Haematologica, Volume 104, Issue 5

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 5: May 2019 Cover Figure Peripheral blood smear from a patient with primary myelofibrosis showing thrombocytosis with platelet anisocytosis and marked morphological alterations. Courtesy of Prof. Rosangela Invernizzi.

Editorials 857

Factor H: a novel modulator in sickle cell disease Wassim El Nemer and Bérengère Koehl

859

A post-transplant optimized transplant-specific risk score in myelodysplastic syndromes Mahasweta Gooptu and John Koreth

862

Precision tyrosine kinase inhibitor dosing in chronic myeloid leukemia?

Giuseppe Saglio et al.

864

Mogamulizumab versus investigator choice in relapsed/refractory adult T-cell leukemia/lymphoma: all for one or none for all?

William Johnson et al.

Perspective Article 868

Next-generation sequencing in the diagnosis and minimal residual disease assessment of acute myeloid leukemia Ross L. Levine and Peter J.M. Valk

Review Articles 872

Evolutionary trajectory of leukemic clones and its clinical implications Amos Tuval and Liran I Shlush

881

CRISPR to fix bad blood: a new tool in basic and clinical hematology Elisa González-Romero et al.

Articles Hematopoiesis

894

Long noncoding RNAs of single hematopoietic stem and progenitor cells in healthy and dysplastic human bone marrow Zhijie Wu et al.

Red Cell Biology & its Disorders

907

Finely-tuned regulation of AMP-activated protein kinase is crucial for human adult erythropoiesis Meriem Ladli et al.

919

Factor H interferes with the adhesion of sickle red cells to vascular endothelium: a novel disease-modulating molecule Elisabetta Lombardi et al.

Myelodysplastic Syndromes

929

Optimized EBMT transplant-specific risk score in myelodysplastic syndromes after allogeneic stem-cell transplantation Nico Gagelmann et al.

Myeloproliferative Neoplasms

937

Ruxolitinib in combination with prednisone and nilotinib exhibit synergistic effects in human cells lines and primary cells from myeloproliferative neoplasms Alicia Arenas Cortés et al.

947

EXPAND, a dose-finding study of ruxolitinib in patients with myelofibrosis and low platelet counts: 48-week follow-up analysis Alessandro M. Vannucchi et al.

Chronic Myeloid Leukemia

955

Imatinib dose reduction in major molecular response of chronic myeloid leukemia: results from the German Chronic Myeloid Leukemia-Study IV Christian Michel et al.

Acute Myeloid Leukemia Haematologica 2019; vol. 104 no. 5 - May 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation Acute Myeloid Leukemia

963

The thymidine dideoxynucleoside analog, alovudine, inhibits the mitochondrial DNA polymerase Îł, impairs oxidative phosphorylation and promotes monocytic differentiation in acute myeloid leukemia Dana Yehudai et al.

973

The small-molecule compound AC-73 targeting CD147 inhibits leukemic cell proliferation, induces autophagy and increases the chemotherapeutic sensitivity of acute myeloid leukemia cells Isabella Spinello et al.

986

Replacing cyclophosphamide/cytarabine/mercaptopurine with cyclophosphamide/etoposide during consolidation/delayed intensification does not improve outcome for pediatric B-cell acute lymphoblastic leukemia: a report from the COG Michael J. Burke et al.

Non-Hodgkin Lymphoma

993

Mogamulizumab versus investigator’s choice of chemotherapy regimen in relapsed/refractory adult T-cell leukemia/lymphoma Adrienne A. Phillips et al.

Chronic Lymphocytic Leukemia

1004

The involvement of microRNA in the pathogenesis of Richter syndrome Katrien Van Roosbroeck et al.

1016

Targeting intermediary metabolism enhances the efficacy of BH3 mimetic therapy in hematologic malignancies Aoula Al-Zebeeby et al.

Plasma Cell Disorders

1026

Efficacy of first-line treatments for multiple myeloma patients not eligible for stem cell transplantation: a network meta-analysis Hedwig M. Blommestein et al.

Platelet Biology & its Disorders

1036

Sphingolipid dysregulation due to lack of functional KDSR impairs proplatelet formation causing thrombocytopenia Tadbir K. Bariana et al.

Coagulation & its Disorders

1036

Prevention of the anti-factor VIII memory B-cell response by inhibition of Bruton tyrosine kinase in experimental hemophilia A Sandrine Delignat et al.

Stem Cell Transplantation

1055

HLA discrepancy between graft and host rather than between that graft and first donor impact the second transplant outcome Yoshinobu Maeda et al.

Cell Therapy & Immunotherapy

1062

Human stem cells transplanted into the rat stroke brain migrate to the spleen via lymphatic and inflammation pathways Kaya Xu et al.

1074

Combination peptide immunotherapy suppresses antibody and helper T-cell responses to the major human platelet autoantigen glycoprotein IIb/IIIa in HLA-transgenic mice Lindsay S. Hall et al.

Quality of life

1084

Randomized controlled trial of individualized treatment summary and survivorship care plans for hematopoietic cell transplantation survivors Navneet S. Majhail et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/104/5.toc

e179

A novel gain-of-function mutation of Piezo1 is functionally affirmed in red blood cells by high-throughput patch clamp Maria G. Rotordam et al. http://www.haematologica.org/content/104/5/e179

e184

The population dynamics of hemoglobins A, A2, F and S in the context of the hemoglobinopathies HbS and a-thalassemia in Kenyan infants Alex W. Macharia et al. http://www.haematologica.org/content/104/5/e184

e187

Detection and characterization of homozygosity of mutated CALR by copy neutral loss of heterozygosity in myeloproliferative neoplasms among cases with high CALR mutation loads or with progressive disease Anna Stengel et al.

Haematologica 2019; vol. 104 no. 5 - May 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation http://www.haematologica.org/content/104/5/e187

e191

Autophagic degradation determines the fate of T315I-mutated BCR-ABL protein Haruka Shinohara et al. http://www.haematologica.org/content/104/5/e191

e195

Clinical and molecular features of acute promyelocytic leukemia with variant retinoid acid receptor fusions Lijun Wen et al. http://www.haematologica.org/content/104/5/e195

e200

Mutational and transcriptomic profiling of acute leukemia of ambiguous lineage reveals obscure but clinically important lineage bias Zhen-Tang Lao et al. http://www.haematologica.org/content/104/5/e200

e204

T-cell acute lymphoblastic leukemias express a unique truncated FAT1 isoform that cooperates with NOTCH1 in leukemia development Charles E. de Bock et al. http://www.haematologica.org/content/104/5/e204

e208

Long-term real-world results of ibrutinib therapy in patients with relapsed or refractory chronic lymphocytic leukemia: 30-month follow up of the Swedish compassionate use cohort Maria Winqvist et al. http://www.haematologica.org/content/104/5/e208

e211

Ibrutinib for the treatment of relapsed/refractory mantle cell lymphoma: extended 3.5-year follow up from a pooled analysis Simon Rule et al. http://www.haematologica.org/content/104/5/e211

e215

A prospective description of HIV-associated multicentric Castleman disease in Malawi Tamiwe Tomoka et al. http://www.haematologica.org/content/104/5/e215

Case Reports Case Reports are available online only at www.haematologica.org/content/104/5.toc

e218

Leukemic presentation of ALK-positive anaplastic large cell lymphoma with a novel partner, poly(A) binding protein cytoplasmic 1 (PABPC1), responding to single-agent crizotinib Dylan Graetz et al. http://www.haematologica.org/content/104/5/e218

e222

Venetoclax penetrates in cerebrospinal fluid and may be effective in chronic lymphocytic leukemia with central nervous system involvement.

Gianluigi Reda et al. http://www.haematologica.org/content/104/5/e222

e224

Venetoclax plus rituximab or obinutuzumab after allogeneic hematopoietic stem cell transplantation in chronic lymphocytic leukemia Othman Al-Sawaf et al. http://www.haematologica.org/content/104/5/e224

Comments Comments are available online only at www.haematologica.org/content/104/5.toc

e227

Identifying potential factors of variable response to mogamulizumab in adult T-cell leukemia/lymphoma between Japanese and other populations Murali Janakiram et al. http://www.haematologica.org/content/104/5/e227

e228

Response to the comment: “Identifying potential factors of variable response to mogamulizumab in adult T-cell leukemia/lymphoma between Japanese and other populations�

Adrienne A. Phillips http://www.haematologica.org/content/104/5/e227

Retraction e229

GRP94 rewires and buffers the FLT3-ITD signaling network and promotes survival of acute myeloid leukemic stem cells Beibei Zhang et al. http://www.haematologica.org/content/104/5/e229

Haematologica 2019; vol. 104 no. 5 - May 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 Factor H: a novel modulator in sickle cell disease Wassim El Nemer1,2,3 and Bérengère Koehl1,2,3,4 1

Biologie Intégrée du Globule Rouge UMR_S1134, Inserm, Univ. Paris Diderot, Sorbonne Paris Cité, Univ. de la Réunion, Univ. des Antilles; Institut National de la Transfusion Sanguine, F-75015; 3Laboratoire d’Excellence GR-Ex and 4Hematology Unit, Sickle Cell Disease Center, Robert Debré Hospital, AP-HP, Paris, France

2

E-mail: WASSIM EL NEMER - wassim.el-nemer@inserm.fr doi:10.3324/haematol.2018.214668

S

ickle cell disease (SCD) is an autosomal recessive disorder caused by a point mutation in the β globin gene that substitutes glutamic acid (GAG) at position 6 of the protein into valine (GTG).1 The resulting mutated hemoglobin (HbS) polymerizes under hypoxic conditions driving sickling of red blood cells (RBC). Sickling and dehydration alter the shape of the RBC, decreasing their deformability and increasing their rigidity, which results in significant intravascular hemolysis. These alterations affect blood rheology and microcirculatory flow as well as blood and endothelial cell functions because of the release of hemoglobin and subsequent free heme in the circulation. In addition to chronic hemolysis and related complications, patients with SCD experience frequent vaso-occlusive crises that are painful episodes caused by obstruction of micro-capillaries, believed to be initiated by abnormally adherent RBC or neutrophils reducing the luminal section of the capillaries with subsequent blockage by rigid, deformed RBC.2,3 Hemolysis and vaso-occlusive crises are critical components of the chronic inflammatory state reported in SCD,4,5 which in turn is responsible for several cellular dysfunctions including activation of neutrophils that contributes to vaso-occlusive crises in a vicious feedback loop.

SCD is a multisystem disease that has been explored for decades. Despite significant efforts to date and recent advances showing a role for neutrophils in vaso-occlusive crises, the pathogenesis of SCD, in terms of the sequence of molecular events underlying the disease, remains only partially understood. Hemolysis and chronic inflammation are features common to SCD and other pathologies in which complement activation has been reported, such as atypical hemolytic uremic syndromes and paroxysmal nocturnal hemoglobinuria.

The complement system is one of the oldest defense mechanisms against infections during evolution.6 It is composed of the classical, alternative and lectin pathways that can be activated by specific chemical components.

Activation of the classical pathway is initiated by the attachment of the first protein of the complement, C1q, to one of its ligands, the most important being the CH2 domain of the IgG Fc fragment and the CH4 domain of IgM. This activation leads to the cleavage of the C4 component present in plasma into a small C4a fragment and a large C4b fragment that binds covalently to the target surface and subsequently forms the C4bC2a complex, called the “classical C3 convertase” because of its ability to cleave C3.7 The lectin pathway is activated by the carbohydrate structure of microorganisms. The recognition protein is MBL (mannan-binding lectin) and is associated with serine proteases called MASP1, -2 or -3. Once activated, MASP acquire the ability to cleave C4 and C2 proteins thus forming the classical C3 convertase C4b2a. The alternative pathway is activated by bacterial products, such as lipopolysaccharides, viruses, and infected, transformed or apoptotic cells; it leads to the formation of the “alternative C3 convertase”. It is initiated by the association of soluble C3b with factor B allowing this latter to be cleaved by a serine protease circulating in active form in the plasma, factor D, producing the Ba and Bb fragments. While Ba is excluded from the complex, Bb remains associated with C3b to form the C3bBb complex, named the “alternative C3 convertase”, capable of catalyzing the cleavage of C3 to C3b, like the C4b2a complex. Activation of the alternative pathway is capable of self-amplification, which is very important for the recognition and elimination of pathogens in the absence of specific antibodies.8 Activation of one of the three pathways leads to succes-

Figure 1. Anti-adhesive role of factor H in sickle cell disease.

Activation of the alternative complement pathway in sickle cell disease drives accumulation of C3 cleavage fragments, iC3b in this figure, on the surface of red blood cells triggering their abnormal adhesion to endothelial cells. Factor H binds iC3b and inhibits adhesion of sickle red blood cells to the vascular wall.

haematologica | 2019; 104(5)

857

Editorials

sive proteolysis of plasma proteins which converges to the central protein of the complement system called C3.

The activation of a C3 convertase (classical or alternative) results in the production of a fragment called C3b which can then initiate different effector pathways: opsonization, recruitment of inflammatory cells, direct destruction of infectious agents by osmotic lysis, elimination of circulating immune complexes and apoptotic cells and modulation of specific immune responses.9 C3b can be dissociated into inactive fragments (iC3b, C3dg and then C3d) by means of plasma cofactors (factor I and factor H) or membrane co-factors (CMP, CD35 or CR1). The C3 cleavage fragments (C3b, iC3b, C3dg and C3d) can interact with different cellular receptors (CR1 or CD35, CR2 or CD21, CR3 or CD11b/CD18, CR4 or CD11c/CD18), thus modulating the response at the surface of the different immune cells: phagocytosis, presentation of the antigen and modulation of specific immune responses.10 As in all activation cascades, a narrow network of circulating or membrane proteins is necessary to closely regulate the different activation pathways. The regulation of the alternative pathway is ensured by factor H which plays a central role in discriminating self from non-self.11 It controls the initiation of the C3bBb complex (alternative C3 convertase) by competing with factor B for C3b binding and accelerates the dissociation of the alternative C3 convertase.

Evidence for altered alternative complement pathway activity in the sera of SCD patients was reported in 1976 by Koethe and collaborators.12 In 1985, Chudwin and collaborators showed that 89% of SCD patients’ sera had elevated concentrations of C3b derivatives indicative of increased alternative pathway activation.13 In 1993, Wang and collaborators showed that altered membrane phospholipid exposure of RBC is a critical element of alternative complement pathway activation in SCD patients.14 Very recently, it was shown that cell-free heme and heme-containing microvesicles resulting from intravascular hemolysis activate complement in SCD.15 In addition, activation of the complement is suspected to be involved in the delayed hemolytic transfusion reaction in SCD, a suspicion recently supported by good outcomes following injections of an anti-C5 monoclonal antibody (eculizumab).16 In this issue of Haematologica, Lombardi and collaborators investigated the activation of the alternative complement pathway as a potential contributor to increased RBC adhesion in SCD patients at steady state and during vasoocclusive crises.17 First, they confirmed complement activation in vivo by showing increased serum levels of complement activation fragment C5a in SCD patients as well as microvascular deposition of another activation marker, C5b-9, in small vessels of skin biopsies from patients but not from healthy subjects. Investigating blood cells, they found higher numbers of RBC carrying C3d-derived opsonins in SCD patients than in healthy subjects, indicative of alternative complement pathway activation occurring directly on sickle RBC. This was associated with higher proportions of RBC exposing phosphatidylserine at their surface, as previously reported.14 The authors hypothesized that C3 fragments deposited on the RBC surface may serve as adhesive sites driving abnormal adhesion of sickle RBC to the endothelial wall. They 858

explored this hypothesis by performing ex vivo adhesion assays under dynamic conditions, in which they found the expected higher levels of adhesion of sickle RBC on tumor necrosis factor-a-activated endothelial cells as compared to control RBC. Pre-incubating RBC with factor H, a soluble regulator of alternative complement pathway activation that circulates in the plasma and binds to C3b/iC3b on self cells, inhibited sickle RBC adhesion in a dose-dependent manner reaching control levels at high concentrations (Figure 1). This was the first evidence that opsonins of the alternative complement pathway, deposited on the surface of sickle RBC, may play a critical role in mediating these cells’ abnormal adhesion to the endothelial wall. The authors tested the inhibitory potential of two factor H fragments and concluded that the factor H 19-20 fragment was sufficient to inhibit sickle RBC adhesion. Finally, using blocking antibodies, the authors showed data suggesting the involvement of P-selectin and Mac-1 in sickle RBC adhesion on the endothelial cell side.

Once activated, the complement pathways play an important role in the induction of tissue lesions, such as recruitment of inflammatory cells (neutrophils, monocytes, macrophages and activated lymphocytes), activation of endothelial cells and platelets, and secretion of pro-inflammatory cytokines. Such dysfunctions are frequent in many pathological situations, including autoimmune diseases,18 ischemia-reperfusion syndrome and septic shock,19 making the complement system a potential therapeutic target in these pathologies.20 The study by Lombardi and collaborators reveals a new role for complement activation in the pathogenesis of SCD, particularly in the adhesive process underlying vaso-occlusive crises.17 It paves the way for future clinical studies in which modulators of the alternative complement pathway, including factor H-based inhibitors, could be tested as potential new therapeutic options in this pathology.

References 1. Pauling L, Itano HA, et al. Sickle cell anemia, a molecular disease.

Science. 1949;109(2835):443.

2. Kaul DK, Fabry ME, Nagel RL. Microvascular sites and characteristics of sickle cell adhesion to vascular endothelium in shear flow conditions: pathophysiological implications. Proc Natl Acad Sci U S A. 1989;86(9):3356-3360.

3. Manwani D, Frenette PS. Vaso-occlusion in sickle cell disease: pathophysiology and novel targeted therapies. Blood. 2013;122(24):38923898.

4. Piel FB, Steinberg MH, Rees DC. Sickle cell disease. N Engl J Med. 2017;376(16):1561-1573.

5. Ware RE, de Montalembert M, Tshilolo L, Abboud MR. Sickle cell disease. Lancet. 2017;390(10091):311-323.

6. Nonaka M, Yoshizaki F. Evolution of the complement system. Mol Immunol. 2004;40(12):897-902.

7. Walport MJ. Complement. First of two parts. N Engl J Med. 2001;344(14):1058-1066.

8. Walport MJ. Complement. Second of two parts. N Engl J Med. 2001;344(15):1140-1144.

9. Ricklin D, Hajishengallis G, Yang K, Lambris JD. Complement: a key system for immune surveillance and homeostasis. Nat Immunol.

2010;11(9):785-797.

10. Ricklin D, Reis ES, Mastellos DC, Gros P, Lambris JD. Complement component C3 - the "Swiss Army Knife" of innate immunity and host defense. Immunol Rev. 2016;274(1):33-58.

11. Medjeral-Thomas N, Pickering MC. The complement factor H-related proteins. Immunol Rev. 2016;274(1):191-201.

12. Koethe SM, Casper JT, Rodey GE. Alternative complement pathway activity in sera from patients with sickle cell disease. Clin Exp

haematologica | 2019; 104(5)

Editorials Immunol. 1976;23(1):56-60.

13. Chudwin DS, Korenblit AD, Kingzette M, Artrip S, Rao S. Increased activation of the alternative complement pathway in sickle cell disease. Clin Immunol Immunopathol. 1985;37(1):93-97.

14. Wang RH, Phillips G Jr, Medof ME, Mold C. Activation of the alternative complement pathway by exposure of phosphatidylethanolamine and phosphatidylserine on erythrocytes from sickle cell disease patients. J Clin Invest. 1993;92(3):1326-1335.

15. Merle NS, Grunenwald A, Rajaratnam H, et al. Intravascular hemolysis activates complement via cell-free heme and heme-loaded microvesicles. JCI Insight. 2018;3(12).

16. Unnikrishnan A, Pelletier JPR, Bari S, et al. Anti-N and anti-Do(a) immunoglobulin G alloantibody-mediated delayed hemolytic trans-

17.

18.

19.

20.

fusion reaction with hyperhemolysis in sickle cell disease treated with eculizumab and HBOC-201: case report and review of the literature. Transfusion. 2019 Feb 15. [Epub ahead of print] Lombardi, A Matte, A.M. Risitano et al.Factor H interferes with the adhesion of sickle red cells to vascular endothelium: a novel diseasemodulating molecule. Haematologica 2019;104(5):919-928.

Toubi E, Vadasz Z. Innate immune-responses and their role in driving autoimmunity. Autoimmun Rev. 2019;18(3):306-311.

Charchaflieh J, Wei J, Labaze G, et al. The role of complement system in septic shock. Clin Dev Immunol. 2012;2012:407324.

Dobo J, Kocsis A, Gal P. Be on target: strategies of targeting alternative and lectin pathway components in complement-mediated diseases. Front Immunol. 2018;9:1851.

A post-transplant optimized transplant-specific risk score in myelodysplastic syndromes Mahasweta Gooptu and John Koreth Division of Hematologic Malignancies, Dana-Farber Cancer Institute, Boston, MA, USA E-mail: MAHASWETA GOOPTU - mahasweta_gooptu@dfci.harvard.edu doi:10.3324/haematol.2018.214452

A

llogeneic hematopoietic stem-cell transplantation (HSCT) remains the only potentially curative therapy for myelodysplastic syndromes (MDS), but treatment risks include relapse and non-relapse mortality (NRM). Whereas relapse following HSCT is typically dictated by disease-related factors, NRM is more influenced by patient- (performance status, co-morbidity, etc.) and transplant-related factors (donor type, conditioning intensity, graft-versus-host disease prophylaxis regimen, etc.).

In order to improve transplant decision-making for the individual MDS patient, better prediction of HSCT outcomes, by including both relapse and NRM predictors in a comprehensive individualized and dynamic risk model, would be optimal. So where do we stand currently?

The prognosis of MDS has historically been based on the International Prognostic Scoring System (IPSS). For transplant decision-making, Markov models based on the IPSS have documented that MDS patients with low- and intermediate-1-risk MDS have better survival outcomes without transplant, whereas transplantation results in better survival outcomes for patients with intermediate2- and high-risk MDS.1,2 The Revised International Prognostic Scoring System (R-IPSS), a refinement of the IPSS, is used to prognosticate MDS at diagnosis, particularly the risk for transformation to acute myeloid leukemia,3 and is often used as part of the decision to proceed to transplantation or not.4 While the IPSS and R-IPSS focus on disease features, they do not consider patient- and transplant related factors relevant to HSCT outcome. Attempts have, therefore, been made to develop MDS transplant-specific risk scores to predict survival better. These scores include the transplantation risk index developed by the Gruppo Italiano Trapianto di Midollo Osseo (GITMO)4 registry using 519 patients as well as a risk score from the Center for International Blood and Marrow Transplant Research (CIBMTR)5, using 1,519 patients. Both of these indices identified similar prognostic variables (including the RIPSS), dividing MDS transplant recipients into four risk haematologica | 2019; 104(5)

groups with overall survival rates ranging from 5-76%.

However, these indices have not been universally adopted in current practice. While the GITMO index has not been externally validated, the CIBMTR index was validated on a distinct subset of patients from within the CIBMTR database. Gagelmann et al. now report on another composite risk score with better predictive ability than the existing indices.6 The authors compiled a cohort of 1,059 adult patients (â&#x2030;Ľ18 years) with MDS from the European Society for Blood and Marrow Transplantation (EBMT) registry who underwent HLA-matched HSCT from a related or unrelated donor between 2000 to 2014. Using a Cox proportional hazards model they identified seven variables with significant impact on overall survival: age >50 years, matched unrelated donor, Karnofsky Performance Status <90%, very poor cytogenetics or monosomal karyotype, positive cytomegalovirus status of the recipient, peripheral blood blasts >1% and platelet count â&#x2030;¤50 x 109/L. Of these, age and cytogenetic risk were the strongest predictors of survival, based on hazard ratios for death, and given more weight than the other factors in the final score. Four prognostic groups were identified (low, intermediate, high and very-high risk) with overall survival rates of 68.7%, 43.2%, 26.6% and 9.5%, respectively.

How does the EBMT score described in the paper by Gagelmann et al. compare to the prior CIBMTR and GITMO scores as well as the R-IPSS itself? One approach would be to compare the concordance or c-statistic (measured as area under the receiver operating curve) of the different indices. The c-statistic is used to compare the goodness of fit of logistic regression models with values that range from 0.5 to 1.0. A c-statistic of 0.5 indicates the predictive ability of the index is no better than chance while a c-statistic in the 0.7-0.8 range has reasonable discriminatory power. Looking at the c-statistic following cross- validation, the EBMT transplant risk index scored 0.609 (95% confidence interval: 0.588 to 0.629), which was better than the CIBMTR (0.555) and GITMO (0.579) 859

Editorials

indices as well as the R-IPSS (0.551). The authors concluded that the EBMT risk index is a better composite prognostic tool for MDS transplantation outcomes than existing indices, albeit acknowledging that the benefit is modest. However, we have caveats. In general, the c-statistic for an index would be expected to decrease slightly when applied to an external validation dataset (in comparison to the parent dataset from which it was derived) and this must be considered when comparing c-statistics for the externally validated GITMO and CIBMTR indices to the non-externally validated EBMT risk index.

Validation of the EBMT risk index in an independent cohort of patients would provide a better estimate of its

discriminatory power. Furthermore, as the authors acknowledge, the Hematopoietic Cell TransplantationComorbidity Index (HCT-CI), a well validated and widely used tool to predict NRM,7 was not part of the variables examined (due to insufficient data) and the EBMT predictive model would be expected to improve if the HCT-CI had been incorporated.

Although the above indices incorporate information on MDS karyotype, they lack MDS genomic data, which are increasingly important for predicting relapse and, to an extent NRM, after transplantation. A large analysis using the CIBMTR database (n=1,514) examined the association between pre-transplant mutational profile and post-

Figure 1. Nomogram adapted from Gagelmann et al.8 showing an example calculation. In this example a 55-year old, cytomegalovirus (CMV)-positive patient with a Karnofsky performance status of 90, platelet count of 150x109/L, very good cytogenetics, <1% blasts and a matched sibling donor would get 160 points with a 2year survival in the 50-60% range. The c-statistic for the EBMT index was 0.609 (95% confidence interval: 0.588 t0 0.629. We also highlight factors not included in the original nomogram â&#x20AC;&#x201C; myelodysplastic syndrome (MDS) mutations, minimal residual disease (MRD) and Hematopoietic Cell Transplantation Comobidity Index (HCT-CI) â&#x20AC;&#x201C; which could enhance the discriminatory power of the index.

