major and minor hemorrhage met internationally defined standards (Online Supplementary Table S2).35-37 Secondary safety outcomes were event-free survival and overall survival. Event-free survival was defined as the time from diagnosis to the date of last follow-up or first event. Events were resistance to therapy, leukemia relapse, secondary neoplasm or death from any cause. Failure to achieve remission due to early death or resistance was considered as an event at time zero. Survival was defined as time from diagnosis to the date of last follow-up or death from any cause.

Statistical analysis The primary objective was to test whether antithrombotic prophylaxis with enoxaparin or antithrombin was superior to that with UFH. The null hypothesis was that there was no difference between enoxaparin or antithrombin versus UFH tested with a one-tailed Fisher exact test at a significance level of P=0.025 each. The main analysis was by intention-to-treat. In

758

order to reach a power of 85% with a significance level of 0.025, 315 patients had to be randomized per group, assuming an event rate of 9% within the UFH group and 3% in the two interventional groups. If both comparisons were significantly different, the thrombosis rates in the enoxaparin and antithrombin arm had to be tested for equivalence (secondary objective). Antithrombin replacement and enoxaparin therapy would be considered equivalent if the two-sided 95% confidence interval (95% CI) of the incidence difference did not exceed ±4%. For the equivalence test, patients were analyzed according to the treatment given (as treated). The Kaplan-Meier method38 was used to estimate survival rates, and differences were compared with the log-rank test.39 A Cox proportional hazards model was used in univariate and multivariate survival analyses.40 Cumulative incidence functions for competing events were constructed by the method of Kalbfleisch and Prentice41 and compared with the Gray test.42 Odds ratios were calculated to compare the risks of thromboembolic events. Except for the confirmatory analyses of the primary study question, all other analyses were exploratory.

Results Patients’ characteristics From December 1, 2002, to December 31, 2011, 1526 patients with ALL treated at one of the 26 study centers in Germany and Switzerland were eligible for randomization (Figure 1). Of these, 577 patients were not randomized, the vast majority because patients and/or parents refused consent to be randomized to the enoxaparin arm as they did not wish to accept a daily subcutaneous injection. Nine hundred and forty-nine patients (the population for the intention-to-treat analyses) were randomly assigned to receive either UFH (n=312), enoxaparin (n=317) or antithrombin (n=320). Randomized and nonrandomized eligible patients did not differ with respect to their initial characteristics (Online Supplementary Table S3). The proportions of patients with a poor response to the prednisone prephase (prednisone poor-responders) and a slow treatment response as assessed by minimal residual disease were significantly higher in the group of non-randomized patients. In the intention-to-treat population, numbers and characteristics of patients were well-balanced between the three randomization arms except for a slight imbalance in the age distribution with fewer children below 6 years in the enoxaparin group (Table 1). Patients’ characteristics were evenly distributed between the randomization arms as treated except for a significantly lower proportion of patients below 6 years of age in the enoxaparin arm (details provided in Online Supplementary Table S4). The proportion of patients who refused antithrombotic treatment as allocated was 3% in patients randomized to UFH (10/312) or antithrombin (11/320), and 33% (105/317) in those assigned to enoxaparin (Figure 1). Rejection of the enoxaparin arm was more frequent in patients below 6 years of age than in older patients [62/157 (39%) versus 42/160 (27%), respectively] with a preferential switch to UFH in the younger cohort (Online Supplementary Table S5). Based on this finding additional exploratory analyses with respect to thromboembolism rate and leukemia-related outcomes were performed, stratified by age and in the as-treated groups. haematologica | 2019; 104(4)

Thromboembolism and thromboprophylaxis in pediatric ALL

Figure 1. Consolidated standards for reporting of trials (CONSORT) diagram. AT: antithrombin; E: enoxaparin; UFH: unfractionated heparin.

Table 1. Patients’ characteristics by thromboprophylaxis group as assigned by randomization.

Study ALL-BFM 2000 AIEOP-BFM ALL 2009 Sex Male Female Age 1≤6 years 6≤10 years ≥ 10 years Central venous catheter CVC in site No CVC WBC at diagnosis (x109/L) < 20 20 ≤ 100 100 ≤200 ≥ 200 CNS status CNS negative CNS positive No information Immunophenotype Non-T-ALL T-ALL No information

Total (n=949) N (%)

UFH (n=312) N (%)

Enoxaparin (n=317) N (%)

Antithrombin (n=320) N (%)

815 (85.9) 134 (14.1)

269 (86.2) 43 (13.8)

272 (85.8) 45 (14.2)

274 (85.6) 44 (13.8)

537 (56.6) 412 (43.4)

173 (55.4) 139 (44.6)

183 (57.7) 133 (42.3)

181 (56.6) 139 (43.4)

512 (54.0) 188 (19.8) 249 (26.2)

174 (55.8) 57 (18.3) 81 (26.0)

157 (49.5) 72 (22.9) 88 (27.8)

181 (56.6) 59 (18.4) 80 (25.0)

896 (94.4) 53 (5.6)

295 (94.6) 17 (5.4)

294 (93.3) 21 (6.7)

303 (95.2) 15 (4.8)

599 (63.1) 249 (26.2) 53 (5.6) 47 (5.0)

199 (63.8) 83 (26.6) 15 (4.8) 15 (4.8)

212 (66.9) 76 (24.0) 14 (4.4) 14 (4.4)

188 (58.8) 90 (28.1) 24 (7.4) 18 (5.6)

872 (91.9) 30 (3.2) 47 (5.0)

278 (89.1) 14 (4.4) 20 (6.4)

298 (94.0) 6 (1.9) 13 (4.1)

296 (92.5) 10 (3.1) 14 (4.4)

827 (87.1) 120 (12.6) 2 (0.2)

264 (84.6) 47 (15.1) 1 (0.3)

298 (89.0) 34 (10.7) 1 (0.3)

281 (87.8) 39 (12.3) 0 (0.0) continued on the next page

haematologica | 2019; 104(4)

759

J. Greiner et al. continued from the previous page

Genetics t(12;21) / TEL-AML1 Negative Positive No information t(9;22) / BCR-ABL Negative Positive No information t(4;11) / MLL-AF4 Negative Positive No information Peripheral blast count on day 8 (prednisone response) < 1x109/L (PGR) ≥ 1x109/L (PPR) No information Risk group Standard Medium High MRD at end of induction Negative < 5 x 10-4 ≥ 5 x 10-3 No information MRD at week 12 Negative < 5 x 10-4 ≥ 5 x 10-3 No information Randomized in induction in AIEOP-BFM ALL 2000* Randomized assigned to prednisone assigned to dexamethasone Not randomized

Total (n=949) N (%)

UFH (n=312) N (%)

Enoxaparin (n=317) N (%)

Antithrombin (n=320) N (%)

722 (76.1) 199 (21.0) 28 (3.0)

235 (75.3) 65 (20.8) 12 (3.8)

245 (77.3) 63 (19.9) 9 (2.8)

242 (75.6) 71 (22.2) 7 (2.2)

924 (97.4) 25 (2.6) 0 (0.0)

303 (97.1) 9 (2.9) 0 (0.0)

309 (97.5) 8 (2.5) 0 (0.0)

312 (97.5) 8 (2.5) 0 (0.0)

942 (99.3) 7 (0.7) 0 (0.0)

311 (99.7) 1 (0.3) 0 (0.0)

314 (99.1) 3 (0.9) 0 (0.0)

317 (99.1) 3 (0.9) 0 (0.0)

880 (92.7) 65 (6.8) 4 (0.4)

291 (93.3) 19 (6.1) 2 (0.6)

295 (93.1) 22 (6.9) 0 (0.0)

294 (91.9) 24 (7.5) 2 (0.6)

301 (31.7) 512 (54.0) 136 (14.3)

97 (31.1) 171 (54.8) 44 (14.1)

101 (32.1) 169 (53.7) 45 (14.3)

101 (31.8) 170 (53.5) 47 (14.8)

303 (31.9) 316 (33.3) 184 (19.4) 146 (15.4)

103 (33.0) 107 (34.2) 57 (18.3) 45 (14.4)

104 (32.8) 113 (35.6) 58 (18.3) 42 (13.2)

96 (30.0) 96 (30.0) 69 (21.6) 59 (18.4)

579 (61.0) 146 (15.4) 43 (4.5) 181 (19.1)

187 (59.9) 53 (17.0) 16 (5.1) 56 (17.9)

202 (63.7) 47 (14.8) 12 (3.8) 56 (17.7)

190 (59.4) 46 (14.4) 15 (4.7) 69 (21.6)

125 (13.2) 136 (14.3) 688 (72.5)

39 (12.5) 45 (14.4) 228 (73.1)

41 (12.9) 45 (14.2) 231 (72.9)

45 (14.1) 46 (14.4) 229 (71.6)

*For details see Figure S2 in the Online Supplementary Appendix and Möricke et al., Blood (2016).19 CNS: central nervous system; CVC: central venous catheter; MRD: minimal residual disease; PGR: prednisone good-response; PPR: prednisone poor-response; UFH: unfractionated heparin; WBC: white blood cell count.

Thromboembolic events Among the 949 randomized patients, 42 thromboembolic events were observed (4.4%; 95% CI: 3.2 to 5.9). Of these events, 20 (47.6%) occurred in the upper deep venous system, seven (16.7) in the lower deep venous system, and 13 (30.9%) in cerebral sinus veins; two patients (4.8%) had a cerebral arterial stroke. Eight of the 42 thromboembolic events (19%) were distant to the site of the CVC. Thirty-three events occurred between treatment day 9 and 36 during induction therapy, the other nine events occurred between treatment day 37 and 52 of induction consolidation. Children below 6 years of age had a significantly lower risk of thromboembolism (14/512, 2.7%) than those aged 6 to 9 years (11/188, 5.9%) or 10 years and older (17/249, 6.8%; P=0.018). Other patients’ characteristics and features, such as gender, initial white blood cell count, 760

immunophenotype or treatment response did not influence the incidence of thromboembolism (data not shown). The incidence of thromboembolism was significantly higher among patients randomized to UFH (25/312; 8.0%) than in the enoxaparin (11/317; 3.5%; P=0.011) or antithrombin group (6/320; 1.9%; P<0.001). The as-treated analysis revealed an incidence of 6.7% in the UFH group (25/372) compared to 3.2% in the enoxaparin (7/216; P=0.089) and 2.6% in the antithrombin group (9/341; P=0.013). The respective cumulative incidences are depicted in Figure 2A,B. The difference between the incidence of thromboembolism in the enoxaparin and antithrombin groups as treated was -0.6%; the lower and upper limits of the 95% CI were -3.5% and +2.3%, respectively (P-values for the corresponding one-sided tests were P=0.01 and P=0.001). Thus, antithrombin and enoxaparin were equally effective. haematologica | 2019; 104(4)

Thromboembolism and thromboprophylaxis in pediatric ALL

Figure 2. Thromboembolic events according to the randomization arms. Results are shown by intention to treat (A, C and E) and by treatment as given (B, D and F) for the total cohort (A and B) and stratified by age <6 years (C and D) and â&#x2030;Ľ6 years (E and F). Events are depicted as cumulative incidence curves. The P values indicated were calculated with the Fisher exact test. CI: confidence interval; OR: odds ratio; TE: thromboembolism; UFH: unfractionated heparin.

Exploratory as-treated analyses stratified by age (Figure 2D,F) demonstrated a significantly reduced risk of thromboembolism in patients 6 years of age or older when treated in one of the experimental arms compared to the risk in the control group [UFH: 18/158, 11.4%; enoxaparin: 5/120, 4.2%, P(versus UFH)=0.001; antithrombin 4/150, 2.7%, P(versus UFH)<0.001]. No significant differences were found among patients below 6 years of age (UFH 7/214, 3.3%; enoxaparin 2/96, 2.1%; antithrombin 5/191, 2.6%). No formal test for interaction was done for the subgroup analysis by age. Applying Fine-Gray models with interaction terms for age older than 6 years and enoxaparin/antithrombin, the interactions were not significant. This, however, does not entirely exclude interactions since the power of such tests is low. haematologica | 2019; 104(4)

Hemorrhage Eight bleeding episodes were documented among the 949 randomized patients (0.9%). Four of them occurred during induction chemotherapy under antithrombotic prophylaxis and four during consolidation after termination of the anticoagulants. All hemorrhages were classified as major (7 gastrointestinal, 1 cerebral). Four patients with hemorrhage were treated in the UFH group (1.1%), three in the antithrombin group [0.9%, P(versus UFH)=1.0] and one patient in the enoxaparin group [0.5%, P(versus UFH)=0.66].

Leukemia outcome and survival The 5-year probability of event-free survival and cumulative incidence of relapse of the THROMBOTECT cohort were comparable to those of the 577 non-random761

J. Greiner et al. A

Figure 3. Outcome of acute lymphoblastic leukemia according to the THROMBOTECT randomization arms. (A,B) Event-free survival and (C,D) cumulative incidence of relapse are shown by intention to treat (A,C) and by treatment as given (B,D). Numbers of patients at risk in the event-free survival graphs also apply to the respective relapse incidence graphs. 5 y-pEFS: 5-year probability of event-free survival; 5 y-CIR: 5-year cumulative incidence of relapse; SE: standard error; UFH: unfractionated heparin.

ized patients (THROMBOTECT cohort: 5-year probability of event-free survival 84.3±1.2%, 5-year cumulative incidence of relapse 11.7±1.1%; non-randomized patients: 5-year probability of event-free survival 84.0±1.6%, 5-year cumulative incidence of relapse 11.8±1.4). Patients randomized to the antithrombin arm had a 5-year probability of event-free survival of 80.9±2.2% compared with those assigned to enoxaparin (86.2±2.0%, P=0.10) or UFH (85.9±2.0%, P=0.06) (Figure 3A) with a hazard ratio of 1.40 (1.02-1.92; P=0.040) for the antithrombin arm versus the remaining patients. The probability of overall survival at 5 years was similar in all three arms (antithrombin 89.8±1.7%, enoxaparin 90.9±1.6%, UFH 92.4±1.5%). The differences observed in the event-free survival were due to a higher incidence of late relapses in the antithrombin group that in the other groups (Figure 3C); the as-treated analyses showed no statistically significant differences between the three groups [hazard ratio for the antithombin group versus the other groups: 1.16 (0.84-1.59); P=0.37) (Figure 3B,D). Retrospective exploratory subgroup analyses revealed a higher incidence of relapse among the antithrombintreated patients, but only within the medium-risk group (Online Supplementary Figure S3). Multivariate Cox regression analyses on event-free survival were performed including risk group according to respective trial criteria, TEL-AML1 status, initial white blood cell count, age and the THROMBOTECT arm as covariates. Hazard ratios 762

for the antithrombin arm were 1.38 (0.99-1.91; P=0.054) for the intention-to-treat analysis and 1.19 (0.86-1.66; P=0.269) for the as-treated analysis and thus comparable with those of the univariate analyses (Online Supplementary Table S6). To test for a potential dose effect of antithrombin, doses given were analyzed in patients treated in the antithrombin arm. Data available for 248 of 341 patients (72.7%) did not disclose a dose-related effect on the relapse incidence (Online Supplementary Figure S4).

Discussion Reliable data on thromboembolism during induction therapy of childhood ALL are scarce. The only randomized interventional trial was the PARKAA trial (Prophylactic antithrombin replacement in kids with ALL treated with L-asparaginase), designed to determine whether there was a trend to efficacy and safety of antithrombin treatment but not powered to prove it.16 To our knowledge, no other data from adequately designed and powered studies have been available so far to provide sufficient evidence that would allow valid recommendations.4,5,9,19,20,23,43,44 The THROMBOTECT trial shows, for the first time, that prophylactic antithrombotic interventions significantly reduce thromboembolism during ALL induction haematologica | 2019; 104(4)

Thromboembolism and thromboprophylaxis in pediatric ALL

therapy as compared to a control. Both interventions, enoxaparin and activity-adapted Antithrombin substitution, were equally effective. Asparaginase-induced antithrombin deficiency is assumed to be the most important mechanism for the development of thromboembolism during ALL induction therapy.45 As a consequence of asparagine depletion, asparaginase therapy leads to intracellular retention of a misfolded antithrombin, resulting in acquired antithrombin deficiency.45,46 The THROMBOTECT trial demonstrated that maintaining antithrombin activity at 80% or higher throughout the induction phase could significantly protect patients from thromboembolism. Thus, correction of low antithrombin activity seems to be one effective way to prevent thromboembolism, this being consistent with clinical and laboratory data on antithrombin supplementation.10,16,18,19,47 A considerable number of patients eligible for the study were not randomized. In this group the rate of prednisone poor-responders was significantly higher than in the THROMBOTECT cohort. This may be attributed to a tendency of the doctors or parents to avoid additional burden from interventions of an add-on trial in particular in those patients with very poor response during the first days of treatment. However, patientsâ&#x20AC;&#x2122; characteristics were comparable between the three randomization groups except for a slight underrepresentation of younger patients assigned to enoxaparin. The main reason for not participating was refusal to accept the daily subcutaneous enoxaparin injections. Not surprisingly, the proportion of patients and parents refusing the assigned enoxaparin was highest in young children. This demonstrates not only their reluctance to receive injections but also underlines a considerable drawback in practical use, irrespective of the antithrombotic efficacy of enoxaparin. Older age proved to be an important risk factor for thromboembolism, as has been reported earlier by others.1,13,48 The best cut-off in our data was the age of 6 years. Exploratory analyses suggested that the benefit from either experimental arm was more pronounced in older patients than in young children. The significant benefit in risk reduction of thromboembolism with either intervention, enoxaparin or antithrombin, as compared to UFH, provides a convincing rationale for thromboprophylaxis in this age group. For younger children, the incidence of thromboembolism was low and comparable in all three randomization arms. The need for thromboprophylaxis in ALL patients below 6 years of age could, therefore, be questioned. However, the study was not powered for subgroup analyses and the lack of statistical difference in the incidences of thromboembolism between the treatment groups in younger children may be due to insufficient power caused by the number of patients as well as the lower incidence of thromboembolism. Furthermore, in younger children thromboembolism may be missed as symptoms are often subtle. This is in line with the findings of the PARKAA study, which showed that children with symptomatic thromboembolism tend to be older than those with clinically asymptomatic thromboembolism.16 Even if clinically not diagnosed, asymptomatic thromboembolism may be associated with significant vessel occlusion.16 This, in turn, can lead to destruction of the vessel wall, causing long-term morbidity in terms of postthrombotic syndrome, likely becoming apparent years after the end of ALL therapy. Whether this applies to haematologica | 2019; 104(4)

young patients with ALL remains unknown.17 Future studies with sufficient statistical power are needed to ascertain whether such interventions in small children are justified. Nevertheless, although the high proportion of patients who refused allocation to the enoxaparin arm may complicate the interpretation of the results in this treatment arm, the reduction of thromboembolism in the global analysis appears to be sufficiently convincing to recommend thromboprophylaxis not only for older patients but for all age groups, all the more as hemorrhage is of no concern. Most thrombotic events occurred between induction treatment day 9 and 36, the latter marking the start of induction consolidation. This confirms our experience that thromboembolism only rarely occurs at the time of ALL diagnosis but rather in the course of induction therapy. Furthermore, not all centers were able to get a CVC inserted at the time of ALL diagnosis. For these reasons, thromboprophylaxis was started after the prednisone prephase on day 8 of induction therapy. The primary objective of the THROMBOTECT trial was to evaluate efficacy and safety of different prophylactic antithrombotic interventions during ALL induction therapy. The duration of thromboprophylaxis was, therefore, limited to induction therapy until day 33. Some of the thromboembolic events occurred after the end of the induction phase. However, only a few of these patients had already started the consolidation phase when the thrombosis was diagnosed. Factors that may have contributed to these late thromboses could be concurrent medical issues such as infections. Given the gradual development of a clot, a still asymptomatic thrombosis might have started to develop towards the end of induction therapy and only become symptomatic in early induction consolidation. Since pegylated asparaginase is presently used more frequently - in the AIEOP-BFM ALL 2009 trial, the second dose of this drug was given on day 26 of induction - late thromboses in induction consolidation might become more relevant as the use of pegylated asparaginase may lead to longer asparagine depletion with disturbed coagulation patterns, including extended dysfunction of antithrombin. Irrespective of possible concomitant prothrombotic risk situations, the hypercoagulable state seems to remain beyond the end of induction therapy. Given the very low rate of hemorrhage it might, therefore, be advisable to extend thromboprophylaxis accordingly. The open label assignment as well as the diagnosis of thromboembolism made on clinical suspicion only are drawbacks of the THROMBOTECT study design. However, masking the antithrombotic intervention would have meant that all patients in all randomization groups would have had to have been given subcutaneous injections, including those in the UFH and antithrombin groups containing placebo. To conduct the study as a double-blinded trial with double dummy subcutaneous injections was not considered feasible in a large pediatric population. Similar concerns apply to the primary outcome defined as thromboembolism based on clinical suspicion. The PARKAA study showed that a high incidence of clinically not recognized thromboses can be found by routine imaging screening.16 To overcome observer bias, various and repeated routine imaging screening for vessel occlusion at all possible anatomical sites would have been mandatory 763

J. Greiner et al.

at predefined time points. This comprises ultrasound but also magnetic resonance imaging which, in young children, often requires general anesthesia. In addition, for the time being the appropriate time points to look for vessel occlusion are not known and hence the possibility of missing a thrombosis at arbitrarily chosen time points would be high. Exposing children to repeated extra anesthesia with a questionable benefit was considered too high an additional burden. The study design chosen was, therefore, in favor of an open-label treatment. Imaging was performed on clinical suspicion despite the acknowledged inherent drawbacks. Evaluation of event-free survival and relapse rate within the THROMBOTECT randomization groups revealed the unexpected finding that patients randomized to the antithrombin group had a higher incidence of relapse compared to those in the enoxaparin and UFH groups. The differences were no longer obvious in the as-treated analysis and were apparent in the medium-risk group only. Although a causal relationship between the cumulative antithrombin dose and the relapse rate could not be established, the possibility that antithrombin substitution might affect leukemia outcome cannot be entirely excluded. In conclusion, the THROMBOTECT study has, for the

References 1 Athale UH, Chan AK. Thrombosis in children with acute lymphoblastic leukemia. Part I. Epidemiology of thrombosis in children with acute lymphoblastic leukemia. Thromb Res. 2003;111(3):125-131. 2 Mitchell L, Hoogendoorn H, Giles AR, Vegh P, Andrew M. Increased endogenous thrombin generation in children with acute lymphoblastic leukemia: risk of thrombotic complications in L-asparaginase-induced antithrombin III deficiency. Blood. 1994;83(2):386-391. 3 Mitchell LG, Sutor AH, Andrew M. Hemostasis in childhood acute lymphoblastic leukemia: coagulopathy induced by disease and treatment. Semin Thromb Hemost. 1995;21(4):390-401. 4 Nowak-Gรถttl U, Ahlke E, Fleischhack G, et al. Thromboembolic events in children with acute lymphoblastic leukemia (BFM protocols): prednisone versus dexamethasone administration. Blood. 2003;101(7):25292533. 5 Caruso V, Iacoviello L, di Castelnuovo A, et al. Thrombotic complications in childhood acute lymphoblastic leukemia: a metaanalysis of 17 prospective studies comprising 1752 pediatric patients. Blood. 2006;108(7):2216-2222. 6 Mitchell LG, Andrew M, Hanna K, et al. A prospective cohort study determining the prevalence of thrombotic events in children with acute lymphoblastic leukemia and a central venous line who are treated with Lasparaginase: results of the Prophylactic Antithrombin Replacement in Kids with Acute Lymphoblastic Leukemia Treated with Asparaginase (PARKAA) study. Cancer. 2003;97(2):508-516. 7 Journeycake JM, Buchanan GR. Catheterrelated deep venous thrombosis and other catheter complications in children with can-

764

first time, demonstrated that activity-targeted antithrombin replacement as well as the use of enoxaparin lead to a significant risk reduction for thromboembolism during ALL induction therapy when compared with low-dose UFH. Bleeding was not a major concern. Thromboprophylaxis during induction therapy can, therefore, be recommended for children and adolescents with ALL. The higher incidence of late relapses in children with medium-risk ALL assigned to the antithrombin group remains to be resolved and leads us to recommend, at present, primarily enoxaparin. Whether thromboprophylaxis contributes to minimize not only clinical but also silent thromboses and by that long-term morbidity in terms of post-thrombotic syndrome remains to be determined. The THROMBOTECT results provide the rationale for new studies, both to elucidate a possible impact of antithrombin on leukemia outcome and to further determine the best practice to prevent thromboembolism during ALL induction chemotherapy. Acknowledgments We thank the investigators, the medical staff, the data managers and the patients of all participating centers for their sustained support of the THROMBOTECT study.

cer. J Clin Oncol. 2006;24(28):4575-4580. 8 Male C, Chait P, Andrew M, Hanna K, Julian J, Mitchell L. Central venous line-related thrombosis in children: association with central venous line location and insertion technique. Blood. 2003;101(11):4273-4278. 9 Athale UH, Chan AK. Thrombosis in children with acute lymphoblastic leukemia. Part II. Pathogenesis of thrombosis in children with acute lymphoblastic leukemia: effects of the disease and therapy. Thromb Res. 2003;111(4-5):199-212. 10 Astwood E, Vora A. Personal practice: how we manage the risk of bleeding and thrombosis in children and young adults with acute lymphoblastic leukaemia. Br J Haematol. 2011;152(5):505-511. 11 Vidal E, Sharathkumar A, Glover J, Faustino EV. Central venous catheter-related thrombosis and thromboprophylaxis in children: a systematic review and meta-analysis: reply. J Thromb Haemost. 2015;13(1):161-162. 12 Wiernikowski JT, Athale UH. Thromboembolic complications in children with cancer. Thromb Res. 2006;118(1):137152. 13 Athale UH, Siciliano SA, Crowther M, Barr RD, Chan AK. Thromboembolism in children with acute lymphoblastic leukaemia treated on Dana-Farber Cancer Institute protocols: effect of age and risk stratification of disease. Br J Haematol. 2005;129(6):803-810. 14 Tuckuviene R, Ranta S, Albertsen BK, et al. Prospective study of thromboembolism in 1038 children with acute lymphoblastic leukemia: a Nordic Society of Pediatric Hematology and Oncology (NOPHO) study. J Thromb Haemost. 2016;14(3):485-494. 15 Nowak-Gรถttl U, Wermes C, Junker R, et al. Prospective evaluation of the thrombotic risk in children with acute lymphoblastic leukemia carrying the MTHFR TT 677 genotype, the prothrombin G20210A variant, and further prothrombotic risk factors. Blood. 1999;93(5):1595-1599.

16 Mitchell L, Andrew M, Hanna K, et al. Trend to efficacy and safety using antithrombin concentrate in prevention of thrombosis in children receiving l-asparaginase for acute lymphoblastic leukemia. Results of the PAARKA study. Thromb Haemost. 2003;90(2):235-244. 17 Avila ML, Duan L, Cipolla A, et al. Postthrombotic syndrome following upper extremity deep vein thrombosis in children. Blood. 2014;124(7):1166-1173. 18 Gugliotta L, D'Angelo A, Mattioli BM, et al. Hypercoagulability during L-asparaginase treatment: the effect of antithrombin III supplementation in vivo. Br J Haematol. 1990;74(4):465-470. 19 Nowak-Gรถttl U, Kuhn N, Wolff JE, et al. Inhibition of hypercoagulation by antithrombin substitution in E. coli Lasparaginase-treated children. Eur J Haematol. 1996;56(1-2):35-38. 20 Elhasid R, Lanir N, Sharon R, et al. Prophylactic therapy with enoxaparin during L-asparaginase treatment in children with acute lymphoblastic leukemia. Blood Coagul Fibrinolysis. 2001;12(5):367-370. 21 Harlev D, Zaidman I, Sarig G, Ben Arush MW, Brenner B, Elhasid R. Prophylactic therapy with enoxaparin in children with acute lymphoblastic leukemia and inherited thrombophilia during L-asparaginase treatment. Thromb Res. 2010;126(2):93-97. 22 Ruud E, Holmstrom H, de Lange C, Natvig S, Albertsen BK, Wesenberg F. Thrombotic effects of asparaginase in two acute lymphoblastic leukemia protocols (NOPHO ALL-1992 versus NOPHO ALL-2000): a single-institution study. Pediatr Hematol Oncol. 2006;23(3):207-216. 23 Meister B, Kropshofer G, Klein-Franke A, Strasak AM, Hager J, Streif W. Comparison of low-molecular-weight heparin and antithrombin versus antithrombin alone for the prevention of symptomatic venous thromboembolism in children with acute

haematologica | 2019; 104(4)

Thromboembolism and thromboprophylaxis in pediatric ALL

25 26

lymphoblastic leukemia. Pediatr Blood Cancer. 2008;50(2):298-303. Goossens GA. Flushing and locking of venous catheters: available evidence and evidence deficit. Nurs Res Pract. 2015;2015: 985686. Hentschel R, Sutor AH. Catheter-related thrombosis and its prevention in infancy. Hamostaseologie. 2002;22(4):167-173. Lersch C. Prevention of catheter-induced thromboses by low molecular weight heparins. Hamostaseologie. 2002;22(4):161166. Randolph AG, Cook DJ, Gonzales CA, Andrew M. Benefit of heparin in central venous and pulmonary artery catheters: a meta-analysis of randomized controlled trials. Chest. 1998;113(1):165-171. Conter V, Bartram CR, Valsecchi MG, et al. Molecular response to treatment redefines all prognostic factors in children and adolescents with B-cell precursor acute lymphoblastic leukemia: results in 3184 patients of the AIEOP-BFM ALL 2000 study. Blood. 2010;115(16):3206-3214. Mรถricke A, Zimmermann M, Valsecchi MG, et al. Dexamethasone vs prednisone in induction treatment of pediatric ALL: results of the randomized trial AIEOP-BFM ALL 2000. Blood. 2016;127(17):2101-2112. Schrappe M, Valsecchi MG, Bartram CR, et al. Late MRD response determines relapse risk overall and in subsets of childhood Tcell ALL: results of the AIEOP-BFM-ALL 2000 study. Blood. 2011;118(8):2077-2084. Massicotte P, Adams M, Marzinotto V, Brooker LA, Andrew M. Low-molecularweight heparin in pediatric patients with thrombotic disease: a dose finding study. J

haematologica | 2019; 104(4)

Pediatr. 1996;128(3):313-318. 32 Punzalan RC, Hillery CA, Montgomery RR, Scott CA, Gill JC. Low-molecular-weight heparin in thrombotic disease in children and adolescents. J Pediatr Hematol Oncol. 2000;22(2):137-142. 33 Dix D, Andrew M, Marzinotto V, et al. The use of low molecular weight heparin in pediatric patients: a prospective cohort study. J Pediatr. 2000;136(4):439-445. 34 Schneppenheim R, Greiner J. Thrombosis in infants and children. Hematology Am Soc Hematol Educ Program. 2006;86-96. 35 Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost. 2005;3(4):692694. 36 Levine MN, Raskob G, Landefeld S, Kearon C. Hemorrhagic complications of anticoagulant treatment. Chest. 2001;119(1 Suppl):108S-121S. 37 Committee for proprietary medicinal products (CPMP). Notes for guidance on clinical investigation of medicinal products for treatment of venous thromboembolic disease. http://www ema europe eu 2016; CPMP/ EWP/563/98 38 Kaplan ES, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53(282):457-481. 39 Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep. 1966;50(3):163-170. 40 Cox DR. Regression models and life-tables. J R Stat Soc Series B Stat Methodol. 1972;42(2):187-220. 41 Kalbfleisch JD, Prentice RL. The statistical

42 43

47 48

analysis of failure time data. 1980;1st163188. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16(3):1141-1154. Santoro N, Colombini A, Silvestri D, et al. Screening for coagulopathy and identification of children with acute lymphoblastic leukemia at a higher risk of symptomatic venous thrombosis: an AIEOP experience. J Pediatr Hematol Oncol. 2013;35(5):348355. Giordano P, Santoro N, Del Vecchio GC, Rizzari C, Masera G, de Mattia D. Timmunophenotype is associated with an increased prevalence of thrombosis in children with acute lymphoblastic leukemia. A retrospective study. Haematologica. 2003;88 (9):1079-1080. Hernandez-Espinosa D, Minano A, Ordonez A, et al. Dexamethasone induces a heat-stress response that ameliorates the conformational consequences on antithrombin of L-asparaginase treatment. J Thromb Haemost. 2009;7(7):1128-1133. Bushman JE, Palmieri D, Whinna HC, Church FC. Insight into the mechanism of asparaginase-induced depletion of antithrombin III in treatment of childhood acute lymphoblastic leukemia. Leuk Res. 2000;24(7):559-565. Payne JH, Vora AJ. Thrombosis and acute lymphoblastic leukaemia. Br J Haematol. 2007;138(4):430-445. Appel IM, Hop WC, van Kessel-Bakvis C, Stigter R, Pieters R. L-asparaginase and the effect of age on coagulation and fibrinolysis in childhood acute lymphoblastic leukemia. Thromb Haemost. 2008;100(2):330-337.

765

ARTICLE Ferrata Storti Foundation

Non-Hodgkin Lymphoma

TIRAP p.R81C is a novel lymphoma risk variant which enhances cell proliferation via NF-kB mediated signaling in B-cells Regula Burkhard,1,2,3 Irene Keller,4 Miroslav Arambasic,1,2,3 Darius Juskevicius,5 Alexandar Tzankov,5 Pontus Lundberg,6 RĂŠmy Bruggmann,4 Stephan Dirnhofer,5 Ramin Radpour1,7* and Urban Novak1,3*

Haematologica 2019 Volume 104(4):766-777

Department of Medical Oncology, Inselspital, Bern University Hospital; 2Division of Experimental Pathology, Institute of Pathology, University of Bern; 3Department for BioMedical Research (DBMR), University of Bern; 4Interfaculty Bioinformatics Unit, Department for BioMedical Research, and Swiss Institute of Bioinformatics, University of Bern; 5Institute of Pathology and Medical Genetics, University of Basel; 6Department of Biomedicine, Experimental Hematology, University Hospital Basel and University of Basel and 7Tumor Immunology, Department for BioMedical Research (DBMR), University of Bern, Switzerland 1

*RR and UN are co-senior authors.

ABSTRACT

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/766

iffuse large B-cell lymphoma is the most common malignant lymphoma in adults. By gene-expression profiling, this lymphoma is divided in three cell-of-origin subtypes with distinct molecular and clinical features. Most lymphomas arise sporadically, yet familial clustering is known, suggesting a genetic contribution to disease risk. Familial lymphoma cases are a valuable tool to investigate risk genes. We studied a Swiss/Japanese family with 2 sisters affected by a primary mediastinal B-cell lymphoma and a non-germinal center diffuse large B-cell lymphoma not otherwise specified, respectively. The somatic landscape of both lymphomas was marked by alterations affecting multiple components of the JAK-STAT pathway. Consequently, this pathway was constitutively activated as evidenced by high pJAK2 as well as increased nuclear pSTAT3 and pSTAT6 in malignant cells. Potential lymphoma risk variants were identified by whole exome sequencing of the germline DNA derived from siblings and unaffected family members. This analysis revealed a pathogenic variant in TIRAP, an upstream regulator of NF-kB, in both affected siblings and their mother. We observed increased B-cell proliferation in family members harboring the TIRAP p.R81C variant. B-cell proliferation correlated with TIRAP and NF-kB target gene expression, suggesting enhanced NF-kB pathway activity in TIRAP p.R81C individuals. TIRAP knockdown reduced B-cell survival and NF-kB target gene expression, particularly in individuals with TIRAP p.R81C. Functional studies revealed significantly increased NF-kB activity and resistance to stress-induced cell-death by TIRAP p.R81C. The identification of an inherited TIRAP variant provides evidence for a novel link between genetic alterations affecting the NF-kB pathway and lymphomagenesis.

ÂŠ2019 Ferrata Storti Foundation

Introduction

Correspondence: URBAN NOVAK urban.novak@insel.ch Received: August 9, 2018. Accepted: October 30, 2018. Pre-published: October 31, 2018. doi:10.3324/haematol.2018.201590

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.

766

Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoma in adults.1 Its molecular subtypes, activated B-cell-like (ABC), germinal center B-celllike (GCB) DLBCL, and primary mediastinal B-cell lymphoma (PMBL) arise from B-cells at distinct differentiation stages.2,3 PMBL is clinically aggressive with bulky mediastinal masses; it accounts for up to 10% of DLBCLs, and preferentially occurs in young female patients. Next-generation sequencing provided insights in genetic lesions of de novo DLBCL and its subtypes.4-10 The genetic hallmarks of PBML are amplifications of the 9p24 locus containing JAK2 and PDL1. Present in 70% of PMBL, this amplification is rare in other DLBCL subtypes.11-13 Constitutive NF-kB pathway activity through various mechanisms is characteristic of PMBL and ABC-DLBCL.14 Until now, the genetic risk factors for DLBCL/PMBLs have remained obscure. haematologica | 2019; 104(4)

TIRAP, a novel familial lymphoma risk gene

Population-based studies reported an increased risk for DLBCL in relatives of individuals with DLBCL, and genome-wide association studies identified several common single nucleotide variants associated with sporadic DLBCL.15-18 Familial clustering provides evidence for Mendelian susceptibility. In very rare cases, familial aggregation is associated with hereditary cancer syndromes,18 but as far as other syndromes are concerned, a heritable basis for DLBCL is not fully understood. A germline variant in MLL described in a Finnish family is still the only reported variant linked to familial PMBL.19 Although familial lymphomas account for less than 5% of cases, these pedigrees are a valuable tool to help identify risk genes that might also contribute to a better understanding of more frequent sporadic cases. Here, we investigate a Swiss/Japanese family in which 2 out of 3 children were diagnosed with aggressive B-cell lymphomas arising in the mediastinum. Whole exome sequencing (WES) on the germline DNA of the affected siblings and healthy family members identified a variant in the TIR-domain-containing adaptor protein (TIRAP). TIRAP engages signals from TLR2 and TLR4 receptors and recruits MyD88 to the plasma membrane mediated through Toll/interleukin-1 receptor (TIR) domain interaction.20 Downstream signaling includes activation of IL-1R-associated kinases (IRAK), ultimately culminating in the activation of the transcription factors NF-kB and AP-1. In this family, we identified an inherited TIRAP p.R81C variant in 2 affected siblings. This variant provided B-cells with increased proliferation and survival through enhanced NF-kB activity. Functional studies revealed that TIRAP p.R81C enhanced NF-kB gene signature and reduced stress-triggered cell death. Collectively, we provide evidence that TIRAP p.R81C may act as a novel lymphoma risk variant and our data suggest that TIRAP should be integrated into the complex network of genes contributing to deregulated NF-kB signaling involved in lymphomagenesis.

Methods Patients Samples from patients, non-affected family members, and healthy donors were collected after informed consent. This study was approved by the local ethics committee (KEK-BE116/11) and was conducted in accordance with the Declaration of Helsinki. The diagnosis of DLBCL/PMBL was made according to the 2017 World Health Organization classification and pathological review by SD and AT confirmed the diagnosis.1 Genomic DNA was extracted from peripheral blood mononuclear cells (PBMCs) using standard methods. Tumor DNA was isolated from formalin-fixed paraffin embedded tissues (FFPE) using phenol-chloroform. In FFPE samples, tumor cell-rich areas were identified on CD20 stained sections and separated from surrounding tissue by laser microdissection.

Whole exome sequencing The quality of extracted DNA was assessed by Bioanalyzer (Agilent) and a PCR fragment size-based assay developed at Fasteris (Geneva, Switzerland). Prior to library preparation with TruSeq DNA Sample Preparation Kit (Illumina), DNA samples were treated with PreCR Repair Mix (New England Biolabs). Exome capturing was performed using TruSeq Exome Enrichment Kit (Illumina) and samples were sequenced on an Illumina haematologica | 2019; 104(4)

HiSeq2500 instrument with 100bp paired-end reads (Fasteris, sequencing performed between 2012 and 2014). As the FFPE lymphoma sample of sister 2 resulted in a low-yield library of poor quality, the library preparation was modified: after PreCR Repair mix treatment, DNA was split, and four libraries were prepared simultaneously using the Nextera Exome Enrichment Kit (Illumina). Before exome enrichment, libraries were pooled and sequencing was performed as described above. Both exome enrichment kits contained the same set of baits, resulting in identical exome coverage. Whole exome sequencing (WES) data has been deposited at the European Nucleotide Archive (http://www.ebi.ac.uk/ena, accession number PRJEB15254). See Online Supplementary Table S3 for exome data metrics.

Analysis of germline and somatic variants Raw sequence read quality was assessed using FastQC. Reads were mapped to the human reference genome hg19 using Bowtie2 v.2.2.1, and duplicated reads were removed by Pi-card-tools v.1.80. Germline variant calling was performed using the Genome Analysis Toolkit (GATK v.3.3.0) best practices workflow using Haplotype Caller and limiting the analysis to enriched targets ±100bp. We used GATK v.3.3.0 to recalibrate the variant quality and refine the genotypes using population (1000 Genomes Project, phase 1 data) and pedigree information. Variants in low complexity regions were removed.21 Germline variants were prioritized as following: 1) good quality genotype in both sisters (Phred quality ≥20); 2) moderate/high impact based on SnpEff v.3.2 prediction; 3) novel/known variant at frequency of less than 1% (not polymorphisms) according to the 1000 Genomes Database and the Exome Variant Server; and 4) the presence of at least one copy of the putatively harmful allele in both siblings. Only variants not present as homozygote in healthy family members were selected. Possible links between genes with germline variants and terms related to cancer and malignant lymphomas were assessed by Ariadne Genomics Pathway Studio v.9 (Elsevier). Alterations with a predicted link to the disease were annotated with PolyPhen-2, SIFT, MutationTaster and GERP++ effect prediction scores using dbNSFP v.2.1, and Combined Annotation Dependent Depletion (CADD) scores v.1.1 (available from https://cadd. gs.washington.edu/). Pathway Studio was used to identify gene networks and canonical pathway enriched for genes containing putatively deleterious variants. The enrichment-scores were calculated using c2 test comparing genes with putatively deleterious mutations to the proportion of background genes in the Gene Ontology group. An enrichment-score ≥3 corresponded to a significant link (P<0.05). See also Online Supplementary Appendix.

Results Clinical and histopathological characterization of the mediastinal B-cell lymphomas The Swiss/Japanese family investigated includes 2 female siblings with lymphomas (Figure 1). The older sister 1 developed a PMBL at 30 years of age and died with primary progressive disease (Online Supplementary Table S1). At 25 years of age, sister 2 was diagnosed with a stage IIA non-germinal center (GC) DLBCL, not otherwise specified (NOS) with a mediastinal mass and cervical lymphadenopathy. Chemo-immunotherapy achieved an ongoing remission. Smoldering myelomas IgGλ were detected in the father and his monozygotic twin at 65 years of age. Other family members are currently healthy, and there are no other hematologic malignancies in the extended family. 767

R. Burkhard et al.

Figure 1. Mediastinal B-cell lymphomas arising in 2 female siblings. Pedigree of the Swiss/Japanese family under study. Circles and squares represent female and male family members, respectively. Open symbols indicate unaffected persons; closed symbols repre-sent the 2 siblings affected by primary mediastinal B-cell lymphoma (PBML) and mediastinal non-germinal center (GC) diffuse large B-cell lymphoma (DLBCL), not otherwise specified (NOS), respectively. The deceased individual (sister 1) is marked by a slash through the symbol. The year of diagnosis is shown in brackets and the year of birth is denoted by *. Both lymphomas were classified according to the World Health Organization 2017 classification.1

Both lymphomas lacked evidence of an Epstein-Barr virus infection (Online Supplementary Table S2), showed a clear cytoplasm and compartmentalizing sclerosis, and were CD20, CD30 and Ki-67 positive. However, the GC markers BCL6, CD10 and GCET1 were only expressed in the PMBL of sister 1, as was CD23 and BCL2. Despite the expression of CD30 and sclerosis, the clinical presentation, morphology and immunophenotype of the tumor in sister 2 were consistent with a non-GC DLBCL NOS.1 Given its genetic (9p24 and 12q13 gains, SOCS1 and STAT6 mutations), and phenotypic characteristics (expression of CD30, overexpression of JAK2-STAT-cascade members), this lymphoma can retrospectively be considered to most probably represent a PMBL, and was initially designated as non-GC DLBCL NOS with features of PMBL.

Analysis of the coding genome of lymphomas The somatic landscape of both lymphomas was analyzed by WES using the Illumina technology on DNA derived from laser-dissected FFPE tissue sections (Online Supplementary Methods). Mutations were validated using Ion Proton sequencing, and their somatic origin was confirmed by the absence in matched normal DNA isolated from PBMCs. A total of 192 and 130 confirmed clonal protein altering mutations were identified in the lymphomas of sisters 1 and 2, respectively (Online Supplementary Table S7). Most of those mutations were missense mutations, with a low number of nonsense and splice site variants (Figure 2A). In addition, somatic copy number alterations (CNA) were analyzed by array comparative genomic hybridization (aCGH). While the tumor of sister 1 contained seven gains and two deletions, three gains were detected in the lymphoma of sister 2. Interestingly, 9p24 and 12q13 gains were present in both lymphomas (Figure 2B). Fluorescence in situ hybridization (FISH) analyses revealed a trisomy at 8q24 (including MYC) in the lymphoma of sister 1, in line with the gain on chromosome 8 by aCGH (Figures 2B and 4B). Taken together, the analysis of CNA 768

and somatic mutations reflected the known complex genetic landscape in those entities. The overall number of somatic lesions in the lymphoma of sister 1 was considerably higher (Figure 2C).

The JAK-STAT pathway has somatic mutations in multiple genes and is constitutively active As mentioned above, aCGH revealed a 9p24 gain in both lymphomas (Figure 3A). The amplification of JAK2, a key target of the 9p24 gain, in both tumors was confirmed by FISH (Figure 3B). Moreover, a gain of 12q13 encompassing STAT2 and STAT6, was detected in both lymphomas (Figure 3C). Besides CNA, we also identified somatic mutations in key genes of the JAK-STAT pathway. Each tumor harbored a private missense mutation within the DNA binding domain of STAT6 (Figure 3D). In addition, SOCS1, encoding a negative regulator of the JAK-STAT pathway, was mutated in the lymphoma of sister 1 (Figure 4A). Collectively, these somatic alterations caused constitutive activation of the JAK-STAT pathway as evidenced by high cytoplasmic expression of phosphorylated JAK2 (pJAK2) and increased nuclear pSTAT3 and pSTAT6 expression in most tumor cells (Figure 3E). In summary, despite distinct pathological and clinical features, these data revealed a shared aberrant activation of JAK-STAT signaling which is a known signature in PMBL.10,22,23

Genetic alterations related to the distinct clinical outcome In contrast to ABC- and GCB-DLBCL, PMBLs have a favorable prognosis when responding to chemoimmunotherapy.2 To investigate genetic lesions associated with the different clinical outcome, we focused on genes with a reported pathogenic role in lymphomas and/or genes which were mutated in more than 10% of DLBCLs.4-10 Mutations in B2M, and TP53, REL and MYC gains as well as a CIITA break apart were confined to the PMBL of sister 1 who died of primary progressive disease (Figure 4A and B). Although amplification of PDL1 resulthaematologica | 2019; 104(4)

TIRAP, a novel familial lymphoma risk gene

Figure 2. Overall load of numerical and structural genomic alterations in the lymphomas of both siblings. (A) Number (N.) of all validated (by IonProton sequencing) non-silent somatic clonal mutations identified through whole exome sequencing in the two tumors. (B) Chromosomal gains and losses detected in the two lymphomas by array comparative genomic hybridization (aCGH). In the aCGH profiles, the normalized log2 ratios are plotted based on their chromosome position, with vertical bars separating the chromosomes. Regions with losses and gains are represented by decreased and increased log2 ratios, respectively. Genomic changes are marked in red (gain) and green (loss). Copy number (CN) alterations that are present in both tumors are underlined. (C) Combined load of somatic non-silent mutations as well as CN gains and losses identified in the two investigated lymphomas.

ing from the 9p24 gain (Figure 4A) is a common feature of both lymphomas, PDL1 was only expressed on the malignant B-cells of sister 1 (Figure 4B). The co-occurrence of genetic alterations involving genes related to immune-cell crosstalk in the lymphoma of sister 1 involving PDL1 expression, B2M p.M1R mutation and genomic alterations of CIITA are an interesting finding that suggests a combined role in escape from immune-surveillance.24,25 In summary, we identified genetic lesions that may collectively contribute to the distinct clinical outcome. Of note, TP53 mutations, MYC gains, CIITA translocation and expression of PDL1 on malignant B cells, all solely present in the lymphoma of sister 1, have been associated with an inferior overall and progression-free survival in DLBCL.26-29 BCL2 expression, as observed in the lymphoma of sister 1, in the absence of a translocation (Online Supplementary Table S2) has a controversial prognostic role.30

Whole exome sequencing identified lymphoma risk genes The low age- and gender-adjusted incidence rate for sporadic DLBCL (0.1/100,000 cases in Switzerland31), and the occurrence of 2 siblings affected by mediastinal B-cell lymphomas suggested a genetic predisposition for lymphomagenesis in this family. Therefore, we performed haematologica | 2019; 104(4)

WES on DNA from PBMCs of both sisters and all the other members of the core family (Figure 1). A total of 547 rare protein altering variants in 444 genes were identified out of 86,000 screened variants in the germline DNA of each sister. After excluding variants that were present as homozygote in unaffected family members, 274 variants in 234 genes remained. To find a potential link between those 234 candidate genes and deregulated proliferation, we performed a comprehensive gene ontology and pathway enrichment analysis. The result indicated a significant link between 45 of these candidate genes with cancer and/or malignant lymphoma. To identify potential deleterious alterations, the pathogenicity of variants in those 45 cancer related genes was examined by five different in silico algorithms. At the end, 15 variants in 15 candidate genes were predicted as deleterious by all algorithms and were considered for further analyses (Online Supplementary Figure S1A). Mutated genes were significantly enriched in processes like proliferation, lymphocyte activation and response to DNA damage, all pathways crucial for tumorigenesis (Online Supplementary Figure S1B). Of note, MLL the only PMBL/DLBCL susceptibility gene reported so far was not found to harbor variants in this family.19 Interestingly, we identified TIRAP among those candidates. TIRAP is an adapter protein that engages signals from TLR2 and 4 and thereby activates the NF-kB path769

R. Burkhard et al. somatic alterations in TIRAP in roughly 0.5% of cases.34,35 However, the role of TIRAP in tumorigenesis has so far not been investigated. Hence, we functionally investigate whether the identified TIRAP variant in this family contributes to lymphomagenesis.

way. Dysregulation of the NF-kB pathway is the oncogenic hallmark of ABC-DLBCL and PMBL.14 Of note, in 420 primary DLBCLs, high TIRAP expression correlated with poor survival and was significantly increased in high risk patients (data generated by SurvExpress32) (Online Supplementary Figure S2A). Besides, amplifications of 11q and 11q24 which also contain TIRAP, have been reported in approximately 20% of DLBCLs and PMBLs.11,33,34 Somatic TIRAP mutations occurring in various cancers including DLBCL are listed in COSMIC and genomic sequencing studies on several hundred DLBCLs revealed

TIRAP p.R81C variant is a potential novel risk factor for lymphomas Whole exome sequencing revealed a heterozygous variant within the coding exon 5 of TIRAP (c.241C>T) in both sisters and their Japanese mother. The variant was absent

Figure 3. The JAK-STAT pathway has somatic mutations in multiple genes and is constitutively active. (A) aCGH probe view of the 9p24 gain. A duplication of the JAK2 locus was detected in both samples, shown as an increase of the average log2 ratio above zero (bold line). Shaded area indicated the extent of a copy number aberration. (B) Fluorescence in situ hybridization (FISH) signals with BAC probes for the 5â&#x20AC;&#x2122; and 3â&#x20AC;&#x2122; regions of JAK2. Arrows indicate examples of cells with multiple FISH signals. Note green autofluorescent sclerosing bundles in the background. Scale bars: 10 mm. (C) Array comparative genomic hybridization (aCGH) probe view of gains in 12p13 region, which among other genes also affect STAT2 and STAT6 as indicated by arrows. (D) (Top) Schematic representation of the human STAT6 gene locus with open and closed boxes indicating non-coding and coding exons, respectively. (Bottom) Confirmed somatic missense mutations located within the DNA binding domain of STAT6. Protein domain annotation according to Pfam. (E) The expression of phosphorylated (p) pJAK2, pSTAT3 and pSTAT6 in the two lymphomas was assessed by immunohistochemistry. Scale bars: 10 mm.

770

haematologica | 2019; 104(4)

TIRAP, a novel familial lymphoma risk gene

in other unaffected family members (Figure 5A). This variant resulted in a substitution of the arginine at amino acid residue 81 to a cysteine (p.R81C) which is highly conserved among species and located in close proximity to the functional TIR domain (Online Supplementary Figure S2B and C). The TIRAP p.R81C variant was predicted to be deleterious by five out of five applied algorithms (Online Supplementary Figure S2D), has a dbSNP identifier (rs138228187) and is reported in COSMIC. It has a global minor allele frequency of 0.00006 and 0.0006 in ExAC and 1000 Genomes, respectively, and 0.0005 in the Japanese population, and is therefore not a polymorphism.36,37 Sanger sequencing of cDNA derived from fresh PBMCs of the family members confirmed the TIRAP p.R81C status and expression of the variant allele in sister 2 and her mother (Figure 5A and Online Supplementary Figure S3). Of note, the p.R81C variant was also identified in the lymphomas of both siblings (data not shown). To investigate the functional consequence of TIRAP p.R81C, we assessed the expression of pIRAK1 and total IRAK4, two downstream kinases and activators of the NFkB signaling pathway. For IRAK4, our analysis was confined to the total protein, as an antibody to reliably determine its phosphorylated form on FFPE tissue was not available. GC B-cells of healthy controls showed little to

no pIRAK1 and IRAK4 expression whereas p.R81C TIRAP carrying malignant B-cells of both sisters were clearly positive for these markers (Figure 5B and C). This suggests that the TIRAP downstream signaling is predominantly active in malignant B-cells with the p.R81C mutation. The relevance of this pathway is underlined by the analysis of 36 primary ABC-DLBCLs that revealed that 63% and 17% expressed pIRAK1 and IRAK4, respectively (Figure 5C). GCB-DLBCLs, however, generally lacked the expression of both kinases. To assess TIRAP/NF-kB pathway activity in non-malignant cells, we determined the gene expression of TIRAP as well as genes involved in cell proliferation and survival, among them several targets of NF-kB in PBMCs of living family members. Unsupervised hierarchical clustering analysis revealed two clusters, separating TIRAP p.R81C and wild-type (WT) samples based on the expression signature of selected genes (Figure 5D). PBMCs carrying the TIRAP p.R81C mutation expressed higher levels of TIRAP as well as genes involved in cell survival, cell cycle and proliferation. In contrast, TIRAP WT PBMCs showed higher expression of CASP9, which is implicated in intrinsic apoptosis. Next, we studied the impact of TIRAP p.R81C on primary B-cells. An increase in proliferating B-cells as deter-

Figure 4. Genetic alterations related to the different clinical outcome. (A) Comparison of the somatic landscape in the two lymphomas implementing the lesions identified by whole exome sequencing, array comparative genomic hybridization (aCGH) and fluorescence in situ hybridization (FISH). Only alterations in genes which are present in more than 10% of diffuse large B-cell lymphoma (DLBCL) cases and/or with a pathogenic significance in lymphoid malignancies were considered for this comparison.4-10 Mutations (M), copy number alterations (CN) and translocations (Tx) are sorted according to whether they were found to be mutated in both tumors or restricted to one lymphoma only. Numbers indicate the total amount of identified somatic mutations. In red, genes which have been associated with worse clinical outcome.9,26-29 (B) (Left) Representative FISH signal patterns using MYC and CIITA break apart assay in the primary mediastinal B-cell lymphoma (PMBL) of sister 1. Arrows indicate examples of cells with MYC (multiple FISH signals) gains and CIITA break apart (split red and green FISH signals), respectively. Scale bars: 10 mm. (Right) Immunohistochemistry analysis of PDL1 protein expression in the lymphoma of sister 1. Scale bar: 50 mm.

haematologica | 2019; 104(4)

771

R. Burkhard et al.

Figure 5. TIRAP p.R81C a potential novel familial lymphoma risk variant. (A) (Top) Schematic representation of the human TIRAP gene locus with open and closed boxes indicating non-coding and coding exons, respectively. (Bottom) Whole exome sequencing data for the affected region of the TIRAP exon 5 visualized in integrative genomic viewer demonstrating a heterozygous variant in both sisters and their mother, whereas homozygous wild-type (WT) sequence was observed in the remaining family members. Sanger sequence data of complementary (c) DNA isolated from fresh peripheral blood mononuclear cells (PBMCs) showing the same variant. Of note, no cDNA was available for sister 1 (deceased). (B) IRAK1 phosphorylation (p) and total IRAK4 expression was assessed by immunohistochemistry (IHC) in the two lymphomas as well as lymphoid tissue of (unmatched) healthy controls. Scale bars: 50 mm. (C) Representation of the percentage of cells expressing pIRAK1 and total IRAK4 in samples described in (B) as well as 36 activated B-cell-like diffuse large B-cell lymphoma (ABC-DLBCLs) and 32 germinal center B-celllike diffuse large B-cell lymphoma (GCB-DLBCLs). The primary samples have been described previously.50 (D) Heatmap showing hierarchical clustering of mRNA levels of genes involved in NF-ÎşB pathway, cell survival and proliferation in peripheral blood mononuclear cells (PBMCs) of mother (M), sister 2 (S2), brother (B) and father (F). The hierarchical cluster analysis (Euclideanâ&#x20AC;&#x2122;s method) reveals two major clusters representing TIRAP p.R81C mutated and WT individuals. Bar chart showing the log2 fold difference in gene expression in TIRAP p.R81C versus WT family members. (E) PBMCs isolated from family members and age- and gender-matched healthy donors (HD) were cultured in the presence (+) or absence (-) of lipopolysaccharide (LPS) for 12 hours. Ki-67 was measured by flow cytometry on CD20+ B-cells. (F) Linear correlation (Pearson correlation) between Ki-67+ B-cells and BCL2L1 (left), NFKB1 (right) or TIRAP (below) expression in PBMCs as measured by flow cytometry and quantitative PCR (normalized to ACTb), respectively.

772

haematologica | 2019; 104(4)

TIRAP, a novel familial lymphoma risk gene

mined by Ki-67 was observed in TIRAP p.R81C compared to TIRAP WT family members (Figure 5E). This was a surprising finding since circulating B-cells are generally quiescent,38,39 which we observed in age- and gender-matched healthy individuals as well as TIRAP WT family members. Sanger sequencing confirmed the absence of TIRAP p.R81C variant in healthy donors (Online Supplementary Figure S3). Activation of TLR

through lipopolysaccharide (LPS) further enhanced B-cell proliferation, preferentially in TIRAP p.R81C individuals (Figure 5E). The higher level of assessed B-cell proliferation in TIRAP p.R81C cases was positively correlated with the increased gene expression levels of TIRAP, as well as BCL2L1 and NFKB1, as main regulators of NF-kB signaling in PBMCs of all family members (Figure 5F). Collectively, our data indicate a potential role of p.R81C

Figure 6. NF-kB signaling mediated by TIRAP is important for B-cell survival. (A) Peripheral blood mononuclear cells (PBMCs) were isolated from family members and transfected with control (CTRL) or TIRAP-directed siRNA. Twenty-four hours following transfection, cell viability of CD20+ B-cells was assessed by Annexin V/DAPI staining and flow cytometry analysis. A representative flow cytometry dot plot and quantification of living B-cells are shown in (A) and (B), respectively. Student t-test, *P<0.05. (C) Heatmap showing hierarchical clustering of mRNA levels of genes involved in NF-kB pathway, cell survival and proliferation in PBMCs of mother (M), sister 2 (S2), brother (B) and father (F) treated as described in (A). Data were clustered using standard Euclideanâ&#x20AC;&#x2122;s method based on the average linkage. (D) Bar chart showing the relative expression levels of NFKB1, CASP9, IL6 and MYC genes normalized to ACTb in family members with wild-type (WT) (brother and father) and p.R81C (sister 2 and mother) TIRAP. Dots represent values of each individual of the investigated family. (E) The difference of gene expression in cells transfected with TIRAP or CTRL siRNA was calculated from the mean values shown (D).

haematologica | 2019; 104(4)

773

R. Burkhard et al.

Figure 7. TIRAP p.R81C enhances NF-kB activity and protects against stress-induced apoptosis. 293T cells were transfected with TIRAP WT-GFP, TIRAP p.R81C-GFP or empty vector (EV)-GFP plasmids. Twenty-four hours (h) post transfection, 293T cells were FACS purified for GFP-positive cells. (A) Heatmap showing hierarchical clustering of mRNA levels of six selected genes involved in NF-kB pathway, cell survival and proliferation within GFP-positive 293T cells. Bar chart showing the log2 fold difference of gene expression in transfected 293T cells (TIRAP p.R81C or WT vs. EV) of a representative experiment. (B) FACS purified GFP-positive 293T were cultured in starvation medium for 48 h, and cell viability was assessed by flow cytometry. Bar chart (mean±Standard Error of Mean) show percentage of viable cells (AnnexinV– and Viability dye–) of 2 independent experiments performed in replicates. One-way ANOVA with Bonferroni post hoc test was used, ***P<0.001.

774

haematologica | 2019; 104(4)

TIRAP, a novel familial lymphoma risk gene

variant in activating NF-kB leading to enhanced B-cell proliferation and survival.

INF-kB signaling mediated by TIRAP is important for B-cell survival To study the effect of TIRAP on cell survival, we performed a siRNA-mediated knockdown of endogenous TIRAP in PBMCs isolated from available family members. Overall, TIRAP knockdown efficiency was 60% at the mRNA level (Online Supplementary Figure S4A). TIRAP knockdown significantly diminished the number of living B-cells (Figure 6A and B). This effect was independent of the p.R81C, indicating that TIRAP is an important determinant for B-cell survival. Furthermore, we profiled the expression of genes important for cell survival and proliferation including NF-kB target genes (Figure 6C). Silencing TIRAP strongly reduced the expression of the NF-kB target genes (NFKB1, IL6 and MYC) in PBMCs, indicating that both WT and p.R81C TIRAP mediate signal through the NF-kB pathway (Figure 6D and Online Supplementary Figure S4). However, downregulation of NF-kB target genes was more pronounced in TIRAP p.R81C PBMCs suggesting that these cells particularly rely on the NF-kB pathway (Figure 6E and Online Supplementary Figure S4). Consistent with the reduced B-cell survival, CASP9 expression increased following TIRAP silencing in PBMCs to a higher extent in p.R81C mutated cells (Figure 6D and E). Interestingly, NF-kB signature was further reduced following stimulation with LPS (Online Supplementary Figure S5), supporting the concept that TIRAP transduces signals from TLR4.20

TIRAP p.R81C drives NF-kB pathway activity and reduces stress-induced cell death To evaluate the functional consequence of TIRAP p.R81C, 293T cells were transfected with bidirectional plasmids encoding for EGFP and TIRAP p.R81C, TIRAP WT or empty vector (control), respectively. Under homeostatic conditions, we did not observe any changes in cell viability 24 hours (h) post transfection (Online Supplementary Figure S6). However, gene expression analysis on GFP-positive transfected cells revealed increased expression of the NF-kB target genes NFKB1, BCL2L1, CDKN2A and MYC by TIRAP p.R81C compared to WT (Figure 7A). Thus, we tested whether these transcriptional changes could affect cell viability upon stress-induced challenge. Therefore, sorted GFP-positive cells were cultured in minimal starving media for 48 h. Surprisingly, cell viability was significantly reduced in control or TIRAP WT transfected cells (Figure 7B). Remarkably, our data indicate that TIRAP p.R81C variant is an upstream activator of NF-kB which leads to a better cell survival/proliferation via enhanced NF-kB activity and decreased stressinduced cell death.

Discussion The etiology of DLBCL is poorly understood. Familial clustering of lymphoma is reported to increase disease risk, indicating a role for genetic factors.15,16 Although familial lymphoma cases are rare, studying such pedigrees might identify disease-causing variations and lead to a better understanding of lymphomagenesis. A Finnish family with 3 siblings affected by PMBL and a cousin with extrahaematologica | 2019; 104(4)

nodal DLBCL has been described.19 These lymphomas segregate with the p.H1845N mutation in MLL. The role of this variant in lymphomagenesis has not been corroborated by functional studies, and it is to the best of our knowledge the only DLBCL/PMBL predisposing mutation that has been described so far. Here, we studied a Swiss/Japanese family with 2 sisters affected by B-cell lymphomas in the mediastinum. Although at initial diagnosis their lymphomas were considered, according to the current WHO classification,1 as distinct DLBCL subtypes, the characterization of the somatic lesions by WES and aCGH revealed noticeable molecular similarities. Their somatic landscape is marked by multiple alterations affecting important players of the JAK-STAT signaling cascade which collectively lead to constitutive pathway activity, known to be crucially implicated in lymphomagenesis.14 Shared gains of 9p24/JAK2 and 12q13 (STAT2 and STAT6) were detected by aCGH. Furthermore, we identified missense hotspot mutations in STAT6 in both lymphomas (p.N417 and p.D417). A significant enrichment of STAT6 mutations in PMBL has been described, with mutations being present in more than 30% of cases.9,22 9p24/JAK2 gains are also recurrent genetic alterations in PMBL, but are not strictly confined to this subtype.13,40 Constitutive activation of the NF-kB pathway is a hallmark of both ABC-DLBCL and PMBL, and promotes survival of malignant cells. Somatic oncogenic mutations in components of the B-cell receptor signaling pathways including CD79A/B and CARD11 activate NF-kB.14 Gainof-function mutations in MYD88 have been described in 29% of ABC-DLBCLs.14 Moreover, TNFAIP3, which negatively regulates the NF-kB pathway, is somatically inactivated in one-third of ABC-DLBCLs and PMBLs.14 Interestingly, germline mutation in TNFAIP3 and CARD11 have been described in lymphomas complicating primary Sjรถgren syndrome and congenital B-cell lymphocytosis, respectively.41,42 We searched for possible risk alleles associated with the lymphomas in the family investigated by WES, and discovered germline variants in TIRAP and IL1R1 (detailed data on IL1R1 not shown). The latter was not among the final candidate genes, as the homozygous mutation was present in all family members. Nevertheless, the presence of germline variants in two upstream regulators of NF-kB in a PMBL family is an interesting finding that confirms the importance of the pathway in lymphogenesis. Of note, our data support the cooperation between rare germline variants and constitutive pathway activation in malignant lymphomas.43 The predicted damaging effect of TIRAP p.R81C variant occurred at a highly conserved amino acid in close proximity to the functional TIR domain that is stabilized by two internal disulfide bonds.44,45 Therefore the substitution of an arginine by a cysteine might have implications on the TIRAP protein interaction with downstream signaling proteins. Of note, an arginine to cysteine mutation in MYD88, another adapter molecule involved in NF-kB signaling, diminished its interaction with TIRAP.46 In mice, deletion of Tirap reduced B-cell proliferation in response to TLR4 signaling.20 In agreement with this, we observed a direct correlation between B-cell proliferation and TIRAP expression. In this family, a high expression of TIRAP correlated with the p.R81C genotype. In 420 primary DLBLC cases, high TIRAP levels correlated with a poor survival and were significantly increased in high-risk 775

R. Burkhard et al. DLBCLs.32,47 Furthermore, we linked the enhanced proliferation of TIRAP p.R81C B cells with an increased expression of NF-kB target genes and genes involved in cell survival and proliferation in PBMCs. In this context, it is important to stress that we determined the lymphocytes to account for more than 65% of cells in the PMBC samples analyzed. In addition, TIRAP knockdown was paralleled by a significant decrease in the gene expression signature of the NF-kB pathway, particularly in PBMCs carrying the p.R81C variant. In contrast, overexpression of TIRAP p.R81C increased NF-kB gene signature in vitro. Moreover, TIRAP p.R81C-expressing cells showed enhanced resistance to stress-induced apoptosis, indicating that TIRAP p.R81C provides a survival advantage under those conditions. Our data link TIRAP p.R81C variant/expression with increased B-cell proliferation as well as survival, and thus add TIRAP to the existing network of lymphoma risk genes that are associated with deregulated NF-kB signaling such as TNFAIP3, CD79A/B, MYD88 and CARD11. Interestingly, all these genes were unmutated in the lymphomas of both sisters. Similar to patients expressing the oncogenic p.L265P MYD88 variant, patients with aberrant TIRAP signaling might benefit from IRAK4-selective kinase inhibitors.48 Diffuse large B-cell lymphoma is a polygenic disease with a complex pathogenesis. Therefore, additional alterations are required for the full malignant transformation of B-cells. Interestingly, all the family members investigated were found to have a homozygous germline loss of GSTT1, a reported risk factor for lymphomas (Online Supplementary Figure S7).18 One can hypothesize that the interplay of the germline TIRAP p.R81C variant and

References 1. Swerdlow SH, World Health Organization, International Agency for Research on Cancer, editors. WHO classification of tumours of haematopoietic and lymphoid tissues. Revised 4th edition. Lyon: International Agency for Research on Cancer; 2017. 2. Rosenwald A, Wright G, Leroy K, et al. Molecular diagnosis of primary mediastinal B cell lymphoma identifies a clinically favorable subgroup of diffuse large B cell lymphoma related to Hodgkin lymphoma. J Exp Med. 2003;198(6):851-862. 3. Savage KJ, Monti S, Kutok JL, et al. The molecular signature of mediastinal large Bcell lymphoma differs from that of other diffuse large B-cell lymphomas and shares features with classical Hodgkin lymphoma. Blood. 2003;102(12):3871-3879. 4. Pasqualucci L, Trifonov V, Fabbri G, et al. Analysis of the coding genome of diffuse large B-cell lymphoma. Nat Genet. 2011; 43(9):830-837. 5. Morin RD, Mendez-Lago M, Mungall AJ, et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature. 2011;476(7360):298-303. 6. Morin RD, Mungall K, Pleasance E, et al. Mutational and structural analysis of diffuse large B-cell lymphoma using whole-genome sequencing. Blood. 2013; 122(7):1256-1265. 7. Lohr JG, Stojanov P, Lawrence MS, et al.

776

10.

11.

12.

GSTT1 loss coupled with additional genomic changes culminated in B-cell transformation in the investigated family. The identification of the TIRAP p.R81C variant in a family with mixed ethnic background, along with the demonstration of distinct targets of recurrent mutations in Chinese DLBCLs,49 might be an important additional aspect. Overall, our findings revealed TIRAP p.R81C to be a potential lymphoma risk variant in a family of mixed ethnic background. Our analysis complements the existing view on the different players of the NF-kB pathway crucially involved in DLBCL. Acknowledgments The authors would like to thank Magne Osteras and his team at Fasteris (Geneva, Switzerland) for their excellent technical assistance. The probe for the TNFAIP3 FISH analysis was a gift from Laura Pasqualucci and Riccardo Dalla-Favera (Institute for Cancer Genetics, Columbia University, New York, NY, USA). The determination of CIITA rearrangement by FISH was kindly performed by Laurence de Leval (Institute of Pathology, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland). We thank Lukas Mager, Deborah Krauer, Mario P. Tschan, Emilly Auma, Adrian F. Ochsenbein, Radek Skoda, Federico Santoni, André Schaller, and Outi Kilpivaara for helpful discussions and critical comments. Funding This work was supported by grants from the "Bernische Krebsliga", the "Werner und Hedy Berger-Janser Stiftung", the "Bernische Stiftung für klinische Krebsforschung", the "Stiftung für klinisch-experimentelle Tumorforschung", and a grant from the Inselspital/Bern University Hospital, all to UN.

Discovery and prioritization of somatic muta-tions in diffuse large B-cell lymphoma (DLBCL) by whole-exome sequencing. Proc Natl Acad Sci U S A. 2012; 109(10):38793884. Zhang J, Grubor V, Love CL, et al. Genetic heterogeneity of diffuse large B-cell lymphoma. Proc Natl Acad Sci. 2013; 110(4):1398-1403. Dubois S, Viailly P-J, Mareschal S, et al. Next-Generation Sequencing in Diffuse Large B-Cell Lymphoma Highlights Molecular Divergence and Therapeutic Opportunities: a LYSA Study. Clin Cancer Res. 2016;22(12):2919-2928. Gunawardana J, Chan FC, Telenius A, et al. Recurrent somatic mutations of PTPN1 in primary mediastinal B cell lymphoma and Hodgkin lymphoma. Nat Genet. 2014;46(4):329-335. Wessendorf S, Barth TFE, Viardot A, et al. Further delineation of chromosomal consensus regions in primary mediastinal B-cell lymphomas: an analysis of 37 tumor samples using high-resolution genomic profiling (array-CGH). Leukemia. 2007; 21(12):24632469. Joos S, Otaño-Joos MI, Ziegler S, et al. Primary mediastinal (thymic) B-cell lymphoma is characterized by gains of chromosomal material including 9p and amplification of the REL gene. Blood. 1996; 87(4):1571-1578.

13. Lenz G, Wright GW, Emre NCT, et al. Molecular subtypes of diffuse large B-cell lym-phoma arise by distinct genetic pathways. Proc Natl Acad Sci U S A. 2008; 105(36):13520-13525. 14. Pasqualucci L, Dalla-Favera R. Genetics of diffuse large B cell lymphoma. Blood. 2018;131(21):2307-2319. 15. Altieri A, Bermejo JL, Hemminki K. Familial risk for non-Hodgkin lymphoma and other lymphoproliferative malignancies by histopathologic subtype: the Swedish Family-Cancer Database. Blood. 2005; 106(2):668-672. 16. Goldin LR, Björkholm M, Kristinsson SY, Turesson I, Landgren O. Highly increased familial risks for specific lymphoma subtypes. Br J Haematol. 2009;146(1):91-94. 17. Cerhan JR, Berndt SI, Vijai J, et al. Genomewide association study identifies multiple susceptibility loci for diffuse large B cell lymphoma. Nat Genet. 2014;46(11):12331238. 18. Skibola CF, Curry JD, Nieters A. Genetic susceptibility to lymphoma. Haematologica. 2007;92(7):960. 19. Saarinen S, Kaasinen E, KarjalainenLindsberg M-L, et al. Primary mediastinal large B-cell lymphoma segregating in a family: exome sequencing identifies MLL as a candi-date predisposition gene. Blood. 2013;121(17):3428-3430. 20. Horng T, Barton GM, Medzhitov R. TIRAP:

haematologica | 2019; 104(4)

TIRAP, a novel familial lymphoma risk gene

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

an adapter molecule in the Toll signaling pathway. Nat Immunol. 2001;2(9):835–841. Li H. Toward better understanding of artifacts in variant calling from high-coverage samples. Bioinformatics. 2014;30(20):28432851. Ritz O, Guiter C, Castellano F, et al. Recurrent mutations of the STAT6 DNA binding domain in primary mediastinal Bcell lymphoma. Blood. 2009;114(6):12361242. Green MR, Monti S, Rodig SJ, et al. Integrative analysis reveals selective 9p24.1 am-plification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma. Blood. 2010;116(17):3268-3277. Mottok A, Woolcock B, Chan FC, et al. Genomic Alterations in CIITA Are Frequent in Primary Mediastinal Large B Cell Lymphoma and Are Associated with Diminished MHC Class II Expression. Cell Rep. 2015;13(7):1418-1431. Challa-Malladi M, Lieu YK, Califano O, et al. Combined Genetic Inactivation of Beta2Microglobulin and CD58 Reveals Frequent Escape from Immune Recognition in Diffuse Large B-cell Lymphoma. Cancer Cell. 2011;20(6):728-740. Xu-Monette ZY, Wu L, Visco C, et al. Mutational profile and prognostic significance of TP53 in diffuse large B-cell lymphoma patients treated with R-CHOP: report from an In-ternational DLBCL Rituximab-CHOP Consortium Program Study. Blood. 2012;120(19):3986-3996. Valera A, López-Guillermo A, CardesaSalzmann T, et al. MYC protein expression and genetic alterations have prognostic impact in patients with diffuse large B-cell lym-phoma treated with immunochemotherapy. Haematologica. 2013;98(10):1554-1562. Steidl C, Shah SP, Woolcock BW, et al. MHC class II transactivator CIITA is a recurrent gene fusion partner in lymphoid cancers. Nature. 2011;471(7338):377-381. Kiyasu J, Miyoshi H, Hirata A, et al. Expression of programmed cell death ligand 1 is associated with poor overall survival in patients with diffuse large B-cell lymphoma. Blood. 2015;126(19):2193-2201. Visco C, Tzankov A, Xu-Monette ZY, et al. Patients with diffuse large B-cell lymphoma

haematologica | 2019; 104(4)

31.

32. 33.

34.

35.

36.

37.

38.

39.

of germinal center origin with BCL2 translocations have poor outcome, irrespective of MYC status: a report from an International DLBCL rituximab-CHOP Consortium Program Study. Haematologica. 2013;98(2): 255-263. Cancer data extracted from the Swiss national dataset managed by the Fondation Na-tional Institute for Cancer Epidemiology and Registration (NICER). Available from: http://www.nicer.org/. Accessed on February 2015. http://www.nicer.org/ de/[Last accessed September 15, 2016]. Lenz G, Wright G, Dave SS, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008;359(22):2313-2323. Bonetti P, Testoni M, Scandurra M, et al. Deregulation of ETS1 and FLI1 contributes to the pathogenesis of diffuse large B-cell lymphoma. Blood. 2013;122(13):22332241. Chapuy B, Stewart C, Dunford AJ, et al. Molecular subtypes of diffuse large B cell lym-phoma are associated with distinct pathogenic mechanisms and outcomes. Nat Med. 2018;24(5):679-690. Schmitz R, Wright GW, Huang DW, et al. Genetics and Pathogenesis of Diffuse Large B-Cell Lymphoma. N Engl J Med. 2018; 378(15):1396-1407. Richards S, Aziz N, Bale S, et al. Standards and Guidelines for the Interpretation of Sequence Variants: A Joint Consensus Recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. Higasa K, Miyake N, Yoshimura J, et al. Human genetic variation database, a reference database of genetic variations in the Japanese population. J Hum Genet. 2016; 61(6):547-553. Caraux A, Klein B, Paiva B, et al. Circulating human B and plasma cells. Age-associated changes in counts and detailed characterization of circulating normal CD138− and CD138+ plasma cells. Haematologica. 2010;95(6):1016-1020. Chong Y, Ikematsu H, Yamaji K, et al. CD27(+) (memory) B cell decrease and apopto-sis-resistant CD27(-) (naive) B cell increase in aged humans: implications for age-related peripheral B cell developmental disturbances. Int Immunol. 2005;17(4):383390.

40. Meier C, Hoeller S, Bourgau C, et al. Recurrent numerical aberrations of JAK2 and de-regulation of the JAK2-STAT cascade in lymphomas. Mod Pathol. 2009; 22(3):476-487. 41. Nocturne G, Boudaoud S, Miceli-Richard C, et al. Germline and somatic genetic variations of TNFAIP3 in lymphoma complicating primary Sjögren’s syndrome. Blood. 2013;122(25):4068-4076. 42. Snow AL, Xiao W, Stinson JR, et al. Congenital B cell lymphocytosis explained by nov-el germline CARD11 mutations. J Exp Med. 2012;209(12):2247-2261. 43. Yang Y, Schmitz R, Mitala J, et al. Essential role of the linear ubiquitin chain assembly complex in lymphoma revealed by rare germline polymorphisms. Cancer Discov. 2014;4(4):480-493. 44. Valkov E, Stamp A, DiMaio F, et al. Crystal structure of Toll-like receptor adaptor MAL/TIRAP reveals the molecular basis for signal transduction and disease protection. Proc Natl Acad Sci. 2011;108(36):1487914884. 45. Lin Z, Lu J, Zhou W, Shen Y. Structural Insights into TIR Domain Specificity of the Bridging Adaptor Mal in TLR4 Signaling. PloS One. 2012;7(4):e34202. 46. Bernuth H von, Picard C, Jin Z, et al. Pyogenic Bacterial Infections in Humans with MyD88 Deficiency. Science. 2008; 321(5889):691-696. 47. Aguirre-Gamboa R, Gomez-Rueda H, Martínez-Ledesma E, et al. SurvExpress: an online biomarker validation tool and database for cancer gene expression data using survival analysis. PloS One. 2013;8(9):e74250. 48. Kelly PN, Romero DL, Yang Y, et al. Selective interleukin-1 receptor–associated kinase 4 inhibitors for the treatment of autoimmune disorders and lymphoid malignancy. J Exp Med. 2015;212(13):21892201. 49. Miranda NFCC de, Georgiou K, Chen L, et al. Exome sequencing reveals novel mutation targets in diffuse large B-cell lymphomas derived from Chinese patients. Blood. 2014;124(16):2544-2553. 50. Juilland M, Gonzalez M, Erdmann T, et al. CARMA1- and MyD88-dependent activation of Jun/ATF-type AP-1 complexes is a hallmark of ABC diffuse large B-cell lymphomas. Blood. 2016;127(14):1780-1789.

777

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):778-788

Non-Hodgkin Lymphoma

Pharmacological modulation of CXCR4 cooperates with BET bromodomain inhibition in diffuse large B-cell lymphoma

Clara Recasens-Zorzo,1 Teresa Cardesa-Salzmann,2 Paolo Petazzi,3 Laia Ros-Blanco,4 Anna Esteve-Arenys,1 Guillem Clot,1 Martina Guerrero-Hernández,1 Vanina Rodríguez,1 Davide Soldini,2 Alexandra Valera,2 Alexandra Moros,1 Fina Climent,5 Eva González-Barca,6 Santiago Mercadal,6 Leonor Arenillas,7 Xavier Calvo,7 José Luís Mate,8 Gonzalo Gutiérrez-García,9 Isolda Casanova,3,10 Ramón Mangues,3,10 Alejandra Sanjuan-Pla,11 Clara Bueno,3 Pablo Menéndez,3,12 Antonio Martínez,1,2 Dolors Colomer,1,2 Roger Estrada Tejedor,4 Jordi Teixidó,4 Elias Campo,1,2 Armando López-Guillermo,1,9 José Ignacio Borrell,4 Luis Colomo,2,7 Patricia Pérez-Galán1 and Gaël Roué1,13

Division of Hemato-Oncology, Institut d'Investigacions Biomediques August Pi I Sunyer (IDIBAPS), CIBERONC, Barcelona; 2Hematopathology Unit, Department of Pathology, Hospital Clinic, Barcelona; 3Josep Carreras Leukemia Research Institute, Department of Biomedicine, School of Medicine, University of Barcelona; 4Grup d’Enginyeria Molecular, IQS School of Engineering, Universitat Ramon Llull, Barcelona; 5Pathology Department, Hospital Universitari de Bellvitge, L’Hospitalet de Llobregat; 6Institut Catalá d’Oncología, Hospital Duran I Reynals, L’Hospitalet de Llobregat; 7Pathology Department, IMIM, Hospital del Mar, Barcelona; 8Pathology Department, Hospital Universitari Germans Trias i Pujol, Badalona; 9Department of Hematology, Hospital Clinic, Barcelona; 10Grup d’Oncogènesi i Antitumorals, lnstitut d’Investigacions Biomèdiques Sant Pau (IIB-Sant Pau) and Centro de Investigación Biomédica en Red CIBER-BBN, Barcelona; 11 Hematology Research Group, Health Research Institute La Fe, Valencia; 12Institucio Catalana de Recerca I Estudis Avançats (ICREA), CIBERONC, Barcelona and 13Laboratory of Experimental Hematology, Department of Hematology, Vall d'Hebron Institute of Oncology, Vall d’Hebron University Hospital, Barcelona, Spain 1

LC, PP-G and GR. share senior authorship.

Correspondence: GAËL ROUÉ groue@vhebron.net PATRICIA PÉREZ-GALÁN pperez@clinic.ub.es Received: September 8, 2017. Accepted: June 25, 2018. Pre-published: June 28, 2018. doi:10.3324/haematol.2017.180505 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/778 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

778

ABSTRACT

onstitutive activation of the chemokine receptor CXCR4 has been associated with tumor progression, invasion, and chemotherapy resistance in different cancer subtypes. Although the CXCR4 pathway has recently been suggested as an adverse prognostic marker in diffuse large B-cell lymphoma, its biological relevance in this disease remains underexplored. In a homogeneous set of 52 biopsies from patients, an antibody-based cytokine array showed that tissue levels of CXCL12 correlated with high microvessel density and bone marrow involvement at diagnosis, supporting a role for the CXCL12-CXCR4 axis in disease progression. We then identified the tetra-amine IQS01.01RS as a potent inverse agonist of the receptor, preventing CXCL12mediated chemotaxis and triggering apoptosis in a panel of 18 cell lines and primary cultures, with superior mobilizing properties in vivo than those of the standard agent. IQS-01.01RS activity was associated with downregulation of p-AKT, p-ERK1/2 and destabilization of MYC, allowing a synergistic interaction with the bromodomain and extra-terminal domain inhibitor, CPI203. In a xenotransplant model of diffuse large Bcell lymphoma, the combination of IQS-01.01RS and CPI203 decreased tumor burden through MYC and p-AKT downregulation, and enhanced the induction of apoptosis. Thus, our results point out an emerging role of CXCL12-CXCR4 in the pathogenesis of diffuse large B-cell lymphoma and support the simultaneous targeting of CXCR4 and bromodomain proteins as a promising, rationale-based strategy for the treatment of this disease.

haematologica | 2019; 104(4)

CXCR4 and MYC dual targeting in DLBCL

Introduction

Methods

Diffuse large B-cell lymphoma (DLBCL) is the most common type of non-Hodgkin lymphoma among adults, accounting for 30-40% of newly diagnosed cases.1 Although the introduction of rituximab into clinical practice has increased the survival of affected patients by 1015%,2 60% of patients with high-risk DLBCL are still not cured by immunochemotherapy and have a dismal outcome. For this subgroup, the development of more effective salvage strategies remains an important objective. Gene expression profiling studies have confirmed the heterogeneity of DLBCL, not otherwise specified, and have recognized two major subtypes according to the putative cell of origin, i.e. activated B-cell (ABC) and germinal center B-cell (GCB).3 These studies have also evidenced the role of the stromal microenvironment in the pathogenesis of the disease, as well as in environmentmediated resistance of DLBCL cells to chemotherapeutics.4 As normal B cells are strongly dependent on soluble cytokines for their development and throughout their whole lifespan, it is not surprising that malignant B cells exploit their microenvironment interaction properties for their own selective advantage.5 The CXCR4 chemokine receptor (fusin, CD184) has a well-known function in normal B-cell development, including homing of hematopoietic stem cells to the bone marrow, B-cell and T-cell lymphopoiesis, leukocyte trafficking, and B-cell positioning in the germinal center, among others.6-11 CXCR4 overexpression has been linked to metastasis in a variety of cancers and recently identified as an adverse prognostic factor in DLBCL.12,13 Accordingly, the CXCR4 ligand, CXCL12 (SDF-1α), is among the genes included in the proangiogenic “stromal 2 gene signature” associated with an unfavorable outcome in DLBCL. 4 This cytokine is secreted by normal and tumor stroma and is a major regulator of cell chemotaxis.14 Leukemia stem cells and other CXCR4-expressing tumors utilize the CXCL12-CXCR4 signaling axis to localize to vascular and endosteal niches normally restricted to hematopoietic stem cells,15 thus obtaining protection from the effects of cytotoxic chemotherapy and making these niches look like a reservoir for minimal residual disease and relapses.16-18 CXCR4 expression allows tumor cell migration, and homing of the neoplastic cells to sites where nonmalignant stromal cells express CXCL12.15 This latter promotes tumor progression by recruiting CD31+ endothelial progenitor cells and consequent tumor angiogenesis.19-21 CXCR4 is expressed in hematologic tumors as diverse as B-cell acute lymphoblastic leukemia, acute myeloid leukemia, chronic lymphocytic leukemia and multiple myeloma, and various ongoing clinical trials for patients with relapsed/refractory hematologic malignancies and recurrent high-grade glioma are evaluating the benefit of targeting the tumor microenvironment through CXCL12-CXCR4.22-27 Here we analyzed the clinical significance of CXCL12 expression level in a homogeneous series of patients with de novo DLBCL. We further characterized a new, potent CXCR4 inhibitor showing in vitro and in vivo combinational activity with a BET bromodomain inhibitor, thereby demonstrating that dual targeting of CXCR4 and MYC represents a promising therapeutic strategy for DLBCL.

Patients’ samples

haematologica | 2019; 104(4)

Fifty-two biopsy specimens from untreated patients with de novo DLBCL from the Catalan lymphoma-study group (GELCAB) were included in this study (see details in Online Supplementary Table S1 and the Online Supplementary Materials and Methods). For functional studies, primary tumor cells from five patients were isolated, cryopreserved, and conserved within the hematopathology collection of our institution (Hospital Clínic-IDIBAPS Biobank R121001-094), as previously described.28 Primary cultures were maintained in complete medium in the presence of the mesenchymal cell line StromaNKTert (Riken BioResource Center)29 at a 4:1 ratio, to prevent spontaneous ex vivo apoptosis. The ethical approvals for this project, including informed consent from the patients, were granted following the guidelines of the Hospital Clínic Ethics Committee (Institutional Review Board, registration number 2012/7498).

Cell lines Thirteen DLBCL cell lines from both GCB and ABC subtypes were used in this study. SUDHL-4, SUDHL-6, SUDHL-8, SUDHL-16, NUDHL-1 and U2932 cell lines were purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ). SUDHL-2 and WSU-DLCL2 were obtained from the American Type Culture Collection (ATCC) cell bank (LGC Standards). OCI-LY8 and Toledo were kindly provided by Dr M. Raffeld (National Cancer Institute, Bethesda, MD, USA) and Dr MA Piris (Fundación Jiménez Díaz, Madrid, Spain). OCI-LY3 and OCI-LY10 cells were provided by Dr A. Staiger (Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany). HBL-1 was provided by Dr E Valls (Transcriptional regulation of gene expression group, IDIBAPS, Barcelona, Spain). Cell lines were authenticated upon reception by short tandem repeat profiling, using an AmpFlSTR identifier kit (Thermo Fisher Scientific), and based on available short tandem repeat profiles. All cell lines were cultured routinely at 37ºC in a humidified atmosphere with 5% carbon dioxide in RPMI 1640 or Iscove modified Dulbecco medium supplemented with 10% to 20% heat-inactivated fetal bovine serum, 2 mM glutamine and 50 mg/mL penicillin-streptomycin (Thermo Fisher). Mycoplasm infection was routinely tested for by polymerase chain reaction. SUDHL-6 cells expressing green fluorescent protein (GFP) and Luciferase reporter genes, were generated as previously reported.30

Immunohistochemistry Fifty-two DLBCL samples were included in tissue microarrays using duplicated cores of 1 mm per tumor sample. CD31+ microvascular density was stained and quantified as previously described.31 Microvascular density values were grouped in quartiles and considered high or low when above or below the 50th percentile. Xenograft tumor samples were stained for phosphohistone H3, cleaved caspase-3 and MYC, as previously described.32 Phospho-Akt was detected using a monoclonal antiphospho-Akt-Ser473 antibody (Cell Signaling Technology). Preparations were evaluated on a DP70 or a BX51 microscope using Cell B Basic Imaging Software (Olympus).

Western blot analysis Whole cell proteins were extracted from 107 DLBCL cells as previously described.28 Proteins (30-50 mg/lane) were subjected to 10-12% sodium dodecylsulfate-polyacrylamide gel electrophoresis, and transferred onto polyvinylidene difluoride

779

C. Recasens-Zorzo et al.

membranes (Immobilon-P; Millipore), or nitrocellulose membranes (Amersham Biosciences). Membranes were then incubated with primary and secondary antibodies (Online Supplementary Materials and Methods) and visualized on a mini-LAS4000 device (Fujifilm) by enhanced chemiluminescence (ECL, Amersham Life Science).

Flow cytometry analysis For the detection of surface CXCR4, DLBCL cells (2x105) were stained with a phycoerythrin (PE)-labeled anti-CXCR4 or isotype control antibody (BD Biosciences). For the CXCR4 occupancy assay, cells were pretreated for 1 h with IQS-01.01, AMD3100 (Sigma-Aldrich) or an anti-human CXCR4 control antibody, followed by sCXCR4 staining as above. For quantification of apoptosis, cells were labeled with Pacific Blue-conjugated annexin V (Thermo Fisher). A total of 10,000 events were acquired and analyzed on an Attune acoustic focusing cytometer (Thermo Fisher). The mean fluorescence intensity ratio was calculated as the ratio between the median fluorescence intensity of the CXCR4-labeled sample and that of the isotype.

Chemotaxis assay DLBCL cells (5x106 cells/mL) were cultured for 1 h in culture medium not containing fetal bovine serum but supplemented with 0.5% bovine serum albumin (Sigma-Aldrich), in the presence or absence of 100 mM AMD3100 or IQS-01.01RS, and analyzed for CXCL12-dependent chemotaxis, as previously described.28 Values are referred to cells cultured without CXCL12.

Cell proliferation assay DLBCL primary cultures and cell lines (5x104 cells) were incubated for 24-48 h with IQS-01.01RS, AMD3100 and/or CPI203 (kindly provided by Constellation Pharmaceuticals) at the indicated doses. MTT (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) reagent (Sigma-Aldrich) was added for 1â&#x20AC;&#x201C;5 additional hours before spectrophotometric measurements. Each measurement was made in triplicate. Values are represented using untreated control cells as the reference. Combination indexes were calculated using Calcusyn software version 2.0 (Biosoft). The interaction between two drugs was considered synergistic when the combination index was less than 0.8.

In vivo assays Prior to efficacy testing, an acute toxicity assay was conducted in NSG mice exposed to IQS-01.01RS salt or AMD3100. Animals (2 adult males and 2 adult females per dose) received a single administration of vehicle, IQS01.01-RS (per os) or plerixafor (intraperitoneally) at doses ranging from 2 to 10 mg/kg, and were monitored during the first 4 h after administration, and daily for 2 weeks, for viability/mortality and vital parameters. This toxicity assay defined a maximum non-lethal dose of 5 mg/kg for AMD3100, while the maximum lethal dose of IQS01.01RS remained unreached.

For the systemic DLBCL model, 8-week old NSG female mice (n=12) were inoculated intravenously with 107 SUDHL6-GFPLuc cells, randomly assigned into three equivalent cohorts, and treated daily with 5 mg/kg AMD3100 (intraperitoneally), 10 mg/kg IQS-01.01RS salt (per os) or vehicle. After 27 days, animals were sacrificed and peripheral blood was collected by submandibular puncture. Erythrocytes were lysed using ACK buffer (Quality Biological) and the percentage of SUDHL-6 cells was evaluated by detection of a GFP signal on an Attune cytometer. For the subcutaneous DLBCL model, SUDHL6-GFP-Luc cells were inoculated subcutaneously in 8-week old NSG female mice as previously.28,32 Animals were randomly assigned into four cohorts of four mice each and were given a twice daily dose of 1.5 mg/kg CPI203 (intraperitoneally), a daily dose of 2 mg/kg IQS-01.01RS (per os), both agents, or an equal volume of vehicle. Tumor engraftment was determined weekly following mice injection with 75 mg/kg D-luciferine (AnaSpec) and bioluminescence imaging (BLI) using an Aequoria Luxiflux device equipped with an ORCA-ER camera (Hamamatsu). The bioluminescence imaging signal was quantified using Image J software. Tumor volume was measured by external calipers twice a week, up to 28 days. Animals were then sacrificed and tumors were harvested and weighed. Animals were handled following protocols approved by the Animal Ethics Committee of the University of Barcelona (protocol #154/16).

Statistical analysis A Wilcoxon rank-sum test, Fisher exact test, Spearman rank correlation or t-test was used to examine the statistical significance of associations between clinico-pathological data and CXCL12-CXCR4 positivity or cytokine expression level, as appropriate. Survival curves were estimated using the KaplanMeier method. A log-rank test was used to compare survival curves between groups. For in vitro and in vivo functional assays, unpaired and paired t-tests were employed to obtain the statistical significance. The Benajmini-Hochberg method was used to adjust P-values for multiple testing. Results were considered statistically significant when the P value was less than 0.05.

Results CXCL12 expression correlates with microvascular density and confers an unfavorable prognosis to patients with diffuse large B-cell lymphoma Gene expression profiling studies have highlighted the prognostic value of the microenvironment and components of several immune regulatory pathways in DLBCL.4 To characterize which cytokines may have a significant impact on the pathogenesis of DLBCL, we performed a fluorescence-based cytokine antibody array using frozen tissue from 12 DLBCL cases, out of an initial cohort of 52 patients (Online Supplementary Table S1), and correlated expression data with clinico-biological characteristics of

Table 1. Top cytokines associated with high microvessel density in diffuse large B-cell lymphoma biopsies.

Cytokine SDF1Îą/CXCL12 IGFBP2 IGFBP1

Mean (MVD=low)

Mean (MVD=high)

Difference

t statistic

Adjusted P

5.207 4.467 5.085

5.637 4.946 5.601

0.430 0.479 0.516

-3.758 -3.744 -3.666

0.000 0.000 0.001

0.032 0.032 0.032

MVD: microvessel density.

780

haematologica | 2019; 104(4)

CXCR4 and MYC dual targeting in DLBCL

the patients. Among the 175 cytokines included in this array, we found that only CXCL12, IGFBP2 and IGFBP3 correlated significantly with high microvascular density (P<0.05) (Table 1), a parameter associated with an unfavorable prognosis in DLBCL.31,33 Additionally, high expression of CXCL12 correlated significantly with bone marrow involvement at diagnosis (P=0.02, data not shown), thus supporting a role for the CXCL12-CXCR4 axis in the progression of DLBCL.

IQS-01.01RS is a novel CXCR4 inhibitor with improved pharmacological properties To evaluate the therapeutic potential of targeting CXCR4 in DLBCL, we took advantage of a novel family of CXCR4 inhibitors with improved pharmacodynamic properties and lower cardiotoxicity than the standard CXCR4 inhibitor AMD3100.34,35 Similarly to AMD3100, IQS-01.01 is a symmetric molecule with a central p-

phenylene group harboring two chiral carbons. The compound thus comports three stereoisomers, namely IQS01.01RR, IQS-01.01SS and IQS-01.01RS (Figure 1A). Nitrogen atoms on each site are intercalated in carbon chains at a similar distance as those of the cyclam in AMD3100 providing IQS-01.01 with positively charged nitrogen atoms at physiological pH which, as with AMD3100, will interact with the lateral chains of acidic amino acids of CXCR4, but with greater flexibility. We first assessed the inhibitory activity of IQS-01.01 (racemic mixture), and its three individually purified stereoisomers, using a CXCR4 antagonist screening study. AMD3100 was used as a reference drug. As shown in Figure 1B, the three stereoisomers had superior CXCR4 antagonist activity to that of AMD3100, with IQS01.01RS (see structure in Figure 1C) being the most potent agent. This compound led to a 181% inhibition of receptor activity, thus resembling an inverse receptor ago-

Figure 1. Design of a new potent inhibitor of CXCR4. (A) Skeletal structure of IQS-01.01. *chiral carbons. (B) Inhibition of CXCL12-mediated intracellular cAMP release was determined in the presence of a racemic mixture of IQS-01.01 and its three individually purified stereoisomers, using AMD3100 blocking activity as a reference control. (C) Ball-and-stick representation of IQS-01.01RS. * chiral carbons. (D) MTT assay showing superior antitumor effect of IQS-01.01RS compared to that of the IQS-01.01 racemic mixture after 48 h. The graph shows mean values obtained from three GCB-DLBCL cell lines (SUDHL6, SUDHL-16, and WSU-DLCL2) and three ABC-DLBCL cell lines (OCI-LY3, OCI-LY10 and SUDHL-2). (E) Inhibition of CXCL12-induced migration upon DLBCL cell treatment with a 100 mM dose of IQS01.01 racemic mixture or IQS-01.01RS. Mean values for the SUDHL-6 and OCI-LY3 cell lines are shown. *P<0.05, **P<0.01, ***P<0.001.

haematologica | 2019; 104(4)

781

C. Recasens-Zorzo et al.

Figure 2. IQS-01.01RS has better pharmacological properties than those of AMD3100. (A) Predicted docking of IQS-01.01RS (red) and AMD3100 (blue) on their target, CXCR4 (orange). CXCL12 is represented in green. (B) A CXCR4 occupancy assay shows the competition between IQS-01.01RS or AMD3100 (100 mM) with a phycoerythrin-labeled anti-CXCR4 antibody for binding to the receptor. A blocking antibody (30 mg/mL, R&D Systems) was used as a control. (C) Inhibition of CXCL12induced migration upon DLBCL cell treatment with increasing doses of IQS-01.01RS or AMD3100. The graphs show mean values from three GCB-DLBCL cell lines (SUDHL8, Toledo and SUDHL-6) and two ABC-DLBCL cell lines (U2932 and OCI-LY3). Data are representative of at least three independent experiments. (D) Mean percentage of tumor B cells detected in blood samples of NSG mice injected intravenously with SUDHL-6-GFP-LUC cells and treated for 27 days with IQS-01.01RS, AMD3100, or vehicle (n=4 animals/group). (E) Time-dependent antitumor effect of IQS-01.01RS (100 mM) and AMD3100 (100 mM) in a panel of 13 DLBCL cell lines; the effect was determined by a MTT assay. Representative results from three experiments are shown. (F) Relative induction of apoptosis in CD19+ tumor B cells upon treatment of primary DLBCL biopsies (n=5) with the indicated doses of IQS-01.01RS for 48 h. The mean viability of untreated primary cells was 79±8%. (G) Western blot analysis of CXCR4 downstream signaling in SUDHL6 and U2932 cells upon 2 h starvation, followed by exposure to recombinant CXCL12 for 1 min, with or without pretreatment with the indicated doses of IQS-01.01RS or AMD3100. β-actin was used as a loading control. (H) CXCR4 downstream signaling and MYC modulation in SUDHL-6 cells at different time points, after 2 h starvation followed by receptor triggering in the presence or absence of 100 mM IQS-01.01RS. β-actin was used as a loading control. Ab: antibody: TM: transmembrane. *P<0.05, ***P<0.001.

782

haematologica | 2019; 104(4)

CXCR4 and MYC dual targeting in DLBCL

Table 2. Sensitivity of diffuse large B-cell lymphoma cell lines to CXCR4 and BET bromodomain inhibition.

Subtype

GCB

ABC

CXCR4 expression (MFI-R)

IQS-01.01RS cytotoxic effect (100 µM)

CPI203 cytotoxic effect (0.5 µM)

Combination index

36 25 41 1 3 52 91 1 4 1 1 46 1

32% 37% 42% 36% 35% 9% 38% 36% 48% 40% 53% 43% 63%

38% 59% 47% 86% 55% 43% 14% 60% 75% 40% 95% 31% 39%

1.1 0.6 0.4 0.5 0.4 0.5 1.1 0.8 0.7 0.8 0.8 0.9 0.6

Cell line OCI-LY8 SUDHL-4 SUDHL-6 SUDHL-16 WSU-DLCL2 Toledo SUDHL-8 NUDHL-1 OCI-LY3 OCI-LY10 SUDHL-2 U2932 HBL-1

ABC: activated B-cell; GCB: germinal center B-cell; MFI-R: median fluorescence intensity ratio.

nist. We then analyzed the antitumor activity of this highly active stereoisomer, together with IQS-01.01 (racemic mixture) in six representative DLBCL cell lines of both GCB and ABC subtypes. As shown in Figure 1D, IQS-01.01RS was significantly more active than the racemic mixture at all the doses tested, with the greatest effect being observed at the 100 mM dose (mean cytotoxicity: 44%; range, 42-49%). This effect was mainly attributed to proliferation blockade, as IQS-01.01RS-treated cultures showed low (< 5%) levels of apoptosis (data not shown). In agreement with this, migration experiments using a CXCL12 gradient further demonstrated that IQS01.01RS was a more potent inhibitor of cell chemotaxis than was the racemic mixture, with a 2-fold improvement of cell migration blockade (Figure 1E). Based on these results, we proceeded to a deeper study of the stereoisomer of choice, IQS-01-01RS. A predictive docking model showed that this stereoisomer localized into the extracellular CXCL12 binding domain of CXCR4, close to the position occupied by AMD3100, and simultaneously interacted with an inner, transmembrane domain of the receptor. This unique feature potentially provides IQS-01.01RS with the ability to inhibit CXCR4 in the presence of the ligand, and to interfere with the catalytic activity of the receptor (Figure 2A). Supporting this model, using the representative cell line SUDHL-6 we observed that IQS-01.01RS and AMD3100 had similar CXCR4 occupancy activity (Figure 2B), and had similar anti-migratory properties against a CXCL12 gradient in vitro, both in ABC- and GCB-DLBCL cell lines (Figure 2C). However, IQS-01.01RS triggered superior mobilization of DLBCL cells into the circulating blood of the animals (Figure 2D) and was able to block the proliferation of a panel of 13 GCB/ABC-DLBCL cell lines in a time-dependent manner, contrasting with the modest antitumor activity of AMD3100 (40% versus 12% mean cell growth inhibition at 48 h) (Figure 2E and Table 2). In addition, in a set of five DLBCL primary cultures, a dosedependent induction of apoptosis was observed upon exposure to IQS-01.01RS (Figure 2F), contrasting with the reported inability of AMD3100 to induce cell death.36 Accordingly, the analysis of CXCR4 downstream signaling showed that the capacity of IQS-01.01RS to inhibit basal and CXCL12-induced phosphorylation of ERK1/2, haematologica | 2019; 104(4)

AKT and the AKT downstream kinase GSK3-β, was superior to that of AMD3100 (Figure 2G). Of special interest, IQS-01.01RS, rather than AMD3100, allowed strong downregulation of the MYC proto-oncogene in ABC- and GCB-DLBCL cells (Figure 2H). This downstream target of CXCR4 with a well-established role in the progression of DLBCL,37 has been reported to depend on AKT phosphorylation level for its stabilization, mediated by a GSK3-β-dependent phosphorylation of its Thr58 residue and consequent proteosomal degradation.38,39 Supporting the hypothesis that the downregulation of MYC may be consequent to inhibition of the CXCR4-AKT axis in IQS-01.01RS-treated cells, the expression of both p-AKT and MYC was almost completely abrogated within a few minutes after the addition of IQS-01.01RS (Figure 2H) and this phenomenon could be maintained for at least 3 h (data not shown). Collectively, these results indicate that the new CXCR4 inhibitor IQS-01.01RS has a unique structure that enables its binding to two distinct domains of the receptor, conferring improved mobilizing properties in vivo, as well as superior antitumor activity in vitro.