860

haematologica | 2019; 104(5)

Editorials

transplant outcomes using targeted mutational analysis with a next-generation sequencing panel. In this cohort, TP53 and RAS pathway mutations were strongly associated with poorer overall survival and earlier relapse independently of transplant conditioning intensity. There was also an indication that patients with JAK2 mutations had increased NRM after ablative conditioning transplants.7 In this context, there has been an attempt to include genomic data into the GITMO index.8 In a cohort of 401 patients, using massively parallel sequencing for mutational analysis, TP53 mutations were again identified as adverse prognostic markers. In addition, spliceosome mutations signifying a secondary-type acute myeloid leukemia phenotype, ASXL1 and RUNX1 were also associated with poorer outcomes. With MDS mutational analysis becoming routine practice in the clinic, it is important that future iterations of transplant risk models incorporate MDS genomic data.

The impact of disease persistence as measurable residual disease − variably measured by multi-parameter flow cytometry, cytogenetics/fluorescence in situ hybridization, and increasingly by next-generation sequencing − has been of great interest as an independent dynamic predictor of relapse. In MDS, the presence of measurable residual disease in the early post-transplant period (assessed by multi-parameter flow cytometry or cytogenetics/fluorescence in situ hybridization) is associated with significantly poorer outcomes.9 Further studies are needed to better define the impact of pre- and post-transplant measurable residual disease, but we expect this to be an important and dynamic predictor for individualized MDS transplant risk prediction in the future.

With regards to individualized risk prediction, the authors of the current study define a user-friendly nomogram for scoring the various elements of the index in finer detail and with greater prognostic power (c-statistic 0.609) (Figure 1).

In Figure 1 we have in addition highlighted three missing variables that would likely add to the discriminatory power of this index (i.e. pre-transplant mutational/genomic analysis, HCT-CI and minimal residual disease status).

We also note that for poor-risk cohorts, failure after transplantation includes both NRM and relapse at equivalent frequency (~40%). This offers opportunities for progress, especially for reducing NRM failures. For instance, cytomegalovirus serostatus of the recipient and its impact on post-transplant survival and immune reconstitution has been an area of increasing research.10 In this study there was a moderately high risk of cytomegalovirus reactivation (39%) with significant impairment of overall survival.6 As the authors point out, optimizing the use of antiviral agents active against cytomegalovirus, such as letermovir, which have been shown to be useful in high-risk settings,11 may improve NRM in a lower-risk HLA-matched cohort. Similarly, avoiding ablative conditioning in TP53- and JAK2-mutant

haematologica | 2019; 104(5)

MDS, in which it offers no benefit and may even be deleterious, may further reduce NRM after transplantation, while ablative conditioning may improve outcomes of RAS pathway-mutant MDS.8 In the future, studies of preemptive immunomodulation strategies (e.g. tumor vaccines, donor lymphocyte infusions) based on individual dynamic risk scoring before and after transplantation may be considered.

In summary, Gagelmann et al. present a new composite risk index to predict MDS transplant survival outcomes which incorporates both disease- and patient-related factors. They document a moderate improvement of predictive power compared to existing indices. A useful nomogram is provided as a step towards individualized outcome prediction. External validation in an independent dataset, and the future incorporation of the HCT CI, MDS genomic data and minimal residual disease status will be important next steps toward the goal of individualized, dynamic MDS transplant outcome prediction and treatment decision-making.

References 1. Koreth J, Pidala J, Perez WS, et al. Role of reduced-intensity conditioning allogeneic hematopoietic stem-cell transplantation in older patients with de novo myelodysplastic syndromes: an international collaborative decision analysis. J Clin Oncol. 2013;31(21):2662-2670.

2. Cutler CS, Lee SJ, Greenberg P, et al. A decision analysis of allogeneic bone marrow transplantation for the myelodysplastic syndromes: delayed transplantation for low-risk myelodysplasia is associated with improved outcome. Blood. 2004;104(2):579-585.

3. Greenberg PL, Tuechler H, Schanz J, et al. Revised International Prognostic Scoring System for myelodysplastic syndromes. Blood.

2012;120(12):2454-2465.

4. Della Porta MG, Alessandrino EP, Bacigalupo A, et al. Predictive factors for the outcome of allogeneic transplantation in patients with MDS stratified according to the revised IPSS-R. Blood. 2014;123(15): 2333-2342.

5. Shaffer BC, Ahn KW, Hu Z-H, et al. Scoring system prognostic of outcome in patients undergoing allogeneic hematopoietic cell transplantation for myelodysplastic syndrome. J Clin Oncol. 2016;34 (16):1864-1871.

6. Gagelmann N. Optimized EBMT transplant-specific risk score in myelodysplastic syndromes after allogeneic stem-cell transplantation. Haematologica.2019;104(5):929-936.

7. Sorror ML, Maris MB, Storb R, et al. Hematopoietic cell transplantation (HCT)-specific comorbidity index: a new tool for risk assessment before allogeneic HCT. Blood. 2005;106(8):2912-2919.

8. Lindsley RC, Saber W, Mar BG, et al. Prognostic mutations in myelodysplastic syndrome after stem-cell transplantation. N Engl J Med. 2017;376(6):536-547.

9. Della Porta MG, Gallì A, Bacigalupo A, et al. Clinical effects of driver somatic mutations on the outcomes of patients with myelodysplastic syndromes treated with allogeneic hematopoietic stem-cell transplantation. J Clin Oncol. 2016;34(30): 3627-3637.

10. Festuccia M, Baker K, Gooley TA, et al. Post-hematopoietic stem cell transplantation minimal residual disease and early relapses in MDS and AML evolving from MDS. Blood. 2015;126(23):2019-2019.

11. Suessmuth Y, Mukherjee R, Watkins B, et al. CMV reactivation drives post-transplant T cell reconstitution and results in defects in the underlying TCRβ repertoire. Blood. 2015;125(25):3835-3850.

12. Marty FM, Ljungman P, Chemaly RF, et al. Letermovir prophylaxis for cytomegalovirus in hematopoietic-cell transplantation. N Engl J Med. 2017;377(25):2433-2444.

861

Editorials

Precision tyrosine kinase inhibitor dosing in chronic myeloid leukemia?

Giuseppe Saglio,1 Carmen Fava1 and Robert Peter Gale2 1

Department of Clinical and Biological Sciences of the University of Turin, Mauriziano Hospital, Italy and 2Haematology Research Center, Division of Experimental Medicine, Department of Medicine, Imperial College London, UK E-mail: GIUSEPPE SAGLIO - giuseppe.saglio@unito.it doi:10.3324/haematol.2018.214445

T

herapy for chronic myeloid leukemia (CML) with tyrosine kinase inhibitors (TKIs) has made this a potentially curable disease.1,2 However, many challenges remain, including: i) defining the best TKI, dose and schedule; ii) how to reduce the frequency and severity of adverse events (AEs); iii) how to increase the number of subjects who can achieve therapy-free remission (TFR), and others. In this issue of Haematologica, using data from the German CML-Study IV,2 Michel et al.3 tackle two of these challenges: the best TKI dose and reducing AEs. They report that subjects randomized to receive high-dose imatinib, 800 mg/day (d), achieving a stable major molecular response (MMR, 0.1% of BCRABL1IS) can have their imatinib dose reduced to 400 mg/d without losing their response, with the additional benefits of reducing AEs and cost, and likely increasing compliance.

Several prior clinical trials tested whether high-dose imatinib, 800 mg/d, was more effective than the approved dose, 400 mg/d.4-6 The primary end point of most of these trials was the proportion of study subjects achieving a MMR at 1 year, a landmark associated with a very low risk of leukemia progression and death from CML-related causes.7 A secondary end point was the time

to MMR achievement. The conclusion of most studies was that high-dose imatinib resulted in faster MMRs but later led to a similar proportion of MMRs after 1 or 2 years.4-6 However, high-dose imatinib was associated with increased rates of â&#x2030;Ľ grade 3 AEs, worse compliance, and higher costs.4-6 Consequently, many study subjects assigned to high-dose imatinib reverted to 400 mg/d.

Recently, a landmark analysis of data from the CMLStudy IV reported that study subjects receiving an optimized high-dose of imatinib (median dose, 600 mg/d) achieved deeper and faster molecular responses (MMR, MR4 and MR4.5) compared with those receiving 400 mg/d, with no increase in â&#x2030;Ľ grade-3 AEs.8 Importantly, the conventional and optimized strategies of giving imatinib resulted in similar event-free survival (EFS), progressionfree survival (PFS), and overall survival (OS).2 There are several caveats to accepting these conclusions including biases associated with landmark analyses and discordances between molecular responses (surrogate end points) and clinically important end points such as EFS, PFS and OS.9,10 Such discordances are common to many, if not most, clinical trials and underscore the limitations of surrogate end points.11 This is not surprising in chronic

FIgure 1. Possible future therapeutic strategy for CML. TKI: tyrosine kinase inhibitors; MMR: major molecular response; MR: molecular response.

862

haematologica | 2019; 104(5)

Editorials

phase CML where more than half of the deaths are not related to CML.2 However, the bottom line at a time of generic imatinib is that a starting dose of 800 or 600 mg/d reduced to 400 mg/d in subjects achieving a stable MMR is probably a safe and effective therapeutic strategy.

The report of Michel et al. recalls an interesting observation made several years ago in the OpTKima study.

There, some older subjects receiving imatinib 400 mg/d, but who stopped therapy every 3rd month maintained a MMR and sometimes even improved the depth of molecular response.12 However, unlike the uniformly stable MMRs reported by Michel et al., approximately 25% of subjects in the OpTKima study lost their MMR. The studies differ, of course, in the fact that, in the OpTKima study, subjects completely stopped imatinib while in the CML-Study IV subjects had an imatinib dose reduction.

What do these data suggest? A reasonable conclusion is that the best strategy is to optimize initial imatinib dose based on the rapidity, depth and stability of a subject’s molecular response rather than using the same dose and schedule for everyone. Alternatively, some subjects who could benefit by starting off directly on a 2nd generation TKI,13 could be moved to lower (and less toxic) dosages of the same drug once they achieved a good molecular response, or eventually, in specific cases, switch to imatinib for maintenance. Studies testing the feasibility and the value of this approach are needed and, indeed, some are already ongoing or planned.14 The regulatory approved dose of imatinib and other TKIs often evolves from results of phase I safety studies designed to determine the maximum tolerated dose (MTD) followed by phase II and III studies of efficacy.15 This developmental scheme assumes the MTD is the maximally effective dose (MED). But is this assumption correct? In the case of CML, the MED is the dose associated with maximal inhibition of P210BCRABL1 that is also safe, especially when given over a long period of time.

Given these considerations, it is easy to see why the MTD and MED might differ for a TKI.15 Another issue is why different subjects respond differently to the same dose of a TKI like imatinib. Many factors could explain this heterogeneity but the most obvious is BCRABL1 mutations.16 Other variables include pharmaco-kinetic and pharmaco-dynamic variables related to TKI absorption and metabolism, susceptibility to AEs, and compliance.17 Also, some subjects in chronic phase CML have subclones with additional mutations in genes other than BCR-ABL1 reflecting the genomic instability typical of CML.18 These subclones are not detected by routine diagnostic procedures and may be important in determining response to TKI-therapy and likelihood of CML progression, obviously confounded outcomes. In this context, it is important to remember that there is a substantial interval between when BCRABL1 is acquired to when CML is diagnosed, leaving ample time for clonal evolution. For example, in the atom bomb survivors, who likely acquired BCRABL1 when the atom bomb exploded, median latency to CML diagnosis was 10 years with a possible range of from <2 to >30 years.19 How can we best reconcile the goal of reducing the risk and severity of AEs with the need to control or eradicate undetected CML subclones that may require a higher TKI haematologica | 2019; 104(5)

dose, different TKIs, or both? One strategy might be to start with what might be called an ‘induction phase’ with a high-dose of a 2nd or even a 3rd generation TKI, or highdose imatinib, followed by switching to a lower dose in a ‘maintenance phase’ in responders. It might also be reasonable to begin with a 2nd or 3rd generation TKI and then switch to imatinib.

The next question is when to transition from the ‘induction’ to the ‘maintenance’ phase. The decision could be based on surrogate end points such as MMR or MR4, but it is also important to remember that end points like MMR or MR4 are predictive rather than prognostic surrogate end points.20 Which TKI is best? Should we reduce the approved dose of newer TKIs or switch to imatinib 400 mg/d? This could depend on the therapeutic goal and this may differ in different subjects. Is it to improve EFS, PFS or survival, achieve TFR, decrease AEs and costs, increase compliance, something else, or a combination of different goals?

When the therapeutic goal is TFR, the rapidity of achieving a deep molecular response (DMR) and its stability and duration are crucial.21 As such, a more intensive initial therapy strategy may be preferable. However, this may not be the goal in other subjects in whom survival is the goal and where less ‘induction’ therapy may be appropriate.

Another way to consider revising TKI therapeutic strategy is to make treatment decisions based on time-toevent end points with the possibility of continually revising strategy according to outcomes using statistical techniques such as Markov or Bayesian adaptive models.22 This can be considered an extension of current European LeukemiaNet recommendations,23 while also considering additional variables, such as TKI, dose, schedule, therapeutic goal, AEs, pharmaco-kinetic and pharmacodynamics, and others, such as the kinetics of decline of BCRABL1 transcripts. It is even conceivable that one might consider potency of suppression of P210BCRABL1 kinase activity in different subjects, and even activity in CML leukemia stem cells.

The bottom line is that it is time to re-think our strategy of using TKIs to treat CML. We suggest testing an individualized, precision-based approach that considers disease, patient and therapeutic goal heterogeneities, and modifying therapy according to the rate, depth, duration and stability of molecular response while acknowledging poor correlations with EFS, PFS and survival. Much work remains to clarify these issues, and this needs to be tested in randomized trials.

References 1. Hochhaus A, Larson RA, Guilhot F, et al. Long-Term Outcomes of Imatinib Treatment for Chronic Myeloid Leukemia. N Engl J Med. 2017;376(10):917-927.

2. Hehlmann R, Lauseker M, Saußele S, et al. Assessment of imatinib as first-line treatment of chronic myeloid leukemia: 10-year survival results of the randomized CML study IV and impact of non-CML determinants. Leukemia. 2017;31(11):2398-2406.

3. Michel C, Burchert A, Hochhaus A, et al. Imatinib dose reduction in major molecular response of chronic myeloid leukemia: results from the German Chronic Myeloid Leukemia-Study IV. Haematologica.

2018 Dec 4. [Epub ahead of print] 4. Baccarani M, Rosti G, Castagnetti F, et al. Comparison of imatinib

863

Editorials

5.

6.

7.

8.

9.

10.

11.

12.

13.

400 mg and 800 mg daily in the front-line treatment of high-risk, Philadelphia-positive chronic myeloid leukemia: a European LeukemiaNet Study. Blood. 2009;113(19):4497-4504.

Cortes JE, Kantarjian HM, Goldberg SL, et al. High-dose imatinib in newly diagnosed chronic-phase chronic myeloid leukemia: high rates of rapid cytogenetic and molecular responses. J Clin Oncol.

2009;27(28):4754-4759.

Corstes JE, Baccarani M, Guilhot F, et al. Phase III, randomized, open-label study of daily imatinib mesylate 400 mg versus 800 mg in patients with newly diagnosed, previously untreated chronic myeloid leukemia in chronic phase using molecular end points: tyrosine kinase inhibitor optimization and selectivity study. J Clin Oncol.

2010;28(3):424-430.

Breccia M, Alimena G. The significance of early, major and stable molecular responses in chronic myeloid leukemia in the imatinib era.

Crit Rev Oncol Hematol. 2011;79(2):135-143.

Hehlmann R, MĂźller MC, Lauseker M, et al. Deep molecular response is reached by the majority of patients treated with imatinib, predicts survival, and is achieved more quickly by optimized high-dose imatinib: results from the randomized CML-study IV. J Clin Oncol. 2014;32(5):415-423.

Giobbie-Hurder A, Gelber RD, Regan MM. Challenges of guaranteetime bias. J Clin Oncol. 2013;31(23):2963-2969.

Dafni U. Landmark analysis at the 25-year landmark point. Circ Cardiovasc Qual Outcomes. 2011;4(3):363-371.

Freidlin B, Sun Z, Gray R, Korn EL. Phase III clinical trials that integrate treatment and biomarker evaluation. J Clin Oncol.

2013;31(25):3158-3161.

Russo D, Martinelli G, Malagola M, et al. Effects and outcome of a policy of intermittent imatinib treatment in elderly patients with chronic myeloid leukemia. Blood. 2013;121(26):5138-5144.

Saglio G, Jabbour E. First-line therapy for chronic phase CML: selecting the optimal BCR-ABL1-targeted TKI. Leuk Lymphoma.

2018;59(7):1523-1538.

14. Clark RE, Polydoros F, Apperley JF wt al. De-escalation of tyrosine kinase inhibitor dose in patients with chronic myeloid leukaemia with stable major molecular response (DESTINY): an interim analysis of a non-randomised, phase 2 trial. Lancet Haematol.

2017;4(7):e310-e316.

15. Takimoto CH. Maximum tolerated dose: clinical endpoint for a bygone era? Target Oncol. 2009;4(2):143-147.

16. Soverini S, Colarossi S, Gnani A, et al. Contribution of ABL kinase domain mutations to imatinib resistance in different subsets of Philadelphia-positive patients: by the GIMEMA Working Party on Chronic Myeloid Leukemia. Clin Cancer Res. 2006;12(24):73747379.

17. Eadie LN, Hughes TP, White DL. Patients with low OCT-1 activity and high ABCB1 fold rise have poor long-term outcomes in response to tyrosine kinase inhibitor therapy. Leukemia. 2018;32(10):22882291.

18. Bransford S, Wang P, Yeung DT, et al. Integrative genomic analysis reveals cancer-associated mutations at diagnosis of CML in patients with high-risk disease. Blood. 2018;132(9):948-961.

19. Hsu WL, Preston DL, Soda M, et al. The incidence of leukemia, lymphoma and multiple myeloma among atomic bomb survivors: 19502001. Radiat Res. 2013;179(3):361-382.

20 Ballman KV. Biomarker: Predictive or Prognostic? J Clin Oncol.

2015;33(33):3968-3971.

21. Saussele S, Richter J, Guilhot J, et al. Discontinuation of tyrosine kinase inhibitor therapy in chronic myeloid leukaemia (EURO-SKI): a prespecified interim analysis of a prospective, multicentre, nonrandomised, trial. Lancet Oncol. 2018;19(6):747-757.

22. Gagniuc PA. Markov chains: from theory to implementation and experimentation. 1st. ISBN 9781119387589 (ePub ebook) 23. Baccarani M, Deininger MW, Rosti G, et al. European LeukemiaNet recommendations for the management of chronic myeloid leukemia: 2013. Blood. 2013;122(6):872-884.

Mogamulizumab versus investigator choice in relapsed/refractory adult T-cell leukemia/ lymphoma: all four one or none for all?

William Johnson,1 Anjali Mishra,1 Adam Binder,1 Alejandro Gru2 and Pierluigi Porcu1 1

Division of Hematologic Malignancies and Hematopoietic Stem Cell Transplantation, Department of Medical Oncology, Sidney Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA and 2Division of Hematopathology, Department of Pathology, University of Virginia, Charlottsville, VA, USA E-mail: PIERLUIGI PORCU - pierluigi.porcu@jefferson.edu doi:10.3324/haematol.2018.214536

T

he human T-cell lymphotropic (or leukemia) virus type-1 (HTLV-1) was isolated by Poiesz et al. in 1980 from the T-cell line Hut-102, established from a patient thought to have cutaneous T-cell lymphoma.1 HTLV-1 causes adult T-cell leukemia/lymphoma (ATL), HTLV-1 associated myelopathy/tropical spastic paresis (HAM/TSP), and other inflammatory disorders.2 ATL is a clinically heterogeneous but often very aggressive mature T-cell neoplasm with dismal survival rates and limited therapeutic options, particularly in the relapsed/refractory (R/R) setting.3 Most cohort studies and clinical trials in ATL come from Japan where the virus is highly endemic in certain regions. Here, investigators have led efforts to define diagnostic criteria, clinical subtypes, prognostic models, and the value of new therapies, including the anti-CCR4 antibody mogamulizumab (KW-0761), approved in Japan for both R/R and chemotherapy-naĂŻve CCR4-positive ATL.4,5 Data on subtype frequency, natural history, and outcome in ATL from non-Japanese endemic regions and from non-endemic regions (North America, Europe) remain very limited, although recent studies have 864

begun to shed some light on this, showing that North American ATL patients present with more aggressive disease and have a worse prognosis (median survival approx. 7 months) compared to Japanese patients.6,7 The availability of mogamulizumab for ATL in Japan provided the impetus to explore its activity in other ATL populations. In this issue of Haematologica, an important study

by Phillips et al.8 significantly advances our understanding of the global therapeutic impact of mogamulizumab in ATL, by reporting results of an international randomized Phase II trial (KW-0761-009) assessing the safety and efficacy of mogamulizumab versus investigator choice of chemotherapy in patients with R/R ATL. HTLV-1 belongs to a group of T-lymphotropic deltaretroviruses, which includes four types of Simian Tlymphotropic viruses (STLV). HTLV-1 is believed to have originated from interspecies transmission between STLV1-infected Old-World monkeys and humans. HTLV-1 is highly endemic in Southwestern Japan, the Caribbean, Northern Iran, and in Aboriginal populations in central Australia.9 HTLV-1 RNA is reverse-transcribed into a haematologica | 2019; 104(5)

Editorials

double-stranded DNA that integrates into the host cell genome as a provirus. The proviral dsDNA is marked at both ends by long terminal repeats (LTRs), which serve as promoters for sense (5′-LTR) and antisense (3′-LTR) transcription.10 Two key oncogenic proteins, Tax and HBZ (HTLV-1 basic leucine zipper factor) (Figure 1), are encod-

ed in the pX region in the 3’ end of the provirus.

An estimated 10-15 million people worldwide are infected with HTLV-1.9 The virus is transmitted vertically (breast milk) and horizontally (sexual contact, blood products), infecting primarily mature CD4+ T cells with a CD25+FOXP3+ regulatory T-cell (Treg) phenotype.11 Direct

Figure 1. Transmission, replication, and oncogenesis of HTLV-1 in adult T-cell leukemia/lymphoma (ATL). Transmission of infected CD4+CD25+FOXP3+ cells occurs via vertical and horizontal routes to a new host. Reverse transcribed HTLV-1 DNA is integrated into the DNA of host cells. and direct cell-to-cell contact and mitosis drives viral replication leading to a clonally diverse population of infected cells. Two transcription regulators, Tax and the HTLV-I basic leucine zipper factor (HBZ) are essential for oncogenesis. Tax up-regulates the P13K/AkT and NFκB pathways including through IL-15, and down-regulates p53. HBZ up-regulates TGFβ, FOXP3, and the C-C chemokine receptor 4 (CCR4) while down-regulating INFα, IL-2, and TNF. After decades of complex interactions between these molecules, together with the acquisition of new mutations, immune dysregulation, and host-specific factors, ATL develops in 2-5% of carriers. The defucosylated monoclonal antibody mogamulizumab binds CCR4 leading to enhanced antibody-dependent cellular cytotoxicity (ADCC).

haematologica | 2019; 104(5)

865

Editorials

cell-to-cell contact is necessary for the infection of new T cells, while the expansion of the HTLV-1 proviral load is achieved by proliferation of infected T cells, which leads to a clonally diverse neoplastic population11 (Figure 1).

Extensive molecular aberrations in HTLV-1-infected T cells, often accumulating over decades, lead to the development of ATL in approximately 3-5% of seropositive carriers. HTLV-1 induced leukemogenesis is a complex, multistep process, driven by Tax and HBZ. Tax-induced upregulation of IL-15, IL-15Ra, and EZH-2 leads to chronic inflammation and polycomb repressive complex 2 (PRC2) hyperactivation, with genome-wide H3K27me3 accumulation.12 Expression of HBZ by HTLV-1 infected T cells results in increased proliferation, impaired apoptosis, and disruption of genomic integrity.13 Analysis of the somatic mutation landscape of ATL reveals common mutations at TP53 and IRF4, and copy number alterations at PD-L1 and CDKN2A.14 HTLV-1 seroprevalence rates mean that ATL predominates in endemic regions, accounting for up to 35% of all T-cell lymphomas in endemic areas in Japan and 15-20% in Peru. However, these figures are only 1-2% in North America and Europe.15 ATL can present with four clinical subtypes: acute, lymphomatous, chronic, and smoldering. A consensus report highlighting the clinical features and treatment guidelines of these subtypes (including an increasingly appreciated fifth subtype: aggressive extranodal primary cutaneous) was recently updated.16 Retrospective studies have described significant clinical and biological differences between Japanese ATL and North American ATL, including a slight female predominance, a younger median age at diagnosis (61-67 vs. 5054 years), and a higher frequency of aggressive subtypes (acute and lymphomatous) (approx. 75% vs. 88-94%).6,7 There are also differences in the mutational landscape, with significantly higher mutation rates for epigenetic regulators, and fewer T-cell receptor/NF-κB pathway alterations in North American ATL compared to Japanese ATL.17 Despite advances in our understanding of the biology of HTLV-1 and ATL, prognosis remains very poor, with median overall survival (OS) of 8.3 months (acute), 10.6 months (lymphomatous), 31.5 months (chronic), and 55 months (smoldering);18 western ATL patients may have a worse prognosis.6,7 Treatment strategies differ significantly between endemic and non-endemic regions. In Japan, the LSG15 regimen was superior to CHOP-14, with higher complete remission (CR) rate and a trend towards improved 3-year OS (24% vs. 13%).19 However, this regimen is not routinely used outside Japan, and the most frequently used chemotherapies in North American ATL are CHOP-like regimens, with overall response rates (ORR) of approximately 60-75% and CR rates of 13-36%.6,7 Consolidation with allogeneic hematopoietic stem cell transplantation (HSCT) is generally recommended for eligible patients with aggressive ATL subtypes, with Japanese studies showing 3-4 year OS ranging between 26% and 36%.20 Unfortunately, most ATL patients relapse, and multiagent salvage chemotherapy is generally ineffective.18 The discovery that C-C chemokine receptor 4 (CCR4) is expressed in over 90% of ATL cases, led to the clinical 866

development of mogamulizumab, a glycoengineered anti-CCR4 monoclonal antibody with a defucosylated Fc region that enhances ADCC. In 2012, mogamulizumab was approved for ATL in Japan in the relapsed setting on the basis of a Phase II trial that showed a 50% ORR,4 and in 2014 was approved for chemotherapy-naïve patients, based on a randomized Phase II trial in combination with the mLSG15 regimen.5 Both studies were quite small (28 and 53 patients, respectively) with ORR as the primary end point. Up-dated outcomes analyses appear to show a real, but relatively modest, benefit for mogamulizumab, with median PFS and OS of 5.2 and 14.4 months for the single arm R/R ATL cohort and 1-year progression-free survival (PFS) 47% and 29% for mLSG15 + mogamulizumab versus mLSG15 in the randomized front-line study.21 In this context, the study by Phillips et al.8 aimed to determine if the incremental, but encouraging, outcome improvements with mogamulizumab in Japanese ATL could be replicated in non-Japanese ATL. This international Phase II study, conducted at 22 centers, randomized (2:1 ratio) 71 patients with R/R ATL with at least one prior line of therapy to either mogamulizumab (n=47) or investigator choice chemotherapy (n=24: GemOx=21; pralatrexate=2; DHAP=1). The primary objective of the study was confirmed overall response rate (cORR), defined as a response sustained for ≥8 weeks. In the mogamulizumab arm, cORRs by investigator and independent review were 15% and 11%, respectively, notably inferior to that of the Japanese registration study.3 Remarkably, the cORR in the investigator’s choice arm was 0%. Concordant with the Japanese Phase II study, the best responses to mogamulizumab by compartment were in blood (54%, all CR) and skin (44%), with no CR in lymph nodes. Responses were observed in all clinical subtypes.