IQS-01.01RS cooperates with BET bromodomain inhibition in vitro and in vivo Based on our observation that IQS-01.01RS had the capacity to downregulate MYC, we further investigated in our 13 DLBCL cell lines whether the compound could cooperate with the BET bromodomain antagonist CPI20332 to block cell proliferation. The combination of each agent induced up to 78% cytoxicity at the optimal doses of 100 mM IQS-01.01RS and 500 nM CPI203, with a mean CI of 0.67, indicative of a synergistic drug interaction in both ABC- and GCB-DLBCL cells (Figure 3A and Table 2). As expected, a dramatic reduction in MYC protein levels was observed after exposure to the combination (Figure 3B). We confirmed by quantitative polymerase chain reaction analysis and cycloheximide protein stability assays that this phenomenon was consequent to a simultaneous CPI203-evoked transcriptional repression of MYC gene and an IQS-01.01RS-mediated destabilization of MYC protein (Figure 3C,D). Confirming that this effect was not due to a BRD4 inhibitor-mediated diminished production of CXCR4, as previously described in T 783

C. Recasens-Zorzo et al.

lymphocytes following JQ1 treatment,40 Figure 3B shows that CPI203 addition did not alter CXCR4 protein levels either in the presence or absence of IQS-01.01RS. Flow cytometry analysis further confirmed that surface levels of CXCR4 were unmodified by BET bromodomain inhibition (data not shown). To further ascertain the role of CXCR4 expression in this drug interaction, we established genetic depletion of CXCR4 in SUDHL-6 cells employing a CRISPR/Cas9 gene editing tool to generate CXCR4 knocked out (KO) cell lines, prior to treatment with CPI203, IQS-01.01RS and the combination of both agents. In parallel, and as a control of non-lymphomatous B malignancy, we used the recently described NALM6CXCR4-KO cell line (kindly provided by Dr Jan Burger, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA). As shown in Online Supplementary Figure S1, CXCR4-KO cell lines

completely lack CXCR4 molecules on their surface. The main difference between these two cell lines was the response to CPI203 as a single agent, which was substantially weaker in NALM6-CXCR4-KO cells. This may illustrate the recent finding that, in acute myeloid leukemia cells, CXCR4 signaling is involved in the regulation of MYC transcription mediated by the downregulation of the miRNA let-7a,41 while in DLBCL cells no variation in MYC mRNA levels are observed following CXCR4 ligation (Online Supplementary Figure S2). Another explanation may come from the lower membrane expression (about one log10) of the receptor observed in parental acute myeloid leukemia cells when compared to parental DLBCL cells, suggesting that NALM6 cells are less dependent on CXCR4 and less susceptible to CXCR4dependent cellular stress than are SUDHL-6 cells. Most importantly, in the acute myeloid leukemia cell line the

Figure 3. IQS-01.01RS synergizes with the BET bromodomain inhibitor CPI203 in vitro. (A) The relative antitumor effect of IQS-01.01RS (100 mM), CPI203 (0.5 mM) and the combination of both was determined by a MTT assay, after 48 h. The data shown are the mean results of the 13 DLBCL cell lines. (B) Cooperation between IQS-01.01RS and CPI203 in the inhibition of CXCR4 downstream signaling, as assessed by western blot analysis of p-AKT and MYC. SUDHL-6 and U2932 cells were starved for 2 h, and treated for 1 h with 100 mM IQS-01.01RS and/or CPI203 (0.5 mM) prior to a 1 min stimulation with 200 ng/mL recombinant CXCL12. Îą-tubulin was used as a loading control. (C) Relative MYC transcript levels in SUDHL-6 and U2932 cells upon 6 h treatment with 100 mM IQS-01.01RS, 0.5 mM CPI203 and the combination of both. Control untreated cells were used as a reference. (D) Time-dependent determination of MYC protein levels in SUDHL-6 cells treated with the translational blocker cycloheximide, as previously described,62 in the presence or absence of 100 mM IQS01.01-RS. Î˛-actin was used as a loading control. CI: combination index; CHX: cycloheximide; COMBO: combination treatment with IQS-01.01RS and CPI203.

784

haematologica | 2019; 104(4)

CXCR4 and MYC dual targeting in DLBCL

loss of CXCR4 was associated with 30% and 24% decreases in the cytotoxic effect of CPI203 and the CPI203/IQS-01.01RS combination, respectively, confirming that CXCR4 expression was required for the activity of the bromodomain inhibitor alone, as well as for its synergistic activity with IQS-01.01RS. In the case of SUDHL6, although the combination was still active in CXCR4-depleted cells, the loss of the receptor induced a 13% reduction in the synergistic activity of the two agents when compared with the parental CXCR4+ cells.

Thus, these results strongly support a crucial role for CXCR4 in the cooperation between CPI203 and IQS01.01RS in malignant B cells. To assess the efficacy of the drug combination in vivo, NSG mice were subcutaneously injected with SUDHL6GFP-Luc cells, and given IQS-01.01RS, CPI203, the combination of both agents, or the equivalent volume of vehicle, for 13 days. Tumor burden was evaluated weekly by bioluminescence signal recording and twice a week by external calipers. At the final time point, tumors were

Figure 4. IQS-01.01RS and CPI203 cooperate to reduce tumor growth in a subcutaneous mouse model of diffuse large B-cell lymphoma. NSG mice were subcutaneously injected with SUDHL6-GFP+Luc+ cells and tumor-bearing mice were randomly assigned to one of the following treatment arms (4 mice per group): IQS01.01RS 2 mg/kg daily (per os), CPI203 1.5 mg/kg BID (intraperitoneally), both agents or equal volume of vehicle, for 2 weeks. (A) Tumor volume was evaluated twice a week using external calipers. (B) Tumor burden was evaluated at week 3 and week 4 by analysis of the bioluminescence signal. Left panel: color maps of two representative animals per group. Right panel: quantification of luciferase activity using Image J software. (C) Mean tumor weight in each treatment group at the final time point. (D) Immunohistochemical labeling of p-histone H3, activated caspase-3, MYC and p-AKT in consecutive tissue sections from four representative tumor specimens (magnification x200). act casp3: activated caspase 3; COMBO/combo: combination treatment with IQS-01.01RS and CPI203; SEM: standard error of mean; ns=not significant; *P<0.05.

haematologica | 2019; 104(4)

785

C. Recasens-Zorzo et al.

extracted and weighed. Figure 4A shows that CPI203 and IQS-01.01RS as single agents induced 27% and 4% reductions in tumor growth, respectively, while the combination of both drugs induced a 38% decrease in tumor burden. Accordingly, reduced luciferase activity as well as a significant (38%) decrease in tumor weight was detected in mice given combination treatment compared to the values in the group administered the vehicle (Figure 4B,C). Histological analysis of the corresponding tumors confirmed a greater reduction of the mitotic index together with an accumulation of apoptotic cells by the combination therapy, as assessed by phospho-histone H3 and activated-caspase-3 staining (Figure 4D). In agreement with the in vitro results, greater reductions in MYC and phospho-Akt were observed in animals treated with the combination of drugs (Figure 5D). Collectively, these results suggest that the combination of IQS-01.01RS with the BET inhibitor CPI203 enhances the antitumor properties of each single agent through the blockade of CXCR4 signaling, followed by the abrogation of MYC expression and the induction of apoptosis.

Discussion The chemokine receptor CXCR4 has a prominent role in homing and retention of tumor cells in their microenvironment, and in the promotion of drug resistance.42 It has been previously reported that CXCR4/CXCL12 expression by tumor cells confers an adverse prognosis to DLBCL patients.12,43 In agreement with this, we observed that CXCL12 expression in tumor biopsies correlated with bone marrow involvement at diagnosis, as well as highly vascularized tumors. These observations support the hypothesis that the CXCL12-CXCR4 axis may play a significant role in neo-angiogenesis in the DLBCL microenvironment. Accordingly, various reports have highlighted a correlation between CXCR4 expression and high serum levels of VEGF in different cancer subtypes.19,21,44 CXCL12 is also one of the genes included in the DLBCL pro-angiogenic and unfavorable â&#x20AC;&#x153;stromal-2 signatureâ&#x20AC;?,4 while activation of the CXCL12-CXCR4 axis is related to tumor angiogenesis and recruitment of endothelial progenitor cells to tumors in myelodysplastic syndrome, glioma and pancreatic cancer.20,45,46 The CXCL12-CXCR4 axis may, therefore, represent a new therapeutic target for DLBCL, as inhibition of CXCR4 and tumor cell mobilization is bound to increase the accessibility of these lymphomas to anticancer therapies. Validating this strategy, the Food and Drug Administration-approved drug AMD3100 has been shown to potentiate the tumor growth inhibition afforded by standard chemotherapeutics and/or irradiation in preclinical models of glioblastoma,26,47 to inhibit the engraftment of B-cell acute lymphoblastic leukemia,48 to mobilize and sensitize acute promyelocytic leukemia cells to cytosine arabinoside,49 and to enhance the response to rituximab and to the anti-CD52 monoclonal antibody alemtuzumab in a mouse model of disseminated lymphoma.50 In the clinical setting, encouraging preliminary data have been obtained from a phase II study using AMD3100 and standard chemotherapy in patients with acute myeloid leukemia.51 In DLBCL, pharmacological inhibition of CXCR4 by AMD3100 impairs the propagation of tumor B cells in a systemic mouse model of the 786

disease.12 Nonetheless, AMD3100 presents serious limitations that could preclude its use as an anticancer drug, such as some degree of cardiotoxicity and adverse pharmacodynamic properties including a positive charge at physiological pH and a limited half-life, impairing its oral bioavailability.52 Thus, the lack of a CXCR4 inhibitor suitable for administration in a standard regimen supports the development of less toxic and more stable CXCR4targeting agents.53 Here, we describe the biological activity of IQS01.01RS, a new potential CXCR4 inhibitor with a higher lethal dose in vivo and better pharmacodynamic properties than AMD3100.35 We report that by interacting with a different CXCR4 domain than that with which AMD3100 interacts, this IQS-01.01RS acts as a putative allosteric CXCR4 inhibitor, impeding the activation of the receptor upon CXCL12 ligation and allowing a more sustained inhibition of the pathway. In vivo, this effect, together with better pharmacokinetic properties, was linked to an improved capacity of the compound to mobilize DLBCL cells. Furthermore, we show that IQS01.01RS-mediated blockade of CXCR4 signaling led to the post-transcriptional downregulation of MYC. Overexpression of this oncogene is associated with shorter survival in DLBCL patients,54 and is known to be stabilized by CXCR4 in cancer cells.55 MYC was considered to be an undruggable factor until (+)-JQ1, a highly potent, selective and cell-permeable inhibitor of BRD4, a member of the BET family of chromatin adaptors, was first shown to have antitumor activity in multiple myeloma xenograft models.56 Further studies with (+)-JQ1 and CPI203, a molecule characterized by a superior bioavailability after oral or intraperitoneal administration,32,57 have validated BET bromodomain targeting and subsequent blockade of MYC, NF-kB or BCL-2 protein family signaling as a promising therapeutic strategy in different subtypes of aggressive B-cell lymphoma, including DLBCL.58-60 BRD4 is a global regulator of gene transcription which selectively recognizes and binds to acetylated lysine residues in histones to activate transcription and mitosis. In DLBCL it preferentially localizes in superenhancers associated with key transcription factors implicated in lymphomagenesis, such as MYC, which are specifically sensitive to BRD4 inhibition thus explaining the selective anti-tumor effect of BRD4 inhibitors.61 It is worth pointing out that BET bromodomain inhibition, as a consequence, downregulates other tumor-related genes apart from MYC.61 Nevertheless, in this work, we focused on the capacity of CPI203 to inhibit MYC, confirming that ABC- and GCB-DLBCL cells are highly sensitive to CPI203 monotherapy, and that the transcriptional downregulation of MYC achieved by this compound allows almost complete abrogation of the protein when the compound is combined with IQS-01.01RS. This dual approach underlies the synergistic interaction of the two compounds and consequent sensitization to apoptosis in vitro and in vivo. In conclusion, the present work shows that, besides CXCR4, the level of CXCL12 may also have an impact on the progression of DLBCL, and describes a new potent orally available CXCR4 inhibitor with antitumor properties. In addition, this study offers the first rational basis for the potential clinical evaluation of a dual approach combining BET bromodain inhibition and CXCR4 blockade in DLBCL. haematologica | 2019; 104(4)

CXCR4 and MYC dual targeting in DLBCL

References 1. Teras LR, DeSantis CE, Cerhan JR, Morton LM, Jemal A, Flowers CR. 2016 US lymphoid malignancy statistics by World Health Organization subtypes. CA Cancer J Clin. 2016 Sep 12. [Epub ahead of print] 2. Miyazaki K. Treatment of diffuse large Bcell lymphoma. J Clin Exp Hematop. 2016;56(2):79-88. 3. Alizadeh AA, Eisen MB, Davis RE, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000;403(6769):503-511. 4. Lenz G, Wright G, Dave SS, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008;359(22):2313-2323. 5. Shain KH, Dalton WS, Tao J. The tumor microenvironment shapes hallmarks of mature B-cell malignancies. Oncogene. 2015;34(36):4673-4682. 6. Allen CDC, Ansel KM, Low C, et al. Germinal center dark and light zone organization is mediated by CXCR4 and CXCR5. Nat Immunol. 2004;5(9):943-952. 7. Blades MC, Manzo A, Ingegnoli F, et al. Stromal cell-derived factor 1 (CXCL12) induces human cell migration into human lymph nodes transplanted into SCID mice. J Immunol. 2002;168(9):4308-4317. 8. Foudi A, Jarrier P, Zhang Y, et al. Reduced retention of radioprotective hematopoietic cells within the bone marrow microenvironment in CXCR4-/- chimeric mice. Blood. 2006;107(6):2243-2251. 9. 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. 10. 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. 11. Scala S. Molecular pathways: targeting the CXCR4-CXCL12 axis--untapped potential in the tumor microenvironment. Clin Cancer Res. 2015;21(19):4278-4285. 12. Moreno MJ, Bosch R, eguez-Gonzalez R, et al. CXCR4 expression enhances diffuse large B cell lymphoma dissemination and decreases patient survival. J Pathol. 2015;235(3):445-455. 13. Zlotnik A, Burkhardt AM, Homey B. Homeostatic chemokine receptors and organ-specific metastasis. Nat Rev Immunol. 2011;11(9):597-606. 14. Bleul CC, Fuhlbrigge RC, Casasnovas JM, Aiuti A, Springer TA. A highly efficacious lymphocyte chemoattractant, stromal cellderived factor 1 (SDF-1). J Exp Med. 1996;184(3):1101. 15. Nagasawa T. The chemokine CXCL12 and regulation of HSC and B lymphocyte development in the bone marrow niche. Adv Exp Med Biol. 2007;602(69-75. 16. Burger JA, Kipps TJ. Chemokine receptors and stromal cells in the homing and homeostasis of chronic lymphocytic leukemia B cells. Leuk Lymphoma. 2002;43(3):461-466. 17. Burger JA, Kipps TJ. CXCR4: a key receptor in the crosstalk between tumor cells and their microenvironment. Blood. 2006;107 (5):1761-1767. 18. Burger JA, Peled A. CXCR4 antagonists: targeting the microenvironment in leukemia and other cancers. Leukemia. 2009;23(1):43-52.

haematologica | 2019; 104(4)

19. Liang Z, Brooks J, Willard M, et al. CXCR4/CXCL12 axis promotes VEGFmediated tumor angiogenesis through Akt signaling pathway. Biochem Biophys Res Commun. 2007;359(3):716-722. 20. Ping YF, Yao XH, Jiang JY, et al. The chemokine CXCL12 and its receptor CXCR4 promote glioma stem cell-mediated VEGF production and tumour angiogenesis via PI3K/AKT signalling. J Pathol. 2011;224(3):344-354. 21. Sun X, Charbonneau C, Wei L, Yang W, Chen Q, Terek RM. CXCR4-targeted therapy inhibits VEGF expression and chondrosarcoma angiogenesis and metastasis. Mol Cancer Ther. 2013;12(7):1163-1170. 22. Cho BS, Kim HJ, Konopleva M. Targeting the CXCL12/CXCR4 axis in acute myeloid leukemia: from bench to bedside. Korean J Intern Med. 2017;32(2):248-257. 23. Kast RE. Profound blockage of CXCR4 signaling at multiple points using the synergy between plerixafor, mirtazapine, and clotrimazole as a new glioblastoma treatment adjunct. Turk Neurosurg. 2010;20(4):425429. 24. Konopleva MY, Jordan CT. Leukemia stem cells and microenvironment: biology and therapeutic targeting. J Clin Oncol. 2011;29(5):591-599. 25. Liu T, Li X, You S, Bhuyan SS, Dong L. Effectiveness of AMD3100 in treatment of leukemia and solid tumors: from original discovery to use in current clinical practice. Exp Hematol Oncol. 2016;5(1):19. 26. Redjal N, Chan JA, Segal RA, Kung AL. CXCR4 inhibition synergizes with cytotoxic chemotherapy in gliomas. Clin Cancer Res. 2006;12(22):6765-6771. 27. Uy GL, Rettig MP, Motabi IH, et al. A phase 1/2 study of chemosensitization with the CXCR4 antagonist plerixafor in relapsed or refractory acute myeloid leukemia. Blood. 2012;119(17):3917-3924. 28. Balsas P, Esteve-Arenys A, Roldan J et al. Activity of the novel BCR kinase inhibitor IQS019 in preclinical models of B-cell nonHodgkin lymphoma. J Hematol Oncol. 2017;10(1):80. 29. Kawano Y, Kobune M, Yamaguchi M, et al. Ex vivo expansion of human umbilical cord hematopoietic progenitor cells using a coculture system with human telomerase catalytic subunit (hTERT)-transfected human stromal cells. Blood. 2003;101(2): 532-540. 30. Bosch R, Dieguez-Gonzalez R, Cespedes MV, et al. A novel inhibitor of focal adhesion signaling induces caspase-independent cell death in diffuse large B-cell lymphoma. Blood. 2011;118(16):4411-4420. 31. Cardesa-Salzmann TM, Colomo L, Gutierrez G, et al. High microvessel density determines a poor outcome in patients with diffuse large B-cell lymphoma treated with rituximab plus chemotherapy. Haematologica. 2011;96(7):996-1001. 32. Moros A, Rodriguez V, Saborit-Villarroya I, et al. Synergistic antitumor activity of lenalidomide with the BET bromodomain inhibitor CPI203 in bortezomib-resistant mantle cell lymphoma. Leukemia. 2014;28(10):2049-2059. 33. Perry AM, Cardesa-Salzmann TM, Meyer PN, et al. A new biologic prognostic model based on immunohistochemistry predicts survival in patients with diffuse large B-cell lymphoma. Blood. 2012;120(11):22902296. 34. PĂŠrez-Nueno VI, Ritchie DW, Rabal O, Pascual R, Borrell JI, TeixidĂł J. Comparison of ligand-based and receptor-based virtual

35.

36.

37. 38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

screening of HIV entry inhibitors for the CXCR4 and CCR5 receptors using 3D ligand shape matching and ligand-receptor docking. J Chem Inf Model. 2008;48(3): 509-533. Ros-Blanco L, Anido J, Bosser R, et al. Noncyclam tetraamines inhibit CXC chemokine receptor type 4 and target glioma-initiating cells. J Med Chem. 2012;55(17):7560-7570. Reinholdt L, Laursen MB, Schmitz A, et al. The CXCR4 antagonist plerixafor enhances the effect of rituximab in diffuse large B-cell lymphoma cell lines. Biomark Res. 2016;4:12. Campo E. MYC in DLBCL: partners matter. Blood. 2015;126(22):2439-2440. An J, Yang DY, Xu QZ, et al. DNA-dependent protein kinase catalytic subunit modulates the stability of c-Myc oncoprotein. Mol Cancer. 2008;7:32. Sears R, Nuckolls F, Haura E, Taya Y, Tamai K, Nevins JR. Multiple Ras-dependent phosphorylation pathways regulate Myc protein stability. Genes Dev. 2000;14(19):2501-2514. Banerjee C, Archin N, Michaels D, et al. BET bromodomain inhibition as a novel strategy for reactivation of HIV-1. J Leukoc Biol. 2012;92(6):1147-1154. Chen Y, Jacamo R, Konopleva M, Garzon R, Croce C, Andreeff M. CXCR4 downregulation of let-7a drives chemoresistance in acute myeloid leukemia. J Clin Invest. 2013;123(6):2395-2407. Zeng Z, Shi YX, Samudio IJ, et al. Targeting the leukemia microenvironment by CXCR4 inhibition overcomes resistance to kinase inhibitors and chemotherapy in AML. Blood. 2009;113(24):6215-6224. Chen J, Xu-Monette ZY, Deng L, et al. Dysregulated CXCR4 expression promotes lymphoma cell survival and independently predicts disease progression in germinal center B-cell-like diffuse large B-cell lymphoma. Oncotarget. 2015;6(8):5597-5614. Bachelder RE, Wendt MA, Mercurio AM. Vascular endothelial growth factor promotes breast carcinoma invasion in an autocrine manner by regulating the chemokine receptor CXCR4. Cancer Res. 2002;62(24):7203-7206. Tseng D, Vasquez-Medrano DA, Brown JM. Targeting SDF-1/CXCR4 to inhibit tumour vasculature for treatment of glioblastomas. Br J Cancer. 2011;104(12): 1805-1809. Zhang Y, Zhao H, Zhao D, et al. SDF1/CXCR4 axis in myelodysplastic syndromes: correlation with angiogenesis and apoptosis. Leuk Res. 2012;36(3):281-286. Kioi M, Vogel H, Schultz G, Hoffman RM, Harsh GR, Brown JM. Inhibition of vasculogenesis, but not angiogenesis, prevents the recurrence of glioblastoma after irradiation in mice. J Clin Invest. 2010;120(3):694705. Juarez J, Dela PA, Baraz R, et al. CXCR4 antagonists mobilize childhood acute lymphoblastic leukemia cells into the peripheral blood and inhibit engraftment. Leukemia. 2007;21(6):1249-1257. Nervi B, Ramirez P, Rettig MP, et al. Chemosensitization of acute myeloid leukemia (AML) following mobilization by the CXCR4 antagonist AMD3100. Blood. 2009;113(24):6206-6214. Hu Y, Gale M, Shields J, et al. Enhancement of the anti-tumor activity of therapeutic monoclonal antibodies by CXCR4 antagonists. Leuk Lymphoma. 2012;53(1):130-138.

787

C. Recasens-Zorzo et al. 51. Stamatopoulos B, Meuleman N, De BC, et al. AMD3100 disrupts the cross-talk between chronic lymphocytic leukemia cells and a mesenchymal stromal or nurselike cell-based microenvironment: pre-clinical evidence for its association with chronic lymphocytic leukemia treatments. Haematologica. 2012;97(4):608-615. 52. Pettersson S, Perez-Nueno VI, Ros-Blanco L, et al. Discovery of novel non-cyclam polynitrogenated CXCR4 coreceptor inhibitors. ChemMedChem. 2008;3(10): 1549-1557. 53. Kashyap MK, Amaya-Chanaga CI, Kumar D, et al. Targeting the CXCR4 pathway using a novel anti-CXCR4 IgG1 antibody (PF-06747143) in chronic lymphocytic leukemia. J Hematol Oncol. 2017;10(1):112. 54. Valera A, Lรณpez-Guillermo A, CardesaSalzmann T, et al. MYC protein expression and genetic alterations have prognostic impact in patients with diffuse large B-cell lymphoma treated with immunochemotherapy. Haematologica. 2013;98(10):

788

1554-1562. 55. Hatano K, Yamaguchi S, Nimura K, et al. Residual prostate cancer cells after docetaxel therapy increase the tumorigenic potential via constitutive signaling of CXCR4, ERK1/2 and c-Myc. Mol Cancer Res. 2013;11(9):1088-1100. 56. Filippakopoulos P, Knapp S. Targeting bromodomains: epigenetic readers of lysine acetylation. Nat Rev Drug Discov. 2014;13(5):337-356. 57. Diaz T, Rodriguez V, Lozano E, et al. The BET bromodomain inhibitor CPI203 improves lenalidomide and dexamethasone activity in in vitro and in vivo models of multiple myeloma by blockade of Ikaros and MYC signaling. Haematologica. 2017;102(10):1776-1784. 58. Boi M, Gaudio E, Bonetti P, et al. The BET bromodomain inhibitor OTX015 affects pathogenetic pathways in preclinical B-cell tumor models and synergizes with targeted drugs. Clin Cancer Res. 2015;21(7):16281638.

59. Ceribelli M, Kelly PN, Shaffer AL, et al. Blockade of oncogenic IkappaB kinase activity in diffuse large B-cell lymphoma by bromodomain and extraterminal domain protein inhibitors. Proc Natl Acad Sci U S A. 2014;111(31):11365-11370. 60. Esteve-Arenys A, Valero JG, ChamorroJorganes A, et al. The BET bromodomain inhibitor CPI203 overcomes resistance to ABT-199 (venetoclax) by downregulation of BFL-1/A1 in in vitro and in vivo models of MYC+/BCL2+ double hit lymphoma. Oncogene. 2018;37(14):1830-1844. 61. Chapuy B, McKeown MR, Lin CY, et al. Discovery and characterization of superenhancer-associated dependencies in diffuse large B cell lymphoma. Cancer Cell. 2013;24(6):777-790. 62. Moros A, Bustany S, Cahu J, et al. Antitumoral activity of lenalidomide in in vitro and in vivo models of mantle cell lymphoma involves the destabilization of cyclin D1/p27KIP1 complexes. Clin Cancer Res. 2014;20(2):393-403.

haematologica | 2019; 104(4)

ARTICLE

Chronic Lymphocytic Leukemia

Deep targeted sequencing of TP53 in chronic lymphocytic leukemia: clinical impact at diagnosis and at time of treatment

Ferrata Storti Foundation

Christian Brieghel,1 Savvas Kinalis,2 Christina W. Yde,2 Ane Y. Schmidt,2 Lars Jønson,2 Michael A. Andersen,1 Caspar da Cunha-Bang,1 Lone B. Pedersen,1 Christian H. Geisler,1 Finn C. Nielsen2 and Carsten U. Niemann1

1 Department of Hematology, Rigshospitalet and 2Center for Genomic Medicine, Rigshospitalet, Copenhagen, Denmark

Haematologica 2019 Volume 104(4):789-796

ABSTRACT

n chronic lymphocytic leukemia, TP53 mutations and deletion of chromosome 17p are well-characterized biomarkers associated with poor progression-free and overall survival following chemoimmunotherapy. Patients harboring low burden TP53 mutations with variant allele frequencies of 0.3-15% have been shown to have similar dismal outcome as those with high burden mutations. We here describe a highly sensitive deep targeted next-generation sequencing assay allowing for the detection of TP53 mutations as low as 0.2% variant allele frequency. Within a consecutive, single center cohort of 290 newly diagnosed patients with chronic lymphocytic leukemia, deletion of chromosome 17p was the only TP53 aberration significantly associated with shorter overall survival and treatment-free survival. We were unable to demonstrate any impact of TP53 mutations, whether high burden (variant allele frequency >10%) or low burden (variant allele frequency ≤10%), in the absence of deletion of chromosome 17p. In addition, the impact of high burden TP53 aberration (deletion of chromosome 17p and/or TP53 mutation with variant allele frequency >10%) was only evident for patients with IGHV unmutated status; no impact of TP53 aberrations on outcome was seen for patients with IGHV mutated status. In 61 patients at time of treatment, the prognostic impact of TP53 mutations over 1% variant allele frequency could be confirmed. This study furthers the identification of a clinical significant limit of detection for robust TP53 mutation analysis in chronic lymphocytic leukemia. Multicenter studies are needed for validation of ultra-sensitive TP53 mutation assays in order to define and implement a technical as well as a clinical lower limit of detection.

Introduction Chronic lymphocytic leukemia (CLL) is a clonal B-cell malignancy characterized by a heterogeneous clinical course. Prognostic and predictive markers of survival and treatment outcome are essential in management of the disease.1 Deletion of chromosome 17p [del(17p)] and TP53 mutation (TP53mut) remain the most important risk factors for progression-free survival (PFS) and overall survival (OS) following chemo- and chemoimmunotherapy (CIT).2-6 In recent years, B-cell receptor pathway inhibitors (idelalisib, ibrutinib and acalabrutinib) and Bcl-2 inhibitors (venetoclax) have demonstrated remarkable response rates and durable remissions in both treatment naïve and previously treated CLL patients with TP53 aberration (TP53ab: either del(17p) or TP53mut).7-11 Randomized clinical trials comparing CIT directly to targeted therapy in a TP53-aberrated population are still awaited. Thus, assessment of TP53ab is recommended prior to any treatment decision.12 Approximately 80% of patients with del(17p) also harbor TP53muts on the remaining allele, while a subset of patients have TP53muts without del(17p).5 haematologica | 2019; 104(4)

Correspondence: CARSTEN U. NIEMANN carsten.utoft.niemann@regionh.dk Received: April 17, 2018. Accepted: November 23, 2018. Pre-published: Decemer 4, 2018. doi:10.3324/haematol.2018.195818 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/789 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

789

C. Brieghel et al.

However, Sanger sequencing and fluorescence in situ hybridization (FISH) fail to detect 4-5% of newly diagnosed and untreated patients with CLL harboring low burden TP53muts (Sanger negative) without concomitant del(17p).13,14 Deep-targeted next-generation sequencing (NGS) of TP53 has shown that low burden TP53muts with a variant allele frequency (VAF) as low as 0.3% have similar outcome to patients with high burden TP53muts (Sanger positive).13,14 However, recent data from the UK CLL4 trial indicated that low burden TP53muts impacted neither OS nor PFS for patients treated with chemotherapy.15 For newly diagnosed patients harboring only one TP53ab, better OS is demonstrated compared to patients with both del(17p) and TP53mut. Similarly, patients with del(17p) and additional low burden TP53mut show better OS compared to patients with additional high burden TP53mut.16,17 Thus, the impact of additional TP53ab warrants further investigation. Upon therapy, the prevalence and size of TP53 clones increase due to clonal evolution and acquisition of new TP53muts.18,19 Targeted therapy is established as the standard of care for patients with TP53 aberrated CLL.20,21 Whether patients with low burden TP53ab may benefit more from targeted therapy compared to standard CIT still remains open for investigation, as the evidence available so far does not allow definitive guidelines to be formulated.12 Thus, studies to elucidate a technical and a clinically significant limit of detection (LOD) for TP53mut are warranted to guide clinical decisions for these patients. We here describe a robust NGS assay for detection of TP53mut as low as 0.2% VAF. In order to investigate a clinically relevant LOD for low burden TP53mut, we assessed the impact of TP53mut at diagnosis and at time of treatment in a single center cohort of patients with CLL.

Methods Patients and materials All consecutive patients diagnosed with CLL from a single center sampled between January 2007 and October 2014 were included in the study (Online Supplementary Figure S1). Samples from patients obtained within 200 days of the diagnostic flow cytometry were considered newly diagnosed.21 To assess the clinical impact of TP53ab at time of treatment, samples obtained up to 200 days before treatment were included for a separate analysis. All available samples considered newly diagnosed and/or sampled at time of treatment were sequenced. Due to the retrospective nature of the study, TP53 analysis was performed on peripheral blood mononuclear cells (PBMCs) and not on purified CLL cells. For 244 patients (81% of the newly diagnosed cohort) with available flow cytometry data at time of sampling, 197 patients (81%) had CLL populations more than 70% of the PBMCs (see Online Supplementary Methods), thus we report VAFs based on PBMCs. Patients’ characteristics and clinical data were obtained from medical records and registries; CLL-International Prognostic Index (CLL-IPI) factors in terms of age (≤65 vs. >65 years), Binet stage (A vs. B or C), beta-2-microglobulin (β2M) (<4.0 mg/L vs. ≥4.0 mg/L), IGHV mutational status (germline identity <98% vs. ≥98%), and TP53ab only by FISH [no del(17p) vs. del(17p)] were included.3,22 Del(17p) was considered positive if present in at least 10% of 200 interphases. The study was approved by the Danish National Committee on Health Research Ethics, the Data Protection Agency and the Health Authorities involved.

790

TP53 mutational analysis by deep-targeted sequencing A high sensitivity TP53 assay was established based on serial 10-fold dilutions of DNA from patient samples with donor DNA. By including a dilution step for each sequenced sample, background noise was filtered and an LOD was established at 0.2% VAF (Online Supplementary Methods, Online Suppplementary Table S1 and Online Suppplementary Figure S2). For each patient, DNA extracted from PBMCs was analyzed undiluted and diluted 20% (dilution factor 5) in DNA derived from the SU-DHL4 cell line. A known TP53mut (p.Arg273Cys) harbored in the cell line DNA acted as internal control of dilution grade. Using 100 ng gDNA per reaction, TP53 exons 2-10 incl. 2 bp intronic overlap for splice sites were amplified with 30 cycles of PCR using Phusion® HSII HighFidelity DNA polymerase (Life Technologies, Waltham, MA, USA). A list of the primers used is provided in Online Supplementary Table S2. In brief, library preparation was performed following manufacturer protocol KAPA DNA Library Preparation (Nimblegen). Using SeqCap Adapter Kit A and B (Roche NimbleGen) or NEXTflex™ DNA Barcodes 96 (Bioo Scientific, Austin, TX, USA), libraries were pooled (24 or 96 samples per lane) and sequenced as paired-end on a HiSeq2500 using HiSeq® SBS Kit v.4 (2x125 base PE, Illumina) to obtain a minimum target read depth of 20,000x.

Bioinformatic workflow A workflow for detection of low burden variants was developed in CLC Biomedical Genomics Workbench 3.0 (CLC BGW, Qiagen, Hilden, Germany) as described in the Online Supplementary Methods. Achieving a median coverage of 144,158 reads (99% of region > 26,217x), we applied both a dilution match algorithm (DMA) and a stereotypic error model (SEM) described in detail in the Online Supplementary Methods. In brief, only variants that diluted correctly were called TP53mut by DMA (Online Supplementary Figure S3), while SEM identified outliers from the position-specific and nucleotide-specific background noise as true TP53mut based on the distribution of stereotypic errors (Online Supplementary Figures S4 and S5).13 Results from both DMA and SEM were compared using contingency tables, and only true positive variants were considered true mutations and used in subsequent analyses (Online Supplementary Table S3 and Online Supplementary Figure S6).

Validation by droplet digital PCR and Capture based targeted next-generation sequencing Droplet digital PCR (ddPCR) was used to validate initial low burden variants. Allele specific Prime Assay™ probes (Bio-Rad, Hercules, CA, USA) were applied for triplicate analyses using QX200™ Droplet Digital™ PCR System and QuantaSoft™ 1.7 (Bio-Rad) according to instructions from the manufacturer. A custom made SeqCap EZ Choice gene panel (Roche Nimblegen) containing TP53 exons 2-10 was used to validate mutations with a VAF of 1% or over, as described in the Online Supplementary Methods.

Statistical analysis Time to event was calculated from date of diagnosis for treatment-free survival (TFS), and from date of diagnosis or first date of treatment for OS. Patients were followed until initiation of CLLspecific treatment or death or end of follow up, which ever came first, defined as TFS, and until death or end of follow up, whichever came first, defined as overall survival (OS). Analyses were performed using the Kaplan-Meier method, and log-rank test was used to compare outcome. TP53mut were stratified into high and low burden mutations (VAF>10% and VAF ≤10%, respectively) haematologica | 2019; 104(4)

TP53 mutations in CLL

Table 1. Patients’ characteristics of the Danish nationwide cohort and study cohorts for newly diagnosed patients and patients at time of treatment.

Variable Age ≤65 years >65 years Binet stage A B/C Unknown β2M ≤4.0 mg/L >4.0 mg/L Unknown IGHV Mutated Unmutated Unknown FISH No del(17p) Del(17p) Unknown

Nationwide N (%)

Newly diagnosed N (%)

Time of treatment N (%)

1017 (28.8) 2513 (71.2)

20 (41.4) 170 (58.6)

35 (57.4) 26 (42.6)

2804 (79.4) 726 (20.6) 0

232 (84.7) 42 (15.3) 16

20 (32.8) 41 (67.2) 0

2233 (86.0) 365 (14.0) 932

213 (86.6) 33 (13.4) 44

27 (73.0) 10 (27.0) 24

1822 (67.9) 861 (32.1) 847

192 (68.1) 90 (31.9) 8*

17 (30.4) 39 (69.6) 5*

2832 (93.9) 185 (6.1) 513

283 (97.6) 7 (2.4) 0

55 (90.2) 6 (9.8) 0

*Indicates inconclusive IGHV analysis. β2M: beta-2-microglobulin; FISH: fluorescence in situ hybridization.

including minor TP53muts (VAF<1%).12,18 Since allogeneic stem cell transplantation is considered the only curative treatment in CLL, follow up was censored upon allogeneic stem cell transplantation for the cohort analyzed at time of treatment. FISH was not repeated at time of treatment in 5 patients for whom the baseline FISH were extrapolated. All analyses downstream of CLC BGW were performed with R version 3.4.1.23

Results Patient cohorts and impact of baseline characteristics A total of 446 patients were included in our study. We excluded 44 patients with unavailable material and 92 patients who were neither newly diagnosed nor sampled at time of treatment. The two final cohorts included 290 newly diagnosed patients and 61 patients sampled at time of treatment, including 50 patients at time of first-line treatment and 11 at time of later lines of treatment (Online Supplementary Figure S1). Median time from date of diagnosis to sample collection was two days [interquartile range (IQR): 1.25-2.00). During a median follow up of 6.0 years (IQR: 3.9-7.9), 97 (33%) patients received treatment and 81 (28%) deaths were registered among newly diagnosed patients. Compared to the Danish nation-wide cohort, fewer patients were older than 65 years (58.6% vs. 71.2%) and a lower prevalence of Binet stage B/C disease (15.3% vs. 20.6%) as well as a lower prevalence of del(17p) (2.4% vs. 6.1%) were seen in this cohort. Except for age, there were more high-risk features in the 61 patients at time of treatment compared to the newly diagnosed patients (Table 1). haematologica | 2019; 104(4)

Improved robustness of low burden TP53 mutation detection Robust detection of low burden TP53muts was ensured by combining a DMA and an SEM. For DMA, the ratio of variants called in both undiluted and diluted samples from the same patient (dilution ratio, DR) and the reference allele frequency of a known cell line mutation used for dilution (dilution grade, DG) were plotted (Online Supplementary Figure S3). The proximity to a line with a slope of one between the DG and adjusted DR defined true mutations for DMA. For SEM, we modeled frequent variants (observed ≥20) according to each unique genomic position and nucleotide change, while infrequent variants (observed <20) were modeled according to each unique nucleotide change only. VAFs were fitted to gamma distributions allowing for identification of true mutations using Bonferroni correction (Online Supplementary Figures S4 and S5). For the full study cohort of 308 patients, 98 and 116 TP53muts were called by DMA and SEM, respectively (Online Supplementary Tables S4 and S5, and Online Supplementary Figure S6). Using an LOD of 1% VAF, we obtained 100% consistency between DMA and SEM for determination of true mutations (Online Supplementary Table S3C). Between 0.3-1% VAF, 32 true positive TP53muts were further identified while excluding four variants only detected by either DMA or SEM (Online Supplementary Table S3B). Reporting TP53muts as low as 0.2% VAF, 10 additional true positive TP53muts could be identified, while 26 mutations only identified by either DMA or SEM were excluded (Online Supplementary Tables S3A and S4). Validating the first 30 low burden TP53muts identified, all were confirmed by ddPCR with high corre791

C. Brieghel et al.

lation between VAF by ddPCR and deep targeted nextgeneration sequencing (tNGS) (r2=0.999) (Online Supplementary Table S6). Consequently, one TP53mut (0.2% VAF) only identified by DMA was excluded based on SEM (Online Supplementary Figure S7C) but was in fact validated by ddPCR. However, the TP53 status remained unchanged as this patient harbored additional high and low burden mutations. Among the 290 newly diagnosed patients, 41 patients (14%) harbored 18 high burden (VAF >10%) and 31 low burden (VAF ≤10%) mutations, including 20 minor TP53muts (VAF <1%). This resulted in 6 patients with both del(17p) and TP53mut, one with del(17p) only, 10 with high burden TP53mut only, and 25 patients with low burden TP53mut only (Figure 1A and Online Supplementary Table S4). Patients harboring only minor TP53mut were mainly older patients (>65 years) but were otherwise characterized as low risk according to the CLL-IPI (Online Supplementary Table S7A), whereas the distribution of high and low burden TP53muts was similar among patients stratified based on IGHV mutational status (Online Supplementary Table S7B). All mutations were located within exons 4-8 and 80% were missense mutations. Multiple high burden TP53muts were seen in 2 patients, while multiple low burden TP53muts were seen in 5 patients. Among 61 patients at time of treatment, we identified 57 mutations in 17 patients (28%): 7 patients with high burden and 10 with low burden TP53muts, including 4 with minor TP53muts (Figure 1B and Online Supplementary Table S4). Forty-three mutations (75%) were observed in 6 out of 11 previously treated patients with a median VAF of 1.01% (IQR: 0.46-2.93). Five of the 6 patients with del(17p) also harbored TP53muts on the remaining allele at time of treatment (Online Supplementary Table S8A). All mutations were located within exons 4-9 and 72% were missense mutations (Figure 1B). Seven patients harbored

multiple TP53muts. Patients’ characteristics are summarized in Online Supplementary Table S8.

Prognostic impact on newly diagnosed patients Stratifying TP53 aberrated patients into high and low burden (TP53wt vs. VAF ≤10% vs. VAF >10%), only high burden TP53ab [including del(17p)] patients showed a trend for worse OS and significantly worse TFS compared to TP53wt (Figure 2A and B) (P=0.06 and P=0.01, respectively). Further stratifying low burden TP53mut patients (VAF <1% vs. VAF 1-10%), still no impact on OS or TFS was seen for either group (Figure 2C and D), whereas combining the group of patients with a VAF above 1% could demonstrate a significant impact on OS and TFS compared to TP53wt (Online Supplementary Figure S8A and B). However, this was fully dependent on del(17p) patients (P=0.004 and P<0.001, respectively) (Figure 2E and F), as TP53 mutated patients without del(17p) demonstrated similar OS and TFS compared to TP53wt patients (P>0.25) (Figure 2E and F and Online Supplementary Figure S8C and D). Multiple TP53muts were observed in 7 newly diagnosed patients without impact on OS or TFS (1 vs. >1 TP53mut; P>0.2) (data not shown), while multiple TP53ab including del(17p) resulted in a significant impact on OS and a trend for TFS (1 vs. >1 TP53ab; P=0.036 and P=0.051, respectively) (data not shown),

Prediction of treatment outcome At time of treatment, TP53ab was significantly associated with a poor OS compared to patients with TP53wt (P=0.005) (Figure 3A and data not shown). Stratifying TP53aberrated patients into high and low burden, only high burden TP53ab patients demonstrated poor OS compared to TP53wt (P<0.001) (Figure 3A). Further stratifying low burden TP53 mutated patients (VAF <1% vs. VAF 1-10%), OS was significantly worse for patients with TP53mut with VAF 1-10% compared to TP53wt (P=0.002) (Figure

Figure 1. Characterization of TP53 mutations. Number of mutations indicated in bar plots [regardless of del(17p)] and the number of patients according to TP53 status indicated in pie charts]. (A) Located within exons 4-8, 49 mutations were detected in 41 of 290 newly diagnosed patients; 6 of 7 del(17p) patients also harbored TP53 mutations. Eighteen (37%) and 31 (63%) mutations classified as high and low burden, respectively. (B) Fifty-seven TP53 mutations within exons 4-9 were detected in 17 of 61 patients at time of treatment; 5 of 6 del(17p) patients also harbored TP53 mutations. Nine (16%) and 48 (84%) mutations classified as high and low burden, respectively. Primarily missense mutations were detected. All percentages indicate variant allele frequencies (VAF). TP53 mutation without del(17p) (TP53mut).

792

haematologica | 2019; 104(4)

TP53 mutations in CLL

3B). Combining the group of patients with TP53ab with VAF 1-100%, the association persisted when omitting del(17p) patients (Pâ&#x2030;¤0.001) (Figure 3C). Four patients with minor TP53muts were still alive and in complete remission at end of follow up (Figure 3B and C). There was no difference in survival between patients harboring one TP53mut with a VAF greater than 1% (VAF 1-100%) and patients with more than one TP53mut (P=0.85) (data not shown). Although both patients receiving first-line treatment (n=50) and patients receiving later lines of treatment (n=11) were included in the cohort at time of treatment, TP53 status demonstrated a similar negative impact on OS in both subcohorts (data not shown).

TP53 status may predict outcome in newly diagnosed patients with unmutated IGHV status As most patients with TP53ab at time of treatment were also IGHV unmutated (IGHV-U) (Online Supplementary Table S8), we explored the synergy between TP53ab and IGHV mutational status for newly diagnosed patients. For patients with mutated IGHV (IGHV-M), TP53ab [whether low burden or high burden, including 4 patients with del(17p)] did not impact OS or TFS. However, for patients with IGHV-U status, high burden TP53ab [including 3 patients with del(17p)] significantly impacted both OS and TFS (P=0.036 and P=0.005, respectively) (Figure 4).

Figure 2. Overall (OS) and treatment-free survival (TFS) in newly diagnosed patients. Kaplan-Meier curves comparing OS (panels A, C, E) and TFS (panels B, D, F) based on (A and B) TP53 aberrations stratified based on variant allele frequencies (VAF) including del(17p) with 10% cut-off or (C and D) 1% and 10% cutoff. (E and F) Del(17p) and subgroups with TP53 mutations without del(17p) [(TP53mut w/o del(17p)] analyzed separately. P-values are indicated in the tables within the panels.

haematologica | 2019; 104(4)

793

C. Brieghel et al.

Figure 3. Overall survival (OS) in patients from time of treatment. Stratifying patients with TP53 aberrations including del(17p) based on (A) variant allele frequencies (VAF) with 10% cut-off or (B) 1% and 10% cut-off. (C) Del(17p) versus subgroups with TP53 mutations without del(17p) [TP53mut w/o del(17p)] with 1% VAF cut-off is shown. Del(17p) status may reflect baseline if a second fluorescence in situ hybridization (FISH) was not performed at time of treatment. P-values indicated in tables within the panels.

Discussion This study demonstrates that neither high nor low burden TP53muts at time of CLL diagnosis independently influenced OS or TFS in a consecutive cohort of newly diagnosed patients. However, patients with del(17p) at time of diagnosis had an inferior outcome. In addition, the subgroup of patients with TP53ab over 10% VAF among patients with IGHV-U status demonstrated inferior OS and TFS. At time of treatment, patients with sole TP53muts over 1% VAF had shorter OS, as had patients with del(17p). In our study, del(17p) in newly diagnosed CLL was rare (2.4%), although still demonstrating a negative prognostic impact in accordance with our previous validation of CLL-IPI3 in a Danish nation-wide cohort.24 The majority of del(17p) patients were, as expected, also TP53 mutated.5 Although we demonstrate a similar prevalence of TP53 mutated patients and a similar distribution of variant allele frequencies, TP53muts without concomitant FISH posi794

tive for del(17p) were more frequent in newly diagnosed patients (10.7% using an LOD of 0.3%) compared to previous publications.13,14,17 In particular, sole low burden TP53muts (7.2%) was highly prevalent, whereas the prevalence of patients with sole high burden TP53muts (3.4%) was similar to previous reports.13,14,17 Despite a high prevalence, and in contrast to reports by Rossi et al.,13 we could not demonstrate impact on OS of neither high nor low burden TP53muts without del(17p) in newly diagnosed patients. Similar to our results, Stengel et al. demonstrated a better OS in newly diagnosed patients with TP53mut only compared to concomitant del(17p) and TP53mut, which may support the lack of impact on OS in our smaller cohort.17 Furthermore, Nadeu et al. reported no impact on time to treatment among newly diagnosed TP53 mutated patients compared to TP53wt patients.14 More prevalent high-risk factors with impact on early need of treatment observed across previous studies may also contribute to the different impact of TP53muts.13,14,17 For example, the previous studies included older patients haematologica | 2019; 104(4)

TP53 mutations in CLL

Figure 4. Stratified analysis in newly diagnosed patients based on TP53 and IGHV status. Kaplan Meier curves for (A) overall survival (OS) and (B) treatment-free survival (TFS) in newly diagnosed patients stratified for IGHV status and for TP53 aberrations based on variant allele frequencies (VAF) â&#x2030;¤10% and >10%. Four and three del(17p) are included for mutated (IGHV-M) and unmutated IGHV (IGHV-U), respectively. P-values indicated in tables within the panels.

with higher Binet stage and more frequent del(17p), resulting in lower frequencies of TP53-mutated patients without del(17p) than in our cohort. Furthermore, patients in our cohort had a lower CLL-IPI score compared to the Danish nation-wide CLL cohort, probably due to varying regional referral patterns. In contrast to previously published data,25 IGHV-U status was not higher in newly diagnosed TP53 mutated patients, which may in part explain the indolence of our cohort. In our study, synergy was demonstrated for IGHV mutational status and TP53ab.25 For patients with IGHV-U status, high burden TP53ab correlated with poor outcome among newly diagnosed patients. However, TP53ab (whether high or low burden) had no negative prognostic impact on the more indolent disease course for patients with IGHV-M status, in accordance with previous studies.26-29 Thus, the less aggressive phenotype in our cohort may diminish any independent impact of TP53muts, especially due to the proportion of IGHV-M status among TP53 mutated patients. High cell proliferation, shorter time to treatment, and a distinct pattern of nucleotide shifts in patients with IGHV-U may contribute to the mechanisms causing this interaction between IGHV mutational status and TP53ab.18,30 Like the majority of NGS studies investigating the clinical impact of TP53muts,13,14,16,18,31,32 we confirm the negative impact on OS of TP53muts over 1% VAF at time of treatment. A recent study, however, was unable to show this association for patients harboring low burden TP53muts only.15 haematologica | 2019; 104(4)

In our study, minor TP53muts were common among pretreated patients. However, minor TP53muts were observed exclusively as the only TP53ab in treatmentnaĂŻve patients. These newly diagnosed patients with only a single minor TP53mut were mainly older patients with an otherwise favorable risk profile and outcome. Even the 4 patients with minor TP53mut requiring initial treatment (3 with IGHV-U status) were still alive and in complete remission at end of follow up. This may indicate that minor TP53muts as the sole TP53ab is an age-related phenomenon of a more benign character, similar to reports on clonal hematopoiesis in myeloid malignancies.33 In accordance with this, a recent study found TP53muts enriched among older CLL patients.17 Current guidelines for assessment of TP53muts prior to treatment recommend an LOD at 10% VAF for clinical decisions, with the option to report low burden mutations down to 5% VAF by NGS as long as the unresolved clinical significance of such mutations is stated.12 The reason for a caveat when reporting TP53muts below 10% VAF results from: 1) low reproducibility between different NGS platforms in this range; and 2) an uncertain clinical significance of low burden mutations.12 To address the technical question of reproducibility, we here report a SEM of both nucleotide and position specific variants from deep tNGS in combination with an algorithm based on the dilution of patient DNA. By this approach, we have developed a technically robust method for detection of TP53mut that could easily be transferred across different platforms and laboratories. For clinical use, we do, howev795

C. Brieghel et al.

er, recommend using a cell line harboring a rare TP53mut predicted to encode functional p53, such as BRG-A (TP53:c.1060C>G), to avoid both risk of contamination and risk of omitting significant low burden mutations.34 We successfully achieved an LOD of 0.3% VAF, applied in previous studies of minor TP53muts,13,14,18 and could even lower the LOD to 0.2% VAF. As we were unable to prove any impact on newly diagnosed patients with IGHV-M, our results support the current guidelines recommending TP53 assessment only prior to treatment.12

References 1. Parikh SA, Shanafelt TD. Prognostic factors and risk stratification in chronic lymphocytic leukemia. Semin Oncol. 2016; 43(2):233240. 2. Pflug N, Bahlo J, Shanafelt T, et al. Development of a comprehensive prognostic index for patients with chronic lymphocytic leukemia. Blood. 2014;124(1):49-62. 3. International CLLIPIwg. An international prognostic index for patients with chronic lymphocytic leukaemia (CLL-IPI): a metaanalysis of individual patient data. Lancet Oncol. 2016;17(6):779-790. 4. Dohner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med. 2000;343(26):1910-1916. 5. Zenz T, Eichhorst B, Busch R, et al. TP53 Mutation and Survival in Chronic Lymphocytic Leukemia. J Clin Oncol. 2010; 28(29):4473-4479. 6. Stilgenbauer S, Schnaiter A, Paschka P, et al. Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):32473254. 7. Byrd JC, O'Brien S, James DF. Ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(13):1278-1279. 8. Farooqui MZ, Valdez J, Martyr S, et al. Ibrutinib for previously untreated and relapsed or refractory chronic lymphocytic leukaemia with TP53 aberrations: a phase 2, single-arm trial. Lancet Oncol. 2015;1 6(2):169-176. 9. Furman RR, Sharman JP, Coutre SE, et al. Idelalisib and Rituximab in Relapsed Chronic Lymphocytic Leukemia. N Engl J Med. 2014;370(11):997-1007. 10. Roberts AW, Davids MS, Pagel JM, et al. Targeting BCL2 with Venetoclax in Relapsed Chronic Lymphocytic Leukemia. N Engl J Med. 2016;374(4):311-322. 11. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):32-42. 12. Malcikova J, Tausch E, Rossi D, et al. ERIC recommendations for TP53 mutation analysis in chronic lymphocytic leukemia-update

796

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23. 24.