Given the study design, with 18 out of 24 patients (75%) on the investigator choice arm crossing over to the investigational arm, it was not possible to assess any OS benefit from mogamulizumab. Median PFS was poor in each arm (0.93 months for mogamulizumab vs. 0.88 months for chemotherapy), much worse than the Japanese pivotal study (PFS, 5.2 months; OS, 14.4 months).3 The authors concluded that the inclusion of primary refractory patients, stricter cORR criteria (8 weeks vs. 4 weeks), and a higher incidence of poor baseline prognostic factors may account for the inferior efficacy of mogamulizumab in this trial compared to the Japanese studies. In addition, 40% of the patients on the mogamulizumab arm of this trial had received prior zidovudine/interferon-Alpha (IFNa) therapy, whereas no patient had received it in the Japanese studies, suggesting that mogamulizumab may be less effective after zidovudine/IFNa failure. Key differences in disease biology between western and Japanese ATL may also explain differences in response. For example, the presence of CCR4 gain-of-function mutations that have been associated with better outcomes following mogamulizumab therapy in some studies22 were not assessed.

Despite the somewhat disappointing results, this is an important study because it gives us the first prospective cohort of homogeneously-treated, non-Japanese ATL haematologica | 2019; 104(5)

Editorials

patients, and it defines an important, if still inadequate, benchmark for mogamulizumab in this patient population.

The study also exemplifies the futility of standard salvage chemotherapy in R/R ATL, highlighting the importance of ATL patients having access to investigational therapies.

Finally, clinically meaningful improvements were evident even after patients had progressed per protocol, underlining shortcomings in the standardized ATL response criteria.

In conclusion, although responses rates were lower than those observed in Japanese studies, performed in a lower risk population with less stringent efficacy end points, the data reported by Phillips et al. support the conclusion that mogamulizumab is a better treatment option in the second line for R/R western ATL compared to standard chemotherapy, and it should be considered when clinical trials are not available.

8.

9.

10.

11.

12.

13.

14.

15.

References 1. Poiesz BJ, Ruscetti FW, Gazdar AF, Bunn PA, Minna JD, Gallo RC. Detection and isolation of type C retrovirus particles from fresh and cultured lymphocytes of a patient with cutaneous T-cell lymphoma.

Proc Natl Acad Sci U S A. 1980;77(12):7415-7419.

2. Bangham CR. HTLV-1 infections. J Clin Pathol. 2000;53(8):581-586.

3. Phillips AA, Harewood JCK. Adult T Cell Leukemia-Lymphoma (ATL): State of the Art. Curr Hematol Malig Rep. 2018;13(4):300-307.

4. Ishida T, Joh T, Uike N, et al. Defucosylated anti-CCR4 monoclonal antibody (KW-0761) for relapsed adult T-cell leukemia-lymphoma: a multicenter phase II study. J Clin Oncol. 2012;10;30(8):837-842.

5. Ishida T, Jo T, Takemoto S, et al. Dose-intensified chemotherapy alone or in combination with mogamulizumab in newly diagnosed aggressive adult T-cell leukaemia-lymphoma: a randomized phase II study. Br J Haematol. 2015;169(5):672-682.

6. Malpica L, Pimentel A, Reis IM, et al. Epidemiology, clinical features, and outcome of HTLV-1-related ATLL in an area of prevalence in the United States. Blood Adv. 2018; 2(6):607-620.

7. Phillips AA, Shapira I, Willim RD, et al. A critical analysis of prognostic factors in North American patients with human T-cell lymphotropic virus type-1-associated adult T-cell leukemia/lymphoma:

haematologica | 2019; 104(5)

16.

17.

18.

19.

20.

21.

22.

a multicenter clinicopathologic experience and new prognostic score. Cancer. 2010;116(14):3438-3446.

Phillips AA, Fields PA, Hermine O, et al. Mogamulizumab versus investigator choice of chemotherapy regimen in relapsed/refractory adult T-cell leukemia/lymphoma. Haematologica. xxx Gessain A, Cassar O. Epidemiological Aspects and World Distribution of HTLV-1 Infection. Front Microbiol. 2012;3:388.

Seiki M, Hattori S, Hirayama Y, Yoshida M. Human adult T-cell leukemia virus: complete nucleotide sequence of the provirus genome integrated in leukemia cell DNA. Proc Natl Acad Sci U S A. 1983;80(12):3618-3622.

Bangham CRM, Matsuoka M. Human T-cell leukaemia virus type 1: parasitism and pathogenesis. Philos Trans R Soc Lond B Biol Sci. 2017;372(1732).

Fujikawa D, Nakagawa S, Hori M, et al. Polycomb-dependent epigenetic landscape in adult T-cell leukemia. Blood. 2016;127(14):17901802.

Baratella M, Forlani G, Accolla RS. HTLV-1 HBZ Viral Protein: A Key Player in HTLV-1 Mediated Diseases. Front Microbiol. 2017;8:2615.

Kataoka K, Iwanaga M, Yasunaga JI, et al. Prognostic relevance of integrated genetic profiling in adult T-cell leukemia/lymphoma.

Blood. 2018;131(2):215-225.

Mehta-Shah N, Ratner L, Horwitz SM. Adult T-Cell Leukemia/Lymphoma. J Oncol Pract. 2017;13(8):487-492.

Cook LB, Fuji S, Hermine O, et al. Revised Adult T-Cell LeukemiaLymphoma International Consensus Meeting Report. J Clin Oncol.

2019;37(8):677-687.

Shah UA, Chung EY, Giricz O, et al. North American ATLL has a distinct mutational and transcriptional profile and responds to epigenetic therapies. Blood. 2018;132(14):1507-1518.

Katsuya H, Ishitsuka K, Utsunomiya A, et al. Treatment and survival among 1594 patients with ATL. Blood. 2015;126(24):2570-2577.

Tsukasaki K, Utsunomiya A, Fukuda H, et al. VCAP-AMP-VECP compared with biweekly CHOP for adult T-cell leukemia-lymphoma: Japan Clinical Oncology Group Study JCOG9801. J Clin Oncol. 2007;25(34):5458-5464.

Ishida T, Hishizawa M, Kato K, et al. Allogeneic hematopoietic stem cell transplantation for adult T-cell leukemia-lymphoma with special emphasis on preconditioning regimen: a nationwide retrospective study. Blood. 2012;120(8):1734-1741.

Ishida T, Utsunomiya A, Jo T, et al. Mogamulizumab for relapsed adult T-cell leukemia-lymphoma: Updated follow-up analysis of phase I and II studies. Cancer Sci. 2017;108(10):2022-2029.

Sakamoto Y, Ishida T, Masaki A, et al. CCR4 mutations associated with superior outcome of adult T-cell leukemia/lymphoma under mogamulizumab treatment. Blood. 2018;132(7):758-761.

867

PERSPECTIVE ARTICLE Ferrata Storti Foundation

Next-generation sequencing in the diagnosis and minimal residual disease assessment of acute myeloid leukemia Ross L. Levine and Peter J.M. Valk

Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center, New York, NY, USA and 2Department of Hematology, Erasmus University Medical Center, Rotterdam, the Netherlands

1

Haematologica 2019 Volume 104(5):868-871

Introduction Risk-stratification of acute myeloid leukemia (AML) based on recurrent somatic abnormalities has evolved substantially in recent years, as illustrated by the current 2017 European LeukemiaNet (ELN) risk stratification.1 These 2017 ELN AML risk stratification recommendations are based on (cyto)genetic aberrations, including hotspot mutations such as those in NPM1, but also small insertions, deletions and point mutations that occur throughout TP53, RUNX1 and ASXL1, the latter being associated with adverse outcome.1 Next-generation sequencing (NGS) enables reliable detection of patient-specific mutations covering complete genes in molecularly heterogeneous diseases such as AML. NGS should, therefore, be incorporated in the routine work-up of preferably bone marrow specimens for accurate risk stratification in AML. Since risk assessment according to 2017 ELN recommendations only requires knowledge of the status of a handful of wellknown driver mutations,1 targeted NGS, easily reaching a sensitivity of 1-2%, is currently the most appropriate and cost-effective approach for routine testing in AML. Targeted NGS using a variety of gene panels has been successfully introduced in routine clinical laboratories; however, several challenges remain.

Gene panels Correspondence: ROSS L. LEVINE leviner@mskcc.org PETER J.M. VALK p.valk@erasmusmc.nl Received: February 1, 2019.

Accepted: February 14, 2019.

Pre-published: March 28, 2019. doi:10.3324/haematol.2018.205955 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/868 Š2019 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.

868

A number of commercially available gene panels focusing on genes frequently mutated in myeloid malignancies have been introduced, e.g., the Illumina TruSight Myeloid panel, the Archer VariantPlex Core Myeloid panel, the Human Myeloid Neoplasms QIASeq DNA Panel and the AmpliSeq for Illumina Myeloid panel among many others. As expected, these panels contain all genes relevant for the 2017 ELN classification and show an enormous overlap in additional mutational hotspots and complete coverage of genes frequently mutated in myeloid diseases.

In addition to these commercial panels, gene panels can be easily configured to meet local requirements. For instance, if AML patients are classified locally according to 2017 ELN, only NPM1, CEBPA, FLT3, RUNX1, ASXL1, and TP53 need to be included in a small and cost-effective gene panel. These types of NGS-based assays are now emerging.2 Importantly, both commercial NGS-based assays and those developed in-house as well as downstream analyses should be thoroughly validated locally before implementation in daily practice can be considered.

Some genes are particularly difficult to sequence with NGS using gene panels.

Bi-allelic mutations in CEBPA characteristically confer a favorable outcome in patients with AML.1 CEBPA is a GC-rich gene which is notoriously difficult to amplify by polymerase chain reaction (PCR) and sequence, and should be given special attention when incorporated in a gene panel. Although some commercial NGS gene panel protocols do now successfully include this single exon gene, other NGS approaches, such as capture-based NGS or custom panels for CEBPA mutation detection could be considered. FLT3 internal tandem duplications (ITD) can be reliably determined by fragment-length PCR following standardized protocols;3 however, the size of the FLT3 ITD and the duplication itself make it challenging to sequence the variably-sized amplicons appropriately by NGS and subsequently to analyze the FLT3 ITD by sequence alignment to reference sequences.

Moreover, the 2017 ELN recommendations require assessment of the size of the FLT3 ITD clone.1 NPM1-mutant AML cases with high FLT3 ITD/FLT3 wildtype ratios (>0.5) are considered at intermediate risk, whereas NPM1 wildtype AML cases with high FLT3 ITD/FLT3 wildtype ratios are seen as adverse. Standardized NGS-based protocols need to be developed not only for the detection of FLT3 ITD, but also for the quantification of FLT3 ITD/FLT3 wildtype ratios. Examples haematologica | 2019; 104(5)

Next-generation sequencing in AML

of sensitive and specific custom-made FLT3 ITD NGSbased mutation detection assays have been published.4,5

Sequencing A selected number of fusion transcripts have been incorporated in the 2017 ELN recommendations and classify AML patients into various risk categories.1 Although the majority of these gene rearrangements are revealed by cytogenetic analyses, molecular approaches are often complementary. Fusion transcript detection was long limited to those transcripts for which standardized assays, generally real-time quantitative PCR, were available.

However, all fusion transcripts relevant for risk stratification of AML can now be detected in a single assay with RNA-based NGS (RNA sequencing). Given that at least one of the partner genes involved in all clinically relevant fusions is known, RNA sequencing analysis can focus on these specific genes, leaving the possibility of revealing novel fusion partners. Such targeted RNA sequencingbased assays are commercially available. However, since the number of clinically relevant fusion transcripts is limited, one could also consider developing customized methods in which the AML-associated transcripts are amplified by (multiplex) PCR and subsequently sequenced by NGS.6 Again, proper validation of these assays at the local site is essential.

Currently, library preparation and sequencing with amplicon-based NGS assays usually require several days, whereas analyses of the limited number of 2017 ELN genes can be done rather quickly. The introduction of novel NGS machines with faster turnaround times, such as the Illumina iSeq100, and the development of customized assays now enable fast library preparation and overnight sequencing, thus allowing for a quick return of test results to the clinic. This is of particular interest when targetable mutations, such as those in FLT3, IDH1 or IDH2, are needed for selection of the appropriate drug for front-line AML therapy or for relapsed patients with a high disease burden.

Since most of the clinically relevant mutations in myeloid malignancies are known, targeted sequencing is currently the method of choice. However, it can be foreseen, when turnaround times and costs are reduced, that whole exome or whole genome sequencing will become the standard approach to genomic characterization of AML at diagnosis. The use of whole exome or whole genome sequencing will allow identification of all somatic coding mutations, including those that are targetable but less frequently present in AML. Moreover, one can prioritize analysis of key AML genes first, such that initial results regarding the clinically most relevant genes can be obtained with a short turnaround and more comprehensive genomic profiling can follow later. Furthermore, whole genome sequencing allows identification of novel biomarkers located outside of protein coding regions, which may be useful not only for proper assessment of the prognosis but also for detection of minimal residual disease (MRD) in AML as they can be used to identify and follow leukemic clones regardless of their role in AML initiation and maintenance.

Minimal residual disease Our improved understanding of the molecular landscape of AML has resulted in better treatment decisions at the time of complete remission after induction treatment. haematologica | 2019; 104(5)

Although the majority of AML patients achieve complete remission, many eventually relapse. Thus, there is still a great need for adequate prediction for subsequent relapse to adapt treatment accordingly and improve the outcomes of patients at high risk of relapse. MRD detection has already proven to have substantial value in predicting relapse and overall survival when applied to AML in complete remission but the use of molecular enumeration of MRD has been limited to only specific, molecularly defined subtypes of AML.7-9 By contrast, flow cytometric analysis of MRD can be done in nearly all AML patients, but is operator- and center-dependent and there is no centrally agreed approach to enumerate flow-based MRD in AML. NGS enables MRD detection by measuring all mutations, including patient-specific persistent mutations, in complete remission. In fact, it has recently been shown that molecular MRD detection by NGS is applicable to virtually every newly diagnosed AML patient because of the frequent prevalence of multiple molecular aberrations among patients with AML.10-13 However, MRD detection based on NGS must overcome several challenges before it can be reliably introduced into clinical practice.

The known oligoclonality of the disease at diagnosis has a clear impact on MRD detection. Molecular markers in small AML subclones at diagnosis could easily be missed by panel-based NGS at lower depth. However, these small populations of cells may be selected during therapy and ultimately result in AML relapse. This issue could be overcome by sensitive detection of all possible mutations frequently present in myeloid malignancies.

However, because of the relatively high error rates of current standard NGS technologies, reliable detection of a multitude of mutations at high sensitivities (<0.01%) is not yet easily achieved. In fact, the currently high intrinsic error rates (1 to 0.1%) impede sensitive MRD monitoring at later time points during therapy as well. At these time points certain targets present at diagnosis can be sequenced individually with single amplicons at high depth, but true residual mutations may still not be reliably discriminated from noise at levels below 0.1%.

Attempts should be made to improve the signal-to-noise ratios in order to detect low-level variants accurately.14 Genomic DNA isolated from bone marrow, peripheral blood or mononuclear cells is generally of high quality, but noise in NGS is subsequently introduced at different levels during library preparation and sequencing.14 The rate of sequencing artefacts can be reduced biochemically, e.g., by using proof-reading polymerases, or computationally; however, these corrections are only modest and cannot attenuate errors/artefacts entirely.

Alternative strategies should be explored. Recently, various error-corrected NGS methodologies using molecular barcoding have been introduced. 14 Error-corrected sequencing is based on barcoding the individual DNA molecules used for NGS library preparation. Using the unique sequence tag all derivative reads, which arise from a common founder, can be recognized after computational NGS, which enables removal of PCR duplicates and false mutation calls. These approaches and protocols15 have been shown to increase the specificity of low-frequency mutation detection. 14 However, whether error-corrected sequencing will improve MRD detection by NGS in AML remains to be demonstrated in large cohorts of AML. 869

R.L. Levine and P.J.M. Valk et al.

Recently, several studies addressed NGS-based MRD detection in relatively large AML cohorts from clinical trials, all demonstrating that NGS-based MRD has a profound prognostic impact in patients with AML.11-13,16 In these studies persistent mutations in complete remission were measured with gene panels,11 capture-based deep sequencing10,12,16 or targeted sequencing.13 The ampliconbased approach was specifically designed for MRD detection by including error-correction,13 which suggests that previous NGS-based MRD studies were not yet optimal.

The results of these initial studies do not allow any firm conclusions to be drawn with regard to the superiority of error-corrected NGS for MRD detection in AML. However, the fact that NGS MRD has consistent prognostic value implies that technological improvements should be made in order to further optimize relapse prediction in AML, assuming that in these initial studies minor AML MRD clones were missed in complete remission. The ELN MRD Working Group is currently aiming to improve and harmonize methodologies for NGS-based detection of MRD in AML. Another successful approach to correct for noise is to use a site-specific error model with a sufficiently large set of reference samples without mutations.11 In such a model the distribution of variants is determined in a reference set without mutations, for example, remission samples. MRD is subsequently defined by those mutations, such as the ones present at diagnosis, which are statistically significantly different from the distribution of variants in the reference set. In this case the detection sensitivity of mutations is variable and dependent on the average coverage for that specific locus for all samples, the observed error variance of the site-specific variant in the reference set (a high variance results in decreased detection sensitivity) and the number of control samples available. A major drawback of this approach is that a set of reference samples to determine MRD has to be available. MRD measurement in a single sample without the dependence of a large reference set is obviously a preferred method since it will be more easily implemented in clinical practice.

Clonal hematopoiesis In the initial NGS-based MRD studies11-13,16 it became clear that gene mutations persisting in complete remission that are well-known to be associated with clonal hematopoiesis17,18 do not have an impact on the risk of relapse, despite the fact that they are among the most common disease-initiating drivers of AML. As a result of high-dose chemotherapy, AML patients with these mutations are apparently brought back into a state of clonal hematopoiesis, in which mutations occurring late in leukemogenesis are irradiated, but mutations also found in clonal hematopoiesis persist. It is clear that these persisting mutations, also known as clonal mutations of indeterminate potential, add another layer of complexity to MRD detection in AML. In studies of molecular MRD, clonal hematopoiesisrelated mutations in DNMT3A, TET2 and ASXL1 (DTA) were considered clonal hematopoiesis rather than residual leukemia. Besides acquired mutations in DTA, other well-known pathogenic mutations such as those in TP53, PPM1D, JAK2, CBL, SRSF2 and SF3B1 are involved in clonal hematopoiesis, however, at lower incidence.17,18 In fact, many of these mutations also persist in complete remission with high variant allele frequencies. It needs to 870

be determined whether and, if so, to what extent persisting mutations other than those in DTA represent true residual leukemia or clonal hematopoiesis, respectively with and without an increased risk of relapse. In a disease as heterogeneous as AML these analyses will require large cohorts of patients. Thus, while the recent developments in NGS-based MRD detection represent major steps forward in predicting relapse, they remain imperfect. It is expected that a better distinction between clonal hematopoiesis and residual leukemia will improve prediction of AML patients at higher risk of relapse.

How can the discrimination between true residual leukemia and clonal populations of cells be improved?

The numbers of AML patients included in the initial studies precluded detailed analyses of rare mutations and indepth analysis of common non-DTA mutations. It is conceivable that non-DTA mutations are a mixture of mutations representative of either true leukemia or clonal hematopoiesis. Improved discrimination of these two conditions by means of types of persisting mutations may have significant value with regards to relapse prediction.

Along the same lines it has recently been shown that it may be feasible to discriminate clonal hematopoiesis from pre-AML in healthy individuals.19 Pre-AML cases were distinct from controls and had more mutations per sample, higher variant allele frequencies, indicating clonal expansion, and showed enrichment in specific genes.19 Similar approaches could possibly better differentiate clonal hematopoiesis from true leukemia after induction treatment.

Sensitivity, timing and tissues for next-generation sequencing How does NGS-based MRD detection perform as compared to the â&#x20AC;&#x2DC;golden standardâ&#x20AC;&#x2122;, multiparameter flow cytometric MRD detection? There are only limited studies with a rigorous comparison between NGS- and multiparameter flow cytometric MRD detection.11,16 These studies demonstrate that there is a 70% concordance with regard to MRD detection using the two technologies, and that those patients who are MRD-positive according to both techniques have the highest risk of developing an AML relapse.11,16 Interestingly, however, those AML cases with discordant results from NGS and flow cytometry have adverse outcome, such that MRD positivity has value whether determined by flow cytometry, molecular techniques, or both.11,16 We need to improve both the sensitivity of our NGS assays and our understanding of the biology of clonal hematopoiesis after high-dose chemotherapy to resolve the discordant cases and determine whether we require both technologies or only one to enumerate MRD. At which time point(s) should NGS-based MRD detection be carried out? In the majority of AML studies NGSbased MRD detection was performed after high-dose induction treatment. Although this time point may be most suitable for choosing the proper consolidation treatment, it is not known whether MRD assessment at other time points may be better prognostic indicators. Few studies have shown that MRD before and after consolidation, such as in the setting of allogenic transplantation, affects clinical outcome.12,16,20 A bone marrow biopsy is an invasive procedure that gives stress and physical discomfort to a patient and creates a risk of infection. Patients with chronic myeloid haematologica | 2019; 104(5)

Next-generation sequencing in AML

leukemia are monitored by measuring BCR-ABL1 levels in peripheral blood. Likewise, in mutant NPM1 AML, response to treatment can be effectively ascertained in peripheral blood. Studies should be carried out to determine whether peripheral blood is also an alternative for NGS-based MRD monitoring in AML for a broader spectrum of molecular alterations.

Conclusions NGS at diagnosis is essential for accurate risk stratification of AML patients according to the 2017 ELN recommendations and has now been implemented in many molecular diagnostic laboratories. Currently, the major limitations of the NGS-based methodology of detecting MRD are related to the limited sensitivity and specificity of the assays and the inability to discriminate correctly between residual leukemia and clonal hematopoiesis.

References 1. Dohner H, Estey E, Grimwade D, et al.

Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel.

Blood.

2017;129(4):424-447.

2. Onecha E, Linares M, Rapado I, et al. Novel deep targeted sequencing method for minimal residual disease monitoring in acute myeloid leukemia.

Haematologica.

2018;104 (2):288-296.

3. Murphy KM, Levis M, Hafez MJ, et al.

Detection of FLT3 internal tandem duplication and D835 mutations by a multiplex polymerase chain reaction and capillary electrophoresis assay. J Mol Diagn. 2003;5 (2):96-102.

4. Schranz K, Hubmann M, Harin E, et al.

Clonal heterogeneity of FLT3-ITD detected by high-throughput amplicon sequencing correlates with adverse prognosis in acute myeloid leukemia. Oncotarget. 2018;9(53): 30128-30145.

5. Levis MJ, Perl AE, Altman JK, et al. A nextgeneration sequencing-based assay for minimal residual disease assessment in AML patients with FLT3-ITD mutations. Blood Adv. 2018;2(8):825-831.

6. Dillon LW, Hayati S, Roloff GW, et al.

Targeted RNA-sequencing for the quantification of measurable residual disease in acute myeloid leukemia. Haematologica.

2018;104(2):297-304.

haematologica | 2019; 104(5)

Improvements need to be made in these areas before NGS-based MRD detection can be successfully implemented in routine practice.

Acknowledgments R.L.L. is on the supervisory board of Qiagen and is a scientific advisor to Loxo, Imago, C4 Therapeutics and Isoplexis, which each include an equity interest. He receives research support from and consulted for Celgene and Roche, he has received research support from Prelude Therapeutics, and he has consulted for Incyte, Novartis, Morphosys and Janssen. He has received honoraria from Lilly and Amgen for invited lectures and from Gilead for grant reviews.

Funding This work was supported in part by MSKCC Support Grant/Core Grant P30 CA008748 and the Netherlands Organization for Health Research and Development ZonMw (846002002).

7. Hourigan CS, Gale RP, Gormley NJ, Ossenkoppele GJ, Walter RB. Measurable residual disease testing in acute myeloid leukaemia. Leukemia. 2017;31(7):14821490.

8. Ivey A, Hills RK, Simpson MA, et al.

Assessment of minimal residual disease in standard-risk AML. N Engl J Med. 2016;374 (5):422-433.

9. Kronke J, Schlenk RF, Jensen KO, et al.

Monitoring of minimal residual disease in NPM1-mutated acute myeloid leukemia: a study from the German-Austrian acute myeloid leukemia study group. J Clin Oncol.

2011;29(19):2709-2716.

10. Klco JM, Miller CA, Griffith M, et al.

Association between mutation clearance after induction therapy and outcomes in acute myeloid leukemia. JAMA. 2015;314 (8):811-822.

11. Jongen-Lavrencic M, Grob T, Hanekamp D, et al. Molecular minimal residual disease in Acute Myeloid Leukemia. N Engl J Med. 2018;378(13):1189-1199.