This study furthers the identification of a clinically significant LOD for TP53muts in CLL. The method proposed here for analysis of minor TP53muts warrants validation across laboratories for a standard technical LOD for TP53muts. Subsequent validation and standardization of TP53 mutation assays within networks such as the European Research Initiative on CLL (ERIC, http://ericll.org) may provide the platform needed for collaborative multicenter analyses seeking to define a validated clinical LOD for TP53muts.

on methodological approaches and results interpretation. Leukemia. 2018; 32(5):10701080. Rossi D, Khiabanian H, Spina V, et al. Clinical impact of small TP53 mutated subclones in chronic lymphocytic leukemia. Blood. 2014;123(14):2139-2147. Nadeu F, Delgado J, Royo C, et al. Clinical impact of clonal and subclonal TP53, SF3B1, BIRC3, NOTCH1, and ATM mutations in chronic lymphocytic leukemia. Blood. 2016;127(17):2122-2130. Blakemore S. The Contribution of Gene Mutations to Long-Term Clinical Outcomes: Data from the Randomised UK LRF CLL4 Trial. Blood. 2017;130(Suppl 1):259. Yu L, Kim HT, Kasar SN, et al. Survival of Del17p CLL Depends on Genomic Complexity and Somatic Mutation. Clin Cancer Res. 2016;23(3):735-745. Stengel A, Kern W, Haferlach T, Meggendorfer M, Fasan A, Haferlach C. The impact of TP53 mutations and TP53 deletions on survival varies between AML, ALL, MDS and CLL: an analysis of 3307 cases. Leukemia. 2017;31(3):705-711. Malcikova J, Stano-Kozubik K, Tichy B, et al. Detailed analysis of therapy-driven clonal evolution of TP53 mutations in chronic lymphocytic leukemia. Leukemia. 2015; 29(4):877-885. Landau DA, Carter SL, Stojanov P, et al. Evolution and impact of subclonal mutations in chronic lymphocytic leukemia. Cell. 2013;152(4):714-726. Brown JR, Kay N. Chemoimmunotherapy Is Not Dead Yet in Chronic Lymphocytic Leukemia. J Clin Oncol. 2017;35(26):29892992. Hallek M, Cheson BD, Catovsky D, et al. iwCLL guidelines for diagnosis, indications for treatment, response assessment, and supportive management of CLL. Blood. 2018;131(25):2745-2760. da Cunha-Bang C, Geisler CH, Enggaard L, et al. The Danish National Chronic Lymphocytic Leukemia Registry. Clin Epidemiol. 2016;8(1):561-565. Team RC. R: A Language and Environment for Statistical Computing. 2017. da Cunha-Bang C, Christiansen I, Niemann CU. The CLL-IPI applied in a population-

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

based cohort. Blood. 2016;128(17):21812183. Sutton LA, Ljungstrom V, Mansouri L, et al. Targeted next-generation sequencing in chronic lymphocytic leukemia: a highthroughput yet tailored approach will facilitate implementation in a clinical setting. Haematologica. 2015;100(3):370-376. Delgado J, Salaverria I, Baumann T, et al. Genomic complexity and IGHV mutational status are key predictors of outcome of chronic lymphocytic leukemia patients with TP53 disruption. Haematologica. 2014;99(11):e231-e234. Jeromin S, Haferlach C, Dicker F, Haferlach T, Kern C. Patients with TP53 disruption and IGHV Mutated Status Show Indolent Clinical Course: A Study on 1,327 Treatment-Naive CLL Cases. Blood. 2016; 128(22):4378. Best OG, Gardiner AC, Davis ZA, et al. A subset of Binet stage A CLL patients with TP53 abnormalities and mutated IGHV genes have stable disease. Leukemia. 2009; 23(1):212-214. Sutton LA, Hadzidimitriou A, Baliakas P, et al. Immunoglobulin genes in chronic lymphocytic leukemia: key to understanding the disease and improving risk stratification. Haematologica. 2017;102(6):968-971. Murphy EJ, Neuberg DS, Rassenti LZ, et al. Leukemia-cell proliferation and disease progression in patients with early stage chronic lymphocytic leukemia. Leukemia. 2017;31(6):1348-1354. Landau DA, Tausch E, Taylor-Weiner AN, et al. Mutations driving CLL and their evolution in progression and relapse. Nature. 2015;526(7574):525-530. Puente XS, Bea S, Valdes-Mas R, et al. Noncoding recurrent mutations in chronic lymphocytic leukaemia. Nature. 2015; 526(7574):519-524. Wong TN, Ramsingh G, Young AL, et al. Role of TP53mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature. 2015;2015(518):552-557. Campomenosi P, Fronza G, Ottaggio L, et al. Heterogeneous p53 mutations in a Burkitt lymphoma from an AIDS patient with monoclonal c-myc and VDJ rearrangements. Int J Cancer. 1997;73(6):816-821.

haematologica | 2019; 104(4)

ARTICLE

Chronic Lymphocytic Leukemia

First-line therapy in chronic lymphocytic leukemia: a Swedish nation-wide real-world study on 1053 consecutive patients treated between 2007 and 2013 Sandra Eketorp Sylvan,1 Anna Asklid,1,2 Hemming Johansson,1 Jenny Klintman,3,4 Jenny Bjellvi,5 Staffan Tolvgård,6 Eva Kimby,7 Stefan Norin,7 Per-Ola Andersson,8 Claes Karlsson,1,9 Karin Karlsson,3 Birgitta Lauri,10 Mattias Mattsson,11 Anna Bergendahl Sandstedt,12 Maria Strandberg,13 Anders Österborg,1,9* and Lotta Hansson1,9*

Department of Oncology-Pathology, Karolinska Institutet, Stockholm; 2Department of Oncology, Karolinska University Hospital, Stockholm; 3Department of Hematology, Oncology and Radiation Physics, Skåne University Hospital, Lund; 4Department of Translational Medicine, Lund University; 5Department of Hematology, Sahlgrenska University Hospital, Gothenburg; 6Department of Internal Medicine, Östersunds Hospital; 7 Department of Internal Medicine Huddinge, Karolinska Institutet, Stockholm; 8 Department of Hematology, South Älvsborg Hospital, Borås; 9Department of Hematology, Karolinska University Hospital, Stockholm; 10Department of Hematology, Sunderby Hospital, Sunderbyn-Luleå; 11Department of Hematology, Uppsala University Hospital; 12Department of Hematology, Linköping University Hospital and 13Department of Medicine, Sundsvall Hospital, Sweden 1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):797-804

*AO and LH are co-senior authors.

ABSTRACT

he aim of this study was to investigate long-term outcome following first-line therapy in consecutive chronic lymphocytic leukemia (CLL) patients in a well-defined geographic area: Sweden. All patients diagnosed with CLL (2007-2013) (n=3672) were identified from national registries, screening of patient files identified all (100%) treated first line (n=1053) and for those, an in-depth analysis was performed. End points were overall response rate, progression-free survival (PFS), overall survival (OS), and safety. Median age was 71 years; 53% had Rai stage III-IV and 97% had performance status grade 0-2. Fluorescence in situ hybridization (FISH) was performed in 57% of patients: 15% had del(17p). Chlorambucil + prednisone was used in 39% (5% also received rituximab). Fludarabine+cyclophosphamide+rituximab or fludarabine+cyclophosphamide was used in 43% and bendamustine + rituximab in 6%. Overall response rate was 64%; chlorambucil 43%, fludarabine+cyclophosphamide+rituximab 84%, fludarabine+cyclophosphamide 75% and bendamustine + rituximab 75%. Median PFS and OS was 24 and 58 months, respectively, both were significantly associated (multivariate analysis) with type of treatment, del(17p), performance status, gender, age and geographical region (OS only). Chlorambucil-treated patients had a median PFS and OS of only 9 and 33 months, respectively. Chlorambucil usage declined gradually throughout the study period, but one-third of patients still received chlorambucil + rituximab in 2013. Infections ≥grade III were significantly associated with treatment; chlorambucil 19% versus fludarabine+cyclophosphamide+rituximab 30%. Richter transformation occurred in 5.5% of the patients, equally distributed across therapies. This is the largest retrospective, real-world cohort of consecutive first-line treated CLL patients with a complete follow up. In elderly patients, an unmet need for more effective, well-tolerated therapies was identified. haematologica | 2019; 104(4)

Correspondence: LOTTA HANSSON lotta.hansson@sll.se Received: June 21, 2018. Accepted: November 19, 2018. Pre-published: November 22, 2018. doi:10.3324/haematol.2018.200204 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/797 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

797

S.E. Sylvan et al.

Introduction

Methods

Chronic lymphocytic leukemia (CLL) is the most common leukemia in the Western world. In Sweden, the incidence is approximately 500 patients per year. The median age at diagnosis is approximately 70 years.1-4 The clinical course is extremely heterogeneous. At diagnosis, most patients are asymptomatic and the disease may be indolent for a long time. However, many patients show disease progression after a few years. When the disease requires treatment, strategies should be individualized.5 Chemoimmunotherapy with fludarabine in combination with cyclophosphamide and rituximab (FCR) resulted in an overall response rate (ORR) of about 90% and improved overall survival (OS),6 and represents the standard treatment in fit patients younger than 65 years. Yet, FCR is less well tolerated in patients over 65 years7 and these patients may instead benefit from bendamustine in combination with rituximab (BR) which has shown response rates similar to those achieved with FCR but with less toxicity.8,9 For elderly fragile patients, chlorambucil in combination with a CD20 monoclonal antibody10,11 could be an alternative; whether BR is to be preferred in old unfit patients remains uncertain.12 The presence of TP53 aberrations [del(17p) or TP53 mutation] is strongly associated with chemotherapy refractoriness, early relapse,13,14 and, until recently, a very dismal prognosis.14-16 Hence, evaluation of TP53 status is strongly recommended before treatment initiation. Brutonsâ&#x20AC;&#x2122;s tyrosine kinase (BTK) inhibitor ibrutinib17-19 has offered new options both for these patients and for relapsed/refractory CLL,20 with an ORR of 85% reported,17,18 and is considered the best available option for patients with TP53 disruptions.3,20,21 New treatments are costly and frequently accepted by regulatory agencies based on trials conducted in selected groups of patients with PFS, not OS, as end point. Hence, long-term results, including OS estimates, in real-life treated patients are important to determine the optimal therapy for patients with CLL.22 Previous data from the United States23 suggest that type of area (rural or urban) and type of hospital may influence response and survival especially in patients with high-risk CLL. However, Swedish results may differ due to the fact that almost all patients are treated within public health care. This means that most treatment decisions are taken at therapy conferences and we have a widespread usage of yearly up-dated national CLL guidelines.3 The Swedish Cancer Registry and the Swedish National CLL-Registry give us a unique opportunity to identify all patients diagnosed with CLL for an in-depth analysis of every single patient file. This provides a complete record of all patients treated within a defined time period on a nation-wide basis. Thus, this study provides high-quality real-world results on CLL first-line treatment that may be used as quality assurance and may help to interpret the cost-effectiveness of new drugs for healthcare providers. It may also serve as a control for clinical trials, selecting patients based on inclusion/exclusion criteria. Given this, the aim of this study was to investigate the outcome following first-line therapy in a well-defined population of consecutive CLL patients, in a setting with complete follow up.

This was a retrospective observational study. All patients diagnosed with CLL according to the World Health Organization criteria from 2007 to 2013 were identified from the National Cancer Registry. A representative physician from each of Swedenâ&#x20AC;&#x2122;s six health-care regions reviewed all the patient files in the region to identify patients who had received first-line CLL treatment due to progressive, symptomatic CLL. Patients who had started therapy before the end of 2013 were included in order to obtain sufficient follow up. Their files were analyzed in detail from the date of diagnosis until death or until the end of the study period (2017), whichever came first. Patients who had only received treatment for autoimmune hemolysis or idiopathic thrombocytopenic purpura (ITP) not related to progressive CLL were excluded. As this was a retrospective observational study, ethics committee approval (2013/952-31/3) was obtained; in Sweden no informed patient consent was required. The study was performed in accordance with the ethical principles of the Declaration of Helsinki24 and in compliance with national laws.

798

Data acquisition and study procedure Data on patientsâ&#x20AC;&#x2122; characteristics, treatment, outcome and toxicity were recorded on case record forms (CRFs). Information on participation in clinical trials, type of hospital (county/rural, regional or university) where the main body of treatment was given, where the decision on treatment was taken, geographical region, and whether choice of first-line therapy was compliant to the actual Swedish national CLL guidelines3 was also recorded. Furthermore, concomitant medication with acetylsalicylic acid (ASA) or statins were recorded since these drugs appear to induce apoptosis on CLL cells25,26 and may improve outcome in FCR-treated CLL.27 Data were incorporated in a specially developed version of the Information Network for Cancer Care (INCA) database and systematically cross-checked and validated for accuracy. Treatment response was evaluated according to the 2008 International Workshop on Chronic Lymphocytic Leukemia (IWCLL) criteria.28 Major infections (grade III-V) and other serious adverse events (SAE) according to the NCI CTCAE 3.0 were recorded. Richter transformation (RT) and secondary tumors were also recorded. The Swedish Cause of Death Registry was used to validate records of death.29

Statistical analysis End points in this study were evaluated according to the IWCLL criteria28 and included: ORR, duration of response (DOR), PFS, OS and safety. In the analysis of PFS, time was calculated from the start date of first-line therapy to the date of progression or date of death, whichever came first. In the analysis of OS, time was calculated from the date of first-line therapy to the date of death. For event-free patients, time was calculated to the date of last clinical visit. The Kaplan-Meier method was used to estimate and graphically display OS and PFS. Proportional hazards regression was used to estimate the effect of risk factors on time to failure. Results from these models are presented as hazard ratios (HR) together with 95% confidence intervals (CI). Reported P-values from these models refer to Wald tests. As FISH analysis was not implemented in the national guidelines until 2010, and cytogenetic status is a strong prognostic and predictive marker,14-16 patients were grouped into an earlier treatment period (2007-2009) and a later period (2010-2013). Multivariate analyses were restricted to the latter cohort. Analysis of the impact of IGHV mutation status did not provide haematologica | 2019; 104(4)

Real-world results of first-line therapy in CLL

sufficient power for this time period (only a small fraction of patients was tested); this was, therefore, excluded from the multivariate analysis. However, regarding IGHV, the prognostic impact of the whole population was analyzed separately. Patients who had received allogeneic stem cell transplant after first-line therapy (n=2) were excluded from PFS and OS analysis.

Results In total, 3672 patients diagnosed with CLL between 2007 and 2013 were identified from the Swedish Cancer registry for whom all (100%) medical files were available for review. Out of these, 1053 patients had started firstline treatment between 2007 and 2013, thus being subject to further in-depth analysis. The six geographical regions included 10-23% of the patients each. Sparsely populated areas included fewer patients (10%) than those with the larger cities. Median follow up for all patients was 4.8 years.

later study period. For all regimens, dosing intensity was similar across geographical regions and type of institution. Treatment was given according to the national guidelines in 80% (n=843) of the patients, 5% (n=49) were included in clinical trials, and in 15% (n=153) the treatment was individual, i.e. neither according to guidelines nor to a clinical protocol. The median age was higher for patients receiving chlorambucil (79 y) and younger for patients receiving FCR (64 y) compared to other chemotherapy-based regimens (F/FC 68 y, B/BR 72 y, CHOP/CVP +/- R 71 y). The median age in the CLB group did not change over the study period. In patients aged 75 y or over, 73% received CLB, F/FC (9%) and B/BR (5%), whereas in patients under 65 years of age, the most commonly used treatment was FCR (53%) followed by F/FC (22%). Those who received CLB also generally had a worse performance status, with 20% in ECOG 2-3 compared to 4% and 5% of those receiving FCR and BR, respectively. Notably, university hospitals used CLB significantly (P=0.01) less often (30%) than other types of

Baseline patientsâ&#x20AC;&#x2122; characteristics Baseline characteristics at start of first-line treatment are shown in Table 1. Median age at first-line treatment was 71 years (range 31-96 years). Thirty-four percent were females and the majority (53%) had advanced disease with Rai stages III-IV. Patients were generally in a good performance status with 97% in Eastern Co-operative Oncology Group (ECOG) grade 0-2. In total, results of cytogenetic assessment were available for 599 patients (57%); more results were available in the latter time period (64% 2010-2013 vs. 47% 2007-2009). Since 2010, when FISH was generally recommended in the national guidelines, there has been a significant difference in the frequency of cytogenetic analysis between the regions (50.8-72.5%; P=0.003) and cytogenetic analysis has been more often performed at university hospitals (80%) than other types of hospitals (55-60%) (P<0.001). The older the patient, the more rarely was the analysis performed. In the younger patient group [<65 years (y)], cytogenetic analysis was available in 87% compared to 75% and 39% in the middle aged (65-74 y) and oldest (â&#x2030;Ľ75 y) groups, respectively. The frequency of del(17p) was 4% and 11% in the earlier and later time periods, respectively, out of all patients tested (n=599). Analysis of IGHV mutational status was, and is still, optional according to the Swedish guidelines and thus was analyzed only in a minority of patients (n=224; 20%): 12% were unmutated and 8% mutated.

Treatment The majority (68%) of patients started first-line treatment in the later time period (2010-2013). Most patients (63%) were treated at County/Rural hospitals (i.e. neither university nor regional hospitals), whereas 32% received their treatment at university hospitals. In almost all cases, the treatment decision was taken at the same institution as that in which the patient was subsequently treated. First-line treatments are shown in Table 2. Type of treatment was unknown in 4 patients and in 4 could not be evaluated. The most frequently used regimens were: chlorambucil (CLB/CLB-R) (39%), FCR (27%) and FC (16%). Nearly all patients (95%) receiving CLB did not receive rituximab. Only 6% of the patients received bendamustin (B) or (BR) and almost all were treated in the haematologica | 2019; 104(4)

Table 1. Baseline characteristics at start of first-line therapy (n=1053).

Factor Age, median [range] Male ECOG performance status 0-1 2 3 Missing Binet stage A-B C Missing Rai stage 0-II III-IV Missing FISH del(13q) Normal Trisomy 12 del(11q) del(17p) Missing Hospital type University Regional County/Rural Missing Treatment Guidelines Clinical trial Individual Unknown

N (%) 71 years [31-96] 691 (66) 916 (87) 102 (10) 19 (2) 16 (2) 520 (49) 499 (47) 34 (3) 474 (45) 556 (53) 23 (2) 198 (19) 112 (11) 103 (10) 94 (9) 92 (9) 454 (43) 341 (32) 51 (5) 660 (63) 1 (0) 843 (80) 49 (5) 153 (15) 8 (1)

n/N: number; ECOG: Eastern Co-operative Oncology Group; FISH: fluorescence in situ hybridization.

799

S.E. Sylvan et al. Table 2. First-line treatment by time period and response/infection rates.

Treatment F/FC FCR Alemtuzumab CLB±R B/BR CHOP/CVP±R Other Rituximab alone Total

2007-2009

Time period (%) 2010-2013

95 (29) 32 (10) 17 (5) 159 (48) 1 (0) 16 (5) 6 (2) 4 (1) 330 (100)

74 (10) 253 (35) 31 (4) 245 (34) 61 (9) 11(2) 26(4) 14 (2) 715 (100)

2007-2013

ORR# (%)

Infection# >Grade III (%)

169 (16) 285 (27) 48 (5) 404 (39) 62 (6) 27 (3) 32 (3) 18 (2) 1045

127 (75) 240 (84) 32 (67) 174 (43) 47 (75) 22 (82) 22 (67) 9 (50) 673 (64)

48 (29) 84 (30) 17 (35) 77 (19) 20 (32) 9 (33) 8 (25) 2 (11) 265 (25)

For the whole time period 2007-2013. ORR: overall response rate; F: fludarabine; FC: fludarabine in combination with cyclophosphamide; FCR: fludarabine in combination with cyclophosphamide and rituximab; CLB±R: chlorambucil and rituximab; B/BR: bendamustine/bendamustin and rituximab; CHOP/CVP±R: cyclophosphamide+hydroxydaunorubicin+vincristine+prednisone / cyclophosphamide+ vincristine+prednisone+rituximab. #

hospitals (43%). The use of CLB declined significantly over the years, with 58% usage in 2007 to 31% in 2013 (P<0.0001). However, by the end of 2013, the CLB firstline usage was still high, varying between regions from 27% to 49%.

Response to first-line therapy The ORR for the study group was 64% (15% CR): CLB 43%, FCR 84%, FC 75% and B/BR 75%. ORR was significantly associated with type of treatment (P<0.001), performance status (P<0.001), del(17p) (P=0.007), age (P<0.001), and compliance to national guidelines (P=0.003), but not with gender, Rai stage or type of hospital (univariate analysis). Patients included in clinical trials showed a numerically better response rate (82%) than those treated according to national guidelines (65%) and compared to patients treated neither according to guidelines nor to a clinical study protocol (54%).

Progression-free and overall survival Median PFS was 24 months (range 20-26 months) and median OS was 58 months (range 40-76 months). At 5-y follow up, 51% of all patients were deceased, and nearly two-thirds had died from CLL or CLL-related infections.

Progression-free survival in relation to type of first-line therapy is shown in Figure 1A. As expected, the longest PFS was observed with FCR, whereas the median PFS in patients who received CLB was only nine months. PFS in relation to FISH results are shown in Figure 1B. Shortest PFS was observed for del(17p). Similar results were observed for type of treatment and FISH data in relation to OS (Figure 2A and B). Notably, CLB-treated patients had a median OS of only 33 months. Survival in relation to type of hospital is shown in Figure 2C and in relation to age in Figure 2D. In a multivariate analysis, both PFS and OS were significantly associated with type of treatment, cytogenetic status, performance status, gender and age (Table 3). OS was also significantly associated with geographical region (P=0.003). There was a tendency but no significant difference (P=0.07) in OS between the two time periods (20072009 vs. 2010-2013). IGHV analysis was only performed in a small fraction of patients and did not provide sufficient power when included in the model (both PFS and OS were non-significant). Thus, this analysis was excluded from multivariate analysis regarding this time period. However, we also analyzed the prognostic impact of IGHV on the whole study population. The results from

Figure 1. Progression-free survival (PFS) after first-line therapy. PFS according to (A) treatment and (B) fluorescence in situ hybridization (FISH) cytogenetic status. F/FC: fludarabine/fludarabine in combination with cyclophosphamide; CLB+/-R: chlorambucil and rituximab; FCR: fludarabine in combination with cyclophosphamide and rituximab; B/BR: bendamustine/bendamustin and rituximab; ALEM: alemtuzumab; CHOP/CVP+/-R: cyclophosphamide+hydroxydaunorubicin+vincristine+prednisone/cyclophosphamid+vincristine+prednisone+rituximab.

800

haematologica | 2019; 104(4)

Real-world results of first-line therapy in CLL

this analysis showed a prognostic impact of IGHV status both on OS (P<0.01) and PFS (P=0.02) (data not shown). In univariate analysis, PFS and OS were also associated with type of hospital (P=0.05 and P=0.04) (Figure 2C), adherence to treatment guidelines (P=0.001 and P=0.006), and Rai stage (P=0.042 and P=0.005). Median PFS and OS at university/regional hospitals versus other hospitals were 28 versus 22 months and 61 versus 54 months, respectively. Those treated according to national guidelines showed a median PFS and OS of 26 months and 58 months, respectively. The correlating time for those treated outside trials or guidelines were 13 months and 44 months, respectively and for those in clinical trials 19 months and 66 months, respectively. Patients on medication with ASA or statins also showed a significantly shorter PFS and OS (P<0.001 and P=0.007) (by univariate analysis only).

Safety Infections of grade III or higher were significantly associated with type of treatment, affecting 19% of the CLBtreated patients and 30% in the FCR group (P=0.006) (Table 2). Richter transformation occurred in 5.5% of the patients, was significantly associated with del(17p) (P=0.04), and was equally distributed between types of first-line therapy. The median time to transformation was three years from diagnosis and 1.5 years from first-line treatment. Secondary malignancies affected 15% of the patients and were equally distributed between types of first-line therapy. About one-third of the secondary

malignancies consisted of basal cell carcinomas. MDS/AML affected only 1% of the patients and the other secondary malignancies were solid tumors.

Discussion Randomized controlled trials (RCTs) remain the scientific ideal for evaluation of novel treatments. However, in studies on malignancies, RCTs are sometimes not sufficient to address the evidentiary requirements of regulating authorities30 and payers as patients are selected on strict inclusion/exclusion criteria and sufficient data on overall survival and long-term follow up is often not provided.31 In addition, the comparative arm in clinical trials may be chosen to favor the treatment of investigation. Therefore, regulating authorities increasingly look for real-world data for additional comparison when evaluating new cost-intensive drug regimens. However, reliable data on consecutive patients in routine health-care may be difficult to obtain. This is the largest retrospective cohort of strictly consecutive real-world patients from a well defined geographical region (Sweden) with a comparatively long complete follow up. By using high-quality Swedish data bases (National Cancer Registry/Swedish CLL-registry) including all patients diagnosed in Sweden within a specified time period, followed by in-depth analysis of each individual medical file, we were able to obtain a complete record of all patients diagnosed with and receiving first-

Figure 2. Overall survival (OS) after first-line therapy. OS according to (A) treatment, (B) fluorescence in situ hybridization cytogenetic status, (C) type of hospital, and (D) age. F/FC: fludarabine/fludarabine in combination with cyclophosphamide; CLB+/R: chlorambucil and rituximab; FCR: fludarabine in combination with cyclophosphamide and rituximab; B/BR: bendamustine/bendamustin and rituximab; ALEM: alemtuzumab; CHOP/CVP+/-R: cyclophosphamide +hydroxydaunorubicin+vincristine+p rednisone/cyclophosphamid+vincristine+prednisone+rituximab.

haematologica | 2019; 104(4)

801

S.E. Sylvan et al.

line CLL treatment in the period 2007-2013. As officebased private medicine is practically non-existent for CLL in Sweden, all patient files were identified, ensuring highquality data and minimal selection bias, which is a key strength of our report. The time period 2007-2013 was selected in order to obtain sufficient follow up. Subjects in this study were older than in previous clini-

cal studies,6,8,32,33 more often had advanced disease but were in good performance status. The median age in our cohort is consistent with median age at diagnosis,3 and the advanced stage combined with good performance status reflects treatment indication and first-line treatment status. The majority of patients were treated in the later time period which possibly suggests that most patients

Table 3. Multivariate analysis* on factors in relation to progression-free survival and overall survival.

Clinical factor (percentage available data) Age, years (100) <65 65-74 >75 Gender (100) Males Females Performance status (ECOG) (98) 0 1 2-3 Rai stage (98) 0-2 3-4 FISH (64) del(13q), normal, tri12 or del(11q) del(17p) Treatment (99) F/FC/FCR Alemtuzumab/CHOP/CVP CLB B/BR Hospital (100) County/Rural University/Regional Region (100) Stockholm/Gotland Uppsala/Örebro Southeast South West North Treatment according to: (99) National Guidelines – Yes Clinical trial – Yes National Guidelines/Clinical trial - No Time to treatment (100) <1 year ≥1 year

Progression-free survival HR (CI 95%)

Overall survival HR (CI 95%)

1 1.41 (1.02-1.93) 1.86 (1.24-2.80)

0.010

1 2.30 (1.39-3.79) 3.76 (2.12-6.67)

<0.001

1 0.70 (0.53-0.94)

0.017

1 0.60 (0.40-0.91)

0.016

1 1.60 (1.18-2.18) 1.87 (1.12-3.13)

0.006

1 2.33 (1.48-3.66) 2.25 (1.10-4.59)

0.001

1 0.92 (0.71-1.19)

0.53

1 1.22 (0.86-1.75)

0.263

1 2.05 (1.43-2.91)

<0.001

1 2.17 (1.38-3.39)

<0.001

1 1.32 (0.74-2.37) 2.40 (1.43-4.04) 1.14 (0.56-2.33)

0.010

0.166

1 1.02 (0.69-1.50)

0.936

0.481

1 0.80 (0.43-1.49) 1.62 (0.85-3.08) 2.42 (1.32-4.39) 2.21 (1.13-4.33) 1.46 (0.70-3.07)

0.003

0.082

1 1.02 (0.45-2.28) 1.05 (0.58-1.89)

0.989

0.652

1 0.89 (0.63-1.27)

0.523

1 1.68 (1.09-2.57) 2.82 (1.86-4.26) 0.79 (0.47-1.33) 1 0.82 (0.61-1.09) 1 0.78 (0.51-1.19) 1.03 (0.65-1.63) 1.22 (0.78-1.92) 0.97 (0.59-1.60) 0.93 (0.54-1.62) 1 1.80 (1.07-3.04) 1.14 (0.72-1.81) 1 0.94 (0.72-1.23)

HR: hazard ratio; CI: confidence interval; ECOG: Eastern Co-operative Oncology Group; FISH: fluorescence in situ hybridization; F: fludarabine; FC: fludarabine in combination with cyclophosphamide; FCR: fludarabine in combination with cyclophosphamide and rituximab; alemtuzumab/CHOP/CVP: alemtuzumab /cyclophosphamide+hydroxydaunorubicin+vincristine+prednisone/cyclophosphamide+vincristine+prednisone; CLB: chlorambucil; B/BR: bendamustine/bendamustin and rituximab. *Model included all listed factors.

802

haematologica | 2019; 104(4)

Real-world results of first-line therapy in CLL

with CLL do not have a treatment indication at the time of diagnosis. Cytogenetic status was only available in 58% of the patients. The analysis was not mandatory until 2010 according to the national guidelines while FISH was more commonly performed in the later time period (2010-2013). Notably, university hospitals performed FISH significantly more often, and the older the patient the more rarely was the analysis performed. The vast majority (80%) were treated according to the national guidelines. Notably, compliance to national guidelines was associated with better response, and PFS and OS (univariate analysis). However, patients included in clinical trials had an even better ORR and OS but a shorter PFS. This may partly be due to our conservative approach on how to interpret response, i.e. if not all variables for a complete or partial response where available responses were considered to be of a lower grade. As patients in studies may have been more thoroughly evaluated, variables for response were possibly more available; this may have resulted in a higher response rate for these patients. Also, selection bias or more effective regimens and a more thorough follow up in clinical trials (which may detect progression at an earlier stage) may have influenced the results. The different types of treatments used may at least partly reflect the development of therapies over time. In fact, FCR was used much more in the later time period and B/BR, which was not introduced until the end of the study period, was used less. Even though CLB use declined during the study period, approximately onethird of all patients still received first-line CLB as late as 2013, although with significant regional differences. Regarding FCR and BR, the response rate was slightly lower than in prospective first-line clinical studies6,8,9 and for CLB it was slightly higher.10,34 As expected, and in line with recently published data,8,32 FCR was more commonly used in younger patients and CLB was the most frequently used treatment overall and in the elderly. The median age for CLB-treated patients did not increase over time, possibly indicating that more modern treatments, such as B/BR, were not always used in patients over 65 years of age. OS differed between the two time periods (2007-2009 vs. 2010-2013), but this difference was not significant. This is in line with our previous findings regarding second-line therapy in the Stockholm region,1 where no improvement in OS over the years was seen. Real-world treatment outcome may differ from that in clinical trials and when compared to data from prospective trials,6,8,9 patients in this study showed a relatively low ORR and short PFS and OS, despite an 80% compliance to national guidelines. In line with previous data, del(17p)14-16 was a negative predictive marker but also type of treatment, age and performance status were independently associated with poorer PFS and OS. Patients in this report were comparatively old versus those in trials,6,8,32 and as many as 39% received CLB and CLB treatment was associated with poorer outcome. In addition, cytogenetic status, the strongest predictor of outcome, was unknown in almost half the patients. These factors may at least partly explain the differences in our results and those reported in trials.6,8,32 Notably, patients treated with single CLB in clinical trials34-36 showed an even shorter PFS; this may be due to a more thorough evaluation in trials. Also, patients treated with B/BR showed a comparatively poor outhaematologica | 2019; 104(4)

come.8 However, these regimens were used only in the elderly, and only FCR and FC showed better outcomes. The difference in OS between the time periods was not significant (P=0.07), and this may possibly be influenced by the low use of FCR and by the fact that bendamustine became available only at the end of the time period. IGHV mutation status is still optional in the Swedish guidelines, probably explaining the low number of patients analyzed. Thus, the power of the analysis was insufficient for the multivariate analysis. However, our findings for the whole population was in line with previous reports37 in showing a significant impact on outcome. The infection rate6,8,10,34 was comparable to previous clinical studies with an expected higher infection rate for FCR. Our study showed an incidence of RT in line with previous data.32,38-40 However, our study did not, in contrast to previous data,38,41 show any association between treatment with fludarabine and RT. The slightly lower incidence of RT within the CLB group may be because the median time to transformation was three years, and the patients in this group were older and had a shorter median OS. Our study indicates that type of hospital may possibly have an impact on outcome but could not confirm previous findings23 regarding outcome in urban versus rural regions. The possible differences between type of hospitals may derive from the fact that FISH analysis was more often performed at university hospitals, and the widespread use of the regularly up-dated Swedish National Guidelines may have minimized the difference between regions as 80% were treated according to guidelines. The regional usage of chlorambucil varied between 27-49%. We still have no explanation for this. Elderly patients and those with a poorer performance status might also be on concomitant medication with ASA or statins, i.e. those with certain comorbidity. This may explain why concomitant ASA or statins showed a significant association in univariate but not in multivariate analysis.42 Also, in some previous reports, ASA or statins do not appear to affect outcome.43,44 A limitation of the study is its retrospective nature and the lack of data from recent years during which different regimens have been used and novel therapies have become available. For example, part of the study was performed before bendamustine was introduced as first-line treatment in CLL and before a CD20 antibody was added to CLB. For a more complete understanding of real-world outcome in CLL patients, an analysis of the outcome of treatment of relapse is warranted. As this requires a longer follow up, we have started a separate project for further investigation. Despite these limitations, in relation to todayâ&#x20AC;&#x2122;s standard-of-care treatment, the results are still important. In summary, our results provide additional information representative of real-world outcome of first-line CLL treatment and provide an important context within which to evaluate the findings obtained from clinical trials of new drugs. We show that outcome in real-world situations differs from that in clinical trials, and that single-agent CLB treatment, as well as age and performance status, were independent factors for poor outcome in multivariate analysis. Notably, the older the patient the more rarely was FISH analysis performed and the more often CLB was chosen as treatment. As CLL and related complications seem to be the major cause of death in 803

S.E. Sylvan et al. patients, regardless of comorbidity,45,46 also elderly, comorbid patients should preferably undergo cytogenetic analysis and receive treatments for adequate disease control. Hence, we conclude that alternative modern, effective first-line treatment alternatives must be offered to elderly comorbid patients. Our study also raises the question as to whether drugs other than CLB, even if combined with a CD20 antibody,47 should be considered as the chemotherapy approach in the standard-of-care arm in pivotal clinical trials. Finally, we have also demonstrated inter-regional differences in drug and FISH usage, and that outcome may vary in different parts of the country despite regular updates on generally available national

References 1. Asklid A, Winqvist M, Eketorp Sylvan S, et al. Outcomes of second-line treatment in chronic lymphocytic leukemia - a population-based study from a well defined geographical region between 2003 and 2013. Leuk Lymphoma. 2017;58(5):1219-1223. 2. Eichhorst B, Dreyling M, Robak T, et al. Chronic lymphocytic leukemia: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2011;22(Suppl 6):vi50-54. 3. Svenska KLL gruppen. https://www.cancercentrum.se/syd/cancerdiagnoser/blod-lymfom-myelom/kronisk-lymfatisk-leukemikll/vardprogram/. Regionalt cancercentrum Stockholm och Gotland. 2016. 4. Siegel RL, Sahar L, Portier KM, Ward EM, Jemal A. Cancer death rates in US congressional districts. CA Cancer J Clin. 2015;65(5):339-344. 5. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for diagnosis, indications for treatment, response assessment and supportive management of chronic lymphocytic leukemia. Blood. 2018;131(25):2745-2760. 6. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomised, open-label, phase 3 trial. Lancet. 2010; 376(9747):1164-1174. 7. Cramer P, Fink AM, Busch R, et al. Secondline therapies of patients initially treated with fludarabine and cyclophosphamide or fludarabine, cyclophosphamide and rituximab for chronic lymphocytic leukemia within the CLL8 protocol of the German CLL Study Group. Leuk Lymphoma. 2013; 54(8):1821-1822. 8. Eichhorst B, Fink AM, Bahlo J, et al. First-line chemoimmunotherapy with bendamustine and rituximab versus fludarabine, cyclophosphamide, and rituximab in patients with advanced chronic lymphocytic leukaemia (CLL10): an international, openlabel, randomised, phase 3, non-inferiority trial. Lancet Oncol. 2016;17(7):928-942. 9. Fischer K, Cramer P, Busch R, et al. Bendamustine in combination with rituximab for previously untreated patients with chronic lymphocytic leukemia: a multicenter phase II trial of the German Chronic Lymphocytic Leukemia Study Group. J Clin Oncol. 2012;30(26):3209-3216. 10. Goede V, Fischer K, Busch R, et al.

804

11.

12.

13.

14.

15.

16.

17.

18.

19. 20.

21.

CLL guidelines; findings which need to be further investigated in detail. Funding This study was a collaboration within the Swedish CLL group and was supported by grants from AFA Insurance (Ref no: 130054), SLL/ALF (Ref no: 20150070), Blodcancerfonden 2016, Dr Åke Olsson Foundation (Ref no: 2-791/2016), SLL/KI Högre klinisk forskare 2018/2019 (K2894-2016), Svenska Läkaresällskapet (SLS-406961), The Gilead Sciences Nordic Fellowship Programme 2015 (LH), The Swedish Cancer Society (Ref no: 150930, 160534). We thank Ms Leila Relander for editorial assistance.

Obinutuzumab plus chlorambucil in patients with CLL and coexisting conditions. N Engl J Med. 2014;370(12):11011110. Hillmen P, Robak T, Janssens A, et al. Chlorambucil plus ofatumumab versus chlorambucil alone in previously untreated patients with chronic lymphocytic leukaemia (COMPLEMENT 1): a randomised, multicentre, open-label phase 3 trial. Lancet. 2015;385(9980):1873-1883. Michallet AS, Aktan M, Hiddemann W, et al. Rituximab plus bendamustine or chlorambucil for chronic lymphocytic leukemia: primary analysis of the randomized, openlabel MABLE study. Haematologica. 2018;103(4):698-706. Stilgenbauer S, Schnaiter A, Paschka P, et al. Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):32473254. Zenz T, Eichhorst B, Busch R, et al. TP53 mutation and survival in chronic lymphocytic leukemia. J Clin Oncol. 2010; 28(29):4473-4479. Dohner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med. 2000;343(26):1910-1916. Rossi D, Rasi S, Spina V, et al. Integrated mutational and cytogenetic analysis identifies new prognostic subgroups in chronic lymphocytic leukemia. Blood. 2013; 121(8):1403-1412. Byrd JC, Brown JR, O'Brien S, et al. Ibrutinib versus ofatumumab in previously treated chronic lymphoid leukemia. N Engl J Med. 2014;371(3):213-223. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):32-42. Furman RR, Cheng S, Lu P, et al. Ibrutinib resistance in chronic lymphocytic leukemia. N Engl J Med. 2014; 370(24):2352-2354. Jones J, Mato A, Coutre S, et al. Evaluation of 230 patients with relapsed/refractory deletion 17p chronic lymphocytic leukaemia treated with ibrutinib from 3 clinical trials. Br J Haematol 2018; 182(4):504-512. O'Brien S, Jones JA, Coutre SE, et al. Ibrutinib for patients with relapsed or refractory chronic lymphocytic leukaemia with 17p deletion (RESONATE-17): a phase 2, open-label, multicentre study. Lancet Oncol. 2016;17(10):1409-1418.

22. Lewis J, Kerridge I, Lipworth W. Use of RealWorld Data for the Research, Development, and Evaluation of Oncology Precision Medicines. JCO Precision Oncology. 2017:111. 23. Loberiza FR Jr, Cannon AJ, Weisenburger DD, et al. Survival disparities in patients with lymphoma according to place of residence and treatment provider: a populationbased study. J Clin Oncol. 2009; 27(32):5376-5382. 24. The World Medical Association. 64th WMA General Assembly Fortaleza Brazil, October 2013. WMA Declaration of Helsinki – Ethical Principles for Medical Research Involving Human Subjects. 2018 [cited; Available from: https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/ 25. Bellosillo B, Pique M, Barragan M, et al. Aspirin and salicylate induce apoptosis and activation of caspases in B-cell chronic lymphocytic leukemia cells. Blood. 1998; 92(4):1406-1414. 26. Chapman-Shimshoni D, Yuklea M, Radnay J, Shapiro H, Lishner M. Simvastatin induces apoptosis of B-CLL cells by activation of mitochondrial caspase 9. Exp Hematol. 2003;31(9):779-783. 27. Chae YK, Trinh L, Jain P, et al. Statin and aspirin use is associated with improved outcome of FCR therapy in relapsed/refractory chronic lymphocytic leukemia. Blood. 2014;123(9):1424-1426. 28. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008; 111(12):5446-5456. 29. Socialstyrelsen. http://www.socialstyrelsen.se/register/dodsorsaksregistret. 30. Lewis J, Kerridge I, Lipworth W. Use of realworld data for the research, revelopment, and evaluation of oncology precision medicines. JCO Precision Oncology. 2017:1-11. 31. Lewis JR, Lipworth WL, Kerridge IH, Day RO. The economic evaluation of personalised oncology medicines: ethical challenges. Med J Aust. 2013;199(7):471-473. 32. Huang SJ, Lee LJ, Gerrie AS, et al. Characterization of treatment and outcomes in a population-based cohort of patients with chronic lymphocytic leukemia referred

haematologica | 2019; 104(4)

Real-world results of first-line therapy in CLL

33.

34.

35.

36.

37.

38.

for cytogenetic testing in British Columbia, Canada. Leuk Res. 2017;55:79-90. Tam CS, O'Brien S, Plunkett W, et al. Longterm results of first salvage treatment in CLL patients treated initially with FCR (fludarabine, cyclophosphamide, rituximab). Blood. 2014;124(20):3059-3064. Burger JA, Tedeschi A, Barr PM, et al. Ibrutinib as Initial Therapy for Patients with Chronic Lymphocytic Leukemia. N Engl J Med. 2015;373(25):2425-2437. Eichhorst BF, Busch R, Stilgenbauer S, et al. First-line therapy with fludarabine compared with chlorambucil does not result in a major benefit for elderly patients with advanced chronic lymphocytic leukemia. Blood. 2009;114(16):3382-3391. Hillmen P, Skotnicki AB, Robak T, et al. Alemtuzumab compared with chlorambucil as first-line therapy for chronic lymphocytic leukemia. J Clin Oncol. 2007; 25(35):56165623. Gentile M, Shanafelt TD, Mauro FR, et al. Comparison between the CLL-IPI and the Barcelona-Brno prognostic model: Analysis of 1299 newly diagnosed cases. Am J Hematol. 2018;93(2):E35-E37. Benjamini O, Jain P, Trinh L, et al. Second

haematologica | 2019; 104(4)

39.

40.

41.

42.

43.

cancers in patients with chronic lymphocytic leukemia who received frontline fludarabine, cyclophosphamide and rituximab therapy: distribution and clinical outcomes. Leuk Lymphoma. 2015;56(6):1643-1650. Tam CS, O'Brien S, Wierda W, et al. Longterm results of the fludarabine, cyclophosphamide, and rituximab regimen as initial therapy of chronic lymphocytic leukemia. Blood. 2008;112(4):975-980. Woyach JA, Ruppert AS, Heerema NA, et al. Chemoimmunotherapy with fludarabine and rituximab produces extended overall survival and progression-free survival in chronic lymphocytic leukemia: long-term follow-up of CALGB study 9712. J Clin Oncol. 2011;29(10):1349-1355. Fischer K, Bahlo J, Fink AM, et al. Long-term remissions after FCR chemoimmunotherapy in previously untreated patients with CLL: updated results of the CLL8 trial. Blood. 2016;127(2):208-215. Thurmes P, Call T, Slager S, et al. Comorbid conditions and survival in unselected, newly diagnosed patients with chronic lymphocytic leukemia. Leuk Lymphoma. 2008;49(1): 49-56. Friedman DR, Magura LA, Warren HA, et al.

Statin use and need for therapy in chronic lymphocytic leukemia. Leuk Lymphoma. 2010;51(12):2295-2298. 44. Shanafelt TD, Rabe KG, Kay NE, et al. Statin and non-steroidal anti-inflammatory drug use in relation to clinical outcome among patients with Rai stage 0 chronic lymphocytic leukemia. Leuk Lymphoma. 2010;51(7):1233-1240. 45. Goede V, Cramer P, Busch R, et al. Interactions between comorbidity and treatment of chronic lymphocytic leukemia: results of German Chronic Lymphocytic Leukemia Study Group trials. Haematologica. 2014;99(6):1095-1100. 46. Strati P, Parikh SA, Chaffee KG, et al. Relationship between co-morbidities at diagnosis, survival and ultimate cause of death in patients with chronic lymphocytic leukaemia (CLL): a prospective cohort study. Br J Haematol. 2017;178(3):394-402. 47. Hillmen P, Janssens A, Babu KG, et al. Healthrelated quality of life and patient-reported outcomes of ofatumumab plus chlorambucil versus chlorambucil monotherapy in the COMPLEMENT 1 trial of patients with previously untreated CLL. Acta Oncol. 2016;55(9-10):1115-1120.

805

ARTICLE Ferrata Storti Foundation

Platelet Biology & its Disorders

Aerobic glycolysis fuels platelet activation: small-molecule modulators of platelet metabolism as anti-thrombotic agents

Paresh P. Kulkarni,1† Arundhati Tiwari,1† Nitesh Singh,1 Deepa Gautam,1 Vijay K. Sonkar,1 Vikas Agarwal2 and Debabrata Dash1

Department of Biochemistry, Institute of Medical Sciences and 2Department of Cardiology, Institute of Medical Sciences, Banaras Hindu University, Varanasi, Uttar Pradesh, India 1

Haematologica 2019 Volume 104(4):806-818

PPK and AT contributed equally to this work.

†

ABSTRACT

Correspondence: DEBABRATA DASH ddash.biochem@gmail.com Received: September 3, 2018. Accepted: October 30, 2018. Pre-published: October 31, 2018. doi:10.3324/haematol.2018.205724 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/806 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

806

latelets are critical to arterial thrombosis, which underlies myocardial infarction and stroke. Activated platelets, regardless of the nature of their stimulus, initiate energy-intensive processes that sustain thrombus, while adapting to potential adversities of hypoxia and nutrient deprivation within the densely packed thrombotic milieu. We report here that stimulated platelets switch their energy metabolism to aerobic glycolysis by modulating enzymes at key checkpoints in glucose metabolism. We found that aerobic glycolysis, in turn, accelerates flux through the pentose phosphate pathway and supports platelet activation. Hence, reversing metabolic adaptations of platelets could be an effective alternative to conventional anti-platelet approaches, which are crippled by remarkable redundancy in platelet agonists and ensuing signaling pathways. In support of this hypothesis, small-molecule modulators of pyruvate dehydrogenase, pyruvate kinase M2 and glucose-6-phosphate dehydrogenase, all of which impede aerobic glycolysis and/or the pentose phosphate pathway, restrained the agonist-induced platelet responses ex vivo. These drugs, which include the anti-neoplastic candidate, dichloroacetate, and the Food and Drug Administration-approved dehydroepiandrosterone, profoundly impaired thrombosis in mice, thereby exhibiting potential as anti-thrombotic agents.

Introduction Platelets play a prominent role in the pathophysiology of acute myocardial infarction and ischemic stroke, which are the major causes of mortality worldwide.1 Anti-platelet drugs remain the mainstay for the prevention of these catastrophic events. Nevertheless, the anti-platelet agents currently in vogue have limited efficacy in preventing thrombotic events without significantly raising bleeding risk.2 There is remarkable redundancy in potential agonists inducing platelet activation, as well as plasticity in the signaling paths downstream of these agonists. Patients on anti-platelet drugs that inhibit platelet responses to specific agonists continue to experience adverse thrombotic episodes, since other potential triggers and parallel signaling cascades can still activate platelets. Hence, it is vital to discover novel anti-platelet strategies to address these limitations. In an intact vasculature, platelets are relatively inactive and are, therefore, described as ‘resting’. However, they are extremely responsive to any breach in the vessel wall and are highly efficient in sealing the defect. Following a break in endothelium, platelets interact with, and adhere to, the now exposed subendothelial matrix proteins, such as collagen, through their cell surface receptors. These receptor-ligand interactions initiate signaling cascades that lead to: (i) a change in the shape of platelets from discoid to ‘spiny-spheres’, (ii) platelet-platelet aggregation through fibrinogen bridges connecting high-affinity integrins, and (iii) degranulation of platelet storage vesicles, the contents of which serve to amplify responses to vessel wall injury. A chain reaction of platelet activation and aggregate formation, as well as conversion of fibrinogen to insoluble polymers of fibrin, culminates haematologica | 2019; 104(4)

Aerobic glycolysis fuels platelet activation

in plugging of the defect in vasculature. This response is termed ‘hemostasis’. A similar but more exaggerated response from platelets upon rupture of an atherosclerotic plaque leads to arterial thrombosis with potentially fatal consequences, such as myocardial infarction and ischemic stroke.3 Platelet responses to agonist stimulation, including adhesion, shape change, integrin activation, aggregation, exocytosis of granule contents and clot retraction are all energy intensive processes.4-7 Nevertheless, platelets accomplish these responses while trapped within the relatively impervious boundaries of a thrombus, with restricted access to nutrients and oxygen. We hypothesized that it is imperative for activated platelets within a thrombotic milieu to adapt their energy metabolism to these challenges in order to sustain and hold the thrombus together. We employed a high-resolution bioenergetics screen to establish aerobic glycolysis and the consequent flux through the pentose phosphate pathway (PPP) as the metabolic signature of agonist-stimulated platelets. Using small-molecule modulators – dichloroacetate (DCA), diarylsulfonamide (DASA-58) (DASA), and dehydroepiandrosterone sulphate (DHEA) – which target these pathways, we demonstrate that rewiring of metabolism is essential for activation of human platelets ex vivo as well as thrombus formation in a murine model in vivo. Considering that a substantial body of information on the safety and pharmacokinetics of DCA and DHEA in humans is already available,8-11 one can expect probable future translation of these drugs into clinical use as antiplatelet agents.