12. Morita K, Kantarjian HM, Wang F, et al.

Clearance of somatic mutations at remission and the risk of relapse in acute myeloid leukemia.. J Clin Oncol. 2018;36(18):17881797.

13. Thol F, Gabdoulline R, Liebich A, et al.

Measurable residual disease monitoring by NGS before allogeneic hematopoietic cell transplantation in AML. Blood.

2018;132(16):1703-1713.

14. Salk JJ, Schmitt MW, Loeb LA. Enhancing the accuracy of next-generation sequencing for detecting rare and subclonal mutations.

Nat Rev Genet. 2018;19(5):269-285.

15. Wong WH, Tong RS, Young AL, Druley TE. Rare event detection using error-corrected DNA and RNA sequencing. J Vis Exp. 2018;(138).

16. Getta BM, Devlin SM, Levine RL, et al.

Multicolor Flow Cytometry and Multigene Multicolor flow cytometry and multigene next-generation sequencing are complementary and highly predictive for relapse in acute myeloid leukemia after allogeneic transplantation. Biol Blood Marrow Transplant. 2017;23(7):1064-1071.

17. Genovese G, Kahler AK, Handsaker RE, et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014;371(26):2477-2487.

18. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371 (26):2488-2498.

19. Abelson S, Collord G, Ng SWK, et al.

Prediction of acute myeloid leukaemia risk in healthy individuals. Nature. 2018;559 (7714):400-404.

20. Kim T, Moon JH, Ahn JS, et al. Next-generation sequencing-based posttransplant monitoring of acute myeloid leukemia identifies patients at high risk of relapse. Blood.

2018;132(15):1604-1613.

871

REVIEW ARTICLE Ferrata Storti Foundation

Evolutionary trajectory of leukemic clones and its clinical implications Amos Tuval1,2 and Liran I Shlush1,3

Department of Immunology, Weizmann Institute of Science, Rehovot; 2Hematology Department, Meir Medical Center, Kfar Saba and 3Hematology Department, Rambam Healthcare Campus, Haifa, Israel

1

Haematologica 2019 Volume 104(5):872-880

ABSTRACT

T

he ontogeny of acute myeloid leukemia is a multistep process. It is driven both by features of the malignant clone itself as well as by environmental pressures, making it a unique process in each individual. The technological advancements of recent years has increased our understanding about the different steps that take place at the genomic level. It is now clear that malignant clones evolve, expand and change even during what seem to be clinically healthy or “cured” periods. This opens a wide window for new therapeutic and monitoring opportunities.

Moreover, prediction and even early prevention have become possible goals to be pursued. The aim of this review is to shed light upon recent observations in leukemia evolution and their clinical implications. We present a critical view of these concepts in order to assist clinicians when interpreting results of the ever growing myriad of genomic diagnostic tests. We wish to help clinicians incorporate genetic tests into their clinical assessment and enable them to provide genetic counseling to their patients.

Correspondence: AMOS TUVAL amos.tuval@weizmann.ac.il LIRAN I SHLUSH liran.shlush@weizmann.ac.il Received: January 21, 2019.

Accepted: April 4, 2019.

Pre-published: April 18 2019. doi:10.3324/haematol.2018.195289 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/872 ©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.

872

Introduction Acute myeloid leukemia (AML) is a clonal disorder that originates in leukemic stem and progenitor cells, termed blasts.1 AML clinically manifests with the accumulation of these immature cells that exhibit uncontrolled growth and decreased apoptosis, and that lack normal differentiation. AML is clinically defined when blasts make up 20% or more of the cellular component of the bone marrow (BM).

These cells inhibit normal hematopoiesis resulting in BM failure.2 Although most of the leukemic cells can be eradicated during the first course of therapy, most patients succumb to disease relapse within the first two years.3 Acute myeloid leukemia is characterized by a relatively small (compared to solid tumors), recurrent set of somatic mutations, designated leukemia driver mutations.4 The various mutations can be used to classify AML subtypes. Genomic analysis of 1540 AML patients identified distinct AML subgroups according to their mutational background. Some mutations (such as in the DNMT3A gene) are shared by few subgroups, some of the mutations co-occur (e.g. DNMT3A, NPM1 and FLT3-ITD), while others are not usually found in the same clone (TP53 and NPM1).5 Most somatic mutations in AML occur stochastically across the genome without any foci of localized hypermutation.6 Theoretically, if the effective size of the stem cell population had been large enough, and it had been given enough replication cycles, it would have been reasonable to assume that almost every possible mutation can be found at the single cell level. Yet, not all mutations are shared by all AML subtypes. Moreover, a specific mutational signature with an elevated rate of C>T transitions was found in AML. This mutational signature was related to spontaneous deamination of 5-methyl-cytosine and was correlated with age.6 This implies that the aging BM niche exerts a selective pressure on the leukemic stem cells and shapes their mutational profile.

Clones become fitter as they accumulate mutations and evolve. However, studies in other malignancies suggest that most mutations that are being accumulated during cancer evolution are deleterious to tumor fitness,7 and are called “passenger haematologica | 2019; 104(5)

Evolutionary trajectory of AML

mutations”. Therefore, it might seem paradoxical that, in some cases, an increased number of somatic mutations predict a worse prognosis and a more rapid evolution.8 It is, therefore, important to stress that it is not only the number of mutations, but also the identity of the specific mutation acquired, that determines progression.9 Only these true “driver” mutations confer an advantageous phenotype.

Over recent years, studies aimed at depicting the clonal structure of AML have been published. These studies performed deep sequencing of primary AML samples taken from patients, patient-derived xenografts, and from in vitro cultures, and used the various somatic variant allele frequencies (VAF) as measures of clonal sizes [assuming no copy number variability (CNV) or loss of heterozygosities (LOH)]. These studies shed light on the selection and expansion that these clones undergo during their evolution and following therapy. Combining this information with clinical data from different time points improves our ability to predict treatment outcomes, enabling us to personalize therapy. Moreover, this approach raises the hope that healthy individuals can be screened for AML, and maybe even preventing the disease; something that was once considered unachievable.

The clonal evolution of most AML subtypes can be viewed as a multistep process. It is now well accepted that this process can be schematically divided into three stages according to the clinical presentation. Each evolutionary stage has a different time frame and is characterized by typical somatic mutations that are, therefore, categorized into three groups according to the timing of their appearance: pre-leukemic, leukemic, and late events (Table 1).

Although acute promyelocytic peukemia (APL), AML with KDM2A translocations, and the core-binding factor (CBF) leukemias do not have a clear pre-leukemic stage, they too develop over time and acquire late events, as discussed below.

This review summarizes our current knowledge regarding somatic mutations in AML, their contribution to clonal fitness under different selective environmental conditions, and their correlation with patients’ clinical outcomes.

Pre-leukemic stage Pre-leukemic mutations are somatic mutations that are found in leukemic blasts as well as in hematopoietic progenitors and mature cells from different lineages that share a common ancestral stem cell. This stem cell, which is still capable of differentiation, is defined as a pre-leukemic hematopoietic stem and progenitor cell (preL-HSPC).10,11 Such preL-HSPCs were isolated from AML patients at diagnosis, remission and relapse.10-12 By definition, the term pre-leukemic can only be inferred retrospectively, after the diagnosis of AML has been made.

Following these findings, somatic mutations were identified in the hematopoietic system of healthy individuals in various allele frequencies increasing with age, a phenomenon that was termed age-related clonal hematopoiesis (ARCH).13,14 In fact, deep sequencing techniques revealed the presence of DNMT3A and TET2 mutations in nearly all individuals; however, most of them were at lower VAF, in comparison to the original reports (median VAF 0.0024), and remained stable over time.15 Such a ubiquitous phenomenon probably represents the haematologica | 2019; 104(5)

general structure of the aging human hematopoietic system, as the same findings could not be replicated among young individuals (aged 20-29 years).16 The large number of human hematopoietic stem cells (HSC) (estimated to be within the range of 50,000-200,000)17 and the number of somatic mutations in each adult single HSC (approx. 1000 mutations) suggest an estimated HSC pool mutation burden of 108,17 and explain why somatic mutations can be present at low VAF in every individual.14 However, with age, an exponential increase in the prevalence of ARCH13,14,16 occurs. In addition, this correlation with age is specific to mutations found in DNMT3A, TET2 and a few additional candidate driver genes,14 suggesting that mutations in these genes confer a selective advantage to HSPCs. The selective advantage that mutations in DNMT3A and TET2 confer is probably introduced during the third or fourth decade of life. A second selective advantage is introduced later, during the fifth decade and onward, reflecting the aging BM selecting for HSC-carrying spliceosome machinery mutations (SRSF2, U2AF1, SF3B1, ZRSR2, DDX41, EZH2, ASXL1, etc.).18,19 The fact that different individuals can carry clones of different sizes at the same age suggests differences in risk factors for clonal expansion.

The presence of such a mutated clone in allele frequency of more than 2% is termed clonal hematopoiesis of indeterminate potential (CHIP).20 It was found to be associated with development of hematologic cancers, cardiovascular morbidity, chronic obstructive pulmonary disease, and with an increase in all-cause mortality.13,14,21 While these clones predict a somewhat dismal prognosis, only a small number of individuals that harbor such a clone will even-

Table 1. Age-related clonal hematopoiesis defining events and the risk that they confer for acute myeloid leukemia progression.9,22

Pre-leukemic / ARCH-defining events Low risk ASXL1 BCOR CALR# CBL# DNMT3A KIT# KMT2D KRAS# NF1 RAD21 SF3B1# TET2$

High risk #

Leukemic mutations

"Late events"

NPM1c#

FLT3-ITD# CEBPA (bi-allelic) KIT# KRAS# NRAS# PTPN11 WT1

Unclear risk

IDH1 BRAF# # IDH2 CEBPA (mono-allelic) JAK2# EZH2@ PHF6 FLT3-TKD# PPM1D$* GATA2 RUNX1 KDM6A SRSF2# KMT2C TP53* NRAS# # U2AF1 PAX5 PTPN11 SMC1A SMC3 STAG2 ZRSR2@

Unless specified, genes that harbor recurrent driver mutations are mentioned (as opposed to specific variants). Some mutations are rarely found, therefore, it is difficult to determine with certainty the risk that they confer. These were designated as "Unclear Risk". Some mutations can appear as an early evolutionary event as well as a "late event" (e.g. KIT, NRAS).

Translocations, such as t(8;21) are not mentioned since they can be missed by targetedsequencing as well as by exome-sequencing methods. Examples for "leukemic mutations" and for "late events" are also presented. *Enriched following chemotherapy for a non-related cancer [other than acute myeloid leukemia (AML)]. @Probably represent myelodysplastic syndromes. $Truncating events only (frameshift and missense mutations). #Only specific hotspots in the gene. ARCH: age-related clonal hematopoiesis.

873

A. Tuval and L.I. Shlush et al.

tually develop AML (approx. 0.1-3%).13,14 Nevertheless, since some of these clones do evolve into full-blown leukemia, these observations can be exploited to build a model for early detection of AML, at its pre-clinical, pre-leukemic phase. In fact, two recently published papers studied large prospective cohorts of healthy individuals and compared benign clonal hematopoiesis (one that did not evolve into AML) with malignant, pre-AML, clonal hematopoiesis. These studies used deep targeted sequencing methods to search for mutations in driver genes in a total of 307 cases (individuals that subsequently developed AML) and 626 age- and gender-matched controls (individuals that did not develop AML) in blood samples obtained 6-9 years before AML diagnosis. These studies found discriminative characteristics between the two cohorts.9,22

Clone size A pre-leukemic state is manifested by an increased incidence of clonal hematopoiesis. Setting a 10% threshold for VAF value significantly discriminated pre-AML from controls. Thirty-nine percent of pre-AML individuals have clones of this size, as opposed to 4% of control individuals. Although statistically significant, there is a large overlap between VAF values of driver mutations found in benign and in malignant ARCH (especially for DNMT3A and TET2 mutations) that precludes it from being a single predictor of a rare disease such as AML.

Number of accumulated mutations Pre-AML individuals have significantly more mutations in driver genes (including a few variants in the same gene, e.g. in DNMT3A22) per individual (not necessarily in the same clone) when compared with controls. This is especially evident among older individuals (>60-65 years of age) underscoring the time frame required for mutations to accumulate. Nevertheless, it is important to note that a substantial number of patients (20-46%) do develop AML without having a mutation in any driver gene prior to the diagnosis. This decreases the negative predictive value of these models.

When measured at a certain time point, these two characteristics indirectly reflect increased clonal fitness, as manifested by increased expansion and increased number of replications with the accumulation of mutations over time. These characteristics were also found to be predictive of progression to myeloid neoplasms when found during the evaluation of unexplained cytopenias.8

Specific high-risk mutations Progression to AML was found to be preceded by accumulation of specific high-risk mutations. These mutations were more prevalent among pre-leukemic clonal hematopoiesis when compared to benign ARCH. Specifically, the presence of spliceosome-machinery mutations in SRSF2 P95R and U2AF1 Q157P as well as in TP53, in IDH1 R132, in IDH2 R140 and in RUNX1 (even at VAF values <10%) confer the highest risk for subsequent AML development in healthy individuals.9,22 When these clones appear at a relatively young age (>50 years) they tend to evolve into AML. The simple explanation for this could be that there is more time for AML transformation to take place. Another explanation could be that the environment that positively selected this clone continues to exert its selective pressure, eventually leading to AML. Table 1 summarizes the various ARCH-defining events 874

and the risk that each of them confers for AML progression.

Temporal progression While some low-risk clones can remain stable over a period of 3-10 years,15 clones that are characterized by high-risk mutations show a more rapid increase in their size, as manifested by an increase in their VAF values over time9,22 (Figure 1). Additional prospective cohorts might better define which clone develops to other hematologic malignancies [e.g. myelodysplastic syndromes (MDS) or myeloproliferative neoplasms (MPN)].

Progression to AML depends on the identity of the initiating mutation and on the identity of additional mutations that are subsequently accumulated. It is conceivable that many factors influence the timing of the appearance of the mutation and the positive selection of such a clone, among which are probably the underlying specific germline background.

In addition, specific environmental pressures confer a selective advantage to HSCs carrying specific mutations; clear examples are TP53 and PPM1D mutations that are enriched following exposure to chemotherapy and radiotherapy. Chemotherapy does not increase the number of somatic single nucleotide variants or the percentage of chemotherapy-related transversions. Rather, it positively selects for pre-existing TP53 and PPM1D mutated clones.23-29 Moreover, a third-generation, single-molecule real-time sequencing assay with long-read length of AML and MDS samples exposed different TP53 variants residing on different alleles in each sample.30 This emphasizes the importance of the environmental conditions that select a certain phenotype, thus enabling the evolution of a few clones in parallel, all sharing similar driver mechanisms (TP53 mutations). It is important to note that chemotherapy exerts a selective pressure regardless of the specific mutation that characterizes the pre-leukemic clone. Following chemotherapy, BM is enriched with pre-leukemic clones and their prevalence increases by 10% or even 30% among younger and elderly individuals, respectively, when compared to their prevalence in the general agematched population.24,13 The clones that were selected can neither be categorized according to a certain mutation, nor according to a certain chemotherapy (with the exception of topoisomerase II inhibitors, for which see below).

However, they can be divided into three groups according to patient age groups, with younger individuals enriched with DNMT3A mutated clones. This holds true also for AML patients in remission that were found to have residual pre-leukemic clones.19,31 This implies that most preleukemic clones have an inherent chemoresistance (Figure 2), a phenomenon that was also shown in both in vivo and in vitro models.32 The time frame of evolution from pre-leukemia to AML depends both on the context (extrinsic factors) and the driver mutations (intrinsic factors). Pre-leukemia in healthy individuals usually progresses slowly with a latency period that can sometimes be as long as 20 years.

Presence of specific mutations was correlated with a shorter timeframe, as in the case of RUNX1 mutations (associated with a rapid progression to AML of <2 years) and of TP53 mutations.22 Specifically, following chemotherapy, TP53 mutated clones, as well as PPM1D mutated clones, evolve to hematologic malignancies withhaematologica | 2019; 104(5)

Evolutionary trajectory of AML

in as few as 6 months to up to 10 years.23,24 Another example is MLL-rearranged AML; this usually develops within 6-18 months following exposure to topoisomerase II inhibitors. In contrast to therapy-related AML with somatic mutations in TP53 or PPM1D, where chemotherapy selects pre-existing mutated clones, MLL-rearrangement is assumed to be induced directly by topoisomerase II inhibitors.19,33 Although IDH mutations can be viewed as high-risk mutations, their presence was not found to be correlated with a shortened AML latency.22 This can be explained by the fact that some of the high-risk mutations occur early along the evolutionary trajectory of the clone; they can be considered as the initiating event (U2AF2, SRSF2, TP53 and RUNX1). Other high-risk mutations (IDH1 and IDH2) tend to appear later and require additional, co-operating, driver mutations in order to progress into AML. Indeed, certain combinations of mutations (when found in the same individual) can shorten the time to AML diagnosis, such as when DNMT3A is found with spliceosomal machinery mutations.22 Importantly, the size of the clone9 and the number of mutations identified9,22 mean a shortened interval for AML progression. Clones with increased fitness, as manifested

A

B

C

haematologica | 2019; 104(5)

by their size (VAF) and replication rate (number of mutations), are prone to acquire additional mutations until transforming, â&#x20AC;&#x2DC;leukemicâ&#x20AC;&#x2122;, mutations occur.

Leukemic stage Molecular analysis of AML reveals a set of somatic mutations that are only present in the leukemic blasts but are absent in normal hematopoietic progenitors, in nonmyeloid, and in mature cells. These mutations were not found in healthy individuals.9,13,14,22 These mutations can be further divided into leukemic mutations, which are shared by all the leukemic blasts, and late events. While the former represent the leukemic transformation, the latter represent subclones, as can be inferred from their lower VAF value (Figure 1).

Since there are already subclones at the time of diagnosis, determining the order of acquisition of the mutations is limited by the sensitivity of the sequencing method used to detect rare clones. It can be achieved by reconstructing the phylogenetic tree of the leukemia with single cell or sub-population sequencing.34 However, there is a general consensus that one mutation can be considered to

Figure 1. High-risk and low-risk pre-leukemic somatic mutations. X: an acquired somatic mutation. Clone size [as manifested by variant allele frequency (VAF) value] is represented by the size of the oval shape. Clonal expansion is represented by the rising curve. (A) Low-risk age-related clonal hematopoiesis (ARCH) mutations, such as DNMT3A or TET2 mutations, are acquired at a relatively young age (marked in white). Most of these clones will not progress to acute myeloid leukemia (AML).

(B) Pre-leukemic clones, characterized by similar low-risk mutations have an increased fitness (as manifested by an increased VAF).

They acquire additional pre-leukemic mutations (marked in yellow and red), not necessarily in the same clone. These cells are hematopoietic stem and progenitor cells (HSPCs), still capable of differentiation and sustain hematopoiesis. Once a clone acquires a leukemic, transforming, mutation (NPM1, for instance, marked in green) it will progress rapidly to an overt AML with loss of differentiation capacity and uncontrolled proliferation.

Retrospectively, its preceding clones are referred to as pre-leukemic. Leukemic mutations are shared by all the leukemic blasts, hence they have a high VAF (50%) in the leukemic clone. Late events (e.g. FLT3-ITD, marked in purple) appear later along the AML evolutionary trajectory, are shared by subclones, and represent the clonal heterogeneity of the leukemia; they have VAFs â&#x2030;¤50%. The exact timing of AML diagnosis can vary.

Therefore, late events are usually already present when the actual diagnosis is made. Single cells or sub-populations have to be sequenced in order to accurately determine the order of acquisition of the mutations. (C) Spliceosomal machinery, and other high-risk mutations (such as SRSF2, U2AF1, IDH1, IDH2 and TP53 that are marked in pink) are usually acquired at a more advanced age. These clones expand more rapidly (as manifested by the rate of increase of their VAF value) and most will lead to AML.

875

A. Tuval and L.I. Shlush et al.

be leukemic; this is a small insertion/duplication in the terminal exon of NPM1 that causes the mutant protein to be aberrantly localized in the cytoplasm (hence, designated NPM1c). Detection of this mutation accompanies a dramatic change in the phenotype of the clone with a rapid proliferation and acquisition of additional mutations.

When individuals were found to harbor an NPM1c mutated clone, they were subsequently diagnosed with NPM1c AML within up to 3 months.22,35 This mutation was most clearly shown to be a marker for leukemic blasts when risk for relapse was found to be correlated with its presence in blood samples of AML patients in remission.36 Targeting cells harboring transforming, leukemic, mutations seems plausible not only for monitoring purposes, but also for therapy.37 In fact, genomic editing of this mutation in cell lines as well as in primary human AML samples using the CRISPR/Cas9 system disrupted the mutant allele and led to nuclear re-localization of the protein. This reverted the leukemic phenotype, resulting in differentiation and a reduced proliferation rate. A nuclear export inhibitor had a similar effect on NPM1c, both at the molecular level as well as at the cellular level, when tested on a cell line and on primary AML samples. It also resulted in prolonged survival of NPM1c-mutated leukemic mice.38

Late events Late events are mutations that appear later on during leukemic evolution, represent clonal selection and heterogeneity, and are not shared by all leukemic blasts.

Examples for such variants are activating mutations in tyrosine kinase receptors (such as FLT3-ITD, KIT, RAS),

emphasizing their role in increasing clonal proliferation capacity.39 Other examples are WT1,39 transcription factor CEBPA bi-allelic mutations,19 and del(7q) in TP53-mutated AML.23 Interestingly, there are some exceptions to these â&#x20AC;&#x153;rulesâ&#x20AC;?: some mutations that were detected in healthy individuals, for example, IDH2, were also described as late events in AML.34 In addition, the FLT3 D835 mutation, usually considered as a late event, was found to be preleukemic.9 Some late events co-occur with certain leukemic events, for example, the FLT3-ITD and NPM1 mutations.4 This strong link between leukemic mutations and late events, rather than between pre-leukemic mutations and late events, was recently demonstrated by HĂśllein et al.40 They described patients with NPM1-mutant AMLs that developed a NPM1 wild-type AML after therapy. Both types of leukemia evolved on a similar pre-leukemic background.

Interestingly, FLT3-ITD was significantly more frequent among NPM1-mutant AML.40 Late events, such as FLT3ITD or KIT mutations, can be found in unique AML subtypes, such as APL and CBF-AML, respectively, as described below.

Targeting cells according to late events can prove beneficial and was employed using tyrosine kinase inhibitors.41 Nevertheless targeting a few subclones can result in a positive selection of other subclones with a different genomic landscape.42 Monitoring residual clones using late events as clonal markers should be done with great care, since this approach might miss subclones lacking these markers.43,44

Unique acute myeloid leukemia subtypes A few AML subtypes, such as CBF-AML and APL, are

Figure 2. Pre-leukemic clones have inherent chemoresistance. Pre-leukemic clones (blue) undergo positive selection by chemotherapy administered for a non-related cancer [other than acute myeloid leukemia (AML)]. They expand and evolve into t-AML (green). Pre-leukemic hematopoietic stem and progenitor cells (HSPCs) have inherent chemoresistance, thus they also survive following AML induction chemotherapy and reconstitute clonal hematopoiesis. Most relapses occur within the first 2 years and originate from residual leukemic clones that can be identified at diagnosis and that were not eradicated by AML therapy. Rare events of second AML (red) stem from (mostly, the same) pre-leukemic clones that evolved again into AML following a more prolonged latency.

876

haematologica | 2019; 104(5)

Evolutionary trajectory of AML

diagnosed by the presence of their typical chromosomal aberrations, regardless of actual blast count.45,46 These chromosomal abnormalities, which result in novel fusion RNA transcripts, are being used for monitoring minimal residual disease following therapy and dictate pre-emptive treatment upon detection.47,48 While this implies that these chromosomal aberrations should be regarded as leukemic events, some of the chromosomal abnormalities of CBFAML were retrospectively detected in blood samples taken more than 10 years before the patients were diagnosed with AML,49 as well as up to 7 years following allogeneic transplantation in patients without evidence of disease.50 In fact, when stem cells and cells from various lineages were sorted from BM of CBF-AML patients in remission, t(8;21) translocations were identified in B cells as well as in myeloid cells. This suggests that this event occurred at the stem cell level, still capable of differentiation.51 A debate continues over the cell of origin of APL. It was shown that APL blasts have a gene expression profile with T-cell lymphoid features52 and that they can engraft immune-deficient mice.53 This raises the possibility that APL blasts (especially blasts of the hypogranular variant) originate in early multipotent progenitors prior to lineage commitment.54 Nevertheless, t(15;17) translocation was identified only in CD34 positive (CD34+) precursor cells of the myeloid lineage and not in B or T lymphocytes.55 Thus, it is generally accepted that the APL transforming event occurs at a more committed progenitor cell.56 In contrast to the aforementioned pre-leukemic HSPCs (e.g. DNMT3A mutated), these cells do not maintain multilineage hematopoiesis.

Leukemic cells that harbor APL and CBF-AML translocations acquire additional mutations as they evolve.

Similarly to other AML subtypes, additional somatic mutations that were identified in these clones involved tyrosine kinase receptor genes (e.g. RAS, FLT3 and KIT).

Clinical implications and future challenges Pre-leukemic stage: prediction and screening A reliable screening strategy for a rare disease, such as AML with an estimated incidence of 4:100,000,57 should have a high positive predictive value. In order to improve current models, specific variants that were described in hematologic malignancies, rather than specific genes, should be used for screening. In addition, a better understanding of the selective pressures and the germline background, under which pre-AML clones evolve, is required.

While such an understanding still remains elusive, the term â&#x20AC;&#x153;fitnessâ&#x20AC;? encompasses both clonal intrinsic factors and environmental selective pressures. Thus, exposing a detrimental, malignant, clonal evolution requires a dynamic, longitudinal follow up of healthy individuals, rather than relying on a single blood test that depicts a static picture of the hematopoietic clonal structure (as suggested by the term CHIP). Therefore, AML screening programs should use a 2% VAF threshold as well as a documentation of clonal temporal evolution (by having at least two assessments of the clonal mutational profile 6-12 months apart). This should be incorporated into a refined definition of ARCH.58 Documenting ARCH-defining events in two consecutive tests will allow a better characterization of the clones and increase the confidence in haematologica | 2019; 104(5)

their status, thus facilitating patient risk stratification.

Screening for therapy-related AML can be performed by detecting pre-leukemic clones at the time of initial chemotherapy treatment (administered for a non-AML tumor), and patients should be monitored at least once again to define ARCH.59 Additional improvement in the positive predictive value of such a model might also require inclusion of yet undescribed non-genomic, evolutionary events that manifest in the epigenome or in post-transcriptional or post-translational landscapes of the pre-malignant clone. It is still not known what influences these events: whether there are cause-and-effect relations between specific mutations and these non-genomic events, or whether they are influenced by the environment itself. As an example of the latter, p53 (wild-type at the genomic level) was shown to acquire a mutant-like post-translational conformation following stimulation by growth factors in AML cells.60 Incorporating clinical data into prediction models is expected to improve their accuracy and enable risk stratification for individuals carrying ARCH.9 However, preAML individuals were shown to have only subtle abnormalities in their blood count measures, often within the normal limits, with a large overlap with values documented in controls.9,13,21

Pre-leukemic stage: prevention Once prediction tools become more reliable, prospective, intervening clinical trials can be initiated. When planning such clinical trials two major challenges arise: 1) the low incidence of AML in the general population; and 2) its prolonged latency. The former can be mitigated by registering a large cohort of participants and by patient selection based on risk stratified according to clonal temporal progression (using the refined ARCH definition) and on their clinical data.8,9,13 Overcoming its prolonged latency requires a long follow up of the participants and might require a prolonged or indefinite treatment period to suppress pre-leukemic clones.

Targeting pre-leukemic clones as a means to prevent AML needs to be performed with caution. This might cause aplasia when the entire hematopoietic system is comprised by this clone (e.g. as is the case with a high VAF TET2 clone). This can also give a selective advantage to a different pre-leukemic clone that resides in the shared microenvironment. An effective intervention must target both the clone with its driver genes and the environment that enabled it to flourish.

Leukemic stage Targeting leukemic mutations might be an effective way to eliminate the malignant clone,38 thus preventing relapse.

This is particularly true in AML because most relapses arise during the first 2 years and stem from the original clone detected at diagnosis.34,61,62 Can pre-leukemic mutations serve as a target for therapy as well? On the one hand, all leukemic cells share these mutations. Targeting IDH1 using a specific inhibitor resulted in a 30% complete remission rate among 125 IDH1-mutated relapsed/refractory AML patients. These responses lasted a median of 8 months. Remission was accompanied by a decrease in IDH1 VAF values.63 Similar results were obtained when using an IDH2 specific inhibitor, with an overall response rate of 40.3% and a median response duration of 5.8 months.64 These results 877

A. Tuval and L.I. Shlush et al.

of phase I trials, using IDH1/2-targeted monotherapy, need to be repeated in larger cohorts of treatment-naĂŻve patients before conclusions can be drawn. On the other hand, the use of early, pre-leukemic, mutations as markers of the malignant clone should be handled with caution since following chemotherapy, the presence of most residual pre-leukemic clones (carrying DNMT3A, TET2 or ASXL1 mutations) do not increase risk for early (4-year) relapse unless accompanied by other mutations.19,31 In addition, one should bear in mind that, in contrast to the pre-leukemic stage, once overt leukemia evolves, targeting pre-leukemic somatic mutations might not prove effective. This can be inferred from the fact that some patients develop relapsed NPM1-mutated leukemia, lacking their DNMT3A pre-leukemic mutations.65 This can represent a biological phenomenon in which the relative contribution of these early driver mutations to clonal fitness diminishes as the clone evolves to overt leukemia.

Indeed, IDH1/2 inhibition induced differentiation of the malignant clone in only 5-7% of the patients and clearance of IDH1 mutated clone was noted only in 21% of clinically responding patients.63 In patients treated with an IDH2 inhibitor, differentiation of the blasts without elimination of the malignant clone was documented,64 in line with previous reports about IDH2 being acquired following AML transformation (as a late event).34 Improving efficacy might be achieved by combining drugs that target both leukemic and late events.

Relapse Most AML patients experience relapse originating in leukemic stem cells (LSC) that belong to the leukemic clone and that can already be identified at the time of diagnosis. It is, therefore, imperative to identify and target these cells. Two main subtypes of AML were identified.

The first AML subtype contains rare stem cells that have a stem/progenitor-like immunophenotype. In the other subtype, relapse originates from the major CD33+ blast population and is more dependent on growth factors when studied in vivo.34,61 Studying gene expression profiles of these subtypes revealed that this division is correlated with French-American-British (FAB) classification. The first subtype is enriched for FAB M4/M5 subtypes and the other is enriched for the less differentiated AML subtypes (M0/M1/M2). Nevertheless, relapse-initiating LSC in both groups had similar gene expression profiles and, as expected, the relapsing clone was found to be characterized by an increased number of LSC.34 Such a "leukemic stemness" transcriptional signature can be used to predict prognosis and to monitor patients in remission.66 Targeting LSC has been studied extensively in xenograft models but less so in clinical trials. A recent study suggests that the combination of Azacitidine and Venetoclax target LSCs, as identified both immune-phenotypically and by their transcriptomics signature.67 The number of participants in this trial was small, and, although this therapy does not induce remission in all patients (only in approx. 67%68), and some relapse while on therapy, this is an important step towards LSC-targeted therapy.

Survivorship Following the elimination of the leukemic clone, chemotherapy-resistant clones, which can sometimes be detected in low VAF values at the time of AML diagnosis,29 expand and re-populate the BM.29,40 Some of these clones 878

harbor ARCH mutations,29,31 and some of these clones are truly pre-malignant since they go on and evolve into MDS69 or a second AML, albeit following a more prolonged latency than the relapse of the original leukemia (median 33.7-43 months vs. 8.6-14 months, respectively).

Second AML should be diagnosed as a distinct entity whenever leukemic mutations that characterize the primary (diagnosis) AML are not identified at relapse (Table 1). Second AML was described to occur in 10-14% of the patients experiencing relapse.40,65 However, second AML should not be considered as a relapse. Sometimes, second AML does not share its pre-leukemic mutations with the primary AML (Table 1). This is underlined by the identification of pre-leukemic clones of second AML that lack the primary AML pre-leukemic DNMT3A, TET2, SRSF2 or RUNX1 mutations.40,70 Nevertheless, most of these secondAML-initiating clones share the same early, pre-leukemic mutations as the primary AML clone,40,65 and some even evolve similarly to the primary clone and acquire a different mutation in the same gene,70,71 emphasizing the role of an environmental selective pressure (Figure 2).

In this regard, environmental influence is best demonstrated when patients that undergo allogeneic stem cell transplantation develop an AML that originates in the donor hematopoietic cells. Two main reasons can lead to this very rare outcome (estimated to occur following 0.08% of transplants72): 1) a pre-existing pre-leukemic clone in the stem cell donation; and 2) evolution of a new leukemic clone in the recipient following the transplantation. Although the stem cell source (BM vs. peripheral blood) did not influence the risk for donor cell leukemia, environmental factors seem to be crucial in promoting the malignant clone. Multivariate analysis revealed three risk factors associated with development of donor cell leukemia: 1) the use of growth factors; 2) in vivo T-cell depletion; and 3) having a previous allograft. These risk factors imply that a reduced immune surveillance and increased replication signals create a more permissive environment that allows the development of the malignant clone. As an emphasis, two different trajectory leukemic evolutions were described following a DNMT3A-mutated pre-leukemic clone donation. While the donor developed NPM1-mutated, FLT3-ITD AML, the recipient developed NPM1 SMC1A-mutated AML.73 Therefore, when monitoring AML patients in remission, predicting a rare, second AML becomes somewhat analogous to predicting transformation from pre-leukemia to AML. Here, too, some residual or newly evolving preleukemic clones confer increased risk for second AML development, heralded by clonal expansion. The exact risk stratification still needs to be validated by appropriately designed studies. These studies need to use broad sequencing panels instead of a panel dictated only by the mutations found at diagnosis.

Conclusions Recently published studies reveal that the evolutionary trajectory of AML begins many years before the patient is actually diagnosed. It is a multistep process characterized by Darwinian evolution with clonal selection and expansion. Much is still unknown regarding the various factors that influence the path that clones in the hematopoietic system follow. They consist of both clonhaematologica | 2019; 104(5)

Evolutionary trajectory of AML

al-intrinsic as well as environmental factors. Both factors are influenced by each patient’s germline background.

We can improve our understanding firstly by depicting the exact route that pre-leukemic clones take on their way to becoming AML. It is important to remember that the trajectory of these clones does not end when patients achieve a complete (even molecular) remission. Residual, as well as new pre-leukemic clones, continue to evolve

References 1. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(7):730-737.

2. Dohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152.

3. Estey E. Acute myeloid leukemia: 2016 update on risk-stratification and management. Am J Hematol. 2016;91(8):824-846.

4. The Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):20592074.

5. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(9):2209-2221.

6. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mutational processes in human cancer. Nature. 2013;500(7463):415421.

7. McFarland CD, Korolev KS, Kryukov GV, Sunyaev SR, Mirny LA. Impact of deleterious passenger mutations on cancer progression. Proc Natl Acad Sci U S A. 2013;110(8):2910-2915.

8. Malcovati L, Gallì A, Travaglino E, et al.

Clinical significance of somatic mutation in unexplained blood cytopenia. Blood.

2017;129(25):3371-3378.

9. Ableson S, Collord G, Ng SWK, et al.

Prediction of acute myeloid leukaemia risk in healthy individuals. Nature. 2018;559 (7714):400-404.

10. Shlush LI, Zandi S, Mitchell A, et al.

Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature.

2014;506(7488):328-333.

11. Corces-Zimmerman MR, Hong WJ, Weissman IL, Medeiros BC, Majeti R. Preleukemic mutations in human acute myeloid leukemia affect epigenetic regulators and persist in remission. Proc Natl Acad Sci U S A. 2014;111(7):2548-2553.

12. Lal R, Lind K, Heitzer E, et al. Somatic TP53 mutations characterize preleukemic stem cells in acute myeloid leukemia. Blood.

2017;129(18):2587-2591.

13. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371 (26):2488-2498.

14. Genovese G. Kähler AK, Handsaker RE, et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014;371(26):2477-2487.

15. Young AL, Challen GA, Birmann BM, Druley TE. Clonal haematopoiesis harbouring AML-associated mutations is ubiquitous in healthy adults. Nat Commun. 2016;7: 12484.

haematologica | 2019; 104(5)

along the same path. Familiarizing ourselves with the ontogeny of AML and incorporating it into clinical practice will expand our therapeutic opportunities and grant us additional time points for intervention. Most importantly, a truly holistic treatment must also consider the environmental pressures under which AML evolved and address them. This will give a comprehensive meaning to the term "cure".

16. Acuna-Hidalgo R, Sengul H, Steehouwer M, et al. Ultra-sensitive Sequencing Identifies High Prevalence of Clonal HematopoiesisAssociated Mutations throughout Adult Life. Am J Hum Genet. 2017;101(1):50-64.

17. Lee-Six H, Øbro NF, Shepherd MS, et al.

Population dynamics of normal human blood inferred from somatic mutations.

Nature. 2018;561(7724):473-478.

18. Mckerrell T, Park N, Moreno T, et al.

Leukemia associated somatic mutations drive distinct patterns of age related clonal hemopoiesis. Cell Rep. 2015;10(8):12391245.

19. Lindsey RC, Mar BG, Mazzola E, et al.

Acute myeloid leukemia ontogeny is defined by distinct somatic mutations.

Blood. 2015;125(9):1367-1376.

20. Busque L, Buscarlet M, Mollica L, Levine RL. Concise review: age-related clonal hematopoiesis: stem cells tempting the devil. Stem Cells. 2018;36(9):1287-1294.

21. Buscarlet, M, Provost S, Zada YF, et al.

DNMT3A and TET2 dominate clonal hematopoiesis and demonstrate benign phenotypes and different genetic predispositions. Blood. 2017;130(6):753-762.

22. Desai P, Mencia-Trinchant N, Savenkov O, et al. Somatic mutations precede acute myeloid leukemia years before diagnosis.

Nat Med. 2018;24(7):1015-1023.

23. Wong TN, Ramsingh G, Young AL, et al.

Role of TP53 mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature. 2015;518(7540):552-555.

24. Gibson CJ, Lindsley RC, Tchekmedyian V, et al. Clonal hematopoiesis associated with adverse outcomes after autologous stem-cell transplantation for lymphoma. J Clin Oncol.

2017;35(14):1598-1605.

25. Coombs CC, Zehir A, Devlin SM, et al.

Therapy-related clonal hematopoiesis in patients with non-hematologic cancers is common and associated with adverse clinical outcomes. Cell Stem Cell. 2017;21(3): 374-382.

26. Hsu JI, Dayaram T, Tovy A, et al. PPM1D mutations drive clonal hematopoiesis in response to cytotoxic chemotherapy. Cell stem Cell. 2018;23(5):700-713.

27. Kahn JD, Miller PG, Silver AJ, et al. PPM1Dtruncating mutations confer resistance to chemotherapy and sensitivity to PPM1D inhibition in hematopoietic cells. Blood.

2018;132(11):1095-1105.

28. Lindsley RC, Saber W, Mar BG, et al.

Prognostic mutations in myelodysplastic syndrome after stem-cell transplantation. N Engl J Med. 2017;376(6):536-547.

29. Wong TN, Miller CA, Klco JM, et al. Rapid expansion of preexisting nonleukemic hematopoietic clones frequently follows induction therapy for de novo AML. Blood.

2016;127(7):893-897.

30. Lodé L, Ameur A, Coste T, et al. Single-

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

molecule DNA sequencing of acute myeloid leukemia and myelodysplastic syndromes with multiple TP53 alterations.

Haematologica. 2018;103(1):e13-e16.

Jongen-Lavrencic M, Grob T, Hanekamp D, et al. Molecular minimal residual disease in acute myeloid leukemia. N Engl J Med. 2018;378(13):1189-1199.

Guryanova OA, Shank K, Spitzer B, et al.

DNMT3A mutations promote anthracycline resistance in acute myeloid leukemia via impaired nucleosome remodeling. Nat Med. 2016;22(12):1488-1495.

Broeker PL, Super HG, Thirman MJ, et al.

Distribution of 11q23 breakpoints within the MLL breakpoint cluster region in de novo acute leukemia and in treatment-related acute myeloid leukemia: correlation with scaffold attachment regions and topoisomerase II consensus binding sites. Blood.

1996;87(5):1912-1922.

Shlush LI, Mitchell A, Heisler L, et al.

Tracing the origins of relapse in acute myeloid leukemia to stem cells. Nature.

2017;547(7661):104-108.

Grove CS, Bolli N, Manes N, et al. Rapid parallel acquisition of somatic mutations after NPM1 in acute myeloid leukemia evolution. Br J Haematol. 2017;176(5):825-829.

Ivy A, Hills RK, Simpson MA, et al.

Assessment of minimal residual disease in standard-risk AML. N Engl J Med. 2016; 374(5):422-433.

Yap TA, Gerlinger M, Futreal PA, Pusztai L, Swanton C. Intratumor heterogeneity: seeing the wood for the trees. Sci Transl Med. 2012;4(127):127ps10.

Brunetti L, Gundry MC, Sorcini D, et al.

Mutant NPM1 maintains the leukemic state through HOX expression. Cancer Cell.

2018;34(3):499-512.

Xie M, Lu C, Wang J, et al. Age-related mutations associated with clonal hematopoietic expansion and malignancies.

Nat Med. 2014;20(12):1472-1478.

Höllein A, Meggendorfer M, Dicker F, et al.

NPM1 mutated AML can relapse with wildtype NPM1: persistent clonal hematopoiesis can drive relapse.

Blood Adv. 2018;2(22):3118-3125.

Stone RM, Mandrekar SJ, Sanford BL, et al.

Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation. N Engl J Med. 2017;377(5):454-464.

Pemovska T, Kontro M, Yadav B, et al.

Individualized systems medicine strategy to tailor treatments for patients with chemorefractory acute myeloid leukemia. Cancer Discov. 2013;3(12):1416-1429.

Shouval R, Shlush LI, Yehudai-Resheff S, et al. Single cell analysis exposes intra-tumor heterogeneity and suggests that FLT3-ITD is a late event in leukemogenesis. Exp Hematol. 2014;42(6):457-463.

Abdelhamid E, Preudhomme C, Helevaut N,

879

A. Tuval and L.I. Shlush et al.

45.

46.

47.

48.

49.

50.

51.

52.

53.

880

et al. Minimal residual disease monitoring based on FLT3 internal tandem duplication in adult acute myeloid leukemia. Leuk Res. 2012;36(3):316-323.

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):23912405.

Swerdlow SH, Campo E, Harris NL, et al, editors. WHO classification of tumours of haematopoietic and lymphoid tissues. 4th ed. Lyon: IARC; 2008.

Grimwade D, Jovanovic JV, Hills RK, et al.

Prospective minimal residual disease monitoring to predict relapse of acute promyelocytic leukemia and to direct pre-emptive arsenic trioxide therapy. J Clin Oncol.

2009;27(22):3650-3658.

Jourdan E, Boissel N, Chevret S, et al.

Prospective evaluation of gene mutations and minimal residual disease in patients with core binding factor acute myeloid leukemia. Blood. 2013;121(12):2213-2223.

Wiemels JL, Xiao Z, Buffler PA, et al. In utero origin of t(8;21) AML1-ETO translocations in childhood acute myeloid leukemia.

Blood. 2002;99(10):3801-3805.

Jurlander J, Caligiuri MA, Ruutu T, et al.

Persistence of the AML1/ETO fusion transcript in patients treated with allogeneic bone marrow transplantation for t(8;21) leukemia. Blood. 1996;88(6):2183-2191.

Miyamoto T, Weissman IL, Akashi K. AML1/ETO-expressing nonleukemic stem cells in acute myelogenous leukemia with 8;21 chromosomal translocation. Proc Natl Acad Sci U S A. 2000;97(13):7521-7526.

Chapiro E, Delabesse E, Asnafi V, et al.

Expression of T-lineage-affiliated transcripts and TCR rearrangements in acute promyelocytic leukemia: implications for the cellular target of t(15;17). Blood. 2006;108(10): 3484–3493.

Ailles LE, Gerhard B, Kawagoe H, Hogge DE. Growth characteristics of acute myelogenous leukemia progenitors that initiate

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

malignant hematopoiesis in nonobese diabetic/severe combined immunodeficient mice. Blood. 1999;94(5):1761-1772.

Grimwade D, Enver T. Acute promyelocytic leukemia: where does it stem from?

Leukemia. 2004;18(3):375-384.

Haferlach T, Löffler H, Nickenig C, et al. Cell lineage specific involvement in acute promyelocytic leukaemia (APL) using a combination of May-Grunwald-Giemsa staining and fluorescence in situ hybridization techniques for the detection of the translocation t(15;17)(q22;q12). Br J Haematol. 1998;103(1):93-99.

Turhan AG, Lemoine FM, Debert C, et al.

Highly purified primitive hematopoietic stem cells are PML-RARA negative and generate nonclonal progenitors in acute promyelocytic leukemia. Blood. 1995;85(8): 2154-2161.

SEER program [Internet]. National Cancer Institute. 2018 - [cited 2019 January 10].

Available from: https://seer.cancer.gov/ Shlush LI. Age related clonal hematopoiesis.

Blood. 2018;131(5):496-504.

Berger G, Kroeze LI, Koorenhof-Scheele TN, et al. Early detection and evolution of preleukemic clones in therapy-related myeloid neoplasms following autologous SCT. Blood. 2018;131(16):1846-1857.

Zhang W, Deisseroth AB. Conformational change of p53 protein in growth factor-stimulated human myelogenous leukemia cells.

Leuk Lymphoma. 1994;14(3-4):251-255.

Ding L, Ley TJ, Larson DE, et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing.

Nature. 2012;481(7382):506-510.

Parkin B, Ouillette P, Li Y, et al. Clonal evolution and devolution after chemotherapy in adult acute myelogenous leukemia. Blood.

2013;121(2):369-377.

Di Nardo CD, Stein EM, de Botton S, et al.

Durable remissions with ivosidenib in IDH1-mutated relapsed or refractory AML. N Engl J Med. 2018;378(25):2386-2398.

Stein EM, Di Nardo CD, Pollyea DA, et al.

65.

66.

67.

68.

69.

70.

71.

72.

73.

Enasidenib in mutant IDH2 relapsed or refractory acute myeloid leukemia. Blood.

2017;130(6):722-731.

Krönke J, Bullinger L, Teleanu V, et al. Clonal evolution in relapsed NPM1-mutated acute myeloid leukemia. Blood. 2013;122(1):100108.

Ng SW, Mitchell A, Kennedy JA, et al. A 17gene stemness score for rapid determination of risk in acute leukaemia. Nature.

2016;540(7633):433-437.

Pollyea DA, Stevens BM, Jones CL, et al.

Venetoclax with azacitidine disrupts energy metabolism and targets leukemia stem cells in patients with acute myeloid leukemia.

Nat Med. 2018;24(12):1859-1866.

Di Nardo CD, Pratz K, Pullarkat V, et al.

Venetoclax combined with decitabine or azacitidine in treatment-naive, elderly patients with acute myeloid leukemia.

Blood. 2019;133(1):7-17.

Herold S, Sockel K, Sayehli C, et al.

Evolution of NPM1-negative therapy-related myelodysplastic syndromes following curative treatment of NPM1- mutant AML. Leukemia. 2017;31(10):2247-2251.

Kohlmann A, Nadarajah N, Alpermann T, et al. Monitoring of residual disease by nextgeneration deep-sequencing of RUNX1 mutations can identify acute myeloid leukemia patients with resistant disease.

Leukemia. 2014;28(1):129-137.

Webersinke G, Kranewitter W, Deutschbauer S, et al. Switch of the mutation type of the NPM1 gene in acute myeloid leukemia (AML): relapse or secondary AML? Blood Cancer J. 2014;4(6):e221.

Engel N, Rovo A, Badoglio M, et al.

European experience and risk factor analysis of donor cell-derived leukaemias/MDS following haematopoietic cell transplantation.

Leukemia. 2019;33(2):508-517.

Hahn CN, Ross DM, Feng J, et al. A tale of two siblings: two cases of AML arising from a single pre-leukemic DNMT3A mutant clone. Leukemia. 2015;29(10):2101-2104.

haematologica | 2019; 104(5)

REVIEW ARTICLE

CRISPR to fix bad blood: a new tool in basic and clinical hematology

Ferrata Storti Foundation

Elisa González-Romero,1 Cristina Martínez-Valiente,1 Cristian García-Ruiz,1 Rafael P. Vázquez-Manrique,2,3 José Cervera4,5 and Alejandra Sanjuan-Pla1

1 Hematology Research Group, Instituto de Investigación Sanitaria La Fe, Valencia; 2Grupo de Investigación en Biomedicina Molecular, Celular y Genómica, Instituto de Investigación Sanitaria La Fe, Valencia; 3CIBER de Enfermedades Raras, Madrid; 4Hematology Department, Hospital Universitari i Politècnic La Fe, Valencia and 5CIBER de Oncología, Madrid, Spain

Haematologica 2019 Volume 104(5):881-893

ABSTRACT

A

dvances in genome engineering in the last decade, particularly in the development of programmable nucleases, have made it possible to edit the genomes of most cell types precisely and efficiently.

Chief among these advances, the clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 system is a novel, versatile and easyto-use tool to edit genomes irrespective of their complexity, with multiple and broad applications in biomedicine. In this review, we focus on the use of CRISPR/Cas9 genome editing in the context of hematologic diseases and appraise the major achievements and challenges in this rapidly moving field to gain a clearer perspective on the potential of this technology to move from the laboratory to the clinic. Accordingly, we discuss data from studies editing hematopoietic cells to understand and model blood diseases, and to develop novel therapies for hematologic malignancies. We provide an overview of the applications of gene editing in experimental, preclinical and clinical hematology including interrogation of gene function, target identification and drug discovery and chimeric antigen receptor T-cell engineering. We also highlight current limitations of CRISPR/Cas9 and the possible strategies to overcome them. Finally, we consider what advances in CRISPR/Cas9 are needed to move the hematology field forward.

Correspondence: ALEJANDRA SANJUAN PLA alejandra_sanjuan@iislafe.es Received: December 20, 2018.

Accepted: February 19, 2019.

Pre-published: March 28, 2019.

Introduction Genome engineering is defined as the deliberate modification of an organism’s genetic material. It has been used since the early 1980s to study the impact of DNA mutations in human disease precisely and has helped to unravel the genetic basis of many malignancies and to advance their diagnosis, prevention, and treatment.

Genome engineering to introduce defined alterations has traditionally employed homologous recombination strategies to modify a gene of interest (gain- or loss-offunction) using segments of exogenous DNA.1 To achieve homologous recombination in the “pre-nuclease” era, large DNA sequences homologous to the target sequence, containing sequence changes designed to produce the desired modification, were introduced into the nucleus of the receiving cell. This technology depends heavily on chance since the DNA construct is expected to interact with the target and induce gene conversion upon recombination of DNA homology arms. The success rate of this technology was historically extremely low, which together with the complexity in designing targeting vectors, and the time and resources required, put it out of reach of some researchers. However, with the advent of highly-specific chimeric nucleases (which are able to recognize 18 or more base pairs) to induce locus-specific DNA double-strand breaks (DSB), the efficiency of homologous recombination rose substantially (e.g., becoming more than 40,000 times more efficient2), depending on the experimental system. The use of such nucleases has increased in recent years with the development of meganucleases,3 zinc-finger nucleases,4 and transcription activator-like effector nucleases haematologica | 2019; 104(5)

doi:10.3324/haematol.2018.211359 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/881 ©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.

881

E. GonzĂĄlez-Romero et al.

(TALEN),5 opening new horizons for genome manipulation (Figure 1A). Nevertheless, designing the aforementioned nucleases to induce DSB in specific loci relies on predicting protein-DNA interactions, which remains technically challenging, and so these nucleases are not practicable in every laboratory. By contrast, the recent breakthrough of clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) technology, which is based on nucleic acid interactions, has enabled specific genome editing in a versatile and uncomplicated manner over previous nucleases, and has

A

B

882

revolutionized the field of genome engineering (Figure 1B, Table 1).