Platelet aggregation and granule secretion Platelets suspended in buffer B or in platelet-rich plasma were stirred (1200 rpm) at 37°C in an optical lumi-aggregometer (Chrono-log model 700-2) for 1 min. Platelet aggregation induced with thrombin or collagen was recorded as percent change in light transmittance where 100% reflects transmittance through a blank (buffer or platelet-poor plasma).12 Release of adenine nucleotides from platelet dense granules was measured with the Chronolume reagent (stock concentration, 0.2 μM luciferase/luciferin). Luminescence generated was monitored using the lumi-aggregometer in parallel with the platelet aggregation measurement.12 Secretion from platelet α-granules was evaluated by quantifying surface expression of P-selectin as described previously.12 Details are provided in the Online Supplement.

Platelet surface integrin activation and GLUT3 expression Platelet surface integrin activation was analyzed as described previously.12 In order to measure GLUT3 surface expression, platelets were fixed with an equal volume of 4% paraformaldehyde, washed and blocked with 2% bovine serum albumin for 40 min before being incubated with a GLUT3-specific antibody for 2 h, and then washed and stained with Alexa Fluor 488-conjugated goat antimouse antibody for 1 h. Cells were analyzed by flow cytometry as described above. Further details are provided in the Online Supplement.

Measurement of intracellular reactive oxygen species

Methods

The intracellular levels of reactive oxygen species (ROS) were determined using a redox-sensitive cell-permeable dye, H2DCFDA as described elsewhere.14 Details are available in the Online Supplement.

Ethical approval

Immunoblotting

The animal experiments were approved by the Central Animal Ethical Committee of the Institute of Medical Sciences, Banaras Hindu University. All efforts were made to minimize the number of animals used, and their suffering. Peripheral venous blood samples were collected from healthy participants after obtaining written informed consent, strictly in accordance with the recommendations and as approved by the ethical committee of the Institute of Medical Sciences, Banaras Hindu University.

Immunoblotting experiments were performed as described previously.12 Details are given in the Online Supplement.

Studies of coagulant activity Externalization of phosphatidylserine, a measure of surface procoagulant activity, in stimulated platelets was assessed from annexin V binding, as described previously.15

Animal experiments Platelet preparation and mitochondrial respirometry Platelets were isolated from fresh human blood by differential centrifugation, as already described.12 Mitochondrial respiration was measured using a high-resolution respirometer (Oxygraph2k, Oroboros Instruments) as previously described.12 Details are provided in the Online Supplement.

Measurements of glucose, lactate and NAPDH Rates of glucose uptake and lactate secretion in thrombin (0.5 U/mL)-stimulated platelets were determined using an YSI 2900D Multiparameter Bioanalytical System (YSI Life Sciences), which uses immobilized enzyme electrodes and electrochemistry-based biosensing.13 The ratio of NADPH to total NADP(H) in washed human platelets was determined using a NADP/NADPH assay kit (Sigma). For this, NADP was decomposed at 60° C for 30 min followed by estimation of the amount of NADPH. To estimate total NADP(H), extract was directly processed for color development following the instructions of the kit’s manufacturer and absorbance was recorded at 450 nm. haematologica | 2019; 104(4)

Intravital microscopy of ferric chloride-induced thrombosis of mice mesenteric arterioles was performed as described elsewhere,16 with some modifications. Tail bleeding time experiments were carried out as described previously,17 with minor modifications. Pulmonary embolism was induced by collagen-epinephrine in 12- to 20-week old Swiss albino mice of either sex as described previously.12 The methods are described in detail in the Online Supplement.

Results Platelets exhibit a higher rate of aerobic glycolysis upon agonist stimulation We measured oxygen flux in platelets (suspended in Tyrode modified buffer under stirring) using a Clark amperometric electrode at high resolution (sampling at 2 s intervals). This system in a cuvette format closely resembles the platelet aggregometry of Born. When we exposed 807

P.P. Kulkarni et al.

platelets to thrombin (0.5 U/mL), a strong physiological agonist, there was a rapid surge in oxygen consumption rate over the basal respiration (within the first 90 s of stimulation) (Figure 1A,B), apparently driven by energydemanding early platelet responses. This upsurge in oxygen consumption rate was, however, followed by an abrupt fall (Figure 1A,B) that, intriguingly, coincided with the exponentially rising phase of the platelet aggregatory response (Online Supplementary Figure S1A). We recorded identical peaking and plunging of oxygen consumption in platelets upon stimulation with collagen, another physiological agonist (Online Supplementary Figure S1B). We sought to identify the reason underlying this rapid plunge in oxygen consumption of stimulated platelets, despite continued ATP requirement for energy-intensive processes such as cytoskeletal reorganization, shedding of extracellular vesicles, and protein synthesis, which are all associated with platelet activation. As aggregate formation could restrict access of oxygen to cells persisting within the core of the aggregate mass, we performed respirometry experiments after pre-treatment with ArgGly-Asp-Ser (RGDS), a tetrapeptide that prevents platelet aggregation. Strikingly, we noticed no difference in polarogram profiles (Online Supplementary Figure S1C), which ruled out any contribution of cell-cell aggregate formation to the observed drop in oxygen consumption. A reduced rate of cellular respiration could reflect dysfunctional mitochondria in agonist-treated platelets.18 We found that oxygen flux following sequential treatment of platelets with oligomycin (an inhibitor of ATPase), carbonyl cyanide m-chlorophenyl hydrazine (an uncoupler), and antimycin A (an inhibitor of respiratory complex III) was consistent with the presence of well-coupled and viable mitochondria in stimulated platelets (Online Supplementary Figure S2A). This prompted us to hypothesize a Warburg-like phenomenon19 in stimulated platelets. Consistent with this proposition, the decline in oxygen flux was obviated when platelets were exposed to DCA (Figure 1A,B), which enhances flux through the tricarboxylic acid (TCA) cycle by promoting pyruvate dehydrogenase activity, or to methylene blue, an alternative electron carrier (Online Supplementary Figure S2B). The Warburg effect, or aerobic glycolysis, entails augmented uptake of glucose from external medium by the cells. We detected 1.8- and 2-fold increases in glucose uptake and lactate generation, respectively, by platelets upon exposure to thrombin (Figure 1C). Enhanced glucose uptake by stimulated platelets is mediated through cell membrane translocation of cytosolic preformed GLUT3.20 As expected, we observed a nearly 41% rise in surface expression of GLUT3 upon stimulation of platelets with thrombin (0.5 U/mL) (Figure 1D,E). The surface mobilization of GLUT3 in neuronal cells is regulated by the activity of AMP-activated protein kinase (AMPK).21 In line with this, pre-treatment of platelets with compound C, which is a specific inhibitor of AMPK, resulted in a significant drop (by ̴ 31%) of GLUT3 expression on thrombin-stimulated platelets (Figure 1D,E).

Stimulated platelets switch to aerobic glycolysis through negative regulation of pyruvate dehydrogenase and pyruvate kinase M2 We next sought to elucidate the mechanism underlying the observed switch to aerobic glycolysis from oxidative phosphorylation in stimulated platelets. Pyruvate dehy808

drogenase (PDH), the ‘gate-keeper’ enzyme, determines the relative fluxes through glycolysis and oxidative phosphorylation. PDH activity is regulated through inhibitory phosphorylation (at S293) by pyruvate dehydrogenase kinase (PDK).22 We examined the expression of phosphorylated PDH in thrombin-stimulated platelets by using a phospho-specific antibody. Thrombin (0.5 U/mL) induced a rise in the level of phosphorylated PDH in platelets (Figure 2A,B), which diverts flux away from the TCA cycle. DCA (20 mM), a pharmacological inhibitor of PDK, almost completely abolished phosphorylation of PDH (Figure 2A,B) and partially restored the drop in oxygen consumption in thrombin-treated platelets (Figure 1A lower panel, 1B). Pre-exposure to DCA also led to significant inhibition of the thrombin-induced rises in the rates of glucose uptake and lactate generation (by 33% and 28%, respectively) (Figure 1C), suggesting PDK-mediated inactivation of PDH and decreased flux through the TCA cycle in stimulated platelets. Interestingly, DCA also triggered a 20% drop in GLUT3 externalization (Figure 1D,E) in thrombin-stimulated platelets. AMPK is known to induce phosphorylation of PDH under conditions of nutrient-deprivation leading to inhibition of PDH activity.23 Exposure to compound C reversed the increases in PDH phosphorylation (Figure 2A,B) as well as glucose uptake/lactate secretion rates (Figure 2E) that were observed in thrombin-stimulated platelets. As AMPK activity in platelets is upregulated by thrombin,24 this kinase is positioned in the thrombin signaling pathway, upstream of PDH. Neoplastic transformation of cells is associated with expression of PKM2, a splice variant of pyruvate kinase, which, unlike PKM1, facilitates aerobic glycolysis through a low-activity dimer state.25 We report that PKM2 is significantly expressed in human platelets (Online Supplementary Figure S3A). Expression of the PKM2 mRNA splice form was remarkably higher than that of its PKM1 counterpart (Online Supplementary Figure S3B). Phosphorylation of PKM2 at Y105 is associated with sustenance of the dimer state with attenuated catalytic activity.26 Using a phosphospecific antibody we observed that exposure to thrombin (0.5 U/mL) evoked appreciably higher phosphorylation (Y105) of PKM2, suggesting low enzymatic activity in stimulated platelets. This phosphorylation was reversed by PP2, an inhibitor of Src family tyrosine kinases, which are activated in stimulated platelets (Figure 2C). PP2 also regressed the thrombin-induced increases in glucose uptake/lactate secretion rates in platelets (Figure 2E). Pretreatment of cells with DASA (200 mM), an activator of PKM2, decreased thrombin-induced GLUT3 externalization by 36% (Figure 1D,E). Thus, our findings suggest that stimulated platelets make a metabolic switch to aerobic glycolysis through post-translational regulation of PDH and PKM2 enzyme activities.

Enhanced flux through the pentose phosphate pathway supports reactive oxygen species-dependent integrin activation Aerobic glycolysis and low PKM2 activity would lead to pooling of glycolytic intermediates upstream of pyruvate, which include glucose-6-phosphate, a substrate for the PPP.25 Hence, we studied metabolic flux through the PPP in thrombin-stimulated platelets, as reflected by the ratio of NADPH to total NADP(H) levels. Thrombin (0.5 U/mL) induced a significant rise (by 50%) in the ratio of NADPH haematologica | 2019; 104(4)

Aerobic glycolysis fuels platelet activation

Figure 1. Platelets exhibit higher rate of aerobic glycolysis upon agonist stimulation. (A) Polarograms exhibiting oxygen flow or oxygen concentration in thrombin (0.5 U/mL)-stimulated platelets suspended in Tyrode modified buffer. The platelets were not pretreated (upper panel) or were pretreated with DCA (20 mM) (lower panel). The blue line represents oxygen concentration within the chamber while the red line traces the rate of oxygen consumption by the cells. (B) Scatter dot plots representing routine, peak (following thrombin addition) and post-peak respiration in platelets. (C) Fold change in the rate of glucose uptake (black circles) and lactate secretion (blue squares) in platelets treated with different reagents as indicated. (D and E) GLUT3 externalization as observed by flow cytometry in platelets treated with different reagents as indicated. Data are presented as the mean Âą standard error of mean. Each dot represents an independent observation. (*P<0.05 as compared to resting platelets; #P<0.05 as compared to thrombin-stimulated platelets). CC, compound C; DCA: dichloroacetate; RP: resting platelets; Thr: thrombin.

haematologica | 2019; 104(4)

809

P.P. Kulkarni et al. A

Figure 2. Regulation of aerobic glycolysis and the pentose phosphate pathway in stimulated platelets. (A and C) Representative immunoblots showing expression of phospho-PDH (Ser 293) and phospho-PKM2 (Tyr 105), respectively, in platelets treated with different reagents (thrombin, 0.5 U/mL; compound C, 50 mM; DCA, 20 mM; PP2, 50 mM) as indicated. (B and D) Corresponding scatter dot plots showing expression of phospho-PDH relative to Î˛-actin and phospho-PKM2 relative to total PKM2, respectively, after densitometry analysis. (E) Glucose uptake and lactate secretion rates in platelets treated with different reagents. (F) Scatter dot plot showing the ratio of NADPH to total NADP(H) in platelets treated with different reagents as indicated. (G and H) ROS generation as detected by H2DCFDA dye in platelets treated with different reagents as indicated. The dose of apocynin was 600 mM. Data are presented as the mean Âą standard error of mean. Each dot represents an independent observation. (*P<0.05 as compared to resting platelets; #P<0.05 as compared to thrombin-stimulated platelets). Apo: apocynin; CC, compound C; DCA: dichloroacetate; DASA: diarylsulfonamide; DHEA: dehydroepiandrosterone sulphate; PDH: pyruvate dehydrogenase; PKM2: pyruvate kinase splice variant; RP: resting platelets; Thr: thrombin.

810

haematologica | 2019; 104(4)

Aerobic glycolysis fuels platelet activation

to total NADP(H), which was reversed by up to ~20% in the presence of DHEA (200 mM) (Figure 2F). DHEA, an endogenous steroid hormone, is an inhibitor of glucose 6phosphate dehydrogenase, the rate-limiting enzyme of the PPP. Small-molecule modulators that facilitate the TCA cycle and check the rate of aerobic glycolysis would also restrict flux of metabolites through the PPP. Predictably, pre-exposure of platelets to DCA (20 mM) or DASA (200 mM) led to reductions in the ratio of NADPH to total NADP(H) of 47% and 12%, respectively (Figure 2F). Activity of NADPH oxidase (NOX) has been widely documented as a significant source of ROS in stimulated platelets.27,28 We investigated whether NADPH accumulated in thrombin-stimulated platelets acts as a substrate for NOX, and leads to enhanced generation of ROS. In keeping with this possibility, pre-incubation of platelets with either DHEA or apocynin (a NOX inhibitor) prior to exposure to thrombin resulted in 63% and 62% attenuation in ROS generation, respectively. Remarkably, pre-treatment with DCA and DASA also led to 59% and 43% decreases, respectively, in ROS production in thrombin-stimulated platelets (Figure 2G,H). These decreases were consistent with the impeded PPP flux observed in the presence of these molecules. There is now considerable evidence to suggest that ROS are important mediators of the platelet activation signaling29 that culminates in the expression of conformationally-active integrin αIIbβIIIa.27 Consistently, DHEA as well as NOX inhibitors – DPI and apocynin – attenuated the binding of PAC-1 (an antibody directed against conformationally-active integrin αIIbβIIIa) to thrombin-stimulated platelets by 59%, 56%, and 62%, respectively (Figure 3A,B). DCA and DASA, which limit flux through the PPP, also inhibited PAC-1 binding by 32% and 65%, respectively. Furthermore, the role of ROS in mediating platelet activation was underscored by near-total abrogation of thrombin-induced PAC-1 binding to platelets in the presence of N-acetyl cysteine (10 mM), a ROS scavenger (Online Supplementary Figure S4). As an active conformer of αIIbβ3 is endowed with high affinity towards fibrinogen, we next studied the association of fluorescently labeled fibrinogen to thrombin-stimulated platelets in the presence of these small-molecule modulators. DCA, DASA, DHEA, DPI and apocynin attenuated fibrinogen binding to stimulated platelets by ̴ 28%, 58%, 37%, 49%, and 63% respectively (Figure 3C,D).

Small-molecule inhibitors of aerobic glycolysis and the pentose phosphate pathway impair platelet responsiveness As we demonstrated that small-molecule inhibitors of aerobic glycolysis and the PPP could prevent platelet surface integrins αIIbβ3 from switching to an active conformation and attenuate fibrinogen binding, we asked whether they could also inhibit activation-initiated platelet responses including aggregation and secretion of granule contents. Pre-treatment with DCA (20 mM), DASA (200 mM) or DHEA (200 mM) led to reductions in platelet aggregation evoked by thrombin (0.5 U/mL) by 61%, 58% and 19%, respectively (Figure 3E,F). A similar inhibition of collagen (2 mg/mL)-induced platelet aggregation, albeit to a much greater extent, was also observed (Figure 3G,H), which was consistent with a previous report that described inhibition of platelet aggregation by DHEA in haematologica | 2019; 104(4)

an Akt/ERK/p38 MAPK-dependent fashion.30 Surface externalization of P-selectin from α-granules and secretion of adenine nucleotides stored in dense granules dropped upon pre-treatment with DCA, DASA or DHEA (Online Supplementary Figure S5). These small molecules also inhibited phosphatidylserine exposure (measured through annexin V binding) on the surface of stimulated platelets (Figure 3I,J), which is critical for the procoagulant activity of the platelets.31 Thrombin elicited shedding of extracellular vesicles (size range, 25-800 nm; predominantly 100250 nm) bearing a phosphatidylserine-rich procoagulant surface32 from platelets, a phenomenon which was also mitigated by ~40% in the presence of DCA (Online Supplementary Figure S6). The foregoing observations strongly underscore the significance of aerobic glycolysis and the PPP-NOX axis in platelet activation and thrombosis. In contrast to DCA, pre-treatment with either antimycin or oligomycin did not significantly inhibit platelet aggregation induced by thrombin (0.2 U/mL) (Online Supplementary Figure S7), suggesting that mitochondrial respiration is relatively dispensable for platelet activation responses. To rule out the possibility that the observed attenuation in platelet activity was a consequence of cell death induced by the small-molecule modulators, we performed a lactate dehydrogenase leak assay. There was no significant release of lactate dehydrogenase activity from platelets when cells were exposed to DCA (20 mM), DASA (200 mM) or DHEA (200 mM) for up to 2 h (Online Supplementary Figure S8).

Small-molecule inhibitors of aerobic glycolysis and the pentose phosphate pathway preclude thrombosis in mice Platelets play a pivotal role in the pathogenesis of arterial thrombosis which underlies occlusive vasculopathies such as acute myocardial infarction and ischemic stroke. In order to establish a link between energy metabolism in activated platelets and arterial thrombosis in vivo, we studied the effect of small-molecule metabolic modulators in a murine model of mesenteric thrombosis induced by ferric chloride. We fluorescently labeled the platelets and induced intramural thrombus in exteriorized mesenteric arterioles of mice administered DCA (200 mg/kg intraperitoneal), DHEA (50 mg/kg intraperitoneal), DASA (40 mg/kg intravenous) or vehicle (control). Intravital imaging of thrombus formation was carried out by epifluorescence video microscopy using a high-speed camera. We documented the time required for initial thrombus formation, rate of thrombus growth and time to occlusion as pointers to the initiation, propagation and stabilization of thrombus, respectively. Strikingly, mice pre-treated with DCA (Online Supplementary Video S2), DHEA (Online Supplementary Video S3) or DASA (Online Supplementary Video S4) had significantly prolonged mean times to form first thrombus compared to vehicle-treated mice (DCA, 9.6 ± 1.2 min; DHEA, 11.8 ± 1.02 min; DASA, 9.67 ± 1.2 min, versus control, 5.9 ± 0.60 min) (Figure 4A,B) and impaired thrombus growth rate (Figure 4A,C,D) compared to vehicle-treated mice (Online Supplementary Video S1). While the mean occlusion time for control mice was 20.4 ± 2.38 min, stable occlusion failed to occur even after 40 min from the time of injury in all DHEA-pretreated mice and most of the mice administered DCA or DASA (except one from each group) (Figure 4E). These results suggest that agents that block metabolic reprogramming 811

P.P. Kulkarni et al. A

Figure 3. Decreased metabolic flux through the pentose phosphate pathway impairs platelet responsiveness to agonist stimulation. (A and B) PAC-1 binding in platelets treated with different reagents as indicated. (C and D) Binding of fluorescent fibrinogen to platelets treated with different reagents as indicated. (E and G) Platelet aggregation induced by thrombin (0.2 U/mL) and collagen (2 mg/mL), respectively in the presence of vehicle (control), DCA (20 mM), DASA (200 mM) or DHEA (200 mM) (tracings 1, 2, 3 and 4 respectively). (F and H) Scatter dot plots showing platelet aggregation. (I and J) Fluorescent-labeled annexin V binding to platelets treated with various reagents as indicated. Data are presented as the mean Âą standard error of mean. Each dot represents an independent observation. (*P<0.05 as compared to resting platelets; #P<0.05 as compared to thrombin-stimulated platelets). Apo: apocynin; DASA: diarylsulfonamide; DCA: dichloroacetate; DHEA: dehydroepiandrosterone sulphate; RP: resting platelets; Thr: thrombin

812

haematologica | 2019; 104(4)

Aerobic glycolysis fuels platelet activation

to aerobic glycolysis and the PPP in stimulated platelets cause profound impairment of arterial thrombosis in vivo. We corroborated the above findings in a mouse model of collagen-epinephrine-induced pulmonary embolism. Mice were pretreated with DCA (200 mg/kg), DHEA (50 mg/kg), DASA (40 mg/kg) or vehicle (control), followed

by intravenous administration of a collagen-epinephrine mixture to induce pulmonary embolism. Hematoxylin & eosin-stained lung sections from mice pretreated with small-molecule inhibitors displayed significantly fewer thrombosed pulmonary vessels [DCA, 4.12 ± 1.4 per low power field (lpf); DHEA, 4.02 ± 0.82/lpf; DASA, 3.03 ±

Figure 4. Small-molecule inhibitors of aerobic glycolysis and the pentose phosphate pathway impair thrombus formation in mice. (A) Representative time lapse images exhibiting thrombus formation in mice, pre-administered either vehicle (control), DCA, DHEA, or DASA captured 5, 10 or 15 min after injury of mesenteric arterioles of >100 mm diameter with ferric chloride. (B and E) Scatter dot plots representing, respectively, time to first thrombus formation and time to stable occlusion in mice pre-administered vehicle (control), DCA, DHEA or DASA. (C and D) Line graph and bar diagram showing thrombus growth rate in different treatment groups, as indicated. Each dot in the scatter plots represents an independent observation. Data are expressed as the mean ± standard error of mean. *P<0.05 with respect to vehicle-treated mice. DASA: diarylsulfonamide; DCA: dichloroacetate; DHEA: dehydroepiandrosterone sulphate.

haematologica | 2019; 104(4)

813

P.P. Kulkarni et al.

0.74 /lpf; versus control 8.1 Âą 0.95/lpf) with erythrocyte extravasation in some fields. These observations establish that prohibition of platelet aerobic glycolysis or the PPP prevents thrombosis, and may impair hemostasis. We next evaluated the effect of metabolic modulators on primary hemostasis in mice by a tail-bleeding assay. Administration of DCA (200 mg/kg, intraperitoneal), DHEA (50 mg/kg, intraperitoneal) or DASA (40 mg/kg, intravenous) to the mice was associated with prolonged bleeding times (Figure 5C) as well as an increase in the

amount of bleeding compared to that of vehicle-treated animals (Online Supplementary Figure S9). These observations establish that aerobic glycolysis and consequent flux through the PPP are also essential for hemostasis.

Discussion The essence of platelet function is response to stimuli. Platelets respond to hemostatic cues with a series of activ-

Figure 5. Small-molecule inhibitors of aerobic glycolysis and the pentose phosphate pathway impair pulmonary embolism and hemostasis. (A) Representative light microscopy images (100X magnification) of hematoxylin & eosin-stained lung sections from mice administered collagen (1 mg/kg) plus epinephrine (10 mg/kg) after pre-treatment with vehicle (control), DCA, DHEA, or DASA as indicated. Black arrows indicate thrombi within the lumen of pulmonary vessels. Yellow arrows indicate extravasated erythrocytes. (B and C) Scatter dot plots showing number of thrombosed pulmonary vessels per low power field and tail-bleeding times, respectively, of mice administered with vehicle, DCA, DHEA or DASA. Data are presented as the mean Âą standard error of mean. Each dot represents an independent observation. DASA: diarylsulfonamide; DCA: dichloroacetate; DHEA: dehydroepiandrosterone sulphate.

814

haematologica | 2019; 104(4)

Aerobic glycolysis fuels platelet activation

ities, both early and late, which are categorically energyintensive.5,6 These activities lead to the generation of macroscopic cell-cell aggregates and fibrin-rich thrombi that can stop bleeding or potentially occlude blood vessels with disastrous consequences to human health. Evidently, platelet energy metabolism sustains a thrombus until it is lysed by the fibrinolytic system. Hence, in our search for an effective anti-thrombotic strategy, we focused on metabolic adaptations of platelets to agonist stimulation. We reasoned that disruption of key metabolic steps would prevent platelet activation, which could be developed into a potent anti-thrombotic measure. When we exposed platelets to physiological agonists, there was sharp rise in cellular oxygen consumption apparently driven by energy-demanding early platelet responses such as shape change, cytoskeletal reorganization, aggregation and granule secretion.4,7,33,34 Intriguingly, this rise was short-lived and was followed by an abrupt drop at a time when platelets continued to aggregate and would possibly be required to discharge resource-intensive late responses such as retraction of fibrin clot, shedding of extracellular vesicles and protein synthesis. We

sought the reason behind this rapid plunge in oxygen consumption despite enhanced ATP requirement. Aggregate formation could restrict access of oxygen to cells persisting within the core of the aggregate mass. However, this possibility was ruled out as oxygen flux in cells pre-treated with RGDS, a tetrapeptide that blocks platelet aggregation, was identical to that in aggregated platelets. Possible mitochondrial dysfunction in stimulated cells was also unlikely as leak respiration and non-mitochondrial oxygen consumption in oligomycin- and antimycin A-treated activated platelets, respectively, were suggestive of viable and well-coupled mitochondria. This prompted us to hypothesize a Warburg effect-like phenomenon in stimulated platelets whereby pyruvate is prevented from oxidation in the TCA cycle leading to a decline in mitochondrial respiration. Consistent with this possibility, the drop in oxygen flux was obviated when platelets were exposed to agents such as DCA (which enhances flux through the TCA cycle), or alternative electron carriers (methylene blue and toluidine blue O). The Warburg effect or aerobic glycolysis would require enhanced uptake of glucose by the cells in order to sustain

Figure 6. Scheme for metabolic flux in stimulated platelets and sites of action of small-molecule modulators. Stimulated platelets switch to aerobic glycolysis through negative regulation of pyruvate kinase M2 and pyruvate dehydrogenase enzyme activities. The consequent increase in flux through the pentose phosphate pathway generates NADPH that fuels ROS generation by NADPH oxidase. ROS signaling in turn mediates platelet activation, thrombosis and hemostasis. Small-molecule modulators that reverse this metabolic adaptation inhibit platelet activation and impair thrombus formation. DASA: diarylsulfonamide; DCA: dichloroacetate; DHEA: dehydroepiandrosterone sulphate; NOX: NADPH oxidase; PDH: pyruvate dehydrogenase; PDK: pyruvate dehydrogenase kinase PKM2: pyruvate kinase M2; PPP: pentose phosphate pathway; ROS: reactive oxygen species.

haematologica | 2019; 104(4)

815

P.P. Kulkarni et al.

adequate ATP generation by glycolysis. In keeping with this and an earlier report,20 we observed surface translocation of cytosolic GLUT3, the major glucose transporter, in thrombin-stimulated platelets. As already observed for neuronal cells,21 GLUT3 externalization in platelets was facilitated by AMPK, whose activity is known to be upregulated by agonist stimulation.24 Enhanced surface expression of GLUT3 was corroborated by a nearly 2-fold rise in the uptake of glucose in thrombin-activated platelets. In parallel there was enhanced secretion of lactate into medium, which was in line with earlier reports of extracellular acidification induced by agonists.34-37 The metabolic checkpoint at the level of pyruvate, which is generated by the enzymatic activity of pyruvate kinase, determines whether glucose will be catabolized to lactate (by glycolysis) or through mitochondrial pathways. The enzyme PDH, part of a larger complex, catalyzes conversion of pyruvate to acetyl-CoA, which is then channeled into the TCA cycle for further oxidation and generation of ATP. Reactions catalyzed by PDH and pyruvate kinase are thus two critical hubs in the landscape of cellular glucose metabolism. PDH activity is regulated through inhibitory serine-293 phosphorylation by the enzyme PDK.22 An increase in the activity of PDK and/or reduced PDH activity restricts flux of pyruvate into the TCA cycle and favors aerobic glycolysis. Inhibitory phosphorylation of PDH is known to be induced by AMPK under conditions of nutrient-deprivation.23 We observed an increase in phosphorylated PDH in agonist-stimulated platelets, which was reversed by inhibition of AMPK activity. Notably, we also discovered that platelets express PKM2, a tissue-specific isoform of pyruvate kinase, which, in low-activity state, facilitates aerobic glycolysis.38 Phosphorylation of PKM2 at Y105 is associated with sustenance of a dimer state with attenuated catalytic activity.26 Interestingly, agonist stimulation of platelets also led to increased expression of phosphorylated PKM2, which could be reduced to a basal level by inhibition of Src family kinases. Thus, we show here that metabolism in stimulated platelets switches to aerobic glycolysis through active regulation of PDH and PKM2; this switch leads to accumulation of glycolytic intermediates upstream of pyruvate and facilitates flux through the PPP.25 Consistent with this we observed significant enrichment in the pool of NADPH, a direct readout of the PPP, in thrombin-stimulated platelets. Increased levels of NADPH in most cells lead to a reductive state through generation of reduced glutathione.25 Yet, in an apparent paradox, NADPH can also serve as a substrate for the enzyme NADPH oxidase generating superoxide free radicals in the phagocytic cells, such as neutrophils, of the innate immune system.39 Platelets are now widely recognized to be key players in immune responses.40,41 Possibly owing to genealogical and now increasingly recognized functional relationships of platelets with innate immunity, these cells, too, express significant levels of NOX activity,28 which is in fact the primary source of ROS in activated platelets.27,28 Our findings were consistent with a rise in the levels of NADPH in stimulated platelets serving to increase the generation of ROS through activity of NOX. There is now compelling evidence indicating that NOX-generated ROS are important mediators of platelet activation signaling,29,42 which culminates in the expression of a conformationally active form of integrins αIIbβIIIa.27,43 Remarkably, pre-treatment of 816

platelets with small-molecule inhibitors of either aerobic glycolysis (DCA or DASA) or the PPP (DHEA) brought about a significant drop in agonist-induced ROS production as well as integrin activation, and was associated with profound impairment in platelet responses to agonists (platelet aggregation, fibrinogen binding and secretion of contents of dense and α granules). The metabolic inhibitors also attenuated phosphatidylserine exposure and extracellular vesicle release, both measures of a procoagulant phenotype31,32 in platelets, thus underlining the significance of aerobic glycolysis and the PPP-NOX axis in platelet activation and thrombosis. In agreement with our findings, genetic deficiency of glucose 6-phosphate dehydrogenase, the key regulatory enzyme in the PPP, is known to be associated with a remarkably lower risk of cardiovascular mortality.44,45 Strikingly, intravenous administration of metabolic modulators significantly delayed thrombus formation in mesenteric arterioles, retarded thrombus growth and prolonged time needed for complete vascular occlusion in murine models. Thus, our findings suggest that aerobic glycolysis and associated flux through the PPP in stimulated platelets play critical roles in arterial thrombosis and underscore the therapeutic potential of targeting metabolic pathways as a novel antithrombotic approach (Figure 6). These results are also consistent with two recent articles by Abel’s group, which establish the role of glucose metabolism in platelet activation using transgenic mice with platelet-specific ablation of glucose transporters.36,37 Currently available anti-platelet agents are plagued by limited efficacy2 which could be attributed to remarkable redundancy in agonist stimuli and downstream signaling pathways that lead to platelet activation. Thus, conventional anti-platelet therapeutic regimens that target specific agonists/activation pathways may not confer absolute protection against thrombotic episodes, as platelets can still be stimulated by other potential triggers and parallel signaling inputs. It is, therefore, vital to discover novel anti-platelet strategies to address these challenges and cater to this largely unmet medical need. The small-molecule metabolic modulators employed in our study block essential metabolic checkpoints in stimulated platelets and effectively prevent platelet activation irrespective of the nature of the agonists and ensuing signaling pathways. We validated our findings in a murine model of thrombosis in which deep vein thrombosis and pulmonary embolism were induced by intravenous administration of collagen and epinephrine. Notably, mice pre-treated with small-molecule metabolic modulators were protected from pulmonary thromboembolism. However, lung sections in these animals revealed extravasated erythrocytes, which was in keeping with prolonged tail-bleeding following administration of metabolic modulators. Thus, these small molecules can impair hemostasis despite conferring significant protection against thrombosis. Therefore, targeting platelet metabolism, much like currently available anti-platelet regimens, would involve a tightrope-walk of dose adjustment for effective prevention of thrombotic events without raising the risk of bleeding complications. Remarkably, metabolic modulators employed in our study have some differential effects on tail bleeding and thrombus stability, which raises hope for preferential targeting of thrombosis over hemostasis in the future by tweaking specific metabolic pathways. haematologica | 2019; 104(4)

Aerobic glycolysis fuels platelet activation

While it would not be possible to predict the off-target adverse effects of the small-molecule metabolic modulators at this stage, the potential for non-selective effects being beneficial to overall cardiovascular health cannot be overemphasized. DCA has been found to be effective in improving cardiac function after ischemia/reperfusion injury46-48 as well as in preventing restenosis after vessel injury49 in preclinical models. There is substantial clinical evidence linking higher serum DHEA levels to decreased cardiovascular mortality.50 Small-molecule activators of PKM2 have been shown to reverse the pro-inflammatory phenotype of monocytes/macrophages isolated from patients with atherosclerotic coronary artery disease.51 Although DASA is only in preclinical stages of development as an antineoplastic drug,52 DCA is already under clinical trials against various cancers10,53 and congenital lactic acidosis8,54 while DHEA was recently granted Food and Drug Administration approval for clinical use in postmenopausal women.9 Hence, considerable information on the safety and pharmacokinetics of the latter drugs in humans is already available, which could in future pave the way for clinical trials of the drugs as potential anti-

References 1. Murray CJ, Lopez AD. Measuring the global burden of disease. N Engl J Med. 2013;369 (5):448-457. 2. Franchi F, Angiolillo DJ. Novel antiplatelet agents in acute coronary syndrome. Nat Rev Cardiol. 2015;12(1):30-47. 3. Furie B, Furie BC. Mechanisms of thrombus formation. N Engl J Med. 2008;359(9):938949. 4. Holmsen H, Kaplan KL, Dangelmaier CA. Differential energy requirements for platelet responses. A simultaneous study of aggregation, three secretory processes, arachidonate liberation, phosphatidylinositol breakdown and phosphatidate production. Biochem J. 1982;208(1):9-18. 5. Holmsen H. Energy metabolism and platelet responses. Vox Sang. 1981;40 (Suppl 1):1-7. 6. Verhoeven AJ, Mommersteeg ME, Akkerman JW. Metabolic energy is required in human platelets at any stage during optical aggregation and secretion. Biochim Biophys Acta. 1984;800(3):242-250. 7. Verhoeven AJ, Mommersteeg ME, Akkerman JW. Quantification of energy consumption in platelets during thrombininduced aggregation and secretion. Tight coupling between platelet responses and the increment in energy consumption. Biochem J. 1984;221(3):777-787. 8. Stacpoole PW, Kerr DS, Barnes C, et al. Controlled clinical trial of dichloroacetate for treatment of congenital lactic acidosis in children. Pediatrics. 2006;117(5):1519-1531. 9. Labrie F, Archer DF, Koltun W, et al. Efficacy of intravaginal dehydroepiandrosterone (DHEA) on moderate to severe dyspareunia and vaginal dryness, symptoms of vulvovaginal atrophy, and of the genitourinary syndrome of menopause. Menopause. 2016;23(3):243-256. 10. Michelakis ED, Sutendra G, Dromparis P, et

haematologica | 2019; 104(4)

11.

12.

13.

14.

15.

16.

17.

18.

19. 20.

platelet/anti-thrombotic agents in the management of thrombotic disorders. In conclusion, this study suggests an indispensable role for platelet energy metabolism in thrombogenesis with potential implications for the development of anti-thrombotic strategies. Future investigations directed at metabolic adaptations of platelets within a developing thrombus in vivo could potentially identify many more therapeutic opportunities. Acknowledgments This research was supported by a JC Bose fellowship and grants received by D. Dash from the Department of Biotechnology (DBT), Science and Engineering Research Board (SERB) and Department of Science and Technology (DST), Government of India, the Indian Council of Medical Research (ICMR) and the Council of Scientific and Industrial Research (CSIR). A.T. and D.G. are recipients of UGC-JRF and CSIRJRF, respectively. We are grateful to the technical staff for their assistance in performing the histopathological analysis of murine lung tissue. We thank Pradyumna Kulkarni for proofreading the manuscript. D.D. acknowledges assistance from the Humboldt Foundation, Germany.

al. Metabolic modulation of glioblastoma with dichloroacetate. Sci Transl Med. 2010;2(31):31ra34. Berendzen K, Theriaque DW, Shuster J, Stacpoole PW. Therapeutic potential of dichloroacetate for pyruvate dehydrogenase complex deficiency. Mitochondrion. 2006;6 (3):126-135. Sonkar VK, Kulkarni PP, Dash D. Amyloid beta peptide stimulates platelet activation through RhoA-dependent modulation of actomyosin organization. FASEB J. 2014;28(4):1819-1829. Gross MI, Demo SD, Dennison JB, et al. Antitumor activity of the glutaminase inhibitor CB-839 in triple-negative breast cancer. Mol Cancer Ther. 2014;13(4):890901. Singh SK, Singh MK, Kulkarni PP, Sonkar VK, Gracio JJ, Dash D. Amine-modified graphene: thrombo-protective safer alternative to graphene oxide for biomedical applications. ACS Nano. 2012;6(3):2731-2740. Nayak MK, Kulkarni PP, Dash D. Regulatory role of proteasome in determination of platelet life span. J Biol Chem. 2013;288 (10):6826-6834. Prakash P, Kulkarni PP, Chauhan AK. Thrombospondin 1 requires von Willebrand factor to modulate arterial thrombosis in mice. Blood. 2015;125(2):399-406. Prakash P, Kulkarni PP, Lentz SR, Chauhan AK. Cellular fibronectin containing extra domain A promotes arterial thrombosis in mice through platelet Toll-like receptor 4. Blood. 2015;125(20):3164-3172. Jobe SM, Wilson KM, Leo L, et al. Critical role for the mitochondrial permeability transition pore and cyclophilin D in platelet activation and thrombosis. Blood. 2008;111(3): 1257-1265. Warburg O. On the origin of cancer cells. Science. 1956;123(3191):309-314. Sorbara LR, Davies-Hill TM, Koehler-Stec EM, Vannucci SJ, Horne MK, Simpson IA.

21.

22.

23.

24.

25. 26.

27.

28.

29.

30.

Thrombin-induced translocation of GLUT3 glucose transporters in human platelets. Biochem J. 1997;328(Pt 2):511-516. Weisova P, Concannon CG, Devocelle M, Prehn JH, Ward MW. Regulation of glucose transporter 3 surface expression by the AMP-activated protein kinase mediates tolerance to glutamate excitation in neurons. J Neurosci. 2009;29(9):2997-3008. Holness MJ, Sugden MC. Regulation of pyruvate dehydrogenase complex activity by reversible phosphorylation. Biochem Soc Trans. 2003;31(Pt 6):1143-1151. Wu CA, Chao Y, Shiah SG, Lin WW. Nutrient deprivation induces the Warburg effect through ROS/AMPK-dependent activation of pyruvate dehydrogenase kinase. Biochim Biophys Acta. 2013;1833(5):11471156. Randriamboavonjy V, Isaak J, Fromel T, et al. AMPK alpha2 subunit is involved in platelet signaling, clot retraction, and thrombus stability. Blood. 2010;116(12):2134-2140. Cairns RA, Harris IS, Mak TW. Regulation of cancer cell metabolism. Nat Rev Cancer. 2011;11(2):85-95. Hitosugi T, Kang S, Vander Heiden MG, et al. Tyrosine phosphorylation inhibits PKM2 to promote the Warburg effect and tumor growth. Sci Signal. 2009;2(97):ra73. Begonja AJ, Gambaryan S, Geiger J, et al. Platelet NAD(P)H-oxidase-generated ROS production regulates alphaIIbbeta3-integrin activation independent of the NO/cGMP pathway. Blood. 2005;106(8):2757-2760. Seno T, Inoue N, Gao D, et al. Involvement of NADH/NADPH oxidase in human platelet ROS production. Thromb Res. 2001;103(5):399-409. Krotz F, Sohn HY, Pohl U. Reactive oxygen species: players in the platelet game. Arterioscler Thromb Vasc Biol. 2004;24 (11):1988-1996. Bertoni A, Rastoldo A, Sarasso C, et al. Dehydroepiandrosterone-sulfate inhibits

817

P.P. Kulkarni et al.

31. 32. 33.

34.

35.

36.

37.

38.

818

thrombin-induced platelet aggregation. Steroids. 2012;77(3):260-268. Lentz BR. Exposure of platelet membrane phosphatidylserine regulates blood coagulation. Prog Lipid Res. 2003;42(5):423-438. Owens AP, 3rd, Mackman N. Microparticles in hemostasis and thrombosis. Circ Res. 2011;108(10):1284-1297. Verhoeven AJ, Mommersteeg ME, Akkerman JW. Comparative studies on the energetics of platelet responses induced by different agonists. Biochem J. 1986;236(3): 879-887. Akkerman JW, Holmsen H. Interrelationships among platelet responses: studies on the burst in proton liberation, lactate production, and oxygen uptake during platelet aggregation and Ca2+ secretion. Blood. 1981;57(5):956-966. Ravi S, Chacko B, Sawada H, et al. Metabolic plasticity in resting and thrombin activated platelets. PloS one. 2015;10(4): e0123597. Fidler TP, Campbell RA, Funari T, et al. Deletion of GLUT1 and GLUT3 reveals multiple roles for glucose metabolism in platelet and megakaryocyte function. Cell Rep. 2017;20(9):2277. Fidler TP, Middleton EA, Rowley JW, et al. Glucose transporter 3 potentiates degranulation and is required for platelet activation. Arterioscler Thromb Vasc Biol. 2017;37(9): 1628-1639. Mazurek S. Pyruvate kinase type M2: a key regulator of the metabolic budget system in tumor cells. Int J Biochem Cell Biol. 2011;43(7):969-980.

39. Lambeth JD, Neish AS. Nox enzymes and new thinking on reactive oxygen: a doubleedged sword revisited. Annu Rev Pathol. 2014;9:119-145. 40. Semple JW, Italiano JE Jr., Freedman J. Platelets and the immune continuum. Nat Rev Immunol. 2011;11(4):264-274. 41. Herter JM, Rossaint J, Zarbock A. Platelets in inflammation and immunity. J Thromb Haemost. 2014;12(11):1764-1775. 42. Dayal S, Wilson KM, Motto DG, Miller FJ, Jr., Chauhan AK, Lentz SR. Hydrogen peroxide promotes aging-related platelet hyperactivation and thrombosis. Circulation. 2013;127(12):1308-1316. 43. Jang JY, Min JH, Chae YH, et al. Reactive oxygen species play a critical role in collagen-induced platelet activation via SHP-2 oxidation. Antioxid Redox Signal. 2014;20 (16):2528-2540. 44. Cocco P, Fadda D, Schwartz AG. Subjects expressing the glucose-6-phosphate dehydrogenase deficient phenotype experience a lower cardiovascular mortality. QJM. 2008;101(2):161-163. 45. Cocco P, Todde P, Fornera S, Manca MB, Manca P, Sias AR. Mortality in a cohort of men expressing the glucose-6-phosphate dehydrogenase deficiency. Blood. 1998;91 (2):706-709. 46. Barak C, Reed MK, Maniscalco SP, Sherry AD, Malloy CR, Jessen ME. Effects of dichloroacetate on mechanical recovery and oxidation of physiologic substrates after ischemia and reperfusion in the isolated heart. J Cardiovasc Pharmacol. 1998;31(3): 336-344.

47. Bersin RM, Stacpoole PW. Dichloroacetate as metabolic therapy for myocardial ischemia and failure. Am Heart J. 1997;134(5 Pt 1):841-855. 48. Wargovich TJ, MacDonald RG, Hill JA, Feldman RL, Stacpoole PW, Pepine CJ. Myocardial metabolic and hemodynamic effects of dichloroacetate in coronary artery disease. Am J Cardiol. 1988;61(1):65-70. 49. Deuse T, Hua X, Wang D, et al. Dichloroacetate prevents restenosis in preclinical animal models of vessel injury. Nature. 2014;509(7502):641-644. 50. Wu TT, Chen Y, Zhou Y, et al. Prognostic value of dehydroepiandrosterone sulfate for patients with cardiovascular disease: a systematic review and meta-analysis. J Am Heart Assoc. 2017;6(5):e004896. 51. Shirai T, Nazarewicz RR, Wallis BB, et al. The glycolytic enzyme PKM2 bridges metabolic and inflammatory dysfunction in coronary artery disease. J Exp Med. 2016;213(3): 337-354. 52. Anastasiou D, Yu Y, Israelsen WJ, et al. Pyruvate kinase M2 activators promote tetramer formation and suppress tumorigenesis. Nat Chem Biol. 2012;8(10):839-847. 53. Dunbar EM, Coats BS, Shroads AL, et al. Phase 1 trial of dichloroacetate (DCA) in adults with recurrent malignant brain tumors. Invest New Drugs. 2014;32(3):452464. 54. Stacpoole PW, Gilbert LR, Neiberger RE, et al. Evaluation of long-term treatment of children with congenital lactic acidosis with dichloroacetate. Pediatrics. 2008;121(5): e1223-1228.

haematologica | 2019; 104(4)

ARTICLE

Coagulation & its Disorders

New insight into antiphospholipid syndrome: antibodies to β2glycoprotein I-domain 5 fail to induce thrombi in rats

Ferrata Storti Foundation

Paolo Durigutto,1 Claudia Grossi,2 Maria Orietta Borghi,2,3 Paolo Macor,1 Francesca Pregnolato,2 Elena Raschi,2 Michael P. Myers,4 Philip G. de Groot,5 Pier Luigi Meroni2 and Francesco Tedesco2

1 Department of Life Sciences, University of Trieste, Italy; 2Istituto Auxologico Italiano, IRCCS, Laboratory of Immuno-Rheumatology, Milan, Italy; 3Department of Clinical Sciences and Community Health, University of Milan, Italy; 4International Centre for Genetic Engineering and Biotechnology, Trieste, Italy and 5Department of Clinical Chemistry and Haematology, University of Utrecht, University Medical Center Utrecht, the Netherlands

Haematologica 2019 Volume 104(4):819-826

PD and CG contributed equally to this work. PLM and FT contributed equally to this work.

ABSTRACT

linical studies have reported different diagnostic/predictive values of antibodies to domain 1 or 4/5 of β2glycoproteinI in terms of risk of thrombosis and pregnancy complications in patients with antiphospholipid syndrome. To obtain direct evidence for the pathogenic role of anti-domain 1 or anti-domain 4/5 antibodies, we analyzed the in vivo pro-coagulant effect of two groups of 5 sera IgG each reacting selectively with domain 1 or domain 5 in lipopolysaccharide (LPS)-treated rats. Antibody-induced thrombus formation in mesenteric vessels was followed by intravital microscopy, and vascular deposition of β2glycoproteinI, human IgG and C3 was analyzed by immunofluorescence. Five serum IgG with undetectable anti-β2glycoproteinI antibodies served as controls. All the anti-domain 1-positive IgG exhibited potent pro-coagulant activity while the anti-domain 5-positive and the negative control IgG failed to promote blood clot and vessel occlusion. A stronger granular deposit of IgG/C3 was found on the mesenteric endothelium of rats treated with anti-domain 1 antibodies, as opposed to a mild linear IgG staining and absence of C3 observed in rats receiving anti-domain 5 antibodies. Purified anti-domain 5 IgG, unlike anti-domain 1 IgG, did not recognize cardiolipin-bound β2glycoproteinI while being able to interact with fluid-phase β2glycoproteinI. These findings may explain the failure of anti-domain 5 antibodies to exhibit a thrombogenic effect in vivo, and the interaction of these antibodies with circulating β2glycoproteinI suggests their potential competitive role with the pro-coagulant activity of anti-domain 1 antibodies. These data aim at better defining “really at risk” patients for more appropriate treatments to avoid recurrences and disability.

Introduction Antiphospholipid syndrome (APS) is a chronic autoimmune disorder characterized by recurrent episodes of vascular thrombosis and adverse pregnancy outcomes in the presence of antibodies to phospholipid-binding proteins (aPL). It occurs either as a primary disease or concomitantly to other connective tissue diseases, particularly systemic lupus erythematosus (SLE).1 Although thrombotic occlusion may affect the vessels of all organs and tissues, common presentations of the syndrome are: a) deep vein thrombosis in the legs often complicated by pulmonary embolism; and b) thrombotic occlusion of cerebral and coronary arteries leading to stroke and myocardial infarction.2 This clinical condition is also associated with pregnancy morbidity, including fetal loss, pre-eclampsia, pre-term delivery, and ‘small for gestational age’ babies.3 These are serious complications that haematologica | 2019; 104(4)

Correspondence: FRANCESCO TEDESCO tedesco@units.it Received: May 22, 2018. Accepted: November 14, 2018. Pre-published: November 15, 2018. doi:10.3324/haematol.2018.198119 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/819 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

819

P. Durigutto et al.

particularly affect young people, and have both social and economic impacts. The disease may sometimes present as catastrophic syndrome, a more severe form of APS characterized by microthrombosis of small vessels in various organs resulting in multiple organ failure.4 Anti-cardiolipin (aCL) and anti-β2glycoprotein I (β2GPI) antibodies and lupus anticoagulant (LA) activity are considered markers of APS and are included among the criteria currently proposed to classify the syndrome.1 Clinical studies have revealed an increased risk of thrombosis and pregnancy complications in patients with medium to high levels of these antibodies and LA present in their plasma.5 The triple positivity of these laboratory markers has also been shown to be associated with more severe forms of APS.5 Conversely, the positivity for a single marker is often associated with a much lower risk of the clinical manifestations of APS.5-9 It has been widely demonstrated that β2GPI is the main antigen recognized by aPL, and the reactivity against the protein has been shown to be responsible for the positivity for aCL and anti-β2GPI assays, and, in part, for the LA phenomenon strongly associated with the clinical manifestations of APS.10 β2GPI mainly circulates in blood in a circular form and is organized into four domains (D1-D4) composed of 60 amino acids with two disulfide bonds and a fifth domain (D5) containing an extra 24 amino acids that interact with anionic phospholipids on the target cells/tissues.11 Besides the classical diagnostic assays measuring antibodies against whole molecule β2GPI, new tests have recently been developed to detect anti-β2GPI antibody subpopulations reacting with different domains of the protein, particularly the combined domains D4/5 and domain 1 (D1).5-7,9,12-14 In APS patients, a large proportion of anti-β2GPI antibodies react with D1 and recognize a cryptic epitope (Arg39–Arg43) in the native molecule exposed after its interaction with anionic phospholipids13,15 or oxidation.1618 Antibodies directed against D1 of β2GPI with or without anti-D4/5 antibodies have frequently been found in APS patients associated with an increased risk of thrombosis and pregnancy complications.7,9,19-24 In contrast, isolated high levels of anti-D4/5 antibodies have been reported in non-APS patients with leprosy, atopic dermatitis, atherosclerosis and in children born to mothers with systemic autoimmune diseases;6 high levels have also been found in asymptomatic aPL carriers although these antibodies are not associated with either vascular or obstetric manifestations of the APS syndrome.7,9 This finding prompted some authors to suggest that the ratio between anti-D1 and anti-D4/5 may be a useful parameter for identifying autoimmune APS and for ranking the patients according to their risk of developing the syndrome.7 An isolated positivity for anti-D4/5 is a rare condition and is usually associated with the absence of aCL and/or LA. In the majority of cases, there is some doubt as to the APS clinical profile and classification/diagnostic criteria are not fulfilled.25 The finding that antibodies with this isolated specificity are observed mainly in the absence of clinical manifestations of hypercoagulable states has suggested that they may not be involved in thrombus formation. The in vivo pathogenic role of aPL has been demonstrated for those directed against the whole molecule and against D1 of β2GPI using animal models of thrombosis developed in rats and mice.26-28 However, at present, there 820

is no direct evidence that antibodies to D4/5 do not play an in vivo pathogenic role in blood clotting, nor is it clear whether they are able to interact with soluble or surfacebound β2GPI. Data indicating that the antibodies are ineffective in causing blood clot due to their failure to recognize bound β2GPI will be reported.

Methods Serum source Two groups of anti-β2GPI positive sera7,27 containing isolated antibodies to either D1 or D4/5 domains6,7 and control sera with undetectable anti-β2GPI antibodies were analyzed. All samples were also tested for aCL antibodies7 and LA activity.29 The antiD1-positive sera were obtained from APS patients.1 The sera were collected after obtaining informed consent and the IgG were purified by a Protein G column (HiTrap Protein G HP, GE Healthcare) as described.27 The local Istituto Auxologico Italiano ethical committee approved the study.

Purification of β2glycoprotein I and generation of recombinant domains D4 and D5

Methods of purification of human β2GPI from pooled normal sera and the generation of D4 and D5 domains have been published previously.12,27,30,31 Sequence analysis was performed as described32 and compared to the published sequence of β2GPI.33 The fine specificity against D4 or D5 was investigated by ELISA.27

Animal model An in vivo model of antibody-induced thrombus formation was established in male Wistar rats (270-300 g) kept under standard conditions in the Animal House of the University of Trieste, Italy, as previously reported in detail.26 The in vivo procedures were performed in compliance with the guidelines of European (86/609/EEC) and Italian (Legislative Decree 116/92) laws and were approved by the Italian Ministry of University and Research and the Administration of the University Animal House. This study was conducted in accordance with the Declaration of Helsinki. Further details are available in the Online Supplementary Methods.

Immunofluorescence analysis The mesenteric tissue was collected from rats at the end of the in vivo experiment.26 Deposits of β2GPI were analyzed using the biotinylated monoclonal antibody MBB2 and FITC-labeled streptavidin (Sigma-Aldrich).27 IgG and C3 were detected using FITClabeled goat anti-human IgG (Sigma-Aldrich) and goat anti-rat C3 (Cappel/MP Biomedicals) followed by FITC-labeled rabbit antigoat IgG (Dako), respectively. The slides were examined using a DM2000 fluorescence microscope equipped with a DFC 490 photo camera and Application Suite software (Leica).

Antibody binding assays

Different concentrations of β2GPI were added to CL-coated plates and the reactivity of IgG with CL-bound β2GPI was measured.7 The interaction of IgG with soluble β2GPI was evaluated by incubating IgG with increasing concentrations of β2GPI or bovine serum albumin (BSA) as unrelated antigen for one hour (h) at 37°C followed by overnight incubation at 4°C in a rotator. The samples were centrifuged at 3000 g for 5 minutes (min) at room temperature and the residual un-complexed antibodies were tested using β2GPI-coated plates (Combiplate EB, Labsystems) as described.7 Further details are available in the Online Supplementary Methods. haematologica | 2019; 104(4)

Anti-β2GPI-D5 are not thrombogenic in animals

Figure 1. Anti-domain (D) 4/5 antibodies specifically react against domain (D) 5 of β2glycoprotein I (β2GPI). Reactivity of 5 anti-D4/5-positive patient sera (P1-P5) against different recombinant human β2GPI domains. (A) Reactivity against the combined D4/5 peptides ( ), in an assay produced for research use (QUANTA Lite β2GPI D4/5 ELISA, INOVA Diagnostics). (B) Reactivity against the recombinant domain D4 ( ) or D5 ( ) antigens separately immobilized on the wells of γ-irradiated polystyrene plates in in-house ELISA. Optical Density (O.D.) values are expressed as mean±Standard Deviation. Data were analyzed with the Student t-test for paired data. The average reactivity against D5 is significantly higher than that against D4 (P=0.0428).

Statistical analysis Statistical analysis was performed using GraphPad Prism 6.0 for Windows. The domain reactivity of the anti-β2GPI D4/5 positive sera was expressed as mean+Standard Deviation (SD) and analyzed with the paired Student t-test. Data from in vivo thrombus formation were compared by Dunnett test. The interaction between IgG and β2GPI bound to CL was analyzed with the Kruskall-Wallis with Dunn post-hoc test. The interaction between IgG and soluble β2GPI was expressed as median and interquartile range and analyzed with the two-way repeated measure ANOVA with Sidak post-hoc test. Probabilities of <0.05 were considered statistically significant.

Results Antibody to phospholipid-binding protein profile of the serum samples Anti-β2GPI IgG titers were comparable in the anti-D4/5and anti-D1-positive samples [1.04±0.26 Optical Density (OD) and 1.46±0.48 OD, mean+SD, respectively]. The isolated anti-D4/5-positive samples displayed anti-D4/5 levels of 50.67±9.86 arbitrary units (AU) (mean±SD) while they were negative for aCL (<10 GPL) and LA. The isolated anti-D1-positive samples showed anti-D1 levels of 75.36±17.15 AU (mean±SD), high titers of IgG aCL (124.4±46.9 GPL, mean±SD), and displayed LA activity. Control samples were negative in all the assays. The purified IgG fractions maintained the antigen specificity shown in the whole serum. Clinical and serological data of all the subjects/patients included in the study are reported in Online Supplementary Table S1.

Fine epitope-specificity of antibodies to domains 4/5 The IgG against D4/5 used in this study were selected for their ability to react with the combined domains obtained from INOVA Diagnostics, but it was unclear whether they recognized one or the other domain or both. To clarify this point, we assessed the reactivity of serum IgG towards recombinant D4 and D5. The amino acid sequences of the two domains are reported in Online Supplementary Figure S1. The results presented in Figure 1 clearly show that all the anti-D4/5 reacted with D5 and did not recognize D4. The difference in the reactivity of haematologica | 2019; 104(4)

the various sera IgG towards D4/5 is essentially similar to that observed in their reaction with D5.

Antibodies to domain 5 fail to cause thrombus formation in vivo To evaluate the pro-coagulant activity of sera containing antibodies to different domains of β2GPI, two groups of serum IgG positive for either D1 or D5 domains were analyzed for their ability to induce thrombus formation followed in vivo by intravital microscopy. IgG from sera negative for antibodies to β2GPI served as a control group. All anti-D1-positive IgG induced blood clots that could be seen from 15 min after serum infusion (Figure 2). Their number progressively increased to reach the highest value after 1 h and was maintained thereafter for up to 90 min. Thrombus formation was associated with vascular occlusion that resulted in a marked decrease, and, in some vessels, in a complete blockage of blood flow. Conversely, the anti-D5-positive IgG did not exhibit pro-coagulant activity and failed to cause reduced blood flow. The latter results were not statistically different from those of anti-β2GPInegative blood donors at each time point. On the contrary, the data of anti-D1 IgG were statistically different from those of anti-β2GPI-negative samples at all times starting from 15 min of analysis (P<0.05).

Antibodies to domain 5 fail to interact with surface-bound β2glycoprotein I

Having observed an absence of intravascular coagulation in rats that had received anti-D5-positive IgG, we decided to investigate whether this was due to the inability of the antibodies to interact with endothelium-bound β2GPI. To this end, samples of ileal mesentery were analyzed for the presence of β2GPI, human IgG and C3. As expected from our previous findings,30 β2GPI was detected on the vessel endothelium of rats primed with LPS (Figure 3), while it was totally absent in unprimed animals (data not shown). A search for IgG and C3 revealed marked granular deposits of both proteins on endothelial cells of rats treated with anti-D1 IgG, while a milder linear staining for IgG and absence of C3 were observed in rats receiving anti-D5 IgG (Figure 3). The animals treated with antiβ2GPI-negative sera showed negligible staining for IgG and undetectable C3 (Figure 3). Since several molecules 821

P. Durigutto et al. other than β2GPI are expressed on the endothelial cell surface and represent potential targets for human IgG, we set out to determine whether the fluorescence was due to the IgG specifically directed against β2GPI. To do this, we set up a β2GPI-dependent CL assay in which the β2GPI supplementation was carried out by adding human purified β2GPI at increasing concentrations instead of fetal calf serum. The system allowed us to test the IgG reactivity with β2GPI added at different concentrations to the CLplates. The anti-D1 IgG reacted with the β2GPI molecule most likely by recognizing the D1 epitope exposed on the β2GPI molecule following its binding to CL (Figure 4). The IgG level detected in the assay varied in different patients and was related to the concentration of β2GPI used to coat CL. In contrast, anti-D5 IgG failed to interact with CLbound β2GPI even at the highest concentration of β2GPI, suggesting that D5 domains were not accessible to the antibodies under these experimental conditions. Like the anti-D5 antibodies, in the assay, the IgG from control sera were negative.

Antibodies to domain 5 interact with soluble β2glycoprotein I

Electron microscopy studies have revealed that β2GPI adopts a circular form in plasma and that this is maintained by the interaction of D1 with D5.34 This special conformation prevents the access of autoantibodies to hidden epitopes on D119 and predicts the presence of cryp-

tic epitopes on D5, though this has not been formally proven.35 We first decided to examine the in vivo interaction of the antibodies with circulating β2GPI and the effect of this interaction on β2GPI bound to vascular endothelium. To this purpose, the in vivo model was slightly modified administering IgG intraperitoneally followed 15 h later by LPS given by the same route; this approach would allow sufficient time for the antibodies to react with the target antigen prior to the binding of β2GPI to vascular endothelium promoted by LPS. The IgG from two sera with relatively high levels of antibodies to D1 and D5, respectively, and from an anti-β2GPI-negative serum were tested and the amount of vascular deposits of β2GPI and IgG was evaluated. As expected, the rat treated with antiD1 developed endovascular thrombi associated with deposition of IgG, both of which were undetectable in animals that received anti-D5-positive or anti-β2GPI-negative IgG (Figure 5). Analysis of the ileal mesentery showed that β2GPI was present on the vascular endothelium of the animals that received the three IgG fractions with no clear difference in the staining intensity observed in the rats treated with anti-D5 and anti-D1 IgG (Figure 5). Since the in vivo data did not provide convincing evidence of the ability of anti-D5 to prevent binding of circulating β2GPI to vascular endothelium, we decided to further investigate this issue using an in vitro inhibition assay. IgG purified from anti-D5-positive, anti-D1-positive or anti-β2GPI-negative sera were incubated with increasing

Figure 2. Anti-domain (D) 5 antibodies fail to induce thrombi in rats. Thrombus formation and vascular occlusion visualized by intravital microscopy in the ileal mesentery of rats that received an intraperitoneal injection of lipopolysaccharide (LPS) (2.5 mg/kg body weight) followed by the injection into the carotid artery of antibodies (10 mg/rat) directed against domain 5 (D5), domain 1 (D1), or anti-β2glycoprotein I (β2GPI)-negative (NHS). The number of thrombi (A) and vessel occlusions (B) were evaluated at various time intervals on 3 rats per each serum. The results are expressed as a ratio between the number of thrombi and the number of microvessels examined and as a percentage of occluded microvessels. Data are reported as mean±Standard Deviation. (C) Sections of the ileal mesentery showing endovascular thrombi in anti-D1-treated rat and undetectable in the vessels of animals receiving anti-D4/5-positive or anti-β2GPI-negative sera. Original magnification 100x. Scale bar 50 mm.

822

haematologica | 2019; 104(4)

Anti-β2GPI-D5 are not thrombogenic in animals

Figure 3. Deposition of β2glycoprotein I (β2GPI), human IgG and C3 on mesenteric vessels of rats treated with antibodies to domain 5 (D5) or domain 1 (D1) of β2GPI. The animals were treated with lipopolysaccharide (LPS) followed by the injection of antibodies directed against domain 5 (D5), domain 1 (D1), or negative for anti-β2GPI (NHS). Mesenteric tissue samples after 90 minutes analyzed for vascular deposition of β2GPI, human IgG and C3 by immunofluorescence. Original magnification 200x. Scale bar 50 mm.

Figure 4. Anti-domain 5 (D5) antibodies fail to interact with β2glycoprotein I (β2GPI) bound to cardiolipin (CL). Reactivity of anti-D5 (aD5) ( ), anti-domain 1 (aD1) ) antibodies (50 mg/mL) against different concentrations of β2GPI bound to cardiolipin. Binding of IgG to: (A) CL alone, (B) 1 ( ) or anti-β2GPI negative (NHS) ( mg/mL CL-bound β2GPI, (C) 5 mg/mL CL-bound β2GPI, (D) 75 mg/mL CL-bound β2GPI. Optical Density (OD) values are expressed as median and interquartile range, and presented as box plots. *P<0.05.

haematologica | 2019; 104(4)

823

P. Durigutto et al.

Figure 5. Deposition of β2glycoprotein I (β2GPI) and IgG on mesenteric vessels of rats treated with patients’ and controls’ serum IgG prior to lipopolysaccharide (LPS) challenge. The animals were treated with antibodies directed against domain 5 (D5), domain 1 (D1), or anti-β2GPI-negative (NHS) (10 mg/rat) before LPS administration (2.5 mg/kg body weight). Mesenteric tissue samples were analyzed for vascular deposition of β2GPI (left) and human IgG (center). Original magnification for immunofluorescence analysis 200x. Scale bar 50 mm. Thrombus formation in mesenteric vessels was monitored by intravital microscopy for 90 minutes and mesenteric tissue was collected at the end of the experiment. Thrombi formed in the vessels are indicated with arrows (right). Original magnification 100x. Scale bar 50 mm.

concentrations of soluble β2GPI and the residual IgG interacting with β2GPI directly bound to the plate wells were measured. The amount of IgG anti-D5 free to bind to solid-phase β2GPI after incubation with the soluble molecule decreased compared to that of the IgG incubated with BSA, particularly at a higher concentration of soluble β2GPI (Figure 6). In contrast, the level of IgG anti-D1 bound to solid-phase β2GPI following incubation with soluble β2GPI was slightly lower, but not significantly different from that of the IgG incubated with BSA.

Discussion Antiphospholipid syndrome is now recognized as an antibody-dependent and complement-mediated syndrome and antibodies to β2GPI have been identified as important players in thrombus formation in APS patients.10 Efforts are being made to determine the clinical relevance of antibodies to D1 and D4/5 domains of the molecule detected in these patients. Clinical studies have suggested that antibodies to D4/5, unlike those directed against D1, do not represent a risk factor for thrombosis and pregnancy complications.7,9,14 The in vivo data presented here focused on the thrombotic aspect of the syndrome and support the clinical observation that the anti-D4/5 antibodies are pathologically irrelevant. The animal model used in this and in previous studies proved to be an invaluable tool to investigate the ability of the anti-β2GPI antibodies to induce blood clots in rats primed with LPS that provides the first hit, followed by the infusion of the antibodies acting as a second hit.10 As expected, all anti-D1 IgG promoted thrombus formation 824

and vascular occlusion, confirming the pathogenicity of these antibodies suggested by clinical observations. It is possible that LA detected in the plasma of these patients may have also contributed to anti-β2GPI-induced blood clots. However, although β2GPI antibody-dependent LA has been shown to correlate with the increased risk of thrombosis,13,14,36 evidence supporting the in vivo prothrombotic activity of LA independently of anti-β2GPI antibody has not yet been provided. Instead, there is good evidence that the antibodies recognizing the D1 domain of β2GPI are directly involved in thrombus formation and vessel occlusion. We have previously shown that a human monoclonal antibody that recognizes D1 induces blood clots and that a CH2-deleted non-complement fixing variant molecule competes with anti-β2GPI antibodies from APS patients and prevents their pro-coagulant activity.27 A similar inhibitory effect was obtained using recombinant D1 to control the thrombus enhancement activity of aPL in mice.37 The in vivo experiments showed that none of the antiD5 IgG exhibited a prothrombotic activity supporting the observations made in clinical studies that these antibodies are pathologically irrelevant.7,14 A possible explanation for this finding is the inability of these antibodies to interact with cell-bound β2GPI. In line with this hypothesis, we showed that anti-D5-positive IgG fractions were unable to react with β2GPI bound to CL-coated plates in vitro because of the shielding of D5 in the β2GPI molecule bound to the CL-coated plate. However, in rats treated with LPS (used to promote binding of β2GPI) and anti-D5 IgG, the mild staining for IgG observed on the endothelium of mesenteric vessels did not allow any definite conclusions to be haematologica | 2019; 104(4)

Anti-β2GPI-D5 are not thrombogenic in animals

Figure 6. Anti-domain 5 (D5) antibodies interact with β2glycoprotein I (β2GPI) in fluid phase. Reactivity of anti-D5 (aD5), anti-D1 (aD1), or anti-β2GPI-negative (NHS) antibodies (50 μg/mL) against purified β2GPI directly coated on ELISA plates, measured after their incubation with (A) 50 mg/mL, (B) 100 mg/mL, and (C) 200 mg/mL ) or BSA ( ) in fluid phase. Optical Density (OD) values are expressed as median and interquartile range, and presented as box plots. of purified β2GPI ( *P<0.05; **P<0.01.

drawn on this issue. It must be emphasized, however, that the staining intensity varied among different sera and was not related to the level of antibodies. The linear deposition of IgG on the mesenteric endothelium from rats treated with anti-D5-positive IgG suggests their interaction with antigens constitutively expressed on endothelial cells. This distribution pattern differs from the irregular staining for IgG seen with the anti-D1-positive IgG most likely explained by their reaction with a plasma-derived molecule, such as β2GPI, bound to the endothelial cell surface. The different distribution of anti-D1 and anti-D5 IgG resembles the well-known difference in the granular and linear distribution patterns of IgG observed in the kidney of patients with SLE and Goodpasture syndrome, respectively. The linear pattern of IgG in Goodpasture is the result of the interaction of the antibodies with their target antigen constitutively expressed on the glomerular basement membrane. In contrast, the granular distribution of IgG in SLE is caused by irregular deposition of circulating immune complexes.38,39 The finding that C3 deposition was undetectable on the vascular endothelium of rats treated with anti-D5 IgG is consistent with the failure of these antibodies to induce thrombus formation. We and others have provided convincing evidence that complement activation is critically involved in the coagulation process induced by anti-β2GPI IgG and in this study by antibodies to the D1 domain.26,27,40-43 The anti-D4/D5 antibodies present in the sera analyzed in this study selectively recognized the recombinant D5 domain and are likely to inhibit deposition of β2GPI on the endothelium by shielding its binding site for the anionic phospholipid on endothelial cells.44 Our attempt to document ex vivo reduced binding of circulating β2GPI to vascular endothelium of the anti-D5-treated rats was unsatisfactory; this was most likely due to a much higher level of serum β2GPI compared to that of injected antibodies in vivo. The in vitro data obtained under more controlled conditions of IgG and β2GPI concentrations showed a fluid phase interaction between anti-D5 IgG and soluble β2GPI, resulting in a significantly reduced reactivity of these antibodies against surface-bound β2GPI (when the molecule was bound to a plate). The finding that anti-D5 IgG have no pro-coagulant effect in our in vivo model has important clinical implications suggesting that individuals with isolated presence of haematologica | 2019; 104(4)

these antibodies should not be considered to be at risk of thrombosis. It should be pointed out, however, that antiD1 and anti-D5 IgG often co-exist in a large proportion of APS patients, and that they are likely to be susceptible to anti-D1-dependent thrombus formation. In view of the ability of the anti-D5 IgG to interact with soluble β2GPI, thus preventing its binding to the target cells, it is tempting to speculate that the anti-D5 IgG may antagonize the pro-coagulant activity of anti-D1 antibodies, according to antibody levels. In accordance with this, we recently published data indicating that patients positive for anti-D1 and anti-D4/5 antibodies have a lower risk of thrombosis if the levels of anti-D4/5 are higher than those of anti-D1 antibodies.7,9 Overall, our experimental findings fit with the clinical observation and offer new tools for stratifying patients into different risk categories. This would help in better preventing recurrences of the clinical manifestations and avoiding overtreatment, thus ultimately improving the patients’ quality of life and sparing them treatment side-effects. In conclusion, the data presented in this work indicate that, unlike the anti-D1 positive sera, those containing antibodies against D5 are unable to induce clot formation and vascular occlusion. The failure of the anti-D5 antibodies to promote coagulation is due mainly to their inability to interact with the target epitopes hidden on the surfacebound molecule, and possibly to the recognition of native β2GPI in plasma that may, to some extent, potentially prevent its binding to the surface of activated endothelial cells. The detection of anti-D5 antibodies in patients with a doubtful APS clinical profile and a single positivity for anti-β2GPI in the absence of a positive aCL assay may offer a valuable tool for ruling out a definite APS diagnosis and for identifying subjects at lower risk of clinical manifestations. Acknowledgments The authors would like to thank Linda Vuch, Luca De Maso and Paola A. Lonati for their valuable technical contribution; Michael Mahler, Gary Norman and Filippo Sarra (INOVA Diagnostics and Werfen Italia) for their support. Funding This work was partially supported by Istituto Auxologico Italiano, Ricerca Corrente 2016 (PLM). 825

P. Durigutto et al.

References 15. 1. Miyakis S, Lockshin MD, Atsumi T, et al. International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost. 2006;4(2):295-306. 2. Gerosa M, Meroni PL, Erkan D. Recognition and management of antiphospholipid syndrome. Curr Opin Rheumatol. 2016;28(1): 51-59. 3. Chighizola CB, Andreoli L, de Jesus GR, et al. The association between antiphospholipid antibodies and pregnancy morbidity, stroke, myocardial infarction, and deep vein thrombosis: a critical review of the literature. Lupus. 2015;24(9):980-984. 4. Rodriguez-Pinto I, Moitinho M, Santacreu I, et al. Catastrophic antiphospholipid syndrome (CAPS): Descriptive analysis of 500 patients from the International CAPS Registry. Autoimmun Rev. 2016;15(12): 1120-1124. 5. Pengo V, Bison E, Denas G, Jose SP, Zoppellaro G, Banzato A. Laboratory Diagnostics of Antiphospholipid Syndrome. Semin Thromb Hemost. 2018; 44(5):439-444. 6. Andreoli L, Nalli C, Motta M, et al. Antibeta(2)-glycoprotein I IgG antibodies from 1-year-old healthy children born to mothers with systemic autoimmune diseases preferentially target domain 4/5: might it be the reason for their 'innocent' profile? Ann Rheum Dis. 2011;70(2):380-383. 7. Andreoli L, Chighizola CB, Nalli C, et al. Clinical characterization of antiphospholipid syndrome by detection of IgG antibodies against beta2 -glycoprotein I domain 1 and domain 4/5: ratio of anti-domain 1 to anti-domain 4/5 as a useful new biomarker for antiphospholipid syndrome. Arthritis Rheumatol. 2015; 67(8):2196-2204. 8. Bertolaccini ML, Sanna G. The Clinical Relevance of Noncriteria Antiphospholipid Antibodies. Semin Thromb Hemost. 2018; 44(5):453-457. 9. Chighizola CB, Pregnolato F, Andreoli L, et al. Beyond thrombosis: Anti-beta2GPI domain 1 antibodies identify late pregnancy morbidity in anti-phospholipid syndrome. J Autoimmun. 2018;90:76-83. 10. Meroni PL, Borghi MO, Raschi E, Tedesco F. Pathogenesis of antiphospholipid syndrome: understanding the antibodies. Nat Rev Rheumatol. 2011;7(6):330-339. 11. de Laat B, Mertens K, de Groot PG. Mechanisms of disease: antiphospholipid antibodies-from clinical association to pathologic mechanism. Nat Clin Pract Rheumatol. 2008;4(4):192-199. 12. Iverson GM, Victoria EJ, Marquis DM. Anti-beta2 glycoprotein I (beta2GPI) autoantibodies recognize an epitope on the first domain of beta2GPI. Proc Natl Acad Sci U S A. 1998;95(26):15542-15546. 13. de Laat B, Derksen RH, Urbanus RT, de Groot PG. IgG antibodies that recognize epitope Gly40-Arg43 in domain I of beta 2-glycoprotein I cause LAC, and their presence correlates strongly with thrombosis. Blood. 2005;105(4):1540-1545. 14. Pengo V, Ruffatti A, Tonello M, et al. Antibodies to Domain 4/5 (Dm4/5) of beta2-Glycoprotein 1 (beta2GP1) in differ-

826

16.

17. 18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

ent antiphospholipid (aPL) antibody profiles. Thromb Res. 2015;136(1):161-163. de Groot PG, Urbanus RT. The significance of autoantibodies against beta2-glycoprotein I. Blood. 2012;120(2):266-274. Ioannou Y, Zhang JY, Passam FH, et al. Naturally occurring free thiols within beta 2glycoprotein I in vivo: nitrosylation, redox modification by endothelial cells, and regulation of oxidative stress-induced cell injury. Blood. 2010;116(11):1961-1970. Passam FH, Rahgozar S, Qi M, et al. Beta 2 glycoprotein I is a substrate of thiol oxidoreductases. Blood. 2010;116(11):1995-1997. Ioannou Y. The Michael Mason Prize: Pathogenic antiphospholipid antibodies, stressed out antigens and the deployment of decoys. Rheumatology (Oxford). 2012; 51(1):32-36. de Laat B, Derksen RH, van Lummel M, Pennings MT, de Groot PG. Pathogenic antibeta2-glycoprotein I antibodies recognize domain I of beta2-glycoprotein I only after a conformational change. Blood. 2006; 107(5):1916-1924. de Laat B, Pengo V, Pabinger I, et al. The association between circulating antibodies against domain I of beta2-glycoprotein I and thrombosis: an international multicenter study. J Thromb Haemost. 2009; 7(11):1767-1773. Mahler M, Norman GL, Meroni PL, Khamashta M. Autoantibodies to domain 1 of beta 2 glycoprotein 1: a promising candidate biomarker for risk management in antiphospholipid syndrome. Autoimmun Rev. 2012;12(2):313-317. Pengo V, Ruffatti A, Tonello M, et al. Antiphospholipid syndrome: antibodies to Domain 1 of beta2-glycoprotein 1 correctly classify patients at risk. J Thromb Haemost. 2015;13(5):782-787. Chaturvedi S, McCrae KR. Clinical Risk Assessment in the Antiphospholipid Syndrome: Current Landscape and Emerging Biomarkers. Curr Rheumatol Rep. 2017;19(7):43. Ioannou Y, Pericleous C, Giles I, Latchman DS, Isenberg DA, Rahman A. Binding of antiphospholipid antibodies to discontinuous epitopes on domain I of human beta(2)glycoprotein I: mutation studies including residues R39 to R43. Arthritis Rheum. 2007;56(1):280-290. Roggenbuck D, Borghi MO, Somma V, et al. Antiphospholipid antibodies detected by line immunoassay differentiate among patients with antiphospholipid syndrome, with infections and asymptomatic carriers. Arthritis Res Ther. 2016;18(1):111. Fischetti F, Durigutto P, Pellis V, et al. Thrombus formation induced by antibodies to beta2-glycoprotein I is complement dependent and requires a priming factor. Blood. 2005;106(7):2340-2346. Agostinis C, Durigutto P, Sblattero D, et al. A non-complement-fixing antibody to beta2 glycoprotein I as a novel therapy for antiphospholipid syndrome. Blood. 2014; 123(22):3478-3487. Pierangeli SS, Vega-Ostertag ME, Raschi E, et al. Toll-like receptor and antiphospholipid mediated thrombosis: in vivo studies. Ann Rheum Dis. 2007;66(10):1327-1333. Pengo V. ISTH guidelines on lupus anticoagulant testing. Thromb Res. 2012;130 Suppl

1:S76-77. 30. Agostinis C, Biffi S, Garrovo C, et al. In vivo distribution of beta2 glycoprotein I under various pathophysiologic conditions. Blood. 2011;118(15):4231-4238. 31. van Os GM, Meijers JC, Agar C, et al. Induction of anti-beta2 -glycoprotein I autoantibodies in mice by protein H of Streptococcus pyogenes. J Thromb Haemost. 2011;9(12):2447-2456. 32. Tomaic V, Gardiol D, Massimi P, Ozbun M, Myers M, Banks L. Human and primate tumour viruses use PDZ binding as an evolutionarily conserved mechanism of targeting cell polarity regulators. Oncogene. 2009;28(1):1-8. 33. Steinkasserer A, Estaller C, Weiss EH, Sim RB, Day AJ. Complete nucleotide and deduced amino acid sequence of human beta 2-glycoprotein I. Biochem J. 1991; 277( Pt 2):387-391. 34. Agar C, van Os GM, Morgelin M, et al. Beta2-glycoprotein I can exist in 2 conformations: implications for our understanding of the antiphospholipid syndrome. Blood. 2010;116(8):1336-1343. 35. de Groot PG, Meijers JC. beta(2) Glycoprotein I: evolution, structure and function. J Thromb Haemost. 2011; 9(7):1275-1284. 36. Pengo V, Testa S, Martinelli I, et al. Incidence of a first thromboembolic event in carriers of isolated lupus anticoagulant. Thromb Res. 2015;135(1):46-49. 37. Ioannou Y, Romay-Penabad Z, Pericleous C, et al. In vivo inhibition of antiphospholipid antibody-induced pathogenicity utilizing the antigenic target peptide domain I of beta2-glycoprotein I: proof of concept. J Thromb Haemost. 2009;7(5):833-842. 38. Agnello V, Koffler D, Kunkel HG. Immune complex systems in the nephritis of systemic lupus erythematosus. Kidney Int. 1973;3(2):90-99. 39. McPhaul JJ Jr, Mullins JD. Glomerulonephritis mediated by antibody to glomerular basement membrane. Immunological, clinical, and histopathological characteristics. J Clin Invest. 1976; 57(2):351-361. 40. Salmon JE, Girardi G. Theodore E. Woodward Award: antiphospholipid syndrome revisited: a disorder initiated by inflammation. Trans Am Clin Climatol Assoc. 2007;118:99-114. 41. Erkan D, Salmon JE. The Role of Complement Inhibition in Thrombotic Angiopathies and Antiphospholipid Syndrome. Turk J Haematol. 2016;33(1):1-7. 42. Oku K, Nakamura H, Kono M, et al. Complement and thrombosis in the antiphospholipid syndrome. Autoimmun Rev. 2016;15(10):1001-1004. 43. Meroni PL, Macor P, Durigutto P, et al. Complement activation in antiphospholipid syndrome and its inhibition to prevent rethrombosis after arterial surgery. Blood. 2016;127(3):365-367. 44. Del Papa N, Sheng YH, Raschi E, et al. Human beta 2-glycoprotein I binds to endothelial cells through a cluster of lysine residues that are critical for anionic phospholipid binding and offers epitopes for antibeta 2-glycoprotein I antibodies. J Immunol. 1998;160(11):5572-5578.

haematologica | 2019; 104(4)

ARTICLE

Stem Cell Transplantation

Asymmetric dimethylarginine serum levels are associated with early mortality after allogeneic stem cell transplantation

Ferrata Storti Foundation

Aleksandar Radujkovic,1 Hao Dai,2 Lambros Kordelas,3 Dietrich Beelen,3 Sivaramakrishna P. Rachakonda,1,2 Carsten MĂźller-Tidow,1 Rajiv Kumar,2 Peter Dreger1 and Thomas Luft1

Department of Internal Medicine V, University Hospital Heidelberg; 2Department of Epidemiology, German Cancer Research Centre (DKFZ), Heidelberg and 3Department of Bone Marrow Transplantation, University Hospital Essen, Germany 1

Haematologica 2019 Volume 104(4):827-834

ABSTRACT

ncreasing evidence suggests that endothelial cell distress is associated with mortality after allogeneic stem cell transplantation and acute graftversus-host disease. Asymmetric dimethylarginine is an endogenous nitric oxide synthase inhibitor that induces endothelial cell dysfunction. We analyzed the impact of pre-transplant serum levels of asymmetric dimethylarginine on outcome after allogeneic stem cell transplantation. Since acute graft-versus-host disease and its treatment are major contributors to posttransplant mortality, the effect of asymmetric dimethylarginine on outcome measures was also assessed after onset of acute graft-versus-host disease. A total of 938 patients allografted at two centers between 2002 and 2013 were included in the retrospective study. In multivariable models, higher pretransplant asymmetric dimethylarginine levels were significantly associated with an increased risk of non-relapse mortality (hazard ratio 1.43 per 1-log2 increase, P=0.005) but not with relapse (hazard ratio 1.21, P=0.109) within the first year after transplantation. This translated into worse overall survival (hazard ratio 1.45, P<0.0001) and shorter progression-free survival (hazard ratio 1.30, P=0.002) in the first year after transplantation. Higher pre-transplant asymmetric dimethylarginine levels were also associated with shorter overall survival (hazard ratio 1.46, P=0.001) and progressionfree survival (hazard ratio 1.32, P=0.010) and higher non-relapse mortality (hazard ratio 1.36, P=0.042) within 1 year after the onset of acute graft-versus-host disease. Taken together, our data indicate an association between pre-transplant asymmetric dimethylarginine status and early non-relapse mortality in allografted patients, both overall and after the onset of acute graft-versus-host disease. These findings underline the relevance of endothelial dysfunction for transplant complications.

Introduction Non-relapse mortality (NRM) represents a challenge for successful allogeneic stem cell transplantation (SCT). NRM is predominantly driven by graft-versus-host disease (GvHD) as a major complication of allogeneic SCT. There is a body of evidence showing that mortality after allogeneic SCT and after acute GvHD is associated with endothelial cell distress. In previous studies, endothelial markers, such as angiopoietin-2 and nitrates, and, on the genomic level, single nucleotide polymorphisms (SNP) in the recipientsâ&#x20AC;&#x2122; thrombomodulin and CD40 ligand genes, were shown to predict overall mortality and GvHD-related mortality already prior to transplantation.1-5 Serum asymmetric dimethylarginine (ADMA) is an endogenous inhibitor of nitric oxide (NO) synthase which competes with arginine and influences the bioavailability of NO.6,7 ADMA may cause endothelial dysfunction and, in patients with a compromised vascular system, elevated ADMA levels can predict cardiovascular events and mortality.8-10 Recently, two large meta-analyses across different haematologica | 2019; 104(4)

Correspondence: THOMAS LUFT thomas.luft@med.uni-heidelberg.de Received: July 19, 2018. Accepted: November 23, 2018. Pre-published: December 4, 2018. doi:10.3324/haematol.2018.202267 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/827 ÂŠ2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

827

A. Radujkovic et al.

patient and general populations demonstrated that circulating ADMA is also an independent risk factor for allcause mortality.11,12 In view of these findings, the present retrospective study investigated the impact of pre-transplant serum ADMA levels on major outcome measures in the setting of allogeneic SCT.

estimated with 95% confidence intervals (95% CI). P values <0.05 were considered statistically significant. Details regarding prophylaxis, diagnosis, and treatment of GvHD, the assessment of serum levels of ADMA and endothelium-related serum factors, SNP analyses in the inducible nitric oxide synthase gene (INOS) and additional statistical methods are described in the Online Supplement.

Methods Results Patients

Adult patients (age ≥18 years) who were allografted between 2002 and 2013 at two centers (Heidelberg, n=518, and Essen, n=420) and who had serum samples available for ADMA measurement (collected directly before the start of conditioning chemotherapy prior to allogeneic SCT) were included in this study. Written informed consent to sample and data collection according to the Declaration of Helsinki was obtained from all patients, and the local ethics committees approved the study.

Assessment of serum levels of asymmetric dimethylarginine and endothelium-related serum factors Serum levels of ADMA and endothelium-related serum factors were assessed by enzyme-linked immunosorbent assay retrospectively in the last serum sample taken before the start of the conditioning treatment.

Statistical analysis Overall survival (OS), progression-free survival (PFS), time to relapse, and NRM (death in the absence of prior relapse) were calculated from the date of allogeneic SCT to the appropriate endpoint. PFS was defined as the time from transplantation to relapse of the underlying disease or death. Since acute GvHD and its treatment are major contributors to post-transplant mortality, OS, PFS and NRM were also assessed after acute GvHD (i.e. from the date of onset of acute GvHD). NRM and recurrence of the underlying malignancy were considered as competing events. For the incidence of acute GvHD, the competing events were relapse and death in remission without acute GvHD. The Kaplan-Meier method was used to estimate distributions of survival times. Follow-up times were calculated by the reverse Kaplan-Meier estimate.13 Cumulative incidence functions were implemented to account for the competing risks,. Due to the relatively wide range of supposedly normal ADMA serum levels, pre-transplant ADMA was analyzed as a continuous variable. Since ADMA serum levels did not follow a normal distribution and the data were log2-transformed, the prognostic effect of pre-transplant ADMA on OS, PFS, relapse, and NRM as a continuous variable was evaluated using Cox regression models. Cox proportional hazards regression modeling was used for OS and PFS (overall and after the onset of acute GvHD). Relapse and NRM (overall and after the onset of acute GvHD) were analyzed by cause-specific Cox models. To illustrate the continuous effect of pre-transplant ADMA on the different endpoints, patients were grouped according to ADMA quartiles, and the observation period was restricted to the first year after transplantation or acute GvHD. To assess the dependency of endothelium-related serum markers [free interleukin (IL)-18, sCD141, and nitrates] on ADMA levels, the Jonckheere trend test was applied comparing serum marker data in four increasing intervals of pre-transplant ADMA levels determined by the quartiles of the ADMA distribution. The Jonckheere trend test is a rank-based nonparametric test which is used to test for an ordered difference in medians.14 All statistical tests were two-sided. Hazard ratios (HR) were 828

Patients’, disease and treatment characteristics and graft-versus-host disease incidence A total of 938 patients met the eligibility criteria for this study. The patients’, disease and transplant characteristics are given in Table 1. A total of 493 patients died after allogeneic SCT, of whom 309 died within the first year after transplantation. The cumulative incidences of relapse and NRM at 1 year after transplantation were 23.0% (95% CI: 20.3%-25.7%) and 18.2% (95% CI: 15.8%-20.7%), respectively. The cumulative incidences of acute GvHD Table 1. Patients’, disease and transplant characteristics.

Parameter

N=938

Age [in years] at allogeneic SCT, median, (IQR) 54 (46-62) Recipient sex, n (%) Female 424 (45) Male 514 (55) Disease, n (%) Myeloida 665 (71) Lymphoidb 273 (29) Disease stage at allogeneic SCTc, n (%) Early 306 (33) Late/intermediate 632 (67) Conditioningd regimen, n (%) Reduced intensity 804 (86) Myeloablative 134 (14) Stem cell source, n (%) Peripheral blood 861 (92) Bone marrow 77 (8) Donor type, n (%) Unrelated 700 (75) Related 238 (25) Donor-recipient HLA-matching, n (%) Mismatched 242 (26) Matched 696 (74) Antithymocyte globulin treatment, n (%) No 225 (24) Yes 713 (76) Donor sex, n (%) Female 311 (33) Male 627 (67) e Pre-transplant ADMA , mM, median (IQR) 0.73 (0.59-0.97) Pre-transplant nitrate levelsf, mM, median (IQR) 22.4 (15.0-35.7) Pre-transplant free IL-18g, pg/mL, median (IQR) 442 (316-683) Pre-transplant soluble CD141h, pg/mL, median (IQR) 4179 (3386-5376) ADMA: asymmetric dimethylarginine; SCT: stem cell transplantation; HLA: human leukocyte antigen; IL-18: interleukin 18; IQR, interquartile range; aMyeloid: acute myeloid leukemia, myelodysplastic and myeloproliferative syndromes. bLymphoid: acute lymphoblastic leukemia, chronic lymphocytic leukemia, lymphoma and multiple myeloma. cAccording to Gratwohl et al.41 dAccording to Bacigalupo et al.42 eBased on 938 serum samples. fBased on 375 serum samples. gBased on 746 serum samples. h Based on 776 serum samples.

haematologica | 2019; 104(4)

ADMA and post-transplant mortality

(any grade) and grade 3-4 acute GvHD on day +100 after transplantation were 54.0% (95% CI: 50.8%-57.3%) and 13.7% (95% CI: 11.1%-16.3%), respectively. The estimated median follow-up time was 64.9 months (95% CI: 62.0-67.9; range, 0.0-167.7).

Pre-transplant serum asymmetric dimethylarginine levels and their impact on outcome measures after allogeneic stem cell transplantation Across all 938 patients, the median pre-transplant ADMA level was 0.73 mM [interquartile range (IQR) 0.590.97]. Initial analyses indicated that the proportional hazards assumption was violated for pre-transplant ADMA, i.e., the effect of pre-transplant ADMA on mortality risks was not constant over time and was limited to the first year after transplantation or after acute GvHD (additional statistical methods are given in the Online Supplement). In multivariable analyses adjusted for known confounders, increasing pre-transplant ADMA levels, as a continuous variable, were significantly associated with worse OS (HR 1.45 per 1- log2 increase, 95% CI: 1.21-1.74, P<0.0001) and shorter PFS (HR 1.30, 95% CI: 1.10-1.54, P=0.002) within the first year after allogeneic SCT, but not thereafter. This was based on an increased risk of NRM (HR 1.43, 95% CI: 1.12-1.83, P=0.005) rather than relapse (HR 1.21, 95% CI: 0.96-1.52, P=0.109) in the year following transplantation. The results of the multivariable analyses with the endpoints OS, PFS, NRM and post-transplant relapse are summarized in Table 2. Increasing pre-transplant ADMA levels were not signif-

icantly associated with the incidence of acute GvHD (univariable: any grade, HR 1.04, 95% CI: 0.90-1.20, P=0.583; univariable: grade 3-4, HR 1.27, 95% CI: 0.93-1.73, P=0.129). However, we observed an association with higher risk of NRM after GvHD onset (HR 1.36, 95% CI: 1.01-1.83, P=0.042). This translated into worse OS (HR 1.46, 95% CI: 1.17-1.83, P=0.001) and shorter PFS (HR 1.32, 95% CI: 1.07-1.63, P=0.010) within the first year after the onset of acute GvHD. The results of the multivariable analyses with the endpoints OS, PFS, and NRM after acute GvHD are given in Table 3. The association of increasing pre-transplant ADMA levels with shorter OS and PFS and higher risk of NRM in the multivariable models, both within the first year aftertransplantation and within the first year after the development of acute GvHD, could also be observed when patients were stratified according to the transplant center applying a meta-analytic approach (Figure 1). In order to illustrate the continuous effect of pre-transplant ADMA levels on all aforementioned endpoints, patients were grouped according to four increasing ranges of pre-transplant ADMA levels determined by the quartiles of the ADMA distribution. The corresponding plots showing OS, PFS, NRM and relapse in the first year after transplantation, and OS, PFS, and NRM in the first year after acute GvHD are given in Figure 2. In addition, the prediction performance of the univariable ADMA models was assessed with regards to OS and PFS (overall and after acute GvHD). As compared to the respective Kaplan-Meier survival estimates, the ADMA

Table 2. Multivariable analysis of predictors of overall survival, progression-free survival, non-relapse mortality and relapse (n=938, complete case analysis).

OS HR (95% CI)

PFS P

HR (95% CI)

NRM* P

HR (95% CI)

Relapse* HR (95% CI) P

Covariate ADMA (per 1-log2 increase): 1.45 (1.21-1.74) 1st year ADMA (per 1-log2 increase): 1.02 (0.74-1.41) 2-3 years ADMA (per 1-log2 increase): 1.03 (0.69-1.55) after 3 years Age at transplantation (per year)1.01 (1.00-1.02) Donor sex Female Ref Male 0.82 (0.68-1.00) Recipient sex Female Ref Male 1.10 (0.92-1.32) Donor-recipient HLA matching Mismatched Ref Matched 0.67 (0.55-0.81) Disease stage† Early Ref Late/intermediate 1.18 (0.97-1.43) Conditioning‡ Myeloablative Ref Reduced intensity 0.96 (0.72-1.28)

<0.0001

1.30 (1.10-1.54)

0.002

1.43 (1.12-1.83)

0.005

1.21 (0.96-1.52)

0.109

0.888

1.00 (0.71-1.42)

0.980

1.18 (0.69-2.04)

0.543

0.92 (0.59-1.42)

0.694

0.878

0.93 (0.58-1.52)

0.781

0.87 (0.48-1.58)

0.645

1.07 (0.47-2.44)

0.877

0.008

1.01 (1.00-1.02)

0.034

1.03 (1.01-1.04)

<0.0001

1.00 (0.96-1.01)

0.395

0.046

Ref 0.83 (0.72-1.04)

0.113

Ref 0.76 (0.58-0.99)

0.043

Ref 0.96 (0.75-1.23)

0.737

0.298

Ref 1.06 (0.89-1.26)

0.507

Ref 1.14 (0.87-1.48)

0.346

Ref 0.99 (0.78-1.25)

0.937

<0.0001

Ref 0.74 (0.62-0.90)

0.002

Ref 0.55 (0.42-0.72)

<0.0001

Ref 0.98 (0.75-1.28)

0.895

0.090

Ref 1.19 (0.99-1.44)

0.060

Ref 0.99 (0.75-1.29)

0.912

Ref 1.43 (1.11-1.86)

0.006

0.785

Ref 1.00 (0.76-1.31)

0.972

Ref 0.87 (0.57-1.36)

0.559

Ref 1.06 (0.74-1.52)

0.749

Number of events: overall survival (OS), n=493; progression-free survival (PFS), n=540; non-relapse mortality (NRM), n=243; relapse, n=297. *Cause-specific hazards from a competing risks analysis for relapse and NRM. †According to Gratwohl et al.41 ‡According to Bacigalupo et al.42 ADMA: asymmetric dimethylarginine; CI: confidence interval; HLA: human leukocyte antigen; HR: hazard ratio.

haematologica | 2019; 104(4)

829

A. Radujkovic et al.

models showed a slightly improved prediction performance. The corresponding prediction error curves representing the Brier score over time are depicted in Online Supplementary Figure S1.

Association of asymmetric dimethylarginine levels with endothelium-related serum factors and high pre-transplant levels in the context of INOS polymorphisms Samples for measurement of serum concentrations of free IL-18, soluble thrombomodulin (sCD141), and nitrates were available for 746, 776 and 375 patients, respectively. The corresponding median pre-transplant serum levels are given in Table 1. Pre-transplant ADMA levels correlated with serum levels of nitrates, free IL-18, and sCD141 (Figure 3A-C). The concept of endothelial vulnerability involves a proinflammatory environment and ADMA is able to modify NO synthase and, in particular, inducible NO synthase (iNOS) activity.7,15 We analyzed whether polymorphisms in INOS modulated the observed effects of ADMA. Genotype data on INOS SNP were available for a total of 386 patients. In order to evaluate the effect of ADMA in the context of INOS polymorphisms, an optimal cut-off with regard to OS in the first year after transplantation was determined, yielding multiple cut-points (maxima) (Online Supplementary Figure S2). The maximum at 0.97 mM, which also represents the upper quartile of the ADMA distribution, was chosen to stratify patients into groups with high (≥0.97 mM) or low (<0.97 mM) ADMA levels in the context of INOS polymorphisms. There was no substantial influence of INOS SNP on the effect of ADMA on OS or NRM within the first year after allogeneic SCT (Online Supplementary Figures S3-5).

Discussion To the best of our knowledge, this is the first report of an association between ADMA levels measured prior to transplant and subsequent risk of mortality in allografted patients. ADMA is a well-characterized cardiovascular risk factor, and dysfunction of the endothelium appears to be a common finding in studies investigating the role of ADMA in (cardio-)vascular disease.10 Elevated ADMA levels have also been reported in autoimmune diseases, linking inflammation with autoimmunity-related vascular complications.16 In renal transplant recipients, associations of increased ADMA levels with acute rejection and mortality have been observed.17,18 Furthermore, high ADMA levels were associated with pro-inflammatory conditions and were shown to predict mortality risk in critically ill and septic patients.19,20 Finally, ADMA was strongly associated with all-cause mortality across different patient and general populations in recent meta-analyses.11,12 Previous studies by us and others2-5,21-23 led us to propose the hypothesis of “endothelial vulnerability” as an important contributor to the main complications of allogeneic SCT. This concept may explain why a proportion of patients with acute GvHD fail to respond to escalating immunosuppressive therapy and ultimately succumb to acute GvHD, its treatment and/or related complications.21 It denotes a risk that, in the setting of allogeneic SCT, is substantiated particularly in the presence of additional challenges, i.e., conditioning treatment and/or allogeneic T-cell attacks in a pro-inflammatory environment, and may result in endothelial damage that leads to perpetuating inflammatory end-organ destruction despite (initially) effective control of T-cell activity. Since this “vulnerability” is, at least in

Table 3. Multivariable analysis of predictors of overall survival, progression-free survival and non-relapse mortality after onset of acute graft-versus-host disease (complete case analysis).

OS after acute GvHD (n=541) HR (95% CI) P Covariate ADMA (per 1-log2 increase): 1st year ADMA (per 1-log2 increase): years 2 and 3 ADMA (per 1-log2 increase): after 3 years Age at transplantation (per year) Donor sex Female Male Recipient sex Female Male Donor-recipient HLA matching Mismatched Matched Disease stage† Early Late/intermediate Conditioning‡ Myeloablative Reduced intensity

PFS after acute GvHD (n=530) HR (95% CI) P

NRM* after acute GvHD (n=530) HR (95% CI) P

1.46 (1.17-1.83) 1.13 (0.73-1.76) 0.82 (0.45-1.48) 1.02 (1.01-1.03)

0.001 0.577 0.507 0.003

1.32 (1.07-1.63) 1.12 (0.70-1.81) 1.06 (0.56-2.02) 1.01 (1.00-1.03)

0.010 0.633 0.860 0.013

1.36 (1.01-1.83) 1.50 (0.76-2.96) 0.65 (0.26-1.64) 1.03 (1.01-1.05)

0.042 0.242 0.364 0.0003

Ref 0.79 (0.62-1.00)

0.054

Ref 0.81 (0.64-1.03)

0.081

Ref 0.71 (0.51-0.99)

0.041

Ref 1.21 (0.95-1.53)

0.119

Ref 1.24 (0.99-1.57)

0.065

Ref 1.30 (0.94-1.80)

0.119

Ref 0.61 (0.48-0.78)

<0.0001

Ref 0.73 (0.57-0.93)

0.010

Ref 0.55 (0.40-0.77)

0.0004

Ref 1.11(0.86-1.42)

0.423

Ref 1.11 (0.88-1.42)

0.378

Ref 0.91 (0.65-1.26)

0.553

Ref 1.08 (0.45-1.48)

0.710

Ref 1.05 (0.71-1.54)

0.847

Ref 1.05 (0.58-1.89)

0.882

Number of events: OS after acute GvHD, n=297; PFS after acute GvHD, n=314; NRM after acute GvHD, n=159; relapse after acute GvHD, n=155. *Cause-specific hazards from a competing risks analysis for relapse and NRM after onset of GvHD. †According to Gratwohl et al.41 ‡According to Bacigalupo et al.42 ADMA: asymmetric dimethylarginine; CI: confidence interval; GvHD: graft-versus-host-disease; HLA: human leukocyte antigen; HR: hazard ratio; NRM: non-relapse mortality; OS: overall survival; PFS: progression-free survival.

830

haematologica | 2019; 104(4)

ADMA and post-transplant mortality

part, a characteristic of the recipient’s endothelial cell system, increased risk of mortality, both overall and after the onset of acute GvHD, may be determined (and predicted) already prior to transplantation. However, it should be noted, that, although most markers of “endothelial vulnerability” are likely to predict outcome after a severe challenge, particularly after acute GvHD, they do not seem to be associated with a higher incidence of acute GvHD. In line with this, in the present study, pre-transplant ADMA was not a biomarker of acute GvHD risk but was associated with early NRM, both overall and following acute GvHD, which translated into shorter survival outcomes. With regards to the endothelial system, higher ADMA levels correlated positively with endothelium-related serum factors, such as sCD141, free IL-18, and nitrates. Thrombomodulin is a surface receptor protein released/shed from the endothelium in an extracellular soluble form (sCD141) during cell distress and elevated serum levels are frequently indicative of inflammatory cell damage.24 IL-18 is a pleiotropic, pro-inflammatory cytokine associated with coronary heart disease25 and de-regulation of the IL-18/IL-18BP pathway26 was shown to cause endothelial cell dysfunction. Interestingly, pre-transplant nitrate levels, which are markers of in vivo NO production, also correlated positively with ADMA, which is a NO synthase inhibitor. This seemingly contradictory finding is in line with some but not all observations made in other clinical settings.27,28

In this context, it should be noted, that higher pre-transplant nitrates levels were also associated with a higher incidence of NRM in a previous study.4 In pro-inflammatory conditions, large amounts of NO are produced by iNOS in numerous cell types.29 In the cardiovascular system, upregulation of iNOS may contribute to endothelial dysfunction.30 In the present study, however, INOS SNP did not influence the effect of ADMA on mortality. Certainly, interactions of ADMA with other NO synthase family members cannot be excluded, and there is also evidence that NO synthase-independent mechanisms may contribute to the detrimental biological effects associated with increases in ADMA.31,32 Consequently, in the setting of allogeneic SCT, elevated levels of both ADMA and nitrate prior to conditioning and immunosuppressive therapy are likely to reflect compromised endothelial homeostasis in the recipient which, particularly in the context of a severe challenge, such as acute GvHD, may result in an impaired outcome. However, since GvHD is an immune-triggered process, an influence of the donor’s ADMA or nitrate levels on the intensity of the subsequent allogeneic immune response in the recipient cannot definitely be ruled out. The median pre-transplant ADMA level in our series was 0.73 mM which is in accordance with literature data that suggest normal serum ADMA levels are in the range of 0.25-0.92 mM when measured by an enzyme-linked immunosorbent assay.33 Importantly, since ADMA is direct-

Figure 1. Higher pre-transplant asymmetric dimethylarginine levels are associated with non-relapse mortality and worse overall and progression-free survival in the first year after transplantation and after acute graft-versus-host disease. Meta-analyses of the multivariable effect of pre-transplant asymmetric dimethylarginine (ADMA) levels as a continuous variable (per 1-log2 increase) for overall survival (OS), progression-free survival (PFS), and non-relapse mortality (NRM) in the first year after transplantation and after acute graft-versus-host disease (GvHD), and relapse in the first year after allografting. CI: confidence interval; HR: hazard ratio; IPD: individual patient data.

haematologica | 2019; 104(4)

831

A. Radujkovic et al. A

G Figure 2. Outcome of patients in the first year after allogeneic stem cell transplantation and after onset of acute graft-versus-host disease when stratified according to pre-transplant asymmetric dimethylarginine quartiles. (A, B) Higher pre-transplant asymmetric dimethylarginine (ADMA) serum levels were associated with lower probabilities of overall survival (OS) in (A) the first year after allogeneic stem cell transplantation and (B) after the onset of acute graft-versus-host disease (GvHD). (C, D) The probabilities of progression-free survival (PFS) in (C) the first year after transplantation and (D) after the onset of acute GvHD were substantially lower in patients with higher pre-transplant ADMA levels. (E, F) Higher pre-transplant ADMA serum levels were associated with higher incidences of non-relapse mortality (NRM) in (E) the first year after allografting and (F) after the onset of acute GvHD. (G) In contrast, the ncidence of relapse during the first post-transplant year was not affected by pre-transplant ADMA levels. Note: Quartiles were chosen for reasons of visualization of the continuous ADMA effect which was observed for every 2-fold change in multivariable analyses (Tables 2 and 3). For this reason, and since pre-transplant ADMA values were not normally distributed, there is an overlap of the second and third quartile. Q1-4, quartiles of the ADMA distribution.