CRISPR sequence repeats were first reported in Escherichia coli6 and were later characterized in Haloferax mediterranei, an archaeon isolated from a hypersaline environment in Alicante (Spain).7 Soon after, these sequence repeats were identified as a part of a primitive adaptive immune system in prokaryotes.8,9 In 2012, Doudna and Charpentier demonstrated the first use of CRISPR/Cas9 to introduce site-specific DSB in target DNA based on the ability of a single guide RNA (gRNA) to direct sequence-

Figure 1. Nucleases used in genome engineering. (A) PreCRISPR nucleases such as meganucleases, zinc-finger nucleases (ZFN) and transcription activator-like effector nucleases (TALEN) are proteins that bind directly to DNA. Meganucleases are naturally occurring restriction enzymes that recognize between 12 to 40 base pair sequences, although they allow for some restricted level of engineering to make them specific to certain loci. Engineered ZFN induce specific double-strand breaks (DSB) acting as dimers. Each monomer is composed of a non-specific cleavage domain from the FokI endonuclease and a zinc-finger protein array where each domain bind three base pairs. ZFN dimers are able to recognize 18â&#x20AC;&#x201C;24 base pairs in the target sequence, allowing for highly specific targeting. TALEN are designed combining the same non-specific endonuclease FokI domain and transcription activator-like effector (TALE) proteins. TALE proteins present a central domain responsible for DNA binding, which interacts specifically with just one nucleotide. One of these domains consists of monomers of 34 amino acid residues, two of which are responsible for nucleotide recognition. This makes the design of TALEN very straightforward in principle. (B) In contrast to the nucleases described in (A), the Cas9 endonuclease of the CRISPR/Cas9 system binds to the target DNA thought the guide RNA (gRNA) by Watson-Crick base pairing. The gRNA is composed of two molecules of RNA: (i) the CRISPR RNA (crRNA) (green nucleotides) of which 20 nucleotides [white bold in top panel in (A), black bold in middle and bottom panels in (A)] show strict homology to the target and (ii) the trans-activating crRNA (tracrRNA), which binds to the crRNA and to the Cas9 nuclease (yellow structure). The gRNA brings Cas9 the target sequence, which is always adjacent to a protospacer adjacent motif (PAM) sequence. The PAM sequence for the most used Cas9, isolated from the bacteria Streptococcus pyogenes, is NGG (TGG in the white box). Notes: white arrows in (A) represent hydrogen bonds between amino acids from proteins and DNA base pairs; thick black arrows point to the site of cleavage of the nucleases.

haematologica | 2019; 104(5)

CRISPR/Cas9 gene editing in hematology

specific Cas9 double-stranded DNA cleavage,10 illustrating the wide-ranging application of CRISPR as a genome-editing technology.11 Indeed, the CRISPR/Cas9 system was first successfully used in human cells in 2013.12-14 The essential components of this technology include a gRNA that binds specifically to a 20-base pair sequence of interest, and the Cas9 enzyme – an endonuclease that introduces a DSB. Additionally, a conserved dinucleotide-containing protospacer adjacent motif sequence upstream of the gRNA-binding site is required by the endonuclease to recognize and cleave the sequence. In the case of Cas9 isolated from Streptococcus pyogenes, the most widely used nuclease, the protospacer adjacent motif sequence is NGG (Figure 2A). If these conditions are fulfilled, CRISPR/Cas9 can be directed to cleave any genomic sequence.

Subsequently, the DSB (in eukaryotic cells) triggers endogenous cellular DNA-repair pathways that can be exploited either to generate gene knock-outs based on the introduction of insertions or deletions (indels) at the DSB by non-homologous end joining (NHEJ)15, or for genome editing, by introducing an engineered template DNA via homology-directed repair (HDR)16 (Figure 2A). In contrast to the protein-DNA interactions of other nuclease editing systems, CRISPR relies on Watson-Crick pairing between RNA and DNA. Thus, researchers keen to perform gene editing require only a basic knowledge of molecular biology to design a targeting system against a locus of choice.

Here, we will focus mainly on work done with the Cas9 nuclease, although it is worth mentioning that the CRISPR/Cas system can include many other enzyme variants with numerous functions that are suitable for applications beyond gene editing17 (Table 2).

In comparison with engineered nucleases, CRISPR/Cas9 is an easy-to-use genome-editing tool, and several different CRISPR/Cas9-component delivery methods are available for in vitro, ex vivo and in vivo applications18 (Table 3).

Generally, Cas9 and gRNA can be introduced into cells in several formats, such as plasmid DNA, lentiviral vectors, mRNA, or more recently by using pre-assembled ribonucleoprotein complexes (Table 4). Indeed, ribonucleoprotein complexes are perhaps the best choice for clinical applications given their high efficiency and short window of action, which reduces the duration of nuclease exposure and, consequently, the possibility of undesired offtarget effects. In the hematopoietic setting, CRISPR/Cas9 gene editing has been applied both in research and in clinical translation studies (Figure 2B). In disease modeling, CRISPR/Cas9 technology coupled to next-generation genomics allows researchers to faithfully recapitulate the genetic mutations seen in patients with clonal hematopoiesis or leukemia.19 In the clinical setting, the main goal is to employ CRISPR/Cas9 to treat diseases of the blood and immune system. With this view, several biotechnology companies have pipelines to develop and

Table 1. Pros and cons of genome engineering tools in mammalian systems.

Gene editing toolsa

Advantages

Disadvantages

Pre-edition era

• Genetic analysis relied on spontaneous, induced random mutations by chemicals or transposons • DSB not usedb • Different gene modifications: knock out, conditional alleles, reporter genes.

• Extremely laborious • No directed gene editing • Highly unpredictable mutations • Extremely laborious • Highly inefficient • Large homology fragments of DNA are needed for homologous recombination • Biallellic changes are difficult to obtain • Difficult to use in hESC and other cell types • Selection markers are necessary • Very low design flexibility • Low specificity of the enzyme/off-target possibility • Off-target effect possible but less than with CRISPR • Harder to design than TALEN nucleases

Conventional gene edition

Meganucleases

ZFN

TALEN

CRISPR/Cas9

• Large recognition site for DNA • Highly specific • DSB repaired by HDR or NHEJ • Possibility of engineering nucleases • Highly efficient • DSB repaired by HDR or NHEJ • Biallelic changes are possible • Works in different cell types and species.

• Easier to design than ZFN • Highly efficient • DSB repaired by HDR or NHEJ • Biallelic changes are possible • Works in different cell types and species • Easy design and optimization • Highest efficiency • DSB repaired by HDR or NHEJ • Biallellic changes obtained with efficiency • Works in different cell types and species

Main dates 1950s

1980s

1988

1996

• Off-target effect possible but less than with CRISPR 2009 • Still harder to design than CRISPR

• More off-target effects than TALEN and ZFN 2013 (though there are ways to reduce them dramatically)c • PAM sequence limits target selection (though, many CRISPR systems available, and more to come)d

a Gene editing of mammalian genomes. bDSB: double-strand break; hESC: human embryonic stem cells; HDR: homology-directed repair; NHEJ: non-homologous end joining; PAM: protospacer adjacent motif; TALEN: transcription activator-like effector nucleases; ZFN: zinc finger nucleases. cThe use of nickases and/or ribonucleoproteins, which reduce the time window in which the nucleases can induce lesions, drastically reduces the probability of off-targets. dMany CRISPR systems have been described from different prokaryotes that use different PAM sequences. This allows for more flexibility when designing a targeting strategy. Relatively few CRISPR systems have been described to date.

haematologica | 2019; 104(5)

883

E. Gonzรกlez-Romero et al.

evaluate therapeutic CRISPR/Cas9 gene editing to correct mutations in pathologies such as autoimmune diseases and multiple myeloma.

Given the exponential output of scientific publications using CRISPR in the last years, there is a clear need to synthesize the latest data on gene editing. With this in mind, here we describe recent advances in the use of CRISPR/Cas9 in hematologic research and clinical translation. We also consider the limitations of CRISPR/Cas9 technology for therapeutic applications, their possible solutions, and how the field of hematology may move forward. While CRISPR/Cas9 gene editing is hoped to be a treatment for many hematologic diseases, large clinical trials are needed to evaluate the efficacy and safety of CRISPR/Cas9 for patients, a promising area that will undoubtedly expand in the near future.

CRISPR/Cas9 in hematology research In vitro gene editing of blood cells Hematopoietic cell lines are a robust model for validating gRNA specificity and CRISPR/Cas9 experimental design because of their easy manipulation and expansion, and enrichment of edited cells. In this respect, cell lines have been employed: (i) to analyze gene function upon NHEJmediated gene disruption; (ii) to insert a point mutation or a DNA fragment; (iii) to correct a point mutation, and (iv) to create chromosomal translocations. Although valuable as a proof-of-concept approach, editing success cannot always be extrapolated to difficult-to-edit primary cells.

Many studies have reported successful CRISPR/Cas9mediated gene editing in hematopoietic cell lines. As a consequence of indels introduced by NHEJ-mediated DSB

A

B

Figure 2. CRISPR/Cas9-mediated genome editing in hematology. (A) Illustration of the CRISPR/Cas9 system. Site-specific DSB are produced

by CRISPR/Cas9 and are either repaired by NHEJ, introducing indels that provoke gene disruption, or by HDR that, in the presence of a DNA template, creates insertions, translocations, or point mutations. gRNA: guide RNA; DSB: doublestrand break; NHEJ: non-homologous end joining; HDR: homologydirected repair; indel: insertions and deletions. (B) Applications of CRISPR/Cas9 technology in hematology research and human therapy.

HIV: human immunodeficiency virus; CAR: chimeric antigen receptor; CHIP: clonal hematopoiesis of indeterminate potential.

884

haematologica | 2019; 104(5)

CRISPR/Cas9 gene editing in hematology

repair, the resulting frameshift or nonsense mutations can give rise to truncated proteins that have a gain- or loss-offunction.20 This is a considerably faster approach than conventional homologous recombination-based gene targeting to create gene knock-outs and, given its simplicity, CRISPR/Cas9 is poised to become the method of choice

for knock-out studies in most cases. Moreover, CRISPR/Cas9 is superior to RNA interference approaches for deciphering gene function, since the latter produce hypomorphic phenotypes that do not always mirror the complete loss-of-function that often occurs with genetic mutations.21

Table 2. Comparison of the most widely used Cas nucleases.

Cas nuclease

Identified from*

Targeted molecule Key features

Cas9

Streptococcus pyogenes

DNA

dCas9

Mutant form of Cas9

DNA

Cas9 nickase

Mutant form of Cas9

DNA

Cas12a (Cpf1)

Acidaminococcus sp.

Lachnospiraceae bacterium

DNA

Cas13a (C2c2)

Leptotrichia wadei Leptotrichia buccalis Leptotrichia shahii

RNA

Most frequent application

• DSB proximal to PAM (blunt ends) • Widely used in genome editing • More off-targets than Cas9 variants (eSp-Cas9, Cas9-HF1, Hypa-Cas9) • Lacks endonuclease activity • Works by recruiting enhancers, silencers, chromatin modifiers • Useful for single base genome mutagenesis • Single-strand break • One inactived nuclease domain • Higher accuracy in gene integration using two nickases • Lower off-targets than Cas9 • DSB distal to PAM (staggered ends) • Cleaves first the non-target strand • No requirement for tracrRNA • Lacks a DNase domain • No requirement for HDR machinery or a PAM • Acts in non-dividing cells • Cleaves additional RNA (only in bacteria)

Knock-out Knock-in

Regulation of gene expression (CRISPRi/ CRISPRa) Knock-in

Knock-out Knock-in Regulation of gene expression

Description of the alternative Cas nucleases employed in genome editing. *Most common organisms in which the Cas nuclease has been isolated from. CRISPRi: CRISPR interference; CRISPRa: CRISPR activation; DSB: double-strand break; tracrRNA: trans‐activating crRNA; crRNA: CRISPR RNA; HDR: homology- directed repair; eSp-Cas9: enhanced specificity Cas9; Cas9-HF1: high fidelity Cas9, Hypa-Cas9: hyper-accurate Cas9.

Table 3. Summary of delivery approaches for CRISPR/Cas9 components.

Delivery vehicle Physical approaches Microinjection Electroporation

Viral-based approaches Lentivirus

Adenovirus

Adeno-associated virus

Non-viral approaches Lipid nanoparticles

Cell-penetrating peptides haematologica | 2019; 104(5)

Advantages

Disadvantages

• Delivered directly into cell of interest • High efficiency • Standardized protocols available • High efficiency with plasmids

• Time-consuming • Requires expertise • Limited to in vitro and ex vivo cells • Cell cytotoxicity • Some cells are not susceptible

• Robust, stable expression • Allows delivery in complex and primary cells • Efficiency variable with construct length

• Immune response, but low • Limited packaging capacity (18 kb) • Random genome integration • Off-targets from Cas9-constitutive expression • Expertise and safety issues • High immune response • Limited packaging capacity (35 kb) • Expertise and safety issues

• No genome integration • Transient expression • Reduced off-targets • High efficiency • No genome integration • Reduced immunogenicity and cytotoxicity • Reduced off-targets • High efficiency • Simple manipulation • Low cost • Reduced off-targets • Deliver intact ribonucleoproteins

• Immune response, but very low • Limited packaging capacity (4.5 kb) • Costly • Expertise and safety issues • Dependent on cell type • Endosomal degradation • Variable penetrating efficiency 885

E. González-Romero et al.

CRISPR/Cas9-mediated cleavage followed by HDR has been employed to introduce point mutations or gene fragments into specific loci using donor template DNA flanked by 3' and 5' sequences homologous to the target region. However, creating a knock-in allele by homologous recombination of a targeting construct using embryonic stem cells (which could be used to produce a mouse model) or by CRISPR/Cas9 is not so different in terms of cost and time.22 CRISPR/Cas9 gene editing can help to elucidate the role of patients’ mutations by generating cellular models carrying these lesions. Along this line, patients with myelodysplastic syndromes frequently have mutations in splicing genes such as the P95H mutation in serine/arginine factor 2 (SRSF2), which regulates pre-mRNA splicing. Zhang et al. developed an SRSF2/P95H cell line using CRISPR/Cas9-mediated HDR, which resulted in a gain-of-function phenotype and changed its RNA-binding preferences, producing splicing misregulation. This illustrates how a mutation associated with myelodysplastic syndromes alters splicing patterns, some of which are relevant for disease and have therapeutic potential.23 In acute myeloid leukemia, driver mutations can also cause and/or maintain leukemia24 and precise AML models are needed to develop novel, targeted therapies. For instance, the R140Q mutation in the Krebs cycle enzyme isocitrate dehydrogenase 2 (IDH2) endows cells with neomorphic enzyme activity, generating an oncometabolite that interferes with epigenetic cell regulation and contributes to malignant transformation.

To study the molecular and functional characteristics of this driver mutation, genome editing was used in K562 cells to introduce the IDH2/R140Q mutation.25 Cells carrying this mutation recapitulated the genetic, epigenetic and functional changes seen in IDH2-mutated patients, offering a suitable model for drug testing.

In addition to modeling disease, CRISPR/Cas9 has been employed to correct mutations in disease-associated genes using single-stranded donor oligonucleotides as DNA donor templates for HDR. For example, a loss-offunction mutation in the Additional sex combs like 1 (ASXL1) gene, frequently mutated in myelodysplastic syndromes, chronic myelomonocytic leukemia, and AML was corrected in a chronic myeloid leukemia cell line.26 Similarly, AML blasts (precursor cells) containing the IDH2R140Q mutation were corrected to restore cell function to wild-type status.25 These results constitute a proof-ofconcept that CRISPR/Cas9 gene correction of primary

hematopoietic cells is feasible. Beyond hematopoietic cells, CRISPR/Cas9 genome editing of human induced pluripotent stem cells (iPSC) has been used to correct disease-relevant mutations. For example, correction of the HCLS1 associated protein X-1 (HAX1) gene by CRISPR/Cas9-mediated HDR reversed the severe congenital neutropenia phenotype in patient-specific iPSC.27 This is important given that iPSC are excellent platforms to model disease and also hold promise for use in patientspecific, cell-based regenerative therapy. Accordingly, hematopoietic cells carrying a mutation could be isolated from the patient, reprogrammed to iPSC, edited, differentiated to hematopoietic stem cells (HSC) and re-introduced by autologous HSC transplantation. However, the capability of iPSC-derived HSC to reconstitute the blood system in the long-term remains a challenge for clinical translation.28 Most pre-clinical models of CRISPR/Cas9-based gene repair have been based on precise but relatively poorly efficient HDR. The greater efficiency of NHEJ-based mutation correction in the absence of donor template DNA has been used successfully to repair frame-shift mutations. For example, in a study on X-linked chronic granulomatous disease, which is caused by mutations in the cytochrome b-245 heavy chain (CYBB) gene,29 patientspecific CYBB point mutations were successfully repaired

by NHEJ – the dominant DSB-repair pathway in hematopoietic stem and progenitor cells (HSPC) – with gene repair efficiency being between 18-25%. The authors of this study assumed that approximately onethird of NHEJ-mediated indels should re-establish the open reading frame disrupted by the disease mutation, leading to a complete or partial recovery of protein function. Importantly, this high-efficiency approach minimizes the number of reagents required to be introduced into patients’ cells and also circumvents homologous donor template delivery, which might be beneficial for translation of HSPC gene editing to the clinic.

In the context of disease modeling, a more complex scenario would be to recreate the fusion proteins resulting from chromosomal rearrangements, a typical hallmark of some leukemias. CRISPR/Cas9-based editing has been successfully used in human cell lines and human HSC to generate chromosomal translocations resembling those described in acute leukemia, such as t(8;21)/RUNX1-ETO, t(4,11)/KMT2A-AFF1/AFF1-KMT2A and t(11;19)/MLLENL.30-32 This achievement is relevant because the model-

Table 4. Comparison of the different formats available for CRISPR/Cas9 components.

CRISPR format option

Advantages

Disadvantages

Cas9 and/or gRNA-encoding plasmids

• Simple-to-use approach • Multiple gRNA can be integrated into the same plasmid • Repositories available • Economical • Lack of genome integration • Less off-targets than integrative plasmids

• Off-targets from Cas9-constitutive expression • Activation of innate immune system against plasmids

Cas9 mRNA and gRNA

Ribonucleoprotein complex

• Few off-target effects due to transient expression • Fast, avoiding cell transcription and/or translation • Highly efficient

• Issues with mRNA stability • Immunogenicity • Expensive • Transient expression not sufficient in some contexts • More expensive than previous options

The main CRISPR component formats are: (i) DNA plasmids encoding the Cas9 protein and a guide RNA (gRNA), either individually or together; (ii) mRNA for Cas9 translation applied to the cell, together with a separate gRNA. (iii) Ribonucleoprotein complexes, formed by pre-assembled Cas9 protein and gRNA. We highlight the most relevant pros and cons for each option. gRNA: guide RNA.

886

haematologica | 2019; 104(5)

CRISPR/Cas9 gene editing in hematology

ing of fusion proteins in hematopoietic cells was previously accomplished by viral expression of fusion protein cDNA cloned from patients or by genomic engineering of mouse DNA to create chromosomal rearrangements using recombination systems (e.g., Cre-loxP), which is complicated.33 These aforementioned studies illustrate that CRISPR/Cas9 technology is a reliable and accurate approach to recreate chromosomal translocations, albeit at low efficiencies, providing a powerful tool for cancer studies.

Another application of CRISPR/Cas9 technology that holds great promise is in the arena of functional genomics, in which it has been employed in genome-wide, loss-offunction screens in mammalian cells. Typically, lentiviral gRNA libraries are used in genetic screens for positive and negative selection,34,35 which have advantages over RNA interference-based screening with inherent incomplete gene knockdown. Another advantage of CRISPR/Cas9 is that it can target non-coding genomic regions, including promoters, enhancer elements, and intergenic regions.

Positive selection studies screen for perturbations conferring enhanced self-renewal/proliferation/survival potential to the interrogated cells, resulting in cell enrichment over time. By contrast, negative selection studies aim to identify genes essential for survival/proliferation that, when targeted, will cause cell depletion over time. In the context of myeloid malignancies, several high-throughput screens have been performed in drug target discovery applications. Using CRISPR/Cas9 to edit protein domains, Shi et al. identified cancer drug targets by screening 192 chromatin regulatory domains in murine AML cells, validating six known drug targets and also revealing additional dependencies.36 In a study aiming to examine mechanisms of cytarabine drug resistance in AML cell lines, CRISPR/Cas9-based screening identified the deoxycytidine kinase gene as the primary contributor to cytarabine resistance.37 In addition to genome-wide CRISPR screens, targeted panel-based screens of previously selected genes would also allow the interrogation of biological processes, for example, cytokine signal transduction, cancer progression or cell migration, which are suspected to be linked to a disease.

Generating mouse models using the CRISPR/Cas9 system In vivo mouse models, usually generated by homologous recombination strategies, have been instrumental in deciphering the role of point mutations, translocations, and DNA sequence indels in the context of a whole organism. CRISPR/Cas9 technology can be used to build both germline (heritable) and somatic mouse models in a fast and precise manner.38

Germline CRISPR/Cas9 mouse models CRISPR/Cas9 has been employed to disrupt the splicing factor ZRSR1 in murine zygotes, resulting in altered erythrocyte function in adult mice, suggesting that ZRSR1-associated minor splicing could have an important role in terminal erythropoiesis.39 More recently, the technology was used for the generation of novel hemophilia mouse models on an immunodeficient NSG (NOD/SCID/IL-2γ−/−) background.40 Hemophilia A and B are congenital, X-linked bleeding disorders caused by mutations in the genes encoding for the blood clotting factor VIII (F8) and factor IX (F9), respectively. haematologica | 2019; 104(5)

CRISPR/Cas9 and gRNA were microinjected into NSG mouse zygotes to generate mice with hemophilia A or hemophilia B. These models should allow the evaluation of the efficacy and safety of novel gene therapy vectors in hemophilia.

Given the importance of reporter mouse lines in biomedical research, it is not surprising that CRISPR/Cas9 technology has been applied in the study of early developmental processes. Recently, a knock-in mouse strain was created for dynamic tracking and enrichment of the MEIS1 transcription factor during hematopoiesis.41 This GFP-HA epitope tag reporter strategy and CRISPR/Cas9 gene editing could be employed to develop new reporter mouse lines to study other transcription factors important for hematopoiesis.

Mice carrying mutations in multiple genes have traditionally been generated by sequential recombination in embryonic stem cells and/or intercrossing of mice with single mutations. CRISPR/Cas9 technology allows the generation of mice bearing different gene mutations in a more affordable, less labor-intensive and time-consuming manner than traditional approaches. Similar to the hemophilia models describe above, mice with bi-allelic mutations in TET1 and TET2 were created by co-injection of targeting gRNA into mouse zygotes, which is a much faster approach compared with traditional techniques and allows one-step generation of animals with precise mutations.42 Accordingly, targeting multiple genes using CRISPR/Cas9 should facilitate, for example, the in vivo study of a family of genes with redundant functions.

Indeed, Cas9 mRNA and multiple gRNA targeting B2M, IL2RG, PRF1, PRKDC, and RAG1 genes were microinjected together into mouse embryos to produce different immunodeficient mouse strains,43 thus generating new valuable tools to advance research in human HSPC xenotransplantation.

There is increasing evidence that the acquisition of somatic mutations in HSC, leading to clonal hematopoiesis, is a cardiovascular risk factor. Indeed, DNMT3A and TET2 somatic mutations are drivers of clonal hematopoiesis of indeterminate potential, a state that predisposes to subsequent development of a hematologic malignancy or cardiovascular death.44 This recent study used CRISPR/Cas9 to inactivate DNMT3A and TET2 genes in HSPC and showed that atherosclerotic plaque size was markedly increased in reconstituted mice.45

Somatic CRISPR/Cas9 mouse models Mouse models with somatic genome editing can be built by CRISPR/Cas9 modification of ex vivo cells followed by transplantation (murine cells) or xenotransplantation (human cells). For instance, the ability to modulate CRISPR/Cas9 activity has been exploited to perform doxycycline CRISPR/Cas9-inducible Trp53-knockout/mutation. When HSPC isolated from a lymphoma transgenic model (Em-Myc) were transplanted, this resulted in accelerated lymphoma development in vivo.

Thus, a highly efficient inducible CRISPR/Cas9 vector system can be used to identify novel gene mutations that drive tumorigenesis or to knock-out essential genes that are required for cell survival in vitro.46 As mentioned in the previous section, one of the unique features of the CRISPR/Cas9 system is its simplicity in enabling simultaneous disruption of several sites in 887

E. González-Romero et al.

the genome.12 Multiplex CRISPR/Cas9 editing of genes mutated in human leukemias has been demonstrated in mouse and human cells using either lentiviral or ribonucleoprotein approaches. Edited cells were then transplanted into conditioned animals and the identity of the disrupted genes was revealed by next-generation sequencing from clones expanded in sick mice.19,35,47,48 Moreover, ex vivo CRISPR/Cas9 gene editing of HSPC is also useful for studying clonal hematopoiesis of indeterminate potential.19 Multiplex ribonucleoprotein-editing and tracking clonal dynamics by high-throughput sequencing revealed the expansion of mutant clones resembling human clonal hematopoiesis of indeterminate potential, some of which continued to expand and cause death, by hematopoietic failure or AML, in transplanted mice.

Accordingly, multiplex CRISPR/Cas9 gene editing is an advantageous tool for functional genomics and for modeling the mutational complexity and co-occurrence patterns observed in hematologic patients at diagnosis, who in the case of AML, carry an average of 2.3 genomic mutations.49 A number of publications on the use of CRISPR/Cas9 gene editing in hematologic research are listed in Table 5.

Gene editing as a therapeutic application in hematologic disorders Allogeneic HSC transplantation is the frontline treatment for many hematologic disorders; however, this

option is only available when a suitable donor exists.

Nevertheless, transplanted patients can develop graft-versus-host disease and die of transplant-associated causes.

In this scenario, ex vivo gene therapy using viral vectors and ex vivo gene editing by TALEN or zinc-finger nucleases in hematopoietic cells followed by autologous HSC transplantation represent therapeutic alternatives that are currently being investigated in clinical trials.50 However, permanent viral integration into the host genome and/or insertional activation of proto-oncogenes that could lead to secondary leukemia are potential pitfalls related to integrative vector-based gene therapy. 51 Site-specific endonucleases, especially CRISPR/Cas9, offer the possibility of delivering non-integrative editing components into target cells, such as mRNA and ribonucleoproteins, constituting a promising approach for HSC gene editing.

Inherited diseases Clinically, CRISPR/Cas9 gene editing holds promise for monogenic hematologic disorders and, thus far, it has been mainly employed in hemoglobinopathies. β-thalassemia is caused by mutations in the human hemoglobin beta (HBB) gene and is characterized by reduced βhemoglobin production, resulting in hemoglobin clumping, hemolytic anemia, and ineffective erythropoiesis.