832

haematologica | 2019; 104(4)

ADMA and post-transplant mortality

Figure 3. Relationship between quartiles of pre-transplant asymmetric dimethylarginine serum concentration and serum concentration of nitrates, free interleukin-18, and soluble thrombomodulin (sCD141). Trend analysis showed that pre-transplant concentrations of (A) nitrates (n=375, (B) soluble thrombomodulin (n=776) and (C) free interleukin-18 (n=746) were positively correlated with quartiles of pre-transplant ADMA concentration (Jonckheere trend test, P=0.010, P<0.0001 and P<0.0001, respectively). ADMA, asymmetric dimethylarginine; IL-18: free interleukin-18; Q1-4, quartiles 1-4 of the ADMA distribution; sCD141: soluble thrombomodulin.

ly involved in endothelial dysfunction, it may be a therapeutic target. However, a specific ADMA-lowering agent is not yet available.34 Supplementation of L-citrulline may be employed to boost intracellular arginine levels and to reduce ADMA production. In a few clinical studies, supplemental L-citrulline in multi-gram doses was well tolerated and was shown to exert various effects suggestive of cardiovascular protection; it could, therefore, be studied in high-risk patients with elevated ADMA.35,36 Given the continuous relation between ADMA serum levels and survival outcomes in our study, treatment of patients irrespective of the individualâ&#x20AC;&#x2122;s pre-transplant ADMA status may be favored and such a clinical study should ideally be placebocontrolled. However, given the necessity of oral L-citrulline ingestion in multi-gram doses (3-10 g/day) and possible drug interactions,36 the first interventional study in the setting of allogeneic SCT setting may rather enroll patients with markedly elevated pre-transplant ADMA serum concentrations, who are at high risk of early mortality according to our retrospective analysis. Based on literature data and our results, in a possible clinical trial setting, ADMA levels â&#x2030;Ľ1 mM could be used to define such a high-risk population of patients. Besides its retrospective nature, some limitations of our study need to be addressed. ADMA was measured at a single time point and thus no information on the serum kinetics of ADMA levels in the post-transplant period can be provided. Furthermore, with regard to the known association of ADMA with cardiovascular pathologies, one could ask whether the NRM events in our study were related to cardiovascular disease. However, it should be noted that in the absence of disease recurrence, the exact cause of death in allografted patients, particularly in the early post-transplant period, is often difficult to determine and for the purposes of categorization, infection, GvHD and (general) organ failure are generally used. In a recently published large registry study,37 the main causes of NRM were (in descending order): infection, GvHD and respiratory disease, all of which may be linked to endothelial distress. The frehaematologica | 2019; 104(4)

quency of (isolated) cardiovascular events after transplantation is rather low with a cumulative incidence of 6% after 15 years.38 However, it should be taken into account that endothelial factors play a crucial role in many important complications of allogeneic SCT, such as transplant-associated microangiopathy, veno-occlusive disease, and refractory GvHD.1,39,40 Finally, like most clinical research on ADMA, our data are observational, describe a relationship and do not allow for interpretation of causality. Our results need to be validated, ideally in a prospective interventional study. As already pointed out above, investigating the effects of Lcitrulline treatment in high-risk patients undergoing allogeneic SCT may be a feasible approach. In summary, this retrospective study shows an association between ADMA levels and outcome after allogeneic SCT. Higher ADMA levels were a risk factor for early mortality in allografted patients, both overall and after the onset of acute GvHD, and were correlated with other biomarkers of endothelial vulnerability. Further studies on the role of the endothelium and markers associated with endothelial distress in the setting of allogeneic SCT are warranted. Acknowledgments The authors would like to thank Michael Hess and Alexandra Hof for their excellent technical assistance and construction of the tissue bank, Maria Gawlik for assistance in the collection of clinical data, and Thorsten Brenner and Stefan Hofer (Department of Anesthesiology, University of Heidelberg) for helpful discussions. We also wish to acknowledge the excellent help of Axel Benner (Department of Biostatistics, German Cancer Research Center) regarding the statistical analyses. Funding This work was supported by grants from the HelmholtzAlliance on Immunotherapy of Cancer, B.L.U.T. e.V. (Weingarten, Germany) and the Wilhelm-Sander-Stiftung (grant n. 2008.068.1). HD was sponsored by the China Scholarship Council. 833

A. Radujkovic et al.

References 1. 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. 2. Dietrich S, Falk CS, Benner A, et al. Endothelial vulnerability and endothelial damage are associated with risk of graft-versus-host disease and response to steroid treatment. Biol Blood Marrow Transplant. 2013;19(1):22-27. 3. Tatekawa S, Kohno A, Ozeki K, et al. A novel diagnostic and prognostic biomarker panel for endothelial cell damage-related complications in allogeneic transplantation. Biol Blood Marrow Transplant. 2016;22(9): 1573-1581. 4. Dietrich S, Okun JG, Schmidt K, et al. High pre-transplant serum nitrate levels predict risk of acute steroid-refractory graft-versushost disease in the absence of statin therapy. Haematologica. 2014;99(3):541-547. 5. Rachakonda SP, Penack O, Dietrich S, et al. Single-nucleotide polymorphisms within the thrombomodulin gene (THBD) predict mortality in patients with graft-versus-host disease. J Clin Oncol. 2014;32(30):3421-3427. 6. Vallance P, Leone A, Calver A, Collier J, Moncada S. Endogenous dimethylarginine as an inhibitor of nitric oxide synthesis. J Cardiovasc Pharmacol. 1992;20 (Suppl 12):S60-62. 7. Vallance P, Leone A, Calver A, Collier J, Moncada S. Accumulation of an endogenous inhibitor of nitric oxide synthesis in chronic renal failure. Lancet. 1992;339 (8793):572-575. 8. Kielstein JT, Impraim B, Simmel S, et al. Cardiovascular effects of systemic nitric oxide synthase inhibition with asymmetrical dimethylarginine in humans. Circulation. 2004;109(2):172-177. 9. Böger RH. Asymmetric dimethylarginine, an endogenous inhibitor of nitric oxide synthase, explains the "L-arginine paradox" and acts as a novel cardiovascular risk factor. J Nutr. 2004;134:(10 Suppl):2842S-2847S. 10. Bouras G, Deftereos S, Tousoulis D, et al. Asymmetric dimethylarginine (ADMA): a promising biomarker for cardiovascular disease? Curr Top Med Chem. 2013;13(2):180200. 11. Schlesinger S, Sonntag SR, Lieb W, Maas R. Asymmetric and symmetric dimethylarginine as risk markers for total mortality and cardiovascular outcomes: a systematic review and meta-analysis of prospective studies. PLoS One. 2016;11(11): e0165811. 12. Zhou S, Zhu Q, Li X, et al. Asymmetric dimethylarginine and all-cause mortality: a systematic review and meta-analysis. Sci Rep. 2017;7:44692. 13. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials. 1996;17(4):343-346. 14. Jonckheere AR. A distribution free k-sample test against ordered alternatives. Biometrika. 1954(1-2):133-145.

834

15. Tsikas D, Bollenbach A, Hanff E, Kayacelebi AA. Asymmetric dimethylarginine (ADMA), symmetric dimethylarginine (SDMA) and homoarginine (hArg): the ADMA, SDMA and hArg paradoxes. Cardiovasc Diabetol. 2018;17(1):1. 16. Chen XM, Hu CP, Li YJ, Jiang JL. Cardiovascular risk in autoimmune disorders: role of asymmetric dimethylarginine. Eur J Pharmacol. 2012;696(1-3):5-11. 17. Esposito C, Grosjean F, Torreggiani M, et al. Increased asymmetric dimethylarginine serum levels are associated with acute rejection in kidney transplant recipients. Transplant Proc. 2009;41(5):1570-1573. 18. Frenay AR, van den Berg E, de Borst MH, et al. Plasma ADMA associates with all-cause mortality in renal transplant recipients. Amino Acids. 2015;47(9):1941-1949. 19. Koch A, Weiskirchen R, Kunze J, et al. Elevated asymmetric dimethylarginine levels predict short- and long-term mortality risk in critically ill patients. J Crit Care. 2013;28(6):947-953. 20. Winkler MS, Kluge S, Holzmann M, et al. Markers of nitric oxide are associated with sepsis severity: an observational study. Crit Care. 2017;21(1):189. 21. Luft T, Dietrich S, Falk C, et al. Steroidrefractory GVHD: T-cell attack within a vulnerable endothelial system. Blood. 2011;118(6):1685-1692. 22. Inamoto Y, Ito M, Suzuki R, et al. Clinicopathological manifestations and treatment of intestinal transplant-associated microangiopathy. Bone Marrow Transplant. 2009;44(1):43-49. 23. Lindås R, Tvedt TH, Hatfield KJ, Reikvam H, Bruserud O. Preconditioning serum levels of endothelial cell-derived molecules and the risk of posttransplant complications in patients treated with allogeneic stem cell transplantation. J Transplant. 2014;2014: 404096. 24. Martin FA, Murphy RP, Cummins PM. Thrombomodulin and the vascular endothelium: insights into functional, regulatory, and therapeutic aspects. Am J Physiol Heart Circ Physiol. 2013;304(12):H1585-1597. 25. Jefferis BJ, Papacosta O, Owen CG, et al. Interleukin 18 and coronary heart disease: prospective study and systematic review. Atherosclerosis. 2011;217(1):227-233. 26. Durpès MC, Morin C, Paquin-Veillet J, et al. PKC-β activation inhibits IL-18-binding protein causing endothelial dysfunction and diabetic atherosclerosis. Cardiovasc Res. 2015;106(2):303-313. 27. Kocak H, Oner-Iyidogan Y, Gurdol F, Oner P, Esin D. Serum asymmetric dimethylarginine and nitric oxide levels in obese postmenopausal women. J Clin Lab Anal. 2011;25(3):174-178. 28. Cvetković T, Veličković-Radovanović R, Stojanović D, et al. Oxidative and nitrosative stress in stable renal transplant recipients with respect to the immunosuppression protocol- differences or similarities? J Med Biochem. 2015;34(3):295-303. 29. Knowles RG, Moncada S. Nitric oxide syn-

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

thases in mammals. Biochem J. 1994;298(Pt 2):249-258. Lind M, Hayes A, Caprnda M, et al. Inducible nitric oxide synthase: good or bad? Biomed Pharmacother. 2017;93:370375. Suda O, Tsutsui M, Morishita T, et al. Asymmetric dimethylarginine produces vascular lesions in endothelial nitric oxide synthase-deficient mice: involvement of reninangiotensin system and oxidative stress. Arterioscler Thromb Vasc Biol. 2004;24(9): 1682-1688. Smith CL, Anthony S, Hubank M, Leiper JM, Vallance P. Effects of ADMA upon gene expression: an insight into the pathophysiological significance of raised plasma ADMA. PLoS Med. 2005;2(10):e264. Németh B, Ajtay Z, Hejjel L, et al. The issue of plasma asymmetric dimethylarginine reference range - a systematic review and meta-analysis. PLoS One. 2017;12(5): e0177493. Tain YL, Hsu CN. Toxic Dimethylarginines: Asymmetric dimethylarginine (ADMA) and symmetric dimethylarginine (SDMA). Toxins (Basel). 2017;9(3):E92. McCarty MF. Asymmetric dimethylarginine is a well established mediating risk factor for cardiovascular morbidity and mortalityshould patients with elevated levels be supplemented with citrulline? Healthcare (Basel). 2016;4(3):E40. Allerton TD, Proctor DN, Stephens JM, Dugas TR, Spielmann G, Irving BA. l-citrulline supplementation: impact on cardiometabolic health. Nutrients. 2018;10 (7):E921. Atsuta Y, Hirakawa A, Nakasone H, et al. Late mortality and causes of death among long-term survivors after allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2016;22(9):1702-1709. Tichelli A, Passweg J, Wójcik D, et al. Late cardiovascular events after allogeneic hematopoietic stem cell transplantation: a retrospective multicenter study of the Late Effects Working Party of the European Group for Blood and Marrow Transplantation. Haematologica. 2008;93 (8):1203-1210. Cooke KR, Jannin A, Ho V. The contribution of endothelial activation and injury to endorgan toxicity following allogeneic hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2008;14(1 Suppl 1):23-32. Tichelli A, Gratwohl A. Vascular endothelium as 'novel' target of graft-versus-host disease. Best Pract Res Clin Haematol. 2008;21(2):139-148. Gratwohl A, Stern M, Brand R, et al. Risk score for outcome after allogeneic hematopoietic stem cell transplantation: a retrospective analysis. Cancer. 2009;115(20): 4715-4726. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant. 2009;15(12):1628-1633.

haematologica | 2019; 104(4)

ARTICLE

Stem Cell Transplantation

Disability related to chronic graft-versus-host disease after alternative donor hematopoietic cell transplantation

Ferrata Storti Foundation

Giancarlo Fatobene,1,2 Barry E. Storer,1 Rachel B. Salit, 1,3Stephanie J. Lee,1,3 Paul J. Martin,1,3 Guang-Shing Cheng,1,3 Paul A. Carpenter,1,3 Gansuvd Balgansuren,3,4 Effie W. Petersdorf,1,3 Colleen Delaney,1,3 Brenda M. Sandmaier,1,3 Filippo Milano1,3 and Mary E. Flowers1,3

Fred Hutchinson Cancer Research Center, Clinical Research Division, Seattle, WA, USA; Universidade de Sao Paulo, Hospital das Clinicas, SP, Brazil; 3University of Washington, Division of Medical Oncology, Seattle, WA, USA and 4Seattle Cancer Care Alliance, Seattle, WA, USA 1 2

Haematologica 2019 Volume 104(4):835-843

ABSTRACT

e determined the incidence of disability related to chronic graftversus-host disease (bronchiolitis obliterans, grade â&#x2030;Ľ2 keratoconjunctivitis sicca, sclerotic features or esophageal stricture) for three categories of alternative donor: cord blood, haplorelated marrow or peripheral blood with post-transplant cyclophosphamide, and unrelated single HLA-allele mismatched peripheral blood. Among 396 consecutive hematopoietic cell transplant recipients, 129 developed chronic graft-versus-host disease with 3-year cumulative incidences of 8% for cord blood, 24% for haplorelated grafts, and 55% for unrelated single HLAallele mismatched peripheral blood. Disability rates were significantly lower for cord blood [hazard ratio (HR) 0.13; 95% confidence interval (CI): 0.1-0.4] and for the haplorelated group (HR 0.31; 95% CI: 0.1-0.7) compared to the rate in the group transplanted with unrelated single HLA-allele mismatched peripheral blood. Cord blood recipients were also >2-fold more likely to return to work/school within 3 years from the onset of chronic graft-versus-host disease (HR 2.54; 95% CI: 1.1-5.7, P=0.02), and the haplorelated group trended similarly (HR 2.38; 95% CI: 1.0-5.9, P=0.06). Cord blood recipients were more likely to discontinue immunosuppression than were recipients of unrelated single HLA-allele mismatched peripheral blood (HR 3.96; 95% CI: 1.9-8.4, P=0.0003), similarly to the haplorelated group (HR 4.93; 95% CI: 2.2-11.1, P=0.0001). Progression-free survival and non-relapse mortality did not differ between groups grafted from different types of donors. Our observations that, compared to recipients of unrelated single HLA-allele mismatched peripheral blood, recipients of cord blood and haplorelated grafts less often developed disability related to chronic graft-versus-host disease, and were more likely to resume work/school, should help better counseling of pre-hematopoietic cell transplant candidates.

Correspondence: MARY E. D. FLOWERS mflowers@fhcrc.org. Received: July 25, 2018. Accepted: November 8, 2018. Pre-published: November 15, 2018. doi:10.3324/haematol.2018.202754 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/835 ÂŠ2019 Ferrata Storti Foundation

Introduction Hematopoietic cell transplantation (HCT) can be accomplished with grafts from alternative donors for patients lacking an HLA-matched related or unrelated donor. The optimal choice of an alternative donor stem cell source remains an open question and is influenced by several factors including chronic graft-versus-host disease (GvHD). Chronic GvHD is a heterogeneous syndrome associated with major morbidity and adverse effects on quality of life and functionality among long-term allogeneic HCT survivors.1-3 Because a diagnosis of chronic GvHD does not always indicate significant morbidity and poor quality of life, the frequency of severe chronic GvHD manifestations (e.g. severe keratoconjunctivitis sicca, bronchiolitis obliterans, cutaneous scleroderma, joint/fasciae features, and esophageal stricture haematologica | 2019; 104(4)

835

G. Fatobene et al.

requiring dilation), the duration of immunosuppressive therapy and resumption of pretransplant activities (e.g., work/school) may serve as better measures of outcome after HCT. The aim of this study was to analyze chronic GvHD manifestations most likely to be associated with disability, requirement of secondary systemic treatment, discontinuation of systemic immunosuppressive therapy and functional outcomes among recipients of grafts from alternative HCT donors. Differences in these clinical outcomes could help inform patients about outcomes after HCT from alternative donors.

Methods Patients and donors This retrospective study included all consecutive adult patients who received a first alternative donor HCT for any underlying diagnosis at Fred Hutchinson Cancer Research Center/Seattle Cancer Care Alliance between 2006 and 2015 and subsequently developed chronic GvHD that required systemic treatment. The alternative grafts included unrelated 4-6/6-HLA-matched single or double umbilical cord blood units (UCB), related HLA-haploidentical bone marrow or mobilized peripheral blood stem cells plus post-transplant cyclophosphamide (Haplo/PTCY), and mobilized peripheral blood stem cells from unrelated donors with a single HLA allele mismatched at an A, B, C or DRB1 locus by high resolution typing (1-mMUD), regardless of whether the mismatch resulted in antigen disparity at the locus. Patients had given written consent allowing the use of medical records for research in accordance with the Declaration of Helsinki, and the institutional review board approved the study.

are often necessary during the initial treatment of chronic GvHD.6 Discontinuation of systemic immunosuppression was defined as cessation of treatment for at least 6 months after resolution of chronic GvHD.

Statistical analysis The main endpoints of this study were chronic GvHD manifestations associated with disability and impaired functional outcomes (i.e., return to work/school after the diagnosis of chronic GvHD, discontinuation of systemic immunosuppressive therapy, and change in Karnovsky Performance Status). We also compared the overall severity of chronic GvHD at initial diagnosis and the incidences of avascular necrosis, new systemic immunosuppression or treatment after first-line therapy for chronic GvHD, nonrelapse mortality and overall survival after chronic GvHD diagnosis. The chronic GvHD characteristics between donor groups were compared using a c2 test for categorical variables and Wilcoxon rank-sum test for continuous variables. Overall survival was estimated by the Kaplan-Meier method. Cumulative incidences of chronic GvHD and of events after the onset of chronic GvHD were estimated by methods for competing risks, as previously described.7 Death was a competing event for all risks except nonrelapse mortality; relapse was a competing event for non-relapse mortality. All comparisons of time-to-event endpoints were performed using Cox regression. The analysis of return to work or school was restricted to patients who were working or in school before HCT, and had not returned to work or school before the onset of chronic GvHD. The analysis of high morbidity was based on the first defining complication. All P-values are two-sided and unadjusted for multiple comparisons.

Results Clinical assessments and definitions Involved sites and types of treatment at the onset of first systemic teraphy of chronic GvHD and treatment changes after initial teraphy were recorded prospectively via the Long-Term FollowUp Program through medical records from our outpatient clinic and local clinics that provided primary care for patients. All patients were screened for evidence of chronic GvHD between days 80 and 100 after HCT, at 1 year after HCT, and whenever clinically indicated to establish the diagnosis of chronic GvHD or to determine treatment. Acute GVHD was graded according to previously described criteria.4 Chronic GvHD was diagnosed using the 2014 National Institutes of Health consensus criteria.5 Disability related to chronic GvHD was defined as 2014 National Institutes of Health consensus grade 2 or 3 keratoconjunctivitis sicca, grade 2 or 3 scleroderma, any grade of bronchiolitis obliterans, grade 2 or 3 joint/fasciae involvement, or grade 3 esophageal stricture requiring dilation. While vulvovaginal chronic GvHD can result in fibrosis, this manifestation is under reported and unlikely to result in disability by itself, and thus not included in our study. National Institutes of Health score 2 or 3 gastrointestinal, oral or hepatic manifestations reflect GvHD activity but are less likely to cause irreversible damage and were also not included in our study. Return to work or school was considered only for patients who were working or in school before the HCT indication was diagnosed and had not resumed those activities before the onset of chronic GvHD. Treatment change was defined as any additional systemic treatment not used for the initial treatment of chronic GvHD. An increase in steroid dose in patients who were initially treated with steroid was not considered as a treatment change, because temporary increases in steroid doses or resumption of steroid treatment 836

We identified 396 alternative donor HCT recipients who received a first allogeneic transplant for any disease between 2006 and 2015 at our center. The median age at HCT was 42 years (range, 18-73) for UCB, 48 years (range, 18-75) for Haplo/PTCY, and 55 years (range, 22-77) for 1mMUD HCT recipients. Acute myeloid leukemia and myelodysplastic syndrome were the most common diagnoses at HCT, 64% and 65% among the UCB and 1mMUD HCT recipients, respectively, while lymphomas were the most common diagnosis (52%) among the Haplo/PTCY HCT recipients. Other demographic characteristics are summarized in Table 1.

Chronic graft-versus-host disease Of the 396 alternative donor HCT recipients, 129 developed chronic GvHD that required systemic treatment and were included in this study. The cases of treatment-requiring chronic GvHD were diagnosed in 29 of the 163 UCB HCT recipients, 21 of 88 Haplo/PTCY HCT recipients and 79 of 145 1-mMUD HCT recipients, for 3-year cumulative incidences of 18%, 24% and 55%, respectively. The rate of chronic GvHD was significantly lower in UCB and Haplo/PTCY recipients than in 1-mMURD recipients [hazard ratio (HR)=0.23; 95% confidence interval (95% CI): 0.2-0.4), P<0.0001, and HR=0.29 (95% CI: 0.2-0.5), P<0.0001, respectively] (Figure 1A). The incidence of chronic GvHD was comparable between patients who received 1-mMUD grafts with either allelic or antigenic mismatches (data not shown). Progression-free survival after HCT did not differ among the three donor groups haematologica | 2019; 104(4)

Severity of chronic GvHD after alternative donor HCT

(Figure 1B). The median follow-up times after the onset of chronic GvHD were 48 (range, 4-121) months for UCB, 60 (range, <1-123) for Haplo/PTCY, and 46 (range, 4-131) for 1-mMUD HCT. The median time from HCT to diagnosis of chronic GvHD was shorter in the UCB recipients than in 1-mMUD HCT recipients [3.9 (range, 2.6-18.2) versus 7.8 (range, 2.7-38.2) months, P=0.001]. As shown in Table 2, the incidence frequencies of overlap chronic GvHD Table 1. Characteristics of the study population according to alternative donor group.

Characteristic

Age at transplant (years), median (range) Female, n. (%) Race, n. (%)* White Other Recipient CMV seropositive, n. (%)** Diagnosis, n. (%) Acute myeloid leukemia Myelodysplastic syndrome Acute lymphocytic leukemia Chronic lymphocytic leukemia Chronic myeloid leukemia Hodgkin lymphoma Non-Hodgkin lymphoma1 Multiple myeloma2 Others3 Conditioning regimen, n. (%)4 Non-myeloablative/reduced intensity Myeloablative GvHD prophylaxis, n. (%) CNI and MMF CNI and MTX Cy posttransplant plus CNI and MMF Other Graft source, n. (%) Peripheral blood Bone marrow Umbilical cord blood Double units infused HLA-match, n. (%) 7/8 4-6/8 5-6/6 4/6 3/6 Follow-up after HCT (months), Median, (range)

Alternative donor group Unrelated Umbilical Related mismatched cord blood haploidentical (N = 145) (N = 163) (N = 88) 55 (22-77)

42 (18-73)

48 (18-75)

54 (37)

83 (51)

35 (40)

113 (82) 24 (18) 90 (62)

85 (56) 68 (44) 103 (64)

59 (69) 26 (31) 52 (59)

53 (37) 31 (28) 17 (12) 9 (6) 11 (8) 1 (1) 12 (8) 8 (6) 3 (2)

82 (50) 23 (14) 37 (23) 2 (1) 7 (4) 0 7 (4) 0 5 (3)

21 (24) 6 (7) 5 (6) 3 (3) 2 (2) 25 (28) 21 (24) 4 (5) 1 (1)

69 (48) 76 (52)

43 (26) 120 (74)

71 (81) 17 (19)

71 (49) 71 (49) 0 3 (2)

163 (100) 0 0 0

0 0 87 (99) 1 (1)

145 (100) -

163 (100) 157 (96)

31 (35) 57 (65) -

145 (100) 0 0 0 0 46 (4-131)

0 0 35 (21) 92 (56) 36 (22) 48 (4-121)

1 (1) 6 (7) 0 1(1) 80 (91) 60 (<1-123)

**Unknown for 21 patients. **Unknown for three patients. 1Includes four cases of prolymphocytic leukemia. 2Includes two cases with plasma cell leukemia. 3Includes mycosis fungoides (n=2), polycythemia vera (n=2), and one each of the following diagnoses: immune deficiency disorder, aplastic anemia, hemophagocytic lymphohistiocytosis, systemic sclerosis and unspecified neoplasm. 4Myeloablative conditioning regimens contained total body irradiation ≥5 Gy single dose or ≥8 Gy fractioned or busulfan >8 mg/kg orally or intravenous equivalent. Non-myeloablative or reduced intensity conditioning consisted of fludarabine + total body irradiation (200-400 cGy) ± cyclophosphamide. CMV: cytomegalovirus; GvHD: graft-versus-host disease; CNI: calcineurin inhibitor; MMF: mycophenolate mofetil; MTX: methotrexate; Cy: cyclophosphamide.

haematologica | 2019; 104(4)

(>80%) and of prior acute GvHD (>70%) were high in all three groups. These findings are consistent with the frequent diagnosis of upper gastrointestinal GvHD at our center.8 The severity of chronic GvHD at diagnosis was significantly lower in the UCB group than in the 1-mMUD group (P=0.008) (Table 2), but the severity of GvHD manifestations at the onset of chronic GvHD did not differ between the Haplo/PTCY and the 1-mMUD groups (P=0.74) (Table 2). According to the National Institutes of Health Global Severity scale, the incidence of moderate or severe chronic GvHD at diagnosis was 62% in the UCB group, 76% in the Haplo/PTCY group and 83% in the 1mMUD group. Table 2 displays additional characteristics of the chronic GvHD according to the alternative donor HCT groups. Sites of chronic GvHD and the presence of eosinophilia at any time during the course of chronic GvHD among the three alternative donor HCT groups are displayed in Figure 2. Mouth and skin were the most common sites of chronic GvHD in the three groups (>80%). Eyes were affected by chronic GvHD of any degree at any time in 75% of the 1-mMUD, 52% of the Haplo/PTCY, and 34% of the UCB HCT recipients. The same pattern was identified for hepatic involvement at any time (56% of 1-

Figure 1. The cumulative incidence of chronic graft-versus-host disease is higher in the mismatched unrelated donor group than in the cord blood or HLA-haploidentical groups, but survival does not differ between the groups. Cumulative incidence of (A) chronic graft-versus-host disease and (B) progression-free survival after transplant according to alternative donor hematopoietic cell transplant group. NIH: National Institutes of Health; cGVHD: chronic graft-versus-host disease; URD Mism: mismatched unrelated donor; Cord: umbilical cord blood; Haplo: haplorelated bone marrow or peripheral blood; PFS: progression-free survival.

837

G. Fatobene et al. Table 2. Characteristics of chronic graft-versus-host disease according to alternative donor group.

Characteristics

Chronic GvHD, n. (%) Classic Overlap Prior late acute GvHD, n. (%) Prior II-IV acute GvHD, n. (%) Time from HCT to diagnosis, months, (range) NIH severity at diagnosis, n. (%) Mild Moderate Severe Type of onset, n. (%) De novo Quiescent Progressive Sites involved and eosinophilia at onset, n. (%) Skin Eyes Mouth Liver Lung (bronchiolitis obliterans) Gastrointestinal tract Joint/fasciae Genitals Eosinophilia N. of GvHD sites at onset, n. (%) 1 or 2 3 >3 KPS <80% at onset, n. (%)* Dose of prednisone at onset, n. (%) None 0.1 - 0.4 mg/kg 0.5 â&#x20AC;&#x201C; 1.0 mg/kg Platelets < 105/mL at onset, n. (%)

Umbilical cord blood (N = 29)

7 (9) 72 (91) 14 (18) 55 (70) 7.8 (2.7-38.2)

3 (10) 26 (90) 3 (10) 29 (100) 3.9 (2.6 -18.2)

13 (16) 46 (58) 20 (25)

11 (38) 17 (59) 1 (3)

5 (24) 11 (52) 5 (24)

0.74

0.008

20 (25) 7 (9) 52 (66)

0 3 (10) 26 (90)

0 5 (24) 16 (76)

0.02

0.01

59 (75) 30 (38) 74 (94) 27 (34) 0 28 (35) 5 (6) 8 (10) 19 (24)

16 (55) 5 (17) 26 (90) 3 (10) 1 (3) 20 (69) 0 1 (3) 1 (3)

0.05 0.04 0.48 0.01 0.10 0.002 0.17 0.27 0.01

16 (76) 5 (24) 16 (76) 2 (10) 1 (5) 6 (29) 0 1 (5) 3 (14)

0.89 0.23 0.02 0.03 0.05 0.55 0.24 0.45 0.34

26 (33) 31 (39) 22 (28) 22 (33)

16 (55) 11 (38) 2 (7) 12 (55)

29 (73) 6 (15) 5 (13) 14 (18)

9 (64) 4 (29) 1 (7) 7 (24)

0.81 0.35 0.0008 0.001

0.03 0.08

0.19 0.46

Related hploidentical (N = 21)

Unrelated mismatched (N =79)

4 (19) 17 (81) 8 (38) 20 (95) 7.5 (2.9-15.4)

13 (62) 7 (33) 1 (5) 6 (35) 7 (64) 3 (27) 1 (9) 2 (10)

0.18 0.05 0.02 0.77

0.02 0.88

0.95 0.36

Cord blood vs. mismatched unrelated. 2Haploidentical vs mismatched unrelated. * Used values within 14 days of onset (pre or posttransplant). KPS at onset available for 66 1mMUD, 22 UCB and 17 haploidentical related HCT recipients. GVHD: graft-versus-host disease; HCT: hematopoietic cell transplant; NIH: National Institutes of Health; KPS: Karnofsky Performance Status.

Figure 2. Patterns of organ involvement differ according to the type of alternative donor. The figure shows the proportions of patients with involved sites or signs of chronic graft-versus-host disease at any time from initial diagnosis to last follow-up according to alternative donor HCT group. Patients could have more than one site involved. GI, gastrointestinal; HCT, hematopoietic cell transplant; UCB: umbilical cord blood.

838

haematologica | 2019; 104(4)

Severity of chronic GvHD after alternative donor HCT Table 3. Outcomes analyzed according to alternative hematopoietic cell transplant donor group.

Outcome

Alternative donor group Cumulative incidence at 3 years Unrelated Umbilical Related mismatched cord blood haploidentical (N = 79) (N = 29) (N = 21)

Off systemic therapy* Return to work1 Change of therapy Overall survival Non-relapse mortality Relapse/ progression

Cord blood vs. unrelated mismatched Hazard ratio P (95% CI)

Haploidentical vs. unrelated mismatched Hazard ratio P (95% CI)

15% 35% 39% 72% 22% 17%

45% 68% 17% 82% 11% 18%

50% 62% 25% 90% 6% 29%

3.96 (1.9-8.4) 2.54 (1.1-5.7) 0.30 (0.1-0.8) 0.59 (0.3-1.4) 0.32 (0.1-1.4) 0.95 (0.3-2.7)

0.0003 0.02 0.01 0.21 0.13 0.93

4.93 (2.2-11.1) 2.38 (1.0-5.9) 0.53 (0.2-1.4) 0.49 (0.2-1.4) 0.20 (0.0-1.5) 1.43 (0.5-4.1)

0.0001 0.06 0.19 0.18 0.12 0.50

Any Kerato conjunctivitis sicca (eyes) Sclerotic features Bronchiolitis obliterans (lung) Joint/fasciae

58% 39% 24% 18% 5%

17% 10% 3% 3% 0

23% 15% 0 5% 0

0.16 (0.1-0.4) 0.17 (0.1-0.5) 0.1 (0.0-0.7) 0.15 (0.0-1.1) 0.0 (undefined)

0.0001 0.003 0.02 0.16 0.08

0.32 (0.1-0.7) 0.35 (0.1-1.0) 0.14 (0.0-1.0) 0.23 (0.0-1.8) 0.0 (undefined)

0.009 0.05 0.05 0.16 0.13

Esophageal stricture

5 (-12 – 44)

11 (-14 – 64)

0.05

0.06

Manifestation of GvHD associated with disability3

Change in Karnofsky Performance Status2 Median (range)

0 (-81 – 46)

*Discontinuation of systemic treatment after resolution of chronic GvHD. 1Among 35, 19, and 13 patients who were working or in school prior to HCT, and had not returned to work or school prior to the onset of chronic GvHD. 2Among 54, 20, and 11 patients who had a value at onset and a value after 6 months. Change is annualized change between onset and last value through 3.5 years. 3Patients could have more than one chronic GvHD manifestation associated with disability defined as .grade 2 or 3 keratoconjunctivitis sicca, scleroderma features, bronchiolitis obliterans, grade 2 or 3 joint/fasciae involvement, or esophageal stricture requiring dilation. KPS: Karnofsky Performance Status; NA: not applicable; GvHD: graft-versus-host disease.

mmUD, 43% of Haplo/PTCY, and 24% of UCB HCT recipients). On the other hand, gastrointestinal tract involvement at any time developed in 72% of the UCB patients, but in 58% of the 1-mmUD and 52% of the Haplo/PTCY HCT recipients. Results of major outcomes according to the alternative donor HCT groups are shown in Table 3.

Chronic graft-versus-host disease manifestations of high morbidity The cumulative incidence of any manifestation of high morbidity at 3 years after the diagnosis of chronic GvHD is shown in Figure 3A and was significantly lower in the UCB (17%) and Haplo/PTCY (23%) groups than in the 1mMUD group (58%) [HR 0.16 (95% CI: 0.1-0.4); P=0.0001, and HR 0.32 (95% CI: 0.1-0.7), P=0.009, respectively], (Figure 3A). Table 3 shows the distribution of chronic GvHD manifestations of high morbidity according to the three HCT donor groups. The most frequent high morbidity was keratoconjunctivitis sicca followed by sclerosis and bronchiolitis obliterans (lungs). Moderate or severe joint/fasciae involvement and esophageal stricture requiring dilation were less frequent high morbidity manifestations. The cumulative incidence of keratoconjunctivitis sicca was significantly lower in the UCB group than in the 1-mMUD group (10% versus 39%, P=0.003) and was also lower in the Haplo/PTCY group (10%, P=0.05). The 3-year cumulative incidence of any high morbidity haematologica | 2019; 104(4)

and the three most frequent high morbidity chronic GvHD manifestations according to the alternative donor groups are displayed in Figure 3.

Duration of immunosuppressive therapy and change in systemic therapy The proportion of patients in each group requiring changes in systemic therapy for control of chronic GvHD at 3 years after first-line treatment was 17% for the UCB group, 25% for the Haplo/PTCY groups and 39% for the 1-mMUD group, and was significantly lower for the UCB group than for the 1-mMUD group [HR 0.30 (95% CI: 0.10.8), P=0.01] (Figure 4A and Table 3). The cumulative incidence of discontinued systemic immunosuppression at 3 years was significantly lower in the 1-mMUD (15%) group than in the UCB (45%) and Haplo/PTC (50%) groups [HR 3.96 (95% CI: 1.9–8.4), P=0.0003, and HR 4.93 (95% CI: 2.2–11.1), P=0.0001, respectively] (Figure 4B).

Functional endpoints A higher proportion of UCB patients than 1-mMUD HCT recipients (68% versus 35%) returned to work or school within 3 years after the onset of chronic GvHD [HR 2.54 (95% CI: 1.1-5.7), P=0.02], and a similar trend was observed in the Haplo/PTCY group [62% versus 35%, HR 2.38 (95% CI: 1.0-5.9), P=0.06] (Figure 4C). Eighteen patients had returned to work or school before the onset 839

G. Fatobene et al.

of chronic GvHD (14 in the 1-mMUD and 4 in the Haplo/PTCY HCT groups). We also found trends suggesting improved annualized Karnovsky Performance Status change from the onset of chronic GvHD to 3.5 years afterwards in the UCB [+5 (range, -12 to +44)] and Haplo/PTCY [+11 (range, -14 to +64)] groups compared to the 1-mMUD group [0 (range, -81 to +46)] (P=0.05 and P=0.06, respectively). We found no difference in avascular necrosis at 3 years among the three alternative donor groups (12% in the 1-mMUD, 12% in the UCB, and 5% in the Haplo/PTCY HCT recipients).

Progression-free survival after HCT did not differ between the three groups (Figure 1B).

Discussion

The cumulative incidence of non-relapse mortality at 3 years among patients with chronic GvHD was 22% in 1mMUD recipients, 11% in UCB recipients, and 6% in the Haplo/PTCY recipients, with no statistically significant differences between the three groups [1-mMUD versus UCB, HR 0.32 (95% CI: 0.1-1.4), P=0.13; 1-mMUD versus Haplo/PTCY, HR 0.20 (95% CI: 0.0-1.5), P=0.12] (Table 3). The cumulative incidence of overall survival at 3 years among patients with chronic GvHD was 72% in 1mMUD recipients, 82% in UCB recipients, and 90% in Haplo/PTCY recipients, with no statistically significant differences between the three groups [1-mMUD versus UCB, HR 0.59 (95% CI: 0.3-1.4), P=0.21; 1-mMUD versus Haplo/PTCY, HR 0.49 (95% CI: 0.2-1.4), P=0.18] (Table 3).

In parallel with the lower rates of overall chronic GvHD, we demonstrated that the cumulative incidence of chronic GvHD manifestations associated with disability was significantly lower after UCB and Haplo/PTCY HCT than that after 1-mMUD HCT. The overall incidence of chronic GvHD in our study was lower in the UCB and Haplo/PTCY HCT cohorts than in the 1-mMUD HCT cohort, similar to results of previous studies.9-17 The overall incidence of chronic GvHD after 1-mMUD HCT was higher in our study than in previous reports.9,11,13,17-21 This difference may be explained by the higher proportion of bone marrow grafts and more frequent use of T-cell depletion in the previous studies.9,11,13,17,18,22 Older age could also be a contributor to the higher rates of chronic GvHD in the 1-mMUD group compared to the UCB and Haplo/PTCY groups. Keratoconjunctivitis sicca was the most common chronic GvHD manifestation of high morbidity. Of note, the incidence of severe keratoconjunctivitis sicca at any time after the diagnosis of chronic GvHD was at least 4-fold higher in the 1-mMUD group than in the UCB group and

Survival endpoints

Figure 3. Patients in the group grafted from a mismatched unrelated donor had a high cumulative incidence of disability caused by morbidity involving the skin, eyes and lungs. Cumulative incidence of: (A) any manifestations of chronic GVHD associated with disability; (B) moderate or severe keratoconjunctivitis sicca, (C) skin sclerosis and (D) bronchiolitis obliterans according to alternative donor HCT group. URD Mism: mismatched unrelated donor; Cord: umbilical cord blood; Haplo: haplorelated bone marrow or peripheral blood; GVHD: graft-versus-host disease.

840

haematologica | 2019; 104(4)

Severity of chronic GvHD after alternative donor HCT

at least 2-fold higher in the 1-mMUD group than in the Haplo/PTCY group. Prospective trials have shown that quality of life is impaired in patients with chronic GvHD involving the eyes.23,24 Sclerotic manifestations of chronic GvHD were the second most common severe morbidity in the three donor groups, but were seen in less than 5% of the UCB and Haplo/PTCY groups as compared to nearly a quarter of the 1-mMUD HCT group. The reported overall cumulative incidence of sclerotic chronic GvHD after HCT is approximately 20%, with lower rates in recipients of HLA-mismatched HCT than in recipients of HLAmatched HCT.25,26 The use of growth factor-mobilized blood cells is a known risk factor for the development of sclerotic GvHD after HCT. The frequent use of mobilized blood cells could explain the 25% incidence of sclerotic manifestations in our 1-mMUD HCT group. The remarkably low rate of sclerotic manifestations in the UCB cohort is consistent with prior reports of outcomes after UCB HCT.26,27 In order to assess protracted chronic GvHD, we evaluated the duration of systemic therapy used to control the disease manifestations and the impact on functional outcomes among the three alternative donor groups. A large proportion of UCB (45%) and Haplo/PTCY (59%) HCT recipients were able to discontinue all systemic treatment for chronic GvHD by 3 years after onset of the condition, in contrast to only 15% of the 1-mMUD group. The use of mobilized blood as the stem cell graft and the involvement of multiple sites at initial diagnosis are risk factors for prolonged immunosuppressive treatment.28-30 The significantly lower incidence of second-line systemic immunosuppressive treatment of chronic GvHD in the UCB and Haplo/PTCY groups than in the 1-mMUD group also supports the conclusion that the severity of chronic GvHD is greater after HCT with 1-mMUD donors than with UCB or HLA-haploidentical related donors. The presence of chronic GvHD has been consistently associated with failure to return to work or school, limited resilience and poor quality of life among HCT survivors.13,31 In our study, a significantly higher proportion of the UCB group were more likely to return to work or school compared to the that of the 1-mMUD group, and a similar benefit trend was also evident for the Haplo/PTCY group, supporting the conclusion that GvHD manifestations associated with disability frequently impaired recovery of pre-transplant function in the 1-mMUD group. Our study has several limitations. First, the three groups were heterogeneous with respect to the patients’ age, gender, race, underlying disease and conditioning regimen intensity. Among these factors, only the patients’ age has been associated with an increased incidence of chronic GvHD. An association of older patients’ age with higher risk of chronic GvHD has not been consistently observed after UCB HCT, and very few data addressing this issue are available for Haplo/PTCY recipients.32-37 A second limitation of the study is that the small numbers of patients with chronic GvHD in the UCB and Haplo/PTCY groups precluded a direct comparison of outcomes between these two groups. Third, we included both one-antigen and allele-mismatched unrelated donor peripheral blood stem cell recipients irrespective of the “direction” of mismatching. However, there was no significant difference in incidence of chronic GvHD between these mismatched unrelated donor groups (data not shown), consistent with the haematologica | 2019; 104(4)

results of a large, previous study from the Center for International Blood and Marrow Transplant Research.38 Fourth, the incidence of chronic GvHD in the 1-mMUD group in our study may not be representative of results with T-cell depleted 1-mMUD grafts.26,34,39,40 Moreover, considering that mobilized blood stem cells were the graft source used for the 1-mMUD HCT group in our study, the risk and morbidity of chronic GvHD with unrelated donor bone marrow grafting might not differ from those for

Figure 4. Patients in the mismatched unrelated donor group had a high cumulative incidence of second-line treatment, and withdrawal of immunosuppressive treatment and return to work or school were delayed. Cumulative incidence of (A) change of systemic therapy after first-line therapy for chronic GVHD; (B) discontinuation of systemic immunosuppressive therapy, and (C) return to work/school after the diagnosis of chronic GVHD according to the alternative HCT group. Rx: treatment; IST: immunosuppressive therapy; URD Mism, mismatched unrelated peripheral blood stem cell; Haplo, related haploidentical; cGVHD, chronic graft-versus-host disease.

841

G. Fatobene et al.

UCB or Haplo/PTCY HCT. Fifth, we were unable to assess quality of life in this study, because very few patients had outcomes based on validated measurement instruments. Sixth, the duration of immunosuppression may be driven ultimately by the selection of stem cell sources associated with a higher risk of severe chronic GvHD (i.e., unrelated mobilized blood cells) or with a lower risk of severe chronic GvHD (i.e., UCB or Haplo/PTCY). Seventh, some manifestations of chronic GvHD may have been underreported in this retrospective study (e.g., genital or lung involvement) Lastly, the analysis of functional endpoints at 3 years is limited by the small numbers. In conclusion, our results show that compared to 1mMUD recipients, UCB and Haplo/PTCY HCT recipients were less likely to develop chronic GvHD, were less likely to develop disability related to chronic GvHD, had a shorter duration of systemic treatment for chronic GvHD and returned to work or school earlier. While our data do not imply superiority of one alternative transplant strategy versus another (e.g. Haplo/PTCY versus 1-mMUD peripheral

References 1. Wong FL, Francisco L, Togawa K, et al. Longterm recovery after hematopoietic cell transplantation: predictors of quality-of-life concerns. Blood. 2010;115(12):2508-2519. 2. Rosenberg AR, Syrjala KL, Martin PJ, et al. Resilience, health, and quality of life among long-term survivors of hematopoietic cell transplantation. Cancer. 2015;121(23):42504257. 3. Lee SJ, Logan B, Westervelt P, et al. Comparison of patient-reported outcomes in 5-year survivors who received bone marrow vs peripheral blood unrelated donor transplantation: long-term follow-up of a randomized clinical trial. JAMA Oncol. 2016;2(12):1583-1589. 4. 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. 5. 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. 6. Martin PJ, Storer BE, Rowley SD, et al. Evaluation of mycophenolate mofetil for initial treatment of chronic graft-versus-host disease. Blood. 2009;113(21):5074-5082. 7. Gooley TA, Leisenring W, Crowley J, Storer BE. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med. 1999;18(6):695-706. 8. Martin PJ, McDonald GB, Sanders JE, et al. Increasingly frequent diagnosis of acute gastrointestinal graft-versus-host disease after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2004;10(5):320-327. 9. Eapen M, Rocha V, Sanz G, et al. Effect of

842

10.

11.

12.

13.

14.

15.

16.

blood versus UCB), the current study provides detailed information regarding disability related to chronic GvHD when it occurs after alternative donor HCT. These findings should help transplant providers to counsel pre-HCT candidates better. Determination of disability associated with chronic GvHD is necessary in prospective studies testing alternative donor graft products such as Haplo/PTCY versus UCB to evaluate the overall efficacy of these strategies. Acknowledgments We thank Chris Davis, Kevin Bray and Aaron Johnson for their help in retrieving and assembling the data. We also thank Helen Crawford for assistance in submission of this paper. Funding This work was supported in part by grants from the National Institutes of Health, National Cancer Institute (CA018029, CA118953, P30 CA015704) and in part by the National Heart, Lung and Blood Institute (HL122173). GF was supported by a grant from the Division of Hematology and Transfusion Medicine of the University of Sao Paulo, SP, Brazil.

graft source on unrelated donor haemopoietic stem-cell tranpslantation in adults with acute leukaemia: a retrospective analysis. Lancet Oncol. 2010;11(7):653-660. Brunstein CG, Fuchs EJ, Carter SL, et al. Alternative donor transplantation after reduced intensity conditioning: results of parallel phase 2 trials using partially HLAmismatched related bone marrow or unrelated umbilical cord blood grafts. Blood. 2011;118(2):282-288. Raiola AM, Dominietto A, di Grazia C, et al. Unmanipulated haploidentical transplants compared with other alternative donors and matched sibling grafts. Biol Blood Marrow Transplant. 2014;20(10):1573-1579. Di Stasi A, Milton DR, Poon LM, et al. Similar transplantation outcomes for acute myeloid leukemia and myelodysplastic syndrome patients with haploidentical versus 10/10 human leukocyte antigen-matched unrelated and related donors. Biol Blood Marrow Transplant. 2014;20(12):1975-1981. Malard F, Milpied N, Blaise D, et al. Effect of graft source on unrelated donor hemopoietic stem cell transplantation in adults with acute myeloid leukemia after reduced-intensity or nonmyeloablative conditioning: a study from the Societe Francaise de Greffe de Moelle et de Therapie Cellulaire. Biol Blood Marrow Transplant. 2015;21(6):1059-1067. Bashey A, Zhang MJ, McCurdy SR, et al. Mobilized peripheral blood stem cells versus unstimulated bone marrow as a graft source for T-cell-replete haploidentical donor transplantation using post-transplant cyclophosphamide. J Clin Oncol. 2017;35(26):30023009. Kanate AS, Mussetti A, Kharfan-Dabaja MA, et al. Reduced-intensity transplantation for lymphomas using haploidentical related donors versus HLA-matched unrelated donors. Blood. 2016;127(7):938-947. Gauthier J, Castagna L, Garnier F, et al. Reduced-intensity and non-myeloablative allogeneic stem cell transplantation from alternative HLA-mismatched donors for Hodgkin lymphoma: a study by the French Society of Bone Marrow Transplantation

17.

18.

19.

20.

21.

22.

23.

24.

25.

and Cellular Therapy. Bone Marrow Transplant. 2017;52(5):689-696. Robin M, Ruggeri A, Labopin M, et al. Comparison of unrelated cord blood and peripheral blood stem cell transplantation in adults with myelodysplastic syndrome after reduced-intensity conditioning regimen: a collaborative study from Eurocord (Cord Blood Committee of Cellular Therapy & Immunobiology Working Party of EBMT) and Chronic Malignancies Working Party. Biol Blood Marrow Transplant. 2015;21(3): 489-495. Malard F, Furst S, Loirat M, et al. Effect of graft source on mismatched unrelated donor hemopoietic stem cell transplantation after reduced intensity conditioning. Leukemia. 2013;27(11):2113-2117. Marks DI, Woo KA, Zhong X, et al. Unrelated umbilical cord blood transplant for adult acute lymphoblastic leukemia in first and second complete remission: A comparison with allografts from adult unrelated donors. Haematologica. 2014;99(2):322-328. Granier C, Biard L, Masson E, et al. Impact of the source of hematopoietic stem cell in unrelated transplants: comparison between 10/10, 9/10-HLA matched donors and cord blood. Am J Hematol. 2015;90(10):897-903. Piemontese S, Ciceri F, Labopin M, et al. A comparison between allogeneic stem cell transplantation from unmanipulated haploidentical and unrelated donors in acute leukemia. J Hematol Oncol. 2017;10(1):24. Fernandes JF, Rocha V, Labopin M, et al. Transplantation in patients with SCID: mismatched related stem cells or unrelated cord blood? Blood. 2012;119(12):2949-2955. Sun YC, Chai X, Inamoto Y, et al. Impact of ocular chronic graft-versus-host disease on quality of life. Biol Blood Marrow Transplant. 2015;21(9):1687-1691. Saboo US, Amparo F, Abud TB, Schaumberg DA, Dana R. Vision-related quality of life in patients with ocular graft-versus-host disease. Ophthalmology. 2015;122(8):1669-1674. Inamoto Y, Storer BE, Petersdorf EW, et al. Incidence, risk factors and outcomes of sclerosis in patients with chronic graft-versus-

haematologica | 2019; 104(4)

Severity of chronic GvHD after alternative donor HCT

26.

27.

28.

29.