One strategy to remedy this defect using CRISPR/Cas9 is to repair the HBB mutation as has been achieved in iPSC from patients with β-thalassemia.52,53 Another strategy is to reactivate the fetal hemoglobin gene via disruption of the BCL11A gene, an erythroid enhancer regulator of the

Table 5. List of studies on CRISPR/Cas9 gene editing in hematologic diseases.

Disease

Gene/s

Aim/Repair pathway

Target cells

Format/Delivery

Reference

Myeloid malignancies TET2, RUNX1, DNMT3A, Knock out/NHEJ LSK Two-vector system/Lentivirus NF1, EZH2 and SMC3 Myeloid malignancies 192 chromatin Knock out/NHEJ RN2 with constitutive One-vector system/Lentivirus regulatory domains Cas9 expression MDS SRSF2 Point mutation/HDR K562 CRISPR vector and ssODN/Electroporation MDS, CMML, AML ASXL1 Mutation correction/HDR KBM5 CRISPR vector and ssODN/Electroporation MLL MLL and AF4 Chromosomal rearrangements/ HDR HEK293 CRISPR vector and template plastmid/Lipofection AML IDH2 Knock in /HDR K562 CRISPR vector and template plasmid/Nucleofection AML IDH2 Mutation correction/HDR Primary AML blasts Two-vector system/Lentivirus SCN HAX1 Mutation correction/HDR iPSC CRISPR vector and ssODN/Lipofectamine Pediatric AML MLL and ENL Chromosomal Human HSPC One-vector system/Lentivirus rearrangements/ NHEJ AML and MDS TET2, ASXL1, DNMT3A, Knock out/NHEJ Human HSPC One-vector system/Lentivirus RUNX1, TP53, NF1, EZH2, STAG2, SMC3, SRSF2 and U2AF1 AML RUNX1 and ETO Chromosomal Human HSPC One-vector system/ Electroporation rearrangements/ NHEJ MDS ASXL1 Knock out/NHEJ U937 Two vector system/Electroporation XCGD CYBB Mutation correction/NHEJ PLB One vector system/Lentivirus CHIP DNMT3A and TET2 Knock out/NHEJ Human HSPC One vector system/Lentivirus CHIP FLT3, DNMT3A, SMC3, Knock out/NHEJ LSK RNP/Electroporation EZH2, RUNX1 and NF1

35

36 23 26 31 25 25 27 32 48

30 20 29 44 19

MDS: myelodysplastic syndromes; CMML: chronic myelomonocytic leukemia; AML: acute myeloid leukemia; MLL: mixed lineage leukemia; SCN: severe congenital neutropenia; XCGD: Xlinked chronic granulomatous disease; CHIP: clonal hematopoiesis of indeterminate potential; NHEJ: non-homologous end joining; HDR: homology-directed repair; LSK: Lin-Sca-1+c-Kit+; iPSC: induced pluripotent stem cells; HSPC: hematopoietic stem and progenitor cells; ssODN: single-stranded donor oligonucleotides; RPN: ribonucleoprotein.

888

haematologica | 2019; 104(5)

CRISPR/Cas9 gene editing in hematology

fetal-to-adult hemoglobin switch and silencer of fetal hemoglobin. BCL11A disruption by CRISPR/Cas9 was shown to facilitate the achievement of threshold levels of functional fetal hemoglobin for treating β-hemoglobinopathies.54 Likewise, sickle cell disease is caused by a major mutation in the HBB gene, resulting in abnormal hemoglobin and the production of malfunctioning erythrocytes. CRISPR/Cas9-mediated gene editing has been employed to correct one HBB allele in iPSC generated from patients with sickle cell disease,55 and to create the hereditary persistence of a fetal hemoglobin genotype in HSPC, which is suggested as an approach for treating βthalassemia and sickle cell disease.56,57 In the clinical setting, CTX001, a gene-edited autologous HSC therapy targeting the erythroid-specific enhancer of the BCL11A gene, is entering clinical trials for β-thalassemia (Europe) and sickle cell disease (USA) (Table 6). Specifically, ex vivo edited patients’ cells will be re-infused into patients to produce fetal hemoglobin-containing erythrocytes and overcome the hemoglobin deficiency. These approaches are remarkable because hemoglobinopathies represent a huge cost to healthcare systems as a consequence of frequent transfusions and hospital admissions. Accordingly, novel therapies for these diseases are in high demand.

CRISPR/Cas9 gene editing has recently been employed in Fanconi anemia, a rare genetic disease characterized by progressive bone marrow failure that results in decreased production of all blood cell types. In 80–90% of cases, Fanconi anemia is caused by mutations in FANCA,

FANCC or FANCG genes. CRISPR/Cas9-gene editing has successfully corrected a FANCC gene mutation in patientderived fibroblasts using Cas9 nickase, obtaining a higher correction frequency than Cas9 nuclease.58 As the name might suggest, nickases introduce a single-strand break or “nick” rather than a DSB. Although clinical trials using other nucleases in Fanconi anemia are ongoing,59 to the best of our knowledge CRISPR/Cas9 gene editing to treat this disease has not been employed thus far.

Immunodeficiencies Primary immunodeficiencies Primary immunodeficiencies are a heterogeneous group of disorders characterized by variable susceptibility to infections due to hereditary defects in the immune system. One such immunodeficiency, X-linked chronic granulomatous disease, is caused by mutations in the CYBB gene encoding gp91phox, a component of the NADPH oxidase in phagocytes which, when mutated, results in fatal infections. HDR-based therapeutic genome editing (zinc-finger nucleases and CRISPR/Cas9) has been employed to correct a CYBB mutation and restore the functional defect in human HSPC.29 Wiskott-Aldrich syndrome (WAS) is a severe X-linked primary immunodeficiency caused by mutations in the WAS gene and characterized by thrombocytopenia, recurrent infections, tumor development, and autoimmune diseases. Recently, CRISPR/Cas9 gene editing of the WAS locus was reported in a leukemic cell line.60 These preclinical studies hold

Table 6. Current clinical trials in hematology using the CRISPR/Cas9 system.

Disease β-thalassemia SCD

Product

Aim/title

Phase

CT identifier

Industry/Academy

CTX001

A safety and efficacy study evaluating CTX001 in subjects with transfusion-dependent β-thalassemia

Enrolling

NCT03655678 2017-003351-38

CRISPR Therapeutics

iHSC treatment group CTX001

iHSC with the gene correction of HBB intervent subjects with β-thalassemia mutations A safety and efficacy study evaluating CTX001 in subjects with severe sickle cell disease Safety of transplantation of CRISPR CCR5 modified CD34+ HSPC in HIV-infected subjects with hematologic malignancies A study evaluating UCART019 in patients with relapsed or refractory CD19+ leukemia and lymphoma

Not yet recruiting

NCT03728322

IND & CTA approved

NCT03745287

Allife Medical Science and Technology CRISPR Therapeutics

Enrolling

NCT03164135

Beijing, China

Phase I/II

NCT03166878

Beijing, China

Phase I/II

NCT03398967

Beijing, China

Initiates in first-half of 2019 Preclinical

NA

CRISPR Therapeutics

NCT03399448

Pennsylvania, USA

Research

NA

CRISPR Therapeutics

Phase I

NCT03690011

Houston, USA

HIV-1 infection

CCR5 gene modification

B-cell leukemia B-cell lymphoma

UCART019 CTX101

CD19+ leukemia CD19+ lymphoma

CTX110 NYCE T Cells

A feasibility and safety study of universal dual specificity CD19 and CD20 or CD22 CAR-T-cell immunotherapy Anti-CD19 allogeneic CAR-T cells with TCR and B2M knocked-out NY-ESO-1-redirected CRISPR (TCR endogenous and PD1) edited T cells

Multiple myeloma CTX120 T-cell ALL CD7.CAR/28zeta T-cell lymphoblastic CAR-T cells lymphoma T-non-Hodgkin lymphoma

Anti-BCMA allogeneic CAR-T cells with TCR and B2M knocked-out Cell therapy for high risk T-cell malignancies using CD7-specific CAR expressed on autologous T cells (CRIMSON)

SCD: sickle cell disease; HIV-1: human immunodeficiency virus type 1; ALL: acute lymphoblastic leukemia; iHSC: induced hematopoietic stem cells; CAR: chimeric antigen receptor; HBB: β hemoglobin; HSPC: hematopoietic stem and progenitor cells; UCART: universal CAR-T cells; TCR: T-cell receptor; B2M: β2-microglobulin; BCMA: B-cell maturation antigen; IND & CTA: investigational new drug and clinical trial authorization; CT: clinical trial; NA: not applicable.

haematologica | 2019; 104(5)

889

E. González-Romero et al.

promise for the clinical translation of CRISPR/Cas9 gene editing therapies for primary immunodeficiencies.

Acquired immunodeficiencies Human immunodeficiency virus type-1 (HIV-1) infection gives rise to acquired immune deficiency syndrome (AIDS). Several different CRISPR/Cas9 gene-editing strategies have been used to target HIV-1 along its replication cycle.61 For example, disruption of the CCR5 receptor gene, which is present in the host cell and is a co-factor for entry of HIV into the cell, represents a promising strategy to combat the disease, and several studies have reported NHEJ-mediated CRISPR/Cas9 inactivation of CCR5 and other co-factors in lymphocytes. In one study, generation of iPSC homozygous for the naturally occurring CCR5Δ32 mutation through CRISPR/Cas9 genome editing conferred resistance to HIV infection.56 Antiretroviral therapy fails to cure HIV-1 infection because of the persistence of HIV reservoirs harboring integrated HIV DNA. CRISPR/Cas9mediated deletion or inactivation of proviral DNA could eliminate this source of HIV persistence, thereby being a potentially curative treatment. In preclinical studies, CRISPR/Cas9 efficiently mutated and deactivated HIV proviral DNA in latently infected Jurkat cells.62 However, complete eradication of HIV latent infection is challenging because of the development of mutations resistant to attack by DNA-shearing enzymes.51 Clinically, the safety of transplantation of CRISPR CCR5-modified CD34+ cells in HIV-infected patients with hematologic malignancies is under evaluation in clinical trials (Table 6).

Cancer immunotherapy using chimeric antigen receptor T cells Cancer immunotherapy can be defined as the induction or enhancement of cancer-specific immune responses against malignant tumors. One approach to this is the ex vivo manipulation of patients’ T cells to express a chimeric antigen receptor (CAR) including an intracellular chimeric signaling domain capable of activating T cells and an extracellular binding domain that recognizes an antigen specific for and strongly expressed on tumor cells. CAR-T cells are re-infused into patients to attack cancer cells in vivo. Currently, CAR-T cells expressing CD19, CD20, CD22, or dual B targeting CAR-T cells are available to treat acute lymphoblastic leukemia, non-Hodgkin lymphoma and chronic lymphocytic leukemia.63 Unfortunately, CAR-T-cell administration can have adverse effects, such as neurotoxicity, cytopenia and cytokine release syndrome, which can be life-threatening.

CRISPR/Cas9 can be utilized to complement CAR-T-cell therapy, for example, via disruption of the endogenous Tcell receptor (TCR). Upon its interaction with engineered, transgenic TCR in patients’ cells, endogenous TCR can alter the antigen specificity of CAR-T cells. In a study using CRISPR/Cas9 to knock out endogenous TCR-β, with simultaneous introduction of CAR-T cells, the authors found that this replacement strategy resulted in more robust transgenic anticancer T cells.64 The CRISPR/Cas9 system has also be applied to eliminate other genes that encode inhibitory T-cell surface receptors, such as programmed cell death protein 1 (PD1), to improve the efficiency of T-cell-based immunotherapy.65 To exploit CAR-T-cell therapy beyond the autologous setting, allogeneic universal T cells derived from healthy donors could be engineered by CRISPR/Cas9 upon disrup890

tion of TCR to prevent graft-versus-host disease, or beta-2microglobulin, to eliminate major histocompatibility complex I (MHC I) expression, or by integrating a CAR precisely at the disrupted T-cell receptor a constant (TRAC) locus to improve safety and potency.66-68 Thus, CRISPR/Cas9 technology holds enormous promise to advance the field of cancer immunotherapy and several clinical trials are running to assess the efficacy of CRISPR/Cas9-edited CAR-T cells (Table 6).

Challenges and opportunities for CRISPR/Cas9 therapeutic applications Delivery of editing tools Delivery platforms that ensure the access of editing components into a large number of target cells are critical for the clinical development of this technique.

Ribonucleoprotein is the cargo format preferred over other transient delivery methods such as mRNA and nonintegrating viral vectors because of its hit-and-run mechanism, which reduces the risk of undesired off-target effects, and also because of its ability to efficiently modify cells with low translation rates.47 Nevertheless, the benefit-to-harm ratio of the CRISPR/Cas9 system must be maximized. Possible solutions include the development of novel approaches to integrate ribonucleoprotein and donor template DNA for gene correction in a unique system.

Safety Off-target DSB can result from non-specific Cas9 cleavage at unwanted genome sites, which is perhaps the major concern regarding therapeutic CRISPR/Cas9 editing. Accordingly, genome-wide sequencing approaches should be employed to thoroughly examine for modifications at unexpected genome locations, or at anticipated off-target sites indicated by in silico prediction tools. The issue of off-target activity is, nevertheless, controversial since studies have yielded contrasting results.69,70 Along this line, several methods have been developed in the last years to detect CRISPR off-target mutations;71 however, there is a lack of consensus on how to predict which putative off-target sites should be examined via deep targeted sequencing. Additionally, the possibility of Cas9induced on-target mutagenesis, including large deletions and rearrangements that may have pathogenic consequences, has been highlighted as another safety concern.72 Accordingly, more research is needed for a definitive understanding of the in vivo genomic effects of CRISPR/Cas9. Indeed, the possibility of producing undesired gene modifications raises concerns about the use of the CRISPR/Cas9 system for therapy in humans. For instance, infused gene-edited HSC could have the potential to expand clonally and induce leukemia, and so clinical gene editing might cause panic. Possible solutions include the substitution of Cas9 with a different nuclease, for example, Cas12a (also known as Cpf1), which prohibits mismatches between the 18 nucleotides next to the protospacer adjacent motif.73 Other alternatives include the use of paired nickases, guided by two different gRNA targeting the same locus but on opposite DNA strands, or “base editors” editing nucleotides without inducing a DNA break.17 haematologica | 2019; 104(5)

CRISPR/Cas9 gene editing in hematology

Efficiency Suboptimal DNA repair outcomes or insufficient target conversion might prevent an intervention from reaching a critical gene-editing threshold necessary to rescue the genetic defect. Strategies to enhance the frequency of HDR in CRISPR/Cas9-mediated transgenesis have been reported and need to be tested in the clinical context.74 Moreover, it should be considered that efficacy could be reduced if the CRISPR/Cas9-induced mutation is detrimental to the cells, having a negative, non-reversible effect.

Immunogenicity The immune system reaction to in vivo administration of gene editing reagents or ex-vivo genetically modified cells is also a cause for concern.75 The presence of antibodies against Cas9, mainly isolated from Staphylococcus aureus or Streptococcus pyogenes, is common in neonates and adults.

Similarly, T lymphocytes against Staphylococcus aureus Cas9 constitute an obstacle to CRISPR/Cas9 therapeutic gene editing.76 Accordingly, the possible immune response must be examined in depth to ascertain whether it could compromise the efficacy of CRISPR-based treatments.

Strategies to minimize/eliminate immunogenicity include the use of nucleases other than Cas9 that have not been exposed to the human immune system, or novel nucleases that do not activate an immune response. Other strategies could be: (i) to design an in silico prediction tool for immunogenic predisposition; (ii) to understand the innate immune mechanism against CRISPR/Cas9 in order to help in vector choice and engineering; (iii) to identify antigenic regions on CRISPR/Cas9 to enable deimmunization and epitope masking; and (iv) to employ immunosuppression by using drugs and/or regulatory T cells to reduce undesired immune reactivity.77

p53-mediated DNA damage response CRISPR/Cas9 genome editing has recently been shown to induce a p53-mediated DNA damage response in some human cell types,78,79 which is in part responsible for the low targeting efficiencies observed in these cells.

Consequently, p53 inhibition may improve the efficiency of genome editing in wild type cells; however, a caveat to this approach would be the increased likelihood of cancerous transformation of cells in which the â&#x20AC;&#x153;guardianâ&#x20AC;? activity of p53 is inhibited. p53 gene sequence and function should, therefore, be monitored closely in cells destined for therapy when developing CRISPR/Cas9 cell-based therapies.

Bioethical regulation CRISPR/Cas9 gene editing is associated with several ethical issues; for example, its application to humans, embryos or germline cells. While the clinical application in human somatic cells to treat hematologic diseases is generally accepted, there is consensus among geneticists that its application in human embryos and germline cells (except for research purposes), in which genetic changes would be inherited by future generations, should be impermissible. That being said, some alarming news was recently reported about the use of CRISPR/Cas9 in human embryos to inactivate the CCR5 receptor and provoke resistance to HIV infection. The biophysicist He Jiankui presented limited (and non-peer-reviewed) data on the birth of twin girls genetically edited with CRISPR/Cas9. haematologica | 2019; 104(5)

This claim, whether true or not, urgently imposes the establishment of strict regulations on human CRISPR/Cas9 genome editing such that it should only be considered for therapeutic uses, but not for human enhancement or eugenics, although it could be used as a research tool to understand early human development or disease pathogenesis. Thus far, no patients have been clinically treated with in vivo CRISPR-based therapy, whereas patients have been given infusions of ex vivo modified T cells (Table 6).

The ethical and regulatory aspects of therapeutic CRISPR/Cas9 genome editing are very complex.80 Given the proven potential of CRISPR/Cas9 to modify the human genome, there are naturally great expectations for future applications. To discuss all these concerns, a multidisciplinary regulatory committee, composed of geneticists, lawyers, society representatives and clinicians, should be created to define a legislative framework to regulate permission or prohibition of CRISPR applications and any genome engineering technique in the future.

Global scientific and biological ethics communities must take the lead and establish standards and procedures that reduce the dangers of these powerful new technologies without forgetting the benefits.

What is necessary to move the hematology field forward As with any new treatment, safety and efficacy are very important. Efforts should be made to develop novel vector systems to maximize delivery to target cells with minimal side effects. In this sense, platforms for systemic and local administration of CRISPR/Cas9 would be highly desirable, including non-integrative vectors for gene therapy or systems based on nanoparticle technology, ribonucleoproteins or other novel approaches.

In the last decade, next-generation sequencing has led to enormous advances in the molecular diagnostics of hematologic malignancies. Somatic mutations have been revealed in many diseases and some may have important prognostic value. The functional impact of these mutations on tumor initiation and/or maintenance needs to be addressed in the next years, and CRISPR/Cas9-based screens in patient-derived cells will be powerful tools to undertake this endeavor. In addition, some hematologic malignancies are characterized by mutations in epigenetic modifiers, proteins that modify DNA at cytosine residues or cause post-translational histone modifications. Some therapies already exploit epigenetic targets, such as DNA methyltransferase 3A (DNMT3A) or histone deacetylase (HDAC), and hypomethylating agents, including the DNA methyltransferase inhibitors azacytidine and decitabine, are used to treat myelodysplastic syndrome and AML. Accordingly, targeted epigenome editing,81 which is the modification of the epigenome at specific sites as opposed to whole-genome modification, could be an area for research development in hematology â&#x20AC;&#x201C; for example, for fine-tuning gene expression by locally modulating DNA methylation or determining the function of specific methylation sites. Because epigenome editing does not involve genetic changes, it may also be less hazardous with respect to off-target effects. The challenges will be how to administer epigenome-editing tools in vivo, to achieve reversible epigenetic modifications at precise sites 891

E. Gonzรกlez-Romero et al.

and to ensure that epigenetic modification is heritable upon cell division.

Last but not least, the contribution of CRISPR/Cas9 to multiplex editing in CAR-T cells to create safer and more effective treatments cannot be underestimated, and it is highly likely that this technique will move forward the field of cancer immunotherapy for personalized T-cellbased therapies. The engineering of novel CAR-T cells by pharmaceutical industries, resulting in costly and unaffordable treatments for the general population, should be accompanied by their production by academia institutions, which could make it easier to tailor CAR-T cells for each patient. Thus, drug regulatory authorities should facilitate their academic production and provide resources for CAR-T- cell manufacturing processes, so that these can be simplified and automated to enable scaling up of these cell products.

Conclusions

other, more time-consuming and expensive approaches such as zinc-finger nucleases or TALEN. In hematology, CRISPR/Cas9 can be used to model diseases using cultured cells or model organisms, but perhaps more importantly, it can be a valuable approach to correct ex vivo mutations and chromosomal aberrations in cells from patients with blood disorders for autologous HSC transplantation. However, many pitfalls and challenges need to be overcome for the translation of CRISPR/Cas9 gene editing to the clinic. For example, we do not know the minimum number of edited cells needed to functionally correct a genetic defect or if gene editing can be applied to treat multigenic diseases. Further research is necessary to implement CRISPR/Cas9 in the clinical context, so that genome editing-based treatments are available to patients.

In conclusion, the CRISPR/Cas9 revolution brings us a specific, efficient and versatile tool for editing genes.

Nowadays, technology is no longer a limitation and scientists can probably do any genetic manipulation they can dream of.

CRISPR/Cas9 technology is a revolutionary approach for genome editing with wide applications in molecular biology, genetics, and medicine. It has great potential for dissecting gene function, modeling diseases and editing human genes for curative treatment. The number of publications in this field has doubled every year since its introduction, and the CRISPR/Cas9 system is now more widely used in biotechnology and research laboratories than

Acknowledgments This work was supported by Ayudas FEDER CIBERONC [CB16/12/00284], Instituto de Salud Carlos III [PI16/01113, PI17/00011, CP16/00011], Conselleria de Educaciรณn, Investigaciรณn, Cultura y Deporte [PROMETEOII/2015/008, ACIF/2018/255], MINECO [RYC-2015-17534, SAF201782171-R], and Beca Leonardo a Investigadores y Creadores Culturales de la Fundaciรณn BBVA.

References 1. Mortensen R. Overview of gene targeting by homologous recombination. Curr Protoc Neurosci. 2007;Chapter 4:Unit 4.29.

2. Porteus MH, Baltimore D. Chimeric nucleases stimulate gene targeting in human cells.

Science. 2003;300(5620):763.

3. Smith J, Grizot S, Arnould S, et al. A combinatorial approach to create artificial homing endonucleases cleaving chosen sequences.

Nucleic Acids Res. 2006;34(22):e149.

4. Urnov FD, Miller JC, Lee YL, et al. Highly efficient endogenous human gene correction using designed zinc-finger nucleases.

Nature. 2005;435(7042):646-651.

5. Miller JC, Tan S, Qiao G, et al. A TALE nuclease architecture for efficient genome editing. Nat Biotechnol. 2011;29(2):143-148.

6. Ishino Y, Shinagawa H, Makino K, Amemura M, Nakata A. Nucleotide sequence of the iap gene, responsible for alkaline phosphatase isozyme conversion in Escherichia coli, and identification of the gene product. J Bacteriol.

1987;169(12):5429-5433.

7. Mojica FJ, Ferrer C, Juez G, RodriguezValera F. Long stretches of short tandem repeats are present in the largest replicons of the Archaea Haloferax mediterranei and Haloferax volcanii and could be involved in replicon partitioning. Mol Microbiol.

1995;17(1):85-93.

8. Mojica FJ, Diez-Villasenor C, GarciaMartinez J, Soria E. Intervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements. J Mol Evol.

2005;60(2):174-182.

892

9. Mojica FJ, Diez-Villasenor C, GarciaMartinez J, Almendros C. Short motif sequences determine the targets of the prokaryotic CRISPR defence system.

Microbiology. 2009;155(Pt3):733-740.

10. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science.

2012;337(6096):816-821.

11. Doudna JA, Charpentier E. Genome editing.

The new frontier of genome engineering with CRISPR-Cas9. Science. 2014;346 (6213):1258096.

12. Cong L, Ran FA, Cox D, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;339(6121):819-823.

13. Mali P, Yang L, Esvelt KM, et al. RNA-guided human genome engineering via Cas9.

Science. 2013;339(6121):823-826.

14. Jinek M, East A, Cheng A, Lin S, Ma E, Doudna J. RNA-programmed genome editing in human cells. Elife. 2013;2:e00471.

15. Bibikova M, Golic M, Golic KG, Carroll D. Targeted chromosomal cleavage and mutagenesis in Drosophila using zinc-finger nucleases. Genetics. 2002;161(3):1169-1175.

16. Capecchi MR. Altering the genome by homologous recombination. Science.

1989;244(4910):1288-1292.

17. Wu WY, Lebbink JHG, Kanaar R, Geijsen N, van der Oost J. Genome editing by natural and engineered CRISPR-associated nucleases. Nat Chem Biol. 2018;14(7):642-651.

18. Lino CA, Harper JC, Carney JP, Timlin JA. Delivering CRISPR: a review of the challenges and approaches. Drug Deliv.

2018;25(1):1234-1257.

19. Shi X, Kitano A, Jiang Y, Luu V, Hoegenauer KA, Nakada D. et al. Clonal expansion and myeloid leukemia progression modeled by multiplex gene editing of murine hematopoietic progenitor cells. Exp Hematol. 2018;64:33-44.e5.

20. Wu ZJ, Zhao X, Banaszak LG, et al.

CRISPR/Cas9-mediated ASXL1 mutations in U937 cells disrupt myeloid differentiation. Int J Oncol. 2018;52(4):1209-1223.

21. Hannon GJ, Rossi JJ. Unlocking the potential of the human genome with RNA interference. Nature. 2004;431(7006):371-378.

22. Liu ET, Bolcun-Filas E, Grass DS, et al. Of mice and CRISPR: the post-CRISPR future of the mouse as a model system for the human condition. EMBO Rep. 2017;18(2): 187-193.

23. Zhang J, Lieu YK, Ali AM, et al. Diseaseassociated mutation in SRSF2 misregulates splicing by altering RNA-binding affinities.

Proc Natl Acad Sci U S A. 2015;112(34): E4726-4734.

24. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209-2221.

25. Brabetz O, Alla V, Angenendt L, et al. RNAguided CRISPR-Cas9 system-mediated engineering of acute myeloid leukemia mutations. Mol Ther Nucleic Acids.

2017;6:243-248.