30.

host disease. Blood. 2013;121(25):50985103. Detrait MY, Morisset S, Peffault de Latour R, et al. Pre-transplantation risk factors to develop sclerotic chronic GvHD after allogeneic HSCT: a multicenter retrospective study from the Societe Francaise de Greffe de Moelle et de Therapie Cellulaire (SFGMTC). Bone Marrow Transplant. 2015;50(2): 253-258. Newell LF, Flowers ME, Gooley TA, et al. Characteristics of chronic GVHD after cord blood transplantation. Bone Marrow Transplant. 2013;48(10):1285-1290. Flowers MED, Parker PM, Johnston LJ, et al. Comparison of chronic graft-versus-host disease after transplantation of peripheral blood stem cells versus bone marrow in allogeneic recipients: long-term follow-up of a randomized trial. Blood. 2002;100(2):415-419. Stewart BL, Storer B, Storek J, et al. Duration of immunosuppressive treatment for chronic graft-versus-host disease. Blood. 2004;104(12):3501-3506. Vigorito AC, Campregher PV, Storer BE, et al. Evaluation of NIH consensus criteria for classification of late acute and chronic GVHD. Blood. 2009;114(3):702-708.

haematologica | 2019; 104(4)

31. Hamilton JG, Wu LM, Austin JE, et al. Economic survivorship stress is associated with poor health-related quality of life among distressed survivors of hematopoietic stem cell transplantation. Psychooncology. 2013;22(4):911-921. 32. Storb R, Prentice RL, Sullivan KM, et al. Predictive factors in chronic graft-versushost disease in patients with aplastic anemia treated by marrow transplantation from HLA-identical siblings. Ann Intern Med. 1983;98(4):461-466. 33. Carlens S, Ringden O, Remberger M, et al. Risk factors for chronic graft-versus-host disease after bone marrow transplantation: a retrospective single centre analysis. Bone Marrow Transplant. 1998;22(8):755-761. 34. Atkinson K, Horowitz MM, Gale RP, et al. Risk factors for chronic graft-versus-host disease after HLA-identical sibling bone marrow transplantation. Blood. 1990;75(12): 2459-2464. 35. Flowers MED, Inamoto Y, Carpenter PA, et al. Comparative analysis of risk factors for acute graft-versus-host disease and for chronic graft-versus-host disease according to National Institutes of Health consensus criteria. Blood. 2011;117(11):3214-3219.

36. Kanda J, Nakasone H, Atsuta Y, et al. Risk factors and organ involvement of chronic GVHD in Japan. Bone Marrow Transplant. 2014;49(2):228-235. 37. Lazaryan A, Weisdorf DJ, DeFor T, et al. Risk factors for acute and chronic graft-versus-host disease after allogeneic hematopoietic cell transplantation with umbilical cord blood and matched sibling donors. Biol Blood Marrow Transplant. 2016;22(1):134140. 38. Lee SJ, Klein J, Haagenson M, et al. High-resolution donor-recipient HLA matching contributes to the success of unrelated donor marrow transplantation. Blood. 2007;110 (13):4576-4583. 39. Socie G, Schmoor C, Bethge WA, et al. Chronic graft-versus-host disease: long-term results from a randomized trial on graft-versus-host disease prophylaxis with or without anti-T-cell globulin ATG-Fresenius. Blood. 2011;117(23):6375-6382. 40. Uhm J, Hamad N, Shin EM, et al. Incidence, risk factors, and long-term outcomes of sclerotic graft-versus-host disease after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2014;20(11): 1751-1757.

843

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(4):844-854

Correspondence: MICHAEL A. PULSIPHER mpulsipher@chla.usc.edu Received: June 21, 2018. Accepted: October 30, 2018. Pre-published: October 31, 2018. doi:10.3324/haematol.2018.200121 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/4/844 ©2019 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

844

Blood Transfusion

Related peripheral blood stem cell donors experience more severe symptoms and less complete recovery at one year compared to unrelated donors

Michael A. Pulsipher,1 Brent R. Logan,2 Deidre M. Kiefer,3 Pintip Chitphakdithai,3 Marcie L. Riches,4 J. Douglas Rizzo,2 Paolo Anderlini,5 0Susan F. Leitman,6 Hati Kobusingye,3 RaeAnne M. Besser,3 John P. Miller,7 Rebecca J. Drexler,3 Aly Abdel-Mageed,8 Ibrahim A. Ahmed,9 Luke P. Akard,10 Andrew S. Artz,11 Edward D. Ball,12 Ruthee-Lu Bayer,13 Carolyn Bigelow,14 Brian J. Bolwell,15 E. Randolph Broun,16 David C. Delgado,17 Katharine Duckworth,18 Christopher C. Dvorak,19 Theresa E. Hahn,20 Ann E. Haight,21 Parameswaran N. Hari,22 Brandon M. Hayes-Lattin,23 David A. Jacobsohn,24 Ann A. Jakubowski,25 Kimberly A. Kasow,26 Hillard M. Lazarus,27 Jane L. Liesveld,28 Michael Linenberger,29 Mark R. Litzow,30 Walter Longo,31 Margarida Magalhaes-Silverman,32 John M. McCarty,33 Joseph P. McGuirk,34 Shahram Mori,35 Vinod Parameswaran,36 Vinod K. Prasad,37 Scott D. Rowley,38 Witold B. Rybka,39 Indira Sahdev,40 Jeffrey R. Schriber,41 George B. Selby,42 Paul J. Shaughnessy,43 Shalini Shenoy,44 Thomas Spitzer,45 William T. Tse,46 Joseph P. Uberti,47 Madhuri Vusirikala,48 Edmund K. Waller,49 Daniel J. Weisdorf,50 Gregory A. Yanik,51 Willis H. Navarro,3 Mary M. Horowitz,2 Galen E. Switzer,52 Dennis L. Confer3,7 and Bronwen E. Shaw2

1 Children's Hospital Los Angeles, CA; 2Center for International Blood and Marrow Transplant Research, Medical College of Wisconsin, Milwaukee, WI; 3Center for International Blood and Marrow Transplant Research, Division of Biostatistics, Minneapolis, MN; 4University of North Carolina Hospitals, Division of Hematology and Oncology Chapel Hill, NC; 5Department of Stem Cell Transplantation and Cell Transplantation and Cellular Therapy, Division of Cancer Medicine, MD Anderson Cancer Center, Houston, TX; 6Department of Transfusion Medicine, National Institutes of Health Clinical Center, Bethesda, MD; 7National Marrow Donor Program/Be The Match, Minneapolis, MN; 8Department of Pediatrics and Human Development, Helen DeVos Children's Hospital, Grand Rapids, MI; 9Department of Hematology and Oncology, Children's Mercy Hospitals and Clinics, Kansas City, MO; 10 Indiana Blood and Marrow Transplantation, Indianapolis, IN; 11University of Chicago Hospitals, IL; 12University of California, San Diego Medical Center, La Jolla, CA; 13North Shore University Hospital, Manhasset, NY; 14University of Mississippi Medical Center, Jackson, MS; 15 Taussig Cancer Institute, Cleveland Clinic Foundation, OH; 16Jewish Hospital, Cincinnati, OH; 17Indiana University Hospital/Riley Hospital for Children, IN; 18Wake Forest University Baptist Medical Center, Winston-Salem, NC; 19Division of Pediatric Blood and Marrow Transplantation, University of California San Francisco Benioff Children’s Hospital, CA; 20 Roswell Park Cancer Institute, Buffalo, NY; 21Aflac Cancer and Blood Disorders Center, Division of Hematology/Oncology-Bone Marrow Pediatric Hematology & Medical Oncology, Children’s Healthcare of Atlanta and Emory University School of Medicine, Atlanta, GA; 22 Department of Medicine, Medical College of Wisconsin, Milwaukee, WI; 23Doernbecher Children's Hospital, Oregon Health & Science University, Portland, OR; 24Children's National Medical Center, Washington DC; 25Memorial Sloan Kettering Cancer Center – Adult, New York, NY; 26Pediatric Hematology Oncology Program, Bone Marrow and Stem Cell Transplantation Program, University of North Carolina Healthcare, Chapel Hill, NC; 27 Seidman Cancer Center-University Hospitals Cleveland Medical Center, OH; 28Strong Memorial Hospital - University of Rochester Medical Center, NY; 29Division of Hematology, Fred Hutchinson Cancer Research Center, Seattle, WA; 30Division of Hematology, Mayo Clinic Rochester, MN; 31University of Wisconsin Hospital and Clinics, Madison, WI; 32Blood and Marrow Transplant Program, University of Iowa Hospitals & Clinics, Iowa City, IA; 33Virginia Commonwealth University Massey Cancer Center Bone Marrow Transplant Program, Richmond, VA; 34University of Kansas, Westwood, KS; 35Florida Hospital Cancer Institute, Florida Center for Cellular Therapy, Orlando, FL; 36Avera McKennan Transplant Institute, Sioux Falls, SD; 37Duke Cancer Institute, Duke University Medical Center, Durham, NC; 38 Hackensack University Medical Center, Washington DC; 39Penn State Health Milton S. Hershey Medical Center, PA; 40Cohen Children’s Medical Center of New York, New Hyde Park, NY; 41Cancer Transplant Institute Honor Health, Scottsdale, AZ; 42HCA Health Services of Oklahoma, Inc., University of Oklahoma, Oklahoma City, OK; 43Texas Transplant Institute, San Antonio, TX; 44Division of Hematology/Oncology, St. Louis Children's Hospital, MO; 45 Massachusetts General Hospital, Boston, MA; 46Ann & Robert H. Lurie Children's Hospital of Chicago, IL; 47Karmanos Cancer Institute, Detroit, MI; 48Division of Hematology/Oncology, University of Texas Southwestern Medical Center, Dallas, TX; 49Emory University Hospital, Atlanta, GA; 50University of Minnesota Medical Center, Fairview, Minneapolis, MN; 51The University of Michigan, Ann Arbor, MI and 52Division of General Internal Medicine, University of Pittsburgh, PA, USA

haematologica | 2019; 104(4)

Toxicity in related versus unrelated HSC donors

ABSTRACT

nlike unrelated donor registries, transplant centers lack uniform approaches to related donor assessment and deferral. To test whether related donors are at increased risk for donation-related toxicities, we conducted a prospective observational trial of 11,942 related and unrelated donors aged 18-60 years. Bone marrow (BM) was collected at 37 transplant and 78 National Marrow Donor Program centers, and peripheral blood stem cells (PBSC) were collected at 42 transplant and 87 unrelated donor centers in North America. Possible presence of medical comorbidities was verified prior to donation, and standardized pain and toxicity measures were assessed pre-donation, peri-donation, and one year following. Multivariate analyses showed similar experiences for BM collection in related and unrelated donors; however, related stem cell donors had increased risk of moderate [odds ratios (ORs) 1.42; P<0.001] and severe (OR 8.91; P<0.001) pain and toxicities (OR 1.84; P<0.001) with collection. Related stem cell donors were at increased risk of persistent toxicities (OR 1.56; P=0.021) and non-recovery from pain (OR 1.42; P=0.001) at one year. Related donors with more significant comorbidities were at especially high risk for grade 2-4 pain (OR 3.43; P<0.001) and non-recovery from toxicities (OR 3.71; P<0.001) at one year. Related donors with more significant comorbidities were at especially high risk for grade 2-4 pain (OR 3.43; P<0.001) and non-recovery from toxicities (OR 3.71; P<0.001) at one year. Related donors reporting grade â&#x2030;Ľ2 pain had significant decreases in Health-Related Quality of Life (HR-QoL) scores at one month and one year post donation (P=0.004). In conclusion, related PBSC donors with comorbidities are at increased risk for pain, toxicity, and non-recovery at one year after donation. Risk profiles described in this study should be used for donor education, planning studies to improve the related donor experience, and decisions regarding donor deferral. Registered at clinicaltrials.gov identifier: 00948636.

Introduction

Methods

Donation of hematopoietic stem cells (HSC) in the form of bone marrow (BM) or peripheral blood stem cells (PBSC) is a commonly performed procedure, with more than 40,000 donations from both volunteer unrelated donors (URD) and related donors (RD) each year.1,2 Over the past decade, donor registries such as the National Marrow Donor Program (NMDP) have published detailed data describing the URD experience, identifying individuals at increased risk for pain and collection-related symptoms, slower recovery, and severe adverse events.3-7 Data describing the RD experience, however, are limited, with only one large recent study.8 This may be because URD are handled by registries that have a mandate to collect and report safety data, whereas RD are cared for by local transplant centers, whose primary focus is care for the recipient and, in most countries, RD safety data are not systematically collected. Inadequate data regarding RD is a cause for concern for many reasons. While URD registries have rigorous standards for donor approval, supported by internal quality initiatives and efforts at international standardization,9 there are no generally accepted guidelines about deferral of a RD. With these concerns in mind, North American Investigators teamed with the National Marrow Donor Program (NMDP) and the Center for International Blood and Marrow Transplant Research (CIBMTR) to conduct a prospective observational trial of RD who donated at 53 transplant centers in the United States between January 2010 and July 2014. This report describes our primary end point comparing pain, toxicities and recovery of RD with URD collected concurrently at 78 BM and 87 PBSC NMDP collection centers.

Prior to donation, RD underwent a medical evaluation including a detailed history, physical examination, blood tests, and additional work up as necessary according to center standards. RD approved for donation were approached for consent for this Institutional Review Board (IRB)-approved study. Unrelated donors provided written informed consent for participation as required by the NMDP IRB. URD were evaluated for medical suitability and comorbidities that would require further evaluation or qualify for deferral for BM or PBSC donation as specified by NMDP standards.10,11

haematologica | 2019; 104(4)

Data collection A pre-donation form including history of pre-existing medical conditions (comorbidities) was completed. Detailed collectionrelated symptoms and pain were collected at five time points: predonation, peri-donation [day +5 from start of granulocyte-colony stimulating factor (G-CSF) for PBSCs and 1-2 days after BM collection], and 1, 6 and 12 months post donation. Toxicity was defined by Common Toxicity Criteria measures for symptoms commonly noted during PBSC and BM collection (fever, fatigue, skin rash, local reactions to an injection, nausea, vomiting, anorexia, insomnia, dizziness, and syncope) and is called the Modified Toxicity Criteria (MTC). This approach was validated by the NMDP and has been published previously.4,5,12,13 Pain was graded from 0-4 as none, mild, moderate, severe, or disabling. Pain and toxicity measures were assessed by the transplant center at pre- and peri-donation time points; the CIBMTR Survey Research Group was responsible for follow-up assessments. A product-specific collection form detailed information on the collection procedure. A subset of donors underwent assessment of long-term psychological recovery by established Health Related Quality of Life instruments (reported previously14-16).

845

M.A. Pulsipher et al. Table 1. Demographics and collection characteristics of first-time bone marrow (BM) donors and first time peripheral blood stem cell (PBSC) donors.

Variable Number of donors Number of centers Age at donation (years) 18 to 29 30 to 39 40 to 49 50 to 60 Median (range) Gender Male Female Weight (kg) N Eval Median (range) BMI (kg/m2) Underweight, <18.5 Normal, 18.5-24.9 Overweight, 25-29.9 Obese, 30+ Unknown Race Caucasian Hispanic African / African American Asian / Pacific Islander Native American Multiple races / other Unknown / declined Comorbidity groupb Comorbidities absent Comorbidities present, acceptable Comorbidities present, indeterminate Comorbidities present, defer Unknown Year of donation 2010 2011 2012 2013 PBSC collection-related Number of days of collection 1 2 3 4 5 Average daily G-CSF dose ( g) N Eval Median (range)

RD N (%)

First-time BM donors URD N (%)

126 37

2553 78

54 (43) 22 (17) 21 (17) 29 (23) 33 (18-61)

1231 (48) 702 (27) 455 (18) 165 (06) 30 (19-60)

67 (53) 59 (47)

1558 (61) 995 (39)

126 83 (50-150)

2553 81 (40-154)

0 40 (33) 34 (28) 47 (39) 5 (N/A)

14 (01) 867 (34) 938 (37) 734 (29) 0 (N/A)

89 (71) 15 (12) 15 (12) 4 (03) 3 (02) 0 0

1637 (64) 332 (13) 188 (07) 146 (06) 21 (01) 208 (08) 21 (01)

RD N (%)

First-time PBSC donors URD N (%)

956 42

8307 87

102 (11) 135 (14) 254 (27) 465 (49) 49 (18-61)

4011 (48) 2091 (25) 1520 (18) 685 (08) 30 (18-61)

537 (56) 419 (44)

5341 (64) 2966 (36)

952 86 (43-198)

8307 82 (37-176)

3 (<1) 209 (23) 325 (36) 368 (41) 51 (N/A)

66 (01) 2884 (35) 3064 (37) 2288 (28) 5 (N/A)

788 (82) 63 (07) 58 (06) 29 (03) 6 (01) 7 (01) 5 (01)

6132 (74) 698 (08) 314 (04) 444 (05) 62 (01) 592 (07) 65 (01)

<0.001

0.070 0.078

0.330 0.069

<0.001

0.003

64 (51) 22 (17) 29 (23) 11 (09) 0 (N/A)

<0.001 <0.001

<0.001

403 (42) 171 (18) 297 (31) 85 (09) 0 (N/A) 0.001

21 (17) 38 (30) 51 (40) 16 (13)

496 (19) 622 (24) 757 (30) 678 (27)

<0.001 113 (12) 301 (31) 375 (39) 167 (17)

1604 (19) 1977 (24) 2370 (29) 2356 (28)

646 (68) 283 (30) 16 (02) 10 (01) 1 (<1)

7478 (90) 829 (10) 0 0 0

918 948 (300-2040)

8183 900 (420-1260)

<0.001

continued on the next page

846

haematologica | 2019; 104(4)

Toxicity in related versus unrelated HSC donors continued from the previous page

Average daily G-CSF dose per donor weight ( g/kg/day) N Eval Median (range) Absolute CD34+ at pre-collection N Eval Median (range) Total blood volume processed Small, <12 L Standard, 12-18 L Large, â&#x2030;Ľ 18 L Unknown Median (range) Central line placement-Male No Yes Unknown Central line placement-Female No Yes Central line site-All donors Femoral Internal jugular Subclavian Other site

914 10.3 (4.7-22.1)

8183 10.6 (5.5-18.7)

539 87.0 (7.1-1342)

8283 80.0 (0.3-2123)

54 (06) 325 (34) 576 (60) 1 (N/A) 20.0 (0.3-112.6)

233 (03) 1882 (23) 6182 (75) 10 (N/A) 21.2 (0.7-45.4)

470 (88) 67 (12) 0 (N/A)

5219 (98) 121 ( 2) 1 (N/A)

259 (62) 160 (38)

2460 (83) 506 (17)

8 ( 4) 208 (92) 11 ( 5) 0

209 (33) 389 (62) 26 ( 4) 3 (<1)

<0.001

<0.001 <0.001

0.040 <0.001

<0.001

RD: related donor; URD: unrelated donor; N: number; N Eval: number evaluated; BMI: Body Mass Index; G-CSF: granulocyte-colony stimulating factor. aPearson c2 test was used for comparing discrete variables; the Kruskal-Wallis test was used for comparing continuous variables. bApplicable to RD only.

End points Pain was assessed for the following sites: back, bones, head, hip, intravenous injection (IV) site, joints, limbs, muscles, neck, throat, or other. Severity of pain was defined as the maximum grade among these pain sites. Body symptoms were assessed using the MTC outlined above and the peak toxicity level across symptoms was analyzed. Recovery to pre-donation levels by one year was defined as a pain or symptom score less than or equal to the score at pre-donation. Pre-donation comorbidity ascertainment included: assessment of bleeding, gastrointestinal, genitourinary, hematologic, hepatic, pulmonary, cardiovascular, psychiatric, central nervous system (CNS), endocrine, autoimmune disorders, or other significant coexisting diseases (Online Supplementary Table S1). We divided comorbidities into three categories: 1) comorbidities that would not result in deferral from URD donation according to NMDP standards;10,11 2) comorbidities that would have resulted in deferral; and 3) comorbidities that could possibly have led to a deferral, but more detailed donor clinical data would be needed to make that judgment.

Statistical analysis Analyses were conducted separately for BM and PB donations. Pre-donation baseline variables were compared between RD and URD groups using the Pearson c2 test for categorical variables and the Kruskal-Wallis test for continuous variables. c2 tests or Fisherâ&#x20AC;&#x2122;s Exact tests as appropriate were used to compare the incidences of skeletal pain and MTC symptoms as well as recovery to pre-donation levels between RD and URD groups. Multivariate analyses using logistic regression models were conducted to compare the RD and URD groups accounting for differhaematologica | 2019; 104(4)

ences in donor characteristics. The following donor characteristics were examined for inclusion in the multivariate model: donor type, race, gender, age, Body Mass Index (BMI), collection year, comorbidity status among related donors, pre-donation counts [white blood cell (WBC) count, platelets, neutrophils, mononuclear cells, hemoglobin], and pre-donation symptoms (skeletal pain or maximum MTC grade). Additional PB donation-specific variables considered were: placement of a central venous line, total blood volume, absolute CD34+ cells and WBC pre-collection, and daily GCSF dose (absolute and per kg). The effects were estimated via odds ratios (OR). In all multivariate models, donor type was forced into the model and stepwise model selection was used to determine additional donor characteristics to be included. Interactions between donor type and each donor characteristic were tested for in all multivariate models.

Results Demographics Table 1 details RD and URD donating BM or PBSC. RD tended to be older than URD, with 23% versus 6% of BM donors and 49% versus 8% of PBSC donors collected aged between 50-60 years. Although males donated more often in both RD and URD groups, a higher percentage of females donated in the RD group. There was a trend toward higher BMI in RD versus URD giving BM (BMI 30+, 39% versus 29%; P=0.07), and a significant difference in obesity in RD versus URD giving PBSC with 41% versus 28% (BMI 30+; P<0.001). There are several notable differences between RD and 847

M.A. Pulsipher et al.

URD involving collection procedures (Table 1). While 90% of URD PBSC donations occurred in a single day, and no collection took more than two days, 30% of RD required two days, and 2% and 1% took three and four days, respectively, with collection for one donor taking place over five days (P<0.001). Notably, more RD were collected with lower volume procedures (<18L, 40% vs. 26%; P<0.001). A major difference in RD versus URD practice was noted in the increase in central venous line placement in RD for both female and male donors (female RD 38%, female URD 17%, P<0.001; male RD 12%, male URD 2%, P=0.001). Of note, the differences were not impacted by age, although obesity had an impact in female donors, and number of collection procedures performed impacted both male and female donors (Online Supplementary Table S2).

Univariate analyses of bone marrow collection, pain and donation-related symptoms Figure 1A and B show rates of grades 1-4 skeletal pain and collection-related symptoms in RD and URD before, peri-

donation, and one year after the BM collection procedure. Online Supplementary Figure S1A-D detail locations of pain and types of symptoms experienced. It is notable that 510% of healthy URDs and 10-20% of RDs reported mild pain or symptoms pre-donation. Almost all donors reported some level of pain or symptoms during the procedure; however, because grade 1 pain and symptoms rarely require intervention, we focused our analyses on higher grades. Univariate analyses showed RD to have higher rates of grade 2-4 pain pre-donation (2.4% vs. 0.6%; P=0.043) (Online Supplementary Table S3). Grade 2-4 pain levels at collection were similar, but grades 3-4 pain were substantially higher in RD (10% vs. 0.6%; P<0.001). At one year, 10% and 5% of RD versus URD reported grade 2-4 pain (P=0.060). Pre-donation, MTC symptoms were similar in RD and URD. At collection, grade 2-4 and 3-4 symptoms were higher in RD versus URD (24% vs. 17% grade 2-4, 3.3% vs. 0.4% grade 3-4; P=0.049 and 0.002, respectively), with higher rates of dizziness, site reactions, nausea, and syncope in RD (P=0.005, 0.008, 0.021, and 0.028, respectively) (Online Supplementary Figure S1C).

Figure 1. Severity of skeletal pain and highest toxicity level across key body symptoms experienced by first time related versus unrelated bone marrow donors at baseline, two days post donation, and one year post donation. (A) Skeletal pain. (B) Highest toxicity level across key body symptoms.

848

haematologica | 2019; 104(4)

Toxicity in related versus unrelated HSC donors

Univariate analyses of peripheral blood stem cell collection, pain and donation-related symptoms Figure 2A and B show rates of grades 1-4 skeletal pain and MTC symptoms in RD and URD before, on day +5 of G-CSF administration (day of peak symptoms), and one year after the PBSC collection procedure. Online Supplementary Figure S2A-D detail locations of pain and types of symptoms experienced by RD and URD undergoing PBSC collection. Online Supplementary Table S4 shows that at pre-donation baseline, day +5 of G-CSF, and one year, all measures of grade 2-4 and 3-4 pain are higher in RD compared to URD (all P<0.001). In addition, 10% fewer RD return to pre-donation levels of pain at one year (P<0.001). Collection-related MTC symptoms are also experienced significantly more and to a higher degree at all time points, and non-recovery to pre-donation levels of these symptoms at one year occurs more often after RD procedures (17% vs. 12%; P<0.001).

Multivariate analyses of bone marrow and peripheral blood donor experiences: related versus unrelated donor Multivariate analysis showed that Grade 2-4 pain after BM collection was similar between RD and URD (Table 2). Grade 2-4 symptoms after BM collection were 1.5 times more likely for RD, but this did not reach significance (P=0.075). Related PBSC donors were at higher risk for grade 2-4 and 3-4 pain (OR 1.42, 8.91, respectively; both P<0.001) and grade 2-4 symptoms (OR 1.84; P<0.001) with collection, as well as the presence of grade 2-4 symptoms at one year (OR 1.56; P=0.021). A notable finding was that RD reporting no comorbidities had a risk of grade 2-4 pain at one year similar to URD. But if RD reported any comorbidities, their risk of grade 2-4 pain was significantly increased, with the highest risk noted in

RD with comorbidities that would have led to deferral by NMDP standards (OR 3.43; P<0.001). Table 2 also describes analyses of failure to recover to pre-donation levels of pain and donation-related symptoms at one year. RD of PBSC had an OR of 1.42 for nonrecovery to pre-donation levels of pain at one year (P=0.001). Recovery to pre-donation levels of symptoms was associated with comorbidity: RD who had no comorbidities were similar to URD, but RD who had comorbidities had a higher risk of non-recovery at one year. Notably, RD identified as having comorbidities that would have led to NMDP deferral had a more than 3-fold increase in risk of non-recovery to pre-donation levels compared to URD (OR 3.71; P<0.001).

Multivariate analysis: other factors affecting risk of pain, symptoms, or non-recovery at one year For BM donation, women were 67% more likely to experience grade 2-4 pain and nearly 3 times more likely to experience grade 2-4 symptoms (P<0.001) (Table 3). Age was an important risk factor for failure to recover to pre-donation levels, as donors aged 50-60 years were more than twice as likely as their younger counterparts to have non-recovery at one year (Table 4). In addition, womenâ&#x20AC;&#x2122;s risk of non-recovery at one year to pre-donation levels of symptoms was twice that of men (P<0.001). Risk factors for pain and MTC symptoms after PBSC collection included both new and previously described clinical characteristics (Tables 3 and 4). A new finding is that high CD34+ counts (â&#x2030;Ľ80.5/mL) prior to day 1 of collection was associated with more grade 2-4 collection pain (OR 1.25; P<0.001). Another novel finding was that there was a dose level of G-CSF above which pain levels increased significantly. If a donor received an average daily dose exceeding 960 mg/day, reported pain levels were

Table 2. Multivariate analysis of collection toxicities and long-term recovery after bone marrow (BM) and peripheral blood stem cell (PBSC) donations showing the effect of donor type. The odds ratios are for comparing related donor (RD) versus unrelated donor (URD).

Event and time point Skeletal pain Grade 2-4 at collection Grade 3-4 at collectiona Grade 2-4 at one yeara RD, comorbidities absent RD, comorbidities present, acceptable RD, comorbidities present, indeterminate RD, comorbidities present, defer Non-recovery to pre-donation level at one year Max MTC symptoms Grade 2-4 at collection Grade 2-4 at one yeara Non-recovery to pre-donation level at one year RD, comorbidities absent RD, comorbidities present, acceptable RD, comorbidities present, indeterminate RD, comorbidities present, defer

BM donors OR (95% CI)

1.19 (0.81-1.74)

0.375

0.96 (0.56-1.63)

0.871

1.50 (0.96-2.36)

0.075

1.08 (0.58-2.01)

0.819

PBSC donors OR (95% CI) 1.42 (1.17-1.72) 8.91 (6.63-12.0) 1.02 (0.63-1.65) 2.66 (1.60-4.42) 1.62 (1.04-2.52) 3.43 (1.86-6.35) 1.42 (1.15-1.74) 1.84 (1.49-2.28) 1.56 (1.07-2.27) 0.94 (0.64-1.37) 1.98 (1.26-3.12) 1.77 (1.24-2.52) 3.71 (2.14-6.45)

P <0.001 <0.001 <0.001 0.944 <0.001 0.033 <0.001 0.001 <0.001 0.021 <0.001 0.743 0.003 0.002 <0.001

OR: odds ratio; CI: Confidence Interval; Max: maximum; MTC: Modified Toxicity Criteria. aNumber of events for BM donors was not sufficient for multivariable analysis.

haematologica | 2019; 104(4)

849

M.A. Pulsipher et al.

higher (OR 1.21; P=0.006). This increase in pain levels was not noted when analyzed by dose per/kg. Females undergoing PBSC collection had twice the risk of moderate and severe pain and MTC symptoms (Table 3). They were also more likely to have persistent symptoms and to fail to recover to pre-donation levels at one year after BM collection (Table 4). The age effect varied, with PBSC donors aged 30-39 years having higher risk for grade 3-4 pain with collection (OR 1.50; P=0.021) (Table 3) and older donors (aged 50-60 years) having lower risks for

grade 2-4 pain with collection (OR 0.61; P<0.00) (Table 3) and higher risks of reporting grade 2-4 pain at one year (aged 50-60 years: OR 2.72; P<0.001) (Table 4). Importantly, donors who started with grade 1 or 2-4 pain or grade 2-4 MTC symptoms were more likely to report higher grades of pain or symptoms with collection (P<0.001) (Table 3). These donors also had higher levels of pain and MTC symptoms at one year, but most of them had returned to pre-donation levels. Obesity was important in pain and MTC symptom risk, as donors with 30+

Table 3. Multivariate analysis showing other key predictors for skeletal pain and Modified Toxicity Criteria (MTC) symptoms associated with bone marrow (BM) and peripheral blood stem cell (PBSC) collections.

Variable BM donation Gender Male Female PBSC donation Gender Male Female Age at collection (years) 18 to 29 30 to 39 40 to 49 50 to 60 Donor BMI Underweight / normal Overweight Obese Unknown Number of days of collection 1 day 2+ days G-CSF dose 0-960 g/day >960 g/day Unknown Skeletal pain pre-donation Grade 0 Grade 1 Grade 2-4 Max MTC pre-donation Grade 0 Grade 1 Grade 2-4 Absolute CD34+ at pre-collection <80.5 â&#x2030;Ľ80.5 Unknown

Grade 2-4 skeletal pain at collection OR (95% CI) P

1.00 1.67 (1.41-1.97)

1.00 1.76 (1.60-1.93) 1.00 1.00 (0.90-1.12) 0.85 (0.75-0.96) 0.61 (0.52-0.71) 1.00 1.13 (1.02-1.26) 1.35 (1.18-1.55) 0.92 (0.52-1.63) 1.00 0.74 (0.64-0.85) 1.00 1.21 (1.06-1.38) 1.37 (0.96-1.97) 1.00 1.57 (1.36-1.82) 2.47 (1.77-3.46)

Grade 3-4 skeletal pain at collection OR (95% CI) P

1.00 2.74 (2.23-3.37)

<0.001

<0.001 <0.001 0.977 0.008 <0.001 <0.001 0.021 <0.001 0.776

Grade 2-4 max MTC symptoms at collection OR (95% CI) P

1.00 2.18 (1.69-2.81) 1.00 1.50 (1.06-2.12) 1.19 (0.83-1.72) 0.77 (0.51-1.16) 1.00 1.30 (0.92-1.84) 2.03 (1.47-2.82) 1.21 (0.48-3.05)

<0.001 0.005 0.021 0.340 0.216 <0.001 0.137 <0.001 0.686

1.00 1.91 (1.65-2.20) 1.00 1.18 (0.98-1.41) 1.26 (1.05-1.53) 0.91 (0.72-1.16) 1.00 1.31 (1.10-1.57) 1.63 (1.36-1.96) 1.11 (0.50-2.45)

<0.001

<0.001 0.013 0.080 0.015 0.452 <0.001 0.003 <0.001 0.806

<0.001 0.007 0.006 0.086 <0.001 <0.001 <0.001

0.004 1.00 1.39 (0.95-2.02) 2.33 (1.37-3.96) 1.00 0.85 (0.54-1.34) 2.67 (1.22-5.82)

0.087 0.002 0.030 0.477 0.014

0.006 1.00 1.31 (1.05-1.64) 1.65 (1.08-2.52) 1.00 1.37 (1.06-1.77) 3.23 (1.86-5.61)

0.016 0.020 <0.001 0.017 <0.001

<0.001 1.00 1.25 (1.14-1.37) 1.14 (0.87-1.48)

<0.001 0.341

OR: odds ratio; CI: Confidence Interval; BMI: Body Mass Index; G-CSF: granulocyte-colony stimulating factor; Max: maximum.

850

haematologica | 2019; 104(4)

Toxicity in related versus unrelated HSC donors

BMI had increased risk of peri-collection grade 2-4 and 34 pain and grade 2-4 MTC symptoms, along with grade 24 pain at one year. Table 4 shows additional factors other than RD/URD status associated with higher levels of late pain/MTC symptoms and lack of recovery to pre-donation levels at one year. Older BM and PBSC donors, and Black and multiple-race PBSC donors were less likely to recover to their pre-donation level of pain. Hispanic and multiple-race PBSC donors were less likely to recover to pre-donation level of MTC symptoms. As might be expected, donors with pre-donation levels of pain or symptoms at grade 1 or grades 2-4 were more likely to recover to that level at one year.

Discussion Unrelated HSC registries have a responsibility to ensure the safety of volunteer donors performing an altruistic act.17 They routinely defer donors with minor health problems, erring on the side of safety. Transplant centers, whose primary task is treatment of patients with cancer and other life-threatening illnesses, must also evaluate the medical fitness of donors and advise them about risk, in

some cases deferring them. Although recent changes in accreditation requirements for transplant centers emphasize donor education and autonomy, requiring an independent donor advocate,18-20 RD may or may not listen to advice to forgo donation, being highly motivated and willing to take medical risks for their family member. Studies have shown that a matched sibling is generally the best HSC donor;21-23 and recent expansion of haploidentical approaches24 have put even more family members into a donor role. Over the past decade, however, improvements in URD procedures have led to comparable outcomes using RD and URD in patients with hematologic malignancies,25-27 offering reasonable HCT alternatives if a RD is unable to donate. With this in mind, when should a transplant center counsel a RD against donation? Our study shows that the choice to donate by a RD with comorbidities can have consequences. We show by multivariate analysis that RD have more intense early pain and toxicities than URD, and because these symptoms are temporally associated with PBSC collection, there is little doubt that the toxicities are related to the donation procedure. There is a question, however, about whether our observation that RD have more pain and non-recovery to pre-donation levels at one year is due to the procedure itself, or other aspects associated with being a RD.

Figure 2. Severity of skeletal pain and highest toxicity level across key body symptoms experienced by first time related versus unrelated peripheral blood stem cell (PBSC) donors at baseline, on the first day of collection prior to apheresis, and one year post donation. (A) Skeletal pain. (B) Highest toxicity level across key body symptoms.

haematologica | 2019; 104(4)

851

M.A. Pulsipher et al. Table 4. Multivariate analysis showing other key predictors for long-term skeletal pain and Modified Toxicity Criteria (MTC) symptoms and recovery to pre-donation level after bone marrow (BM) and peripheral blood stem cells (PBSC) donations.

Variable

Grade 2-4 skeletal pain at one year OR (95% CI) P

BM donation Gender Male Female Age at collection 18 to 29 30 to 39 40 to 49 50 to 59 Skeletal pain pre-donation Grade 0 Grade 1-4 Max MTC symptoms pre-donation Grade 0 Grade 1-4 PBSC donation Gender Male Female Age at collection (years) 18 to 29 1.00 30 to 39 1.81 (1.19-2.74) 40 to 49 1.85 (1.22-2.82) 50 to 60 2.72 (1.77-4.16) Donor BMI Underweight / normal 1.00 Overweight 1.36 (0.95-1.95) Obese 1.79 (1.25-2.57) Unknown 0.96 (0.31-2.94) Donor race Caucasian 1.00 Hispanic 1.45 (0.90-2.33) African / African American 2.91 (1.74-4.84) Asian / Pacific Islander 2.18 (1.26-3.77) Multiple races / other 1.88 (1.09-3.22) Year of collection 2010-2011 2012-2013 Skeletal pain pre-donation Grade 0 1.00 Grade 1 3.17 (2.27-4.43) Grade 2-4 3.09 (1.71-5.60) Max MTC symptoms pre-donation Grade 0 1.00 Grade 1 0.74 (0.46-1.20) Grade 2-4 3.04 (1.10-8.42)

Grade 2-4 max MTC symptoms at one year OR (95% CI) P

Skeletal pain non-recovery to pre-donation level at one year OR (95% CI) P

Max MTC symptoms non-recovery to pre-donation level at one year OR (95% CI) P

1.00 1.92 (1.35-2.74)

<0.001

1.00 0.29 (0.11-0.72)

0.008

0.012 1.00 1.13 (0.78-1.63) 1.42 (0.95-2.12) 2.21 (1.34-3.64)

0.521 0.091 0.002

1.00 0.39 (0.20-0.74)

0.004

1.00 1.61 (1.15-2.25) 0.005 <0.001

<0.001 1.00 1.50 (1.21-1.86) <0.001 1.78 (1.43-2.22) <0.001 2.18 (1.71-2.78) <0.001

0.005 0.004 <0.001 0.013 0.097 0.002 0.943 <0.001

0.001 1.00 1.14 (0.85-1.54) 0.377 2.05 (1.40-3.00) <0.001 1.22 (0.87-1.71) 0.243 1.52 (1.11-2.08) 0.009

0.127 <0.001 0.006 0.023

0.015 1.00 1.54 (1.09-2.18) 1.42 (0.87-2.32) 1.25 (0.81-1.91) 1.62 (1.11-2.38) 1.00 0.74 (0.61-0.91)

<0.001 <0.001 <0.001 0.040 0.225 0.033

0.044 1.00 1.69 (1.04-2.72) 0.033 1.94 (0.86-4.34) 0.109 0.003 1.00 1.00 (0.55-1.82) 0.998 5.28 (2.04-13.60) 0.001

<0.001 1.00 0.53 (0.40-0.71) <0.001 0.22 (0.11-0.45) <0.001

1.00 1.67 (1.23-2.27) 1.35 (0.71-2.59) 1.00 0.27 (0.15-0.47) 0.17 (0.02-1.31)

0.014 0.166 0.310 0.013

0.004 0.004 0.001 0.362 <0.001 <0.001 0.089

OR: odds ratio; CI: Confidence Interval; BMI: Body Mass Index; Max: maximum.

852

haematologica | 2019; 104(4)

Toxicity in related versus unrelated HSC donors

Although attempting to link observed pain at one year directly to PBSC donation was not one of our objectives, an observation that we made may shed light on this question. We performed additional assessments of RD at one and six months; we noted that the donation pain levels do not fully recover at one month, and remain at heightened levels at six and 12 months (P>0.001) (Online Supplementary Figure S3), suggesting that persistently elevated pain levels are a consequence of donation. Given that elevated post-donation pain levels were only grade 1 or 2 (mild/moderate), are these findings clinically significant? This is an important question because selfreported pain from individuals can vary based on characteristics such as gender or cultural differences. To define whether the persistent pain we detected was clinically meaningful, we performed an analysis of the relation of reported pain to donor HR-QoL. In a companion study imbedded in our protocol, 186 RD and URD were randomly chosen for assessment of HR-QoL. We noted that at one month and one year after donation, those reporting grade 2 pain or toxicities had significantly lower physical scores measured by the SF36 multidimensional HR-QoL measure compared to those not reporting pain or toxicities [P=0.002 (1 month) and 0.004 (1 year)] (Online Supplementary Figure S4). The findings of our HR-QoL companion study (reported separately) support the outcomes we report (e.g. RD reported the donation to be more painful than URD; at 1 year RD were less likely to feel back to normal and reported a longer period of recovery). These observations allow us to conclude that the persistent pain is clinically meaningful, but the cause of higher levels of persistent pain in RD is unclear. Future studies could explore potential contributing factors, such as the possibility of G-CSF increasing inflammation in donors with comorbid conditions or the relationship of persistent pain to psychological stressors experienced by family donors. With this in mind, should RD at highest risk of pain or non-recovery consider deferral? Deferring RD who would have been deferred by the NMDP (our highest risk group) is in line with a recent Worldwide Network for Blood and Marrow Transplantation task force recommendation to screen RD using URD registry standards.9,28 It is likely that unless transplant centers collecting RD accept limits on screening and collection similar to URD registries, RD will remain at higher risk for pain/toxicity and lack of recovery. But should pre-donation standards for deferral of a RD be similar to those for URD? Although there is no clear medical benefit from donation, there is evidence that both URD and RD may experience psychosocial benefits, including feelings of enhanced self-worth. For RD, there are additional benefits of alleviating the suffering or saving the life of a loved one and closer family relationships.29-31 While some family members may willingly accept increased medical risk in exchange for psychosocial benefits, others may hesitate and feel coerced by family obligations. Striking a balance is a challenge, as transplant centers should support the wishes of RD who are ambivalent about donation and protect those in whom donation could be a serious risk. But at the same time, RD should

haematologica | 2019; 104(4)

have the choice as to whether to shoulder some level of increased risk. This study identified a series of risk factors that could either motivate a transplant center to recommend against use of a given donor, or allow a donor with multiple risk factors to understand their risk and choose to forgo donation (Tables 3 and 4). A desired outcome from this study is to motivate transplant centers to test interventions aimed at minimizing discomfort or preventing persistent pain or symptoms experienced by high-risk RD. The data on risks presented herein should be shared with RD as part of their counseling regarding the donation process. In summary, this study showed for the first time that adult RD of PBSC are at increased risk for higher levels of pain and symptoms in the short-term after a collection procedure and one year later compared to URD. The presence of comorbidities in a prospective donor heightens this risk, and comorbidities in combination with other factors described in this study should be carefully considered as transplant teams and individuals make decisions regarding BM or PBSC donation. Funding The study was funded by R01 HL085707 through the NHLBI. Additional funding for MAP was provided by 2UG1HL069254 (NHLBI/NCI) and the Johnny Crisstopher Children’s Charitable Foundation St. Baldrick’s Consortium Grant. The CIBMTR is supported primarily by Public Health Service Grant/Cooperative Agreement 5U24CA076518 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 1U24HL138660 from NHLBI and NCI; a contract HHSH250201700006C with Health Resources and Services Administration (HRSA/DHHS); two Grants N00014-17-12388, N00014-17-1-2850 and N00014-18-1-2045 from the Office of Naval Research; and grants from Adaptive Biotechnologies; *Amgen, Inc.; Anonymous donation to the Medical College of Wisconsin; Astellas Pharma US; Atara Biotherapeutics, Inc.; Be the Match Foundation; *bluebird bio, Inc.; *Bristol Myers Squibb Oncology; *Celgene Corporation; *Chimerix, Inc.; *CytoSen Therapeutics, Inc.; Fred Hutchinson Cancer Research Center; Gamida Cell Ltd.; Gilead Sciences, Inc.; HistoGenetics, Inc.; Immucor; *Incyte Corporation; Janssen Scientific Affairs, LLC; *Jazz Pharmaceuticals, Inc.; Karius, Inc.; Karyopharm Therapeutics, Inc.; *Kite Pharma, Inc.; Medac, GmbH; *Mediware; The Medical College of Wisconsin; *Merck & Co, Inc.; *Mesoblast; MesoScale Diagnostics, Inc.; Millennium, the Takeda Oncology Co.; *Miltenyi Biotec, Inc.; Mundipharma EDO; National Marrow Donor Program; Novartis Pharmaceuticals Corporation; PCORI; *Pfizer, Inc; *Pharmacyclics, LLC; PIRCHE AG; *Sanofi Genzyme; *Seattle Genetics; Shire; Spectrum Pharmaceuticals, Inc.; St. Baldrick’s Foundation; Swedish Orphan Biovitrum, Inc.; *Takeda Oncology; and University of Minnesota. The views expressed in this article do not reflect the official policy or position of the National Institute of Health, the Department of the Navy, the Department of Defense, Health Resources and Services Administration (HRSA) or any other agency of the U.S. Government. *Corporate Members.

853

M.A. Pulsipher et al.

References 1. Gratwohl A, Baldomero H, Aljurf M, et al. Hematopoietic stem cell transplantation: a global perspective. JAMA. 2010; 303(16):1617-1624. 2. Passweg JR, Baldomero H, Bader P, et al. Hematopoietic stem cell transplantation in Europe 2014: more than 40 000 transplants annually. Bone Marrow Transplant. 2016; 51(6):786-792. 3. Pulsipher MA, Chitphakdithai P, Logan BR, et al. Donor, recipient, and transplant characteristics as risk factors after unrelated donor PBSC transplantation: beneficial effects of higher CD34+ cell dose. Blood. 2009;114(13):2606-2616. 4. Pulsipher MA, Chitphakdithai P, Logan BR, et al. Acute toxicities of unrelated bone marrow versus peripheral blood stem cell donation: results of a prospective trial from the National Marrow Donor Program. Blood. 2013;121(1):197-206. 5. Pulsipher MA, Chitphakdithai P, Logan BR, et al. Lower risk for serious adverse events and no increased risk for cancer after PBSC vs BM donation. Blood. 2014;123(23):36553663. 6. Lee MH, Jang JH, Min HJ, et al. Predictors of general discomfort, limitations in activities of daily living and intention of a second donation in unrelated hematopoietic stem cell donation. Bone Marrow Transplant. 2017;52(2):258-263. 7. Holig K, Kramer M, Kroschinsky F, et al. Safety and efficacy of hematopoietic stem cell collection from mobilized peripheral blood in unrelated volunteers: 12 years of single-center experience in 3928 donors. Blood. 2009;114(18):3757-3763. 8. Kodera Y, Yamamoto K, Harada M, et al. PBSC collection from family donors in Japan: a prospective survey. Bone Marrow Transplant. 2014;49(2):195-200. 9. Lown RN, Philippe J, Navarro W, et al. Unrelated adult stem cell donor medical suitability: recommendations from the World Marrow Donor Association Clinical Working Group Committee. Bone Marrow Transplant. 2014;49(7):880-886. 10. NMDP Standards. Available at: https:// bethematch.org/workarea/downloadasset.aspx?id=7711: National Marrow Donor Program; 2015. 11. Miller J. Hematopoietic Progenitor Cell Donation Evaluation. In: Wingard J, Gastineau DA, Leather H, Snyder DL, Szczepiorkowski ZM, eds. Hematopoietic Stem Cell Transplantation: A Handbood

854

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

For Clinicians. Bethesda, MD: AABB; 2015:93-108. Miller JP, Perry EH, Price TH, et al. Recovery and safety profiles of marrow and PBSC donors: experience of the National Marrow Donor Program. Biol Blood Marrow Transplant. 2008;14(9 Suppl):29-36. Pulsipher MA, Chitphakdithai P, Miller JP, et al. Adverse events among 2408 unrelated donors of peripheral blood stem cells: results of a prospective trial from the National Marrow Donor Program. Blood. 2009;113(15):3604-3611. Switzer GE, Bruce J, Kiefer DM, et al. Health-Related Quality of Life among Older Related Hematopoietic Stem Cell Donors (>60 Years) Is Equivalent to That of Younger Related Donors (18 to 60 Years): A Related Donor Safety Study. Biol Blood Marrow Transplant. 2017;23(1):165-171. Switzer GE, Bruce J, Kiefer DM, et al. Health-Related Quality of Life among Pediatric Hematopoietic Stem Cell Donors. J Pediatr. 2016;178:164-170. Switzer GE, Bruce J, Pastorek G, et al. Parent versus child donor perceptions of the bone marrow donation experience. Bone Marrow Transplant. 2017;52(9):13381341. Shaw BE, Ball L, Beksac M, et al. Donor safety: the role of the WMDA in ensuring the safety of volunteer unrelated donors: clinical and ethical considerations. Bone Marrow Transplant. 2010;45(5):832-838. van Walraven SM, Nicoloso-de Faveri G, Axdorph-Nygell UA, et al. Family donor care management: principles and recommendations. Bone Marrow Transplant. 2010;45(8):1269-1273. O'Donnell PV, Pedersen TL, Confer DL, et al. Practice patterns for evaluation, consent, and care of related donors and recipients at hematopoietic cell transplantation centers in the United States. Blood. 2010;115(24):5097-5101. Anthias C, Shaw BE, Kiefer DM, et al. Significant Improvements in the Practice Patterns of Adult Related Donor Care in US Transplantation Centers. Biol Blood Marrow Transplant. 2016;22(3):520-527. Shaw PJ, Kan F, Woo Ahn K, et al. Outcomes of pediatric bone marrow transplantation for leukemia and myelodysplasia using matched sibling, mismatched related, or matched unrelated donors. Blood. 2010;116(19):4007-4015. Weisdorf DJ, Anasetti C, Antin JH, et al. Allogeneic bone marrow transplantation

23.

24.

25.

26.

27.

28.

29.

30.

31.

for chronic myelogenous leukemia: comparative analysis of unrelated versus matched sibling donor transplantation. Blood. 2002;99(6):1971-1977. Hows JM, Passweg JR, Tichelli A, et al. Comparison of long-term outcomes after allogeneic hematopoietic stem cell transplantation from matched sibling and unrelated donors. Bone Marrow Transplant. 2006;38(12):799-805. Luznik L, O'Donnell PV, Symons HJ, et al. HLA-haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant. 2008;14(6):641-650. Saber W, Opie S, Rizzo JD, et al. Outcomes after matched unrelated donor versus identical sibling hematopoietic cell transplantation in adults with acute myelogenous leukemia. Blood. 2012;119(17):3908-3916. Majhail NS, Chitphakdithai P, Logan B, et al. Significant improvement in survival after unrelated donor hematopoietic cell transplantation in the recent era. Biol Blood Marrow Transplant. 2015;21(1):142-150. Bashey A, Zhang X, Sizemore CA, et al. Tcell-replete HLA-haploidentical hematopoietic transplantation for hematologic malignancies using post-transplantation cyclophosphamide results in outcomes equivalent to those of contemporaneous HLA-matched related and unrelated donor transplantation. J Clin Oncol. 2013;31(10):1310-1316. Worel N, Buser A, Greinix HT, et al. Suitability Criteria for Adult Related Donors: A Consensus Statement from the Worldwide Network for Blood and Marrow Transplantation Standing Committee on Donor Issues. Biol Blood Marrow Transplant. 2015;21(12):2052-2060. DiMartini A, Dew MA, Liu Q, et al. Social and Financial Outcomes of Living Liver Donation: A Prospective Investigation Within the Adult-to-Adult Living Donor Liver Transplantation Cohort Study 2 (A2ALL-2). Am J Transplant. 2017; 17(4):1081-1096. Dew M, Boneysteele G, DiMartini A. Unrelated Donors. In: Steel J, ed. Living Donor Advocacy: An Evolving Role Within Transplantation. New York: Springer; 2014:149-167. Simmons RG, Klein SD, Simmons RL. Gift of Life: The Social and Psychological Impact of Organ Transplantation. Am J Sociol. 1979;85(2):179-481.

haematologica | 2019; 104(4)

Obituary

Obituary Francesco Lo Coco, a distinguished hematologist and a friend Professor Francesco Lo Coco passed away on March 3rd. The hematology community has lost a dear friend, a respected colleague and an enlightened scientist. Francesco was a brilliant student of Franco Mandelli and made a substantial contribution to the field of translational and clinical research, especially the advancement of the treatment of acute promyelocytic leukemia and molecular diagnostics in acute leukemia. He was one of the leaders of the Italian Hemato-Oncology Cooperative Group GIMEMA, member of the Executive Board of the Italian Society of Hematology and President of the Italian Society of Experimental Hematology from 2000 to 2002. Prof. Lo Coco received much recognition and many awards for his work, in particular the JosĂŠ Carreras Award from the European Hematology Association in 2018. Francesco was truly a friend in the best sense. He was warm, sensitive, open-minded and helpful. His friends know that he was especially proud of his son Gaetano, a young orchestral conductor, with whom he enjoyed a particularly close relationship. On behalf of the SIE, SIES and GIMEMA: Prof. Paolo Corradini, President of SIE Prof. Pellegrino Musto, President of SIES Prof. Marco Vignetti, President of GIMEMA doi: 10.3324/haematol.2019.222356

haematologica | 2019; 104(4)

Francesco Lo Coco.

855