26. Valletta S, Dolatshad H, Bartenstein M, et al.

ASXL1 mutation correction by CRISPR/Cas9 restores gene function in leukemia cells and increases survival in mouse xenografts. Oncotarget. 2015;6(42): 44061-44071.

haematologica | 2019; 104(5)

CRISPR/Cas9 gene editing in hematology

27. Pittermann E, Lachmann N, MacLean G, et al. Gene correction of HAX1 reversed Kostmann disease phenotype in patient-specific induced pluripotent stem cells. Blood Adv. 2017;1(14):903-914.

28. Kaufman DS. Tu-mor(e) blood cells from human pluripotent stem cells. Blood.

2013;12188):1245-1246.

29. Sürün D, Schwäble J, Tomasovic A, et al.

High efficiency gene correction in hematopoietic cells by donor-template-free CRISPR/Cas9 genome editing. Mol Ther Nucleic Acids. 2018;10:1-8.

30. Torres R, Martin MC, Garcia A, Cigudosa JC, Ramirez JC, Rodriguez-Perales S. Engineering human tumour-associated chromosomal translocations with the RNA-guided CRISPR-Cas9 system. Nat Commun.

2014;5:3964.

31. Castaño J, Herrero AB, Bursen A, et al.

Expression of MLL-AF4 or AF4-MLL fusions does not impact the efficiency of DNA damage repair. Oncotarget. 2016;7(21):3044030452.

32. Reimer J, Knöß, S, Labuhn M, et al. CRISPRCas9-induced t(11;19)/MLL-ENL translocations initiate leukemia in human hematopoietic progenitor cells in vivo. Haematologica.

2017;102(9):1558-1566.

33. Rabbitts TH, Appert A, Chung G, et al.

Mouse models of human chromosomal translocations and approaches to cancer therapy. Blood Cells Mol Dis. 2001;27(1): 249-259.

34. Zhou Y, Zhu S, Cai C, et al. High-throughput screening of a CRISPR/Cas9 library for functional genomics in human cells. Nature.

2014;509(7501):487-491.

35. Heckl D, Kowalczyk MS, Yudovich D, et al.

Generation of mouse models of myeloid malignancy with combinatorial genetic lesions using CRISPR-Cas9 genome editing.

Nat Biotechnol. 2014;32(9):941-946.

36. Shi J, Wang E, Milazzo JP, Wang Z, Kinney JB, Vakoc CR. Discovery of cancer drug targets by CRISPR-Cas9 screening of protein domains. Nat Biotechnol. 2015;33(6):661667.

37. Rathe SK, Moriarity BS, Stoltenberg CB, et al. Using RNA-seq and targeted nucleases to identify mechanisms of drug resistance in acute myeloid leukemia. Sci Rep. 2014; 4:6048.

38. Mou H, Kennedy Z, Anderson DG, Yin H, Xue W. Precision cancer mouse models through genome editing with CRISPR-Cas9.

Genome Med. 201;7(1):53.

39. Horiuchi K, Perez-Cerezales S, Papasaikas P, et al. Impaired spermatogenesis, muscle, and erythrocyte function in U12 intron splicingdefective zrsr1 mutant mice. Cell Rep. 2018;23(1):143-155.

40. Yen CT, Fan MN, Yang YL, Chou SC, Yu IS, Lin SW. Current animal models of hemophilia: the state of the art. Thromb J. 2016;14(Suppl 1):22 eCollection 2016.

41. Xiang P, Wei W, Hofs N, et al. A knock-in mouse strain facilitates dynamic tracking and enrichment of MEIS1. Blood Adv. 2017;1(24):2225-2235.

42. Wang H, Yang H, Shivalila CS, et al. Onestep generation of mice carrying mutations in multiple genes by CRISPR/Cas-mediated genome engineering. Cell. 2013;153(4):910918.

43. Zhou J, Shen B, Zhang W, et al. One-step generation of different immunodeficient mice with multiple gene modifications by CRISPR/Cas9 mediated genome engineering. Int J Biochem Cell Biol. 2014;46:49-55.

haematologica | 2019; 104(5)

44. Sano S, et al. CRISPR-Mediated gene editing to assess the roles of Tet2 and Dnmt3a in clonal hematopoiesis and cardiovascular disease. Circ Res. 2018;123(3):335-341.

45. Fuster JJ, MacLauchlan S, Zuriaga MA, et al.

Clonal hematopoiesis associated with TET2 deficiency accelerates atherosclerosis development in mice. Science. 2017;355(6327): 842-847.

46. Aubrey BJ, Kelly GL, Kueh AJ, et al. An inducible lentiviral guide RNA platform enables the identification of tumor-essential genes and tumor-promoting mutations in vivo. Cell Rep. 2015;10(8):1422-1432.

47. Gundry MC, Brunetti L, Lin A, et al. Highly Efficient Genome Editing of Murine and Human Hematopoietic Progenitor Cells by CRISPR/Cas9. Cell Rep. 2016;17(5):14531461.

48. Tothova Z, Krill-Burger JM, Popova KD, et al. Multiplex CRISPR/Cas9-based genome editing in human hematopoietic stem cells models clonal hematopoiesis and myeloid neoplasia. Cell Stem Cell. 2017;21(4):547555.e8.

49. Hussaini MO, Mirza AS, Komrokji R, Lancet J, Padron E, Song J. Genetic landscape of acute myeloid leukemia interrogated by next-generation sequencing: a large cancer center experience. Cancer Genomics Proteomics. 2018;15(2):121-126.

50. Dunbar CE, High KA, Joung JK, Kohn DB, Ozawa K, Sadelain M. Gene therapy comes of age. Science. 2018;359(6372).

51. Zhang H, McCarty N. CRISPR editing in biological and biomedical investigation. J Cell Biochem. 2017;118(12):4152-4162.

52. Xie F, Ye L, Chang JC, et al. Seamless gene correction of beta-thalassemia mutations in patient-specific iPSCs using CRISPR/Cas9 and piggyBac. Genome Res. 2014;24(9): 1526-1533.

53. Song B, Fan Y, He W, et al. Improved hematopoietic differentiation efficiency of gene-corrected beta-thalassemia induced pluripotent stem cells by CRISPR/Cas9 system. Stem Cells Dev. 2015;24(9):1053-1065.

54. Canver MC, Smith EC, Sher F, et al. BCL11A enhancer dissection by Cas9-mediated in situ saturating mutagenesis. Nature.

2015;527(7577):192-197.

55. Huang X, Wang Y, Yan W, et al. Production of gene-corrected adult beta globin protein in human erythrocytes differentiated from patient iPSCs after genome editing of the sickle point mutation. Stem Cells. 2015; 33(5):1470-1479.

56. Ye L, Wang J, Tan Y, et al. Genome editing using CRISPR-Cas9 to create the HPFH genotype in HSPCs: an approach for treating sickle cell disease and beta-thalassemia. Proc Natl Acad Sci U S A. 2016;113(38): 1066110665.

57. Traxler EA, Yao Y, Wang YD, et al. A genome-editing strategy to treat beta-hemoglobinopathies that recapitulates a mutation associated with a benign genetic condition.

Nat Med. 2016;22(9):987-990.

58. Osborn M, Lonetree CL, Webber BR, et al.

CRISPR/Cas9 targeted gene editing and cellular engineering in Fanconi anemia. Stem Cells Dev. 2016;25(20):1591-1603.

59. Rio P, Navarro S, Bueren JA. Advances in gene therapy for Fanconi anemia. Hum Gene Ther. 2018;29(10):1114-1123.

60. Gutierrez-Guerrero A, Sanchez-Hernandez S, Galvani G, et al. Comparison of zinc finger nucleases versus CRISPR-specific nucleases for genome editing of the WiskottAldrich syndrome locus. Hum Gene Ther.

2018;29(3):366-380.

61. Wang G, Zhao N, Berkhout B, Das, AT. CRISPR-Cas based antiviral strategies against HIV-1. Virus Res. 2018;244:321-332.

62. Zhu W, Lei R, Le Duff Y, et al. The CRISPR/Cas9 system inactivates latent HIV1 proviral DNA. Retrovirology. 2015;12:22.

63. Paczesny S, Pavletic SZ, Bollard CM. Introduction to a review series on emerging immunotherapies for hematologic diseases.

Blood. 2018;131(24):2617-2620.

64. Legut M, Dolton G, Mian AA, Ottmann OG, Sewell AK. CRISPR-mediated TCR replacement generates superior anticancer transgenic T cells. Blood. 2018;131(3):311-322.

65. Xia AL, He QF, Wang JC, et al. Applications and advances of CRISPR-Cas9 in cancer immunotherapy. J Med Genet. 2019;56(1):49.

66. Ren J, Liu X, Fang C, Jiang S, June CH, Zhao Y. Multiplex genome editing to generate universal CAR T cells resistant to PD1 inhibition. Clin Cancer Res. 2017;23(9):2255-2266.

67. Eyquem J, Mansilla-Soto J, Giavridis T, et al.

Targeting a CAR to the TRAC locus with CRISPR/Cas9 enhances tumour rejection.

Nature. 2017;543(7643):113-117.

68. Liu X, Zhao Y. CRISPR/Cas9 genome editing: Fueling the revolution in cancer immunotherapy. Curr Res Transl Med. 2018;66(2):39-42.

69. Aryal NK, Wasylishen AR, Lozano G. CRISPR/Cas9 can mediate high-efficiency off-target mutations in mice in vivo. Cell Death Dis. 2018;9(11):1099.

70. Iyer V, Shen B, Zhang W, et al. Off-target mutations are rare in Cas9-modified mice.

Nat Methods. 2015;12(6):479.

71. Tsai SQ, Joung JK. Defining and improving the genome-wide specificities of CRISPRCas9 nucleases. Nat Rev Genet.

2016;17(5):300-312.

72. Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPRCas9 leads to large deletions and complex rearrangements. Nat Biotechnol. 2018;36(8): 765-771.

73. Zetsche B, Gootenberg JS, Abudayyeh OO, et al. Cpf1 is a single RNA-guided endonuclease of a class 2 CRISPR-Cas system. Cell.

2015;163(3):759-771.

74. Devkota S. The road less traveled: strategies to enhance the frequency of homologydirected repair (HDR) for increased efficiency of CRISPR/Cas-mediated transgenesis.

BMB Rep. 2018;51(9):437-443.

75. Maeder ML, Gersbach CA. Genome-editing technologies for gene and cell therapy. Mol Ther. 2016;24(3):430-446.

76. Charlesworth CT, Deshpande PS, Dever DP, et al. Identification of pre-existing adaptive immunity to Cas9 proteins in humans. bioRxiv. 2018;243345.

77. Chew WL. Immunity to CRISPR Cas9 and Cas12a therapeutics. Wiley Interdiscip Rev Syst Biol Med. 2018;10(1).

78. Haapaniemi E, Botla S, Persson J, Schmierer B, Taipale J. CRISPR-Cas9 genome editing induces a p53-mediated DNA damage response. Nat Med. 2018;24(7):927-930.

79. Ihry RJ, Worringer KA, Salick MR, et al. p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat Med. 2018;24(7):939-946.

80. Kohn DB, Porteus MH, Scharenberg AM. Ethical and regulatory aspects of genome editing. Blood. 2016;127(21):2553-2560.

81. Kungulovski G, Jeltsch A. Epigenome editing: state of the art, concepts, and perspectives.

Trends Genet. 2016;32(2):101-113.

893

ARTICLE Ferrata Storti Foundation

Hematopoiesis

Long noncoding RNAs of single hematopoietic stem and progenitor cells in healthy and dysplastic human bone marrow Zhijie Wu,1* Shouguo Gao,1* Xin Zhao,1 Jinguo Chen,2 Keyvan Keyvanfar,1 Xingmin Feng,1 Sachiko Kajigaya1 and Neal S. Young1

Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health and 2Trans-NIH Center for Human Immunology, Autoimmunity, and Inflammation, National Institutes of Health, Bethesda, MD, USA

1

Haematologica 2019 Volume 104(5):894-906

*ZW and SG contributed equally to this work.

ABSTRACT

L

Correspondence: ZHIJIE WU zhijie.wu@nih.gov Received: October 12, 2018.

Accepted: November 22, 2018.

Pre-published: December 13, 2018.

ong noncoding RNAs (lncRNAs) are regulators of cell differentiation and development. The lncRNA transcriptome in human hematopoietic stem and progenitor cells is not comprehensively defined. We investigated lncRNAs in 979 human bone marrow-derived CD34+ cells by single cell RNA sequencing followed by de novo transcriptome reconstruction. We identified 3,173 lncRNAs in total, among which 2,365 were previously unknown, and we characterized lncRNA stem, differentiation, and maturation signatures. lncRNA expression exhibited high cell-to-cell variation, which was only apparent in single cell analysis. lncRNA expression followed a lineage-specific and highly dynamic pattern during early hematopoiesis. lncRNAs in hematopoietic cells closely correlated with protein-coding genes of known functions in the regulation of hematopoiesis and cell fate decisions, and the potential regulatory roles of lncRNAs in hematopoiesis were imputed by projection from protein-coding genes with a “guilt-by-association” approach. We characterized lncRNAs preferentially expressed in hematopoietic stem cells and in various downstream differentiated lineage progenitors. We also profiled lncRNA expression in single cells from patients with myelodysplastic syndromes and in aneuploid cells in particular. Our study provides a global view of lncRNAs in human hematopoietic stem and progenitor cells. We observed a highly ordered pattern of lncRNA expression and participation in regulation of early hematopoiesis, and coordinate aberrant messenger RNA and lncRNA transcriptomes in dysplastic hematopoiesis. (Registered at clinicaltrials.gov with identifiers: 00001620, 00001397)

doi:10.3324/haematol.2018.208926 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/894 ©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.

894

Introduction Long noncoding RNAs (lncRNAs), which are defined as a subclass of noncoding RNAs, are longer than 200 nucleotides and lack protein coding capacity. lncRNAs are newly recognized as regulators of gene expression, transcriptionally and posttranscriptionally.1-3 Unlike messenger RNAs (mRNAs), which localize specifically to the cytoplasm, lncRNAs can occupy various nuclear compartments and/or the cytoplasm. lncRNAs function via RNA-DNA, RNA-RNA, and RNA-protein interactions.2-6 As a result, they affect multiple stages of gene regulation, including placement of chromatin marks, mRNA biogenesis, and signaling pathways. lncRNA expression is tissue- and cell type-specific5,7-9 but less conserved across species than is mRNA expression.10,11 lncRNAs have been linked to the development of several lineages in hematopoiesis and in the immune response. Some lncRNAs were found to be enriched in hematopoietic stem cells (HSCs)12 or dynamically expressed during erythropoiesis.13,14 RNA interference studies have revealed that lncRNAs control HSC self-renewal and differentiation,12 erythroid precursor maturation,14 and granulocytic differentiation of hematopoietic stem and progenitor cells (HSPCs).15 Intergenic lncRNA signatures exhibit subset-specificity in T and B lymphocytes.16-18 lincR-Ccr2-5’AS, together with GATA3, is essential in the regulation haematologica | 2019; 104(5)

Single cell lncRNAs in hematopoiesis

of gene expression and migration of Th2 cells.16 Downregulation of linc-MAF-4 skews T-cell differentiation towards the Th2 phenotype.17 TMEVPG1, a Th1-specific intergenic lncRNA, controls the expression of interferon-γ together with the Th1-specific transcription factor T-bet, and is critical in modulating susceptibility to infection with Theiler virus.19,20 Expression of lncRNAs in pro-B and mature B cells is regulated by PAX5, a transcriptional factor required to specify B-cell lineage.18 Despite these many examples of specific functions for either stem cells or differentiated lineages, the repertoire of lncRNAs in human HSPCs has not been fully described.

Whole transcriptome sequencing allows large scale profiling of lncRNAs in tissues and diseases and, therefore, enables the identification of many putative lncRNAs.5,21,22 lncRNAs in general are expressed at much lower levels3,4,23,24 but are more cell type-specific than are mRNAs.9,25 Until recently, lncRNA expression was assessed by averaging transcriptomes of bulk RNA extracted from mixed cell populations, which limits the sensitivity to detect lncRNA expression in small cell populations and thus to resolve diversity within a cell type. With recent advances in single cell transcriptome profiling methods, many seemingly homogeneous cell populations have shown unexpected variability in gene expression. Recently published studies profiling lncRNAs at the single cell level have revealed the cell-specific expression of these RNAs.5,26-30 In the current work, we performed single cell RNA sequencing (scRNA-seq) of 979 freshly isolated bone marrow-derived human CD34+ cells from both healthy donors and patients with myelodysplastic syndrome (MDS).

Using de novo transcriptome reconstruction, we identified a total of 3,173 lncRNAs, including 2,365 potential novel lncRNAs not reported in public databases. We further characterized the features and expression patterns of lncRNAs in CD34+ cells, revealing stage- and lineagespecificity of lncRNA expression and putative functions in normal hematopoiesis. Expression and lineage-specificity of almost 40 lncRNAs, including those novel lncRNAs, were validated by quantitative real-time polymerase chain reaction (RT-PCR). We also profiled lncRNAs in MDS cells, and aneuploid cells in particular. Our study provides a global assessment of lncRNA biology in early human hematopoiesis.

Methods Subjects and samples Bone marrow samples from seven healthy donors and five MDS patients were obtained after written informed consent in accordance with the Declaration of Helsinki and under protocols (www.clinicaltrials.gov NCT00001620 and NCT00001397) approved by the Institutional Review Boards of the National Heart, Lung, and Blood Institute. Of the five patients with MDS, patients 1, 2, and 5 had evolved to MDS from aplastic anemia while patients 3 and 4 had de novo MDS. Fluorescence activated cell sorting (FACS) was performed using the FACSAria II Cell Sorter (BD Biosciences) after isolation of bone marrow mononuclear cells. The gating strategies are shown in Online Supplementary Figure S1A. CD34+CD38- and CD34+CD38+ cells from four healthy donors and patient 4 were sequenced separately, while only the CD34+ populations of patients 1, 2, 3, and 5 were sequenced due to limited cell numbers (Online Supplementary Figure S1B). The clinical characteristics of these patients have been published.31 Another haematologica | 2019; 104(5)

set of bone marrow cells from a further three healthy donors was used for quantitative RT-PCR (Online Supplementary Figure S2).

Single cell RNA sequencing The C1 Single-cell Auto Prep System (Fluidigm) was employed to perform SMARTer (Clontech) whole transcriptome amplification on as many as 96 individual cells, according to the manufacturer’s protocols (www.fluidigm.com). Whole transcriptome amplification products were converted to Illumina sequencing libraries using the Nextera XT DNA Sample Preparation Kit (Illumina). Final cDNA libraries were quantified using High Sensitivity DNA Kits (Agilent) and sequenced on a HiSeq 2500 or 3000 (Illumina), using the paired-end 75-bp protocol, as described previously.31 RNA-seq data from this study have been deposited at the National Center for Biotechnology Information Gene Expression Omnibus (accession number GSE99095), and updated with intermediate and result files from the lncRNA analysis.

Aliquots of whole transcriptome amplification products were used for quantitative RT-PCR analysis.

Bioinformatic analysis Total reads were mapped to the reference genome (hg19) with RSubreader and gene-level read counts were calculated using featureCounts.32 Only data from high-quality cells with captured genes were utilized further. The schematic pipeline has been published.31 Aneuploidy was evaluated by three independent methods, including a sliding window analysis of copy number variations, chromosome relative expression value distribution, and analysis of the degree of loss of heterozygosity.

Identification and classification of long noncoding RNAs After filtering computationally for quality,31 single cells were used to define lncRNAs with a pipeline adopted from published methods of identifying high-confidence gene models.13,14,16,17,28 Fastq files of cells from each subject were merged. Reads were mapped to human genome hg19 with Tophat2 and assembled using Cufflinks packages.33 The assembled transcripts from all subjects were merged with Cuffmerge33 before removing genes with <200 nucleotides or containing single exons in order to obtain long transcripts. Assembled genes overlapping with known protein-coding genes were excluded, and we removed those with low expression (FPKM<2) to improve the reliability of the model. We investigated the coding potential of the remaining genes using three independent algorithms: (i) protein database homology with BlastX and Pfam 31.0 (hmmer2.0); (ii) codon potential assessment with CPAT;34 and (iii) presence of long open reading frames >100 amino acids with EMBOSS GetORF.35 Defined lncRNAs were compared with annotated databases from Ensembl, University of California Santa Cruz (UCSC) Genome Browser, and GENCODE:36 overlapping lncRNAs were defined as “annotated lncRNAs” and the others as putative “novel lncRNAs”.

If supported by cap analysis of gene expression (CAGE) data,37 lncRNA transcripts obtained by the same filtering pipeline, but with medium expression levels (FPKM 0.1-2) were also defined to be expressed in human CD34+ cells (Online Supplementary Methods and Results).

Results Identification and characterization of long noncoding RNAs in human CD34+ hematopoietic cells To assess lncRNA expression in human HSPCs, we purified CD34+ cells from the marrow of four healthy donors and five MDS patients. We then analyzed polyadenylated 895

Z. Wu et al.

RNA by scRNA-seq. After filtering, 391 cells from healthy donors and 588 cells from MDS patients were retained for analysis, with over 9.1 billion 75 bp paired-end mapped reads in total and 7.7 million reads per cell on average.

Using a published strategy,31 a total of 10,791 protein-coding genes were captured, 3,777 per cell on average.

To obtain reliable models of lncRNA expression, we followed a de novo transcript assembly pipeline (Figure 1A), in which â&#x20AC;&#x153;high-confidenceâ&#x20AC;? transcriptomes13,14,16,17,28 from CD34+ single cells of all nine subjects were merged in order to undergo multi-step filtering for: (i) overlap with known mRNA exon annotations, (ii) size and multiexonic selection, (iii) known protein domains, (iv) low levels of expression, and (v) predicted coding potential. Using this conservative multilayered analysis, we identified a total of 2,892 lncRNAs across 979 single human CD34+ cells. To assign lncRNAs to specific classes, we examined their overlap with annotated noncoding genes present in public databases: 808 lncRNAs were previously annotated and

2,084 were putative novel lncRNAs (Figure 1B and Online Supplementary File 1). In addition, transcripts that were expressed at medium levels and supported by CAGE data37 were also defined to be lncRNAs (n=281) expressed in human CD34+ cells (Online Supplementary File 2).

Defined lncRNAs exhibited similarly low protein-coding potential (relative to protein-coding genes) as had previously annotated lncRNAs in the GENCODE database (Figure 1C). Such defined lncRNAs in single human CD34+ cells were distributed across all chromosomes, at much lower average abundance than were protein-coding transcripts. Compared with protein-coding genes, lncRNAencoding genes had fewer exons, were shorter and less well conserved. In general, lncRNA-encoding genes were enriched in 4-kb regions around the transcriptional start sites of their neighboring protein-coding genes, in agreement with previous work,38 suggesting that they share promoter regions [lncRNA-encoding genes show higher co-expression with protein-coding neighbors than do pro-

A

B

896

C

Figure 1. Identification of long noncoding RNAs expressed in single human CD34+ cells. (A) Bioinformatics pipeline for identification of long noncoding RNAs (lncRNAs). Single cell RNA-sequencing (scRNAseq) data from nine subjects were processed and filtered before further analysis of messenger RNA (mRNA) and lncRNA expression. mRNA transcriptome analysis including cell clustering, cell type assignment, and identification of monosomy 7 cells was described32 and employed to analyze gene expression patterns among cell types, functional imputation of lncRNAs, and differentiation trajectory analysis in the current study. scRNA-seq data were processed by de novo genome-based transcriptome reconstruction for the quantification of lncRNAs expressed in human CD34+ cells through the multi-step filtering bioinformatic pipeline. Numbers of remaining transcripts after each filtering step are indicated. (B) By comparing defined lncRNA transcripts in de novo transcript assembly with transcripts in the GENCODE database, 808 lncRNAs were previously annotated while 2,084 were classified as potential novel lncRNAs. (C) Comparison of coding potential among previously annotated lnRNAs, novel lncRNAs, and mRNAs. x axis, coding probability calculated with CPAT; y axis, cumulative distribution function (CDF).

haematologica | 2019; 104(5)

Single cell lncRNAs in hematopoiesis

tein-coding gene pairs (see Online Supplementary Results â&#x20AC;&#x153;Characterization of lncRNAs defined in human CD34+ hematopoietic cellsâ&#x20AC;?; Online Supplementary Figure S3)].

Detection of long noncoding RNAs with single cell RNA-sequencing Expression of lncRNAs showed more variation among single cells than did the expression of coding transcripts (Figure 2A). Across all percentiles of gene expression levels, lncRNAs were expressed in smaller proportions of cells than were mRNAs (Figure 2B). Low overall expres-

A

B

sion of lncRNAs in bulk samples was likely partly attributable to limited but high expression of lncRNAs in a minority of cells or in small cell populations. Seven bulk samples of the CD34+ population from the nine individuals studied were sequenced in parallel with single cells. We sought to compare the maximum abundance of mRNAs or lncRNAs versus housekeeping genes in bulk samples and individual cells,28 to quantify the power of gene expression detection by these different technical approaches. mRNAs were detected at a similar ratio to housekeeping genes in both bulk samples and single cells,

C

D

Figure 2. Detection of long noncoding RNAs by single cell RNA sequencing. (A) Variance of long noncoding RNA (lncRNA) and messenger RNA (mRNA) expression among single cells. x axis, Log (TPM+1); y axis, variance. (B) Proportion of CD34+ cells (individual dots) that express individual lncRNAs (blue) and mRNAs (red), separated by expression quantile of the set of all transcripts (lncRNAs and mRNAs combined). x axis, average expression level quantiles; y axis, proportion of cells. (C) Comparison of single cell and bulk tissue maximum expression levels of mRNAs and lncRNAs. Gray, housekeeping genes; green, mRNAs; red, lncRNAs. Projected density plots summarized expression levels of scatter plots along the single-cell (horizontal) and bulk tissue (vertical) axes. Short lines noted alongside the histogram plots represent the difference of the median expression of lncRNAs or mRNAs to the median expression of housekeeping genes in single cell or bulk tissue RNA-seq.

(D) Gene-ontology semantic similarity matrix of protein-coding genes defined by a guilt-by-association approach of lncRNAs in human CD34+ cells. Gene ontology terms involved in a similar functional matrix were adjacent and formed a block with Pearson R values ranging from -1 to 1. Terms noted on the right side depict common biological processes of the block of gene-ontology terms.

haematologica | 2019; 104(5)

897

Z. Wu et al.

but the ratio of maximum expression of lncRNAs relative to housekeeping genes was about 4-fold higher in single cells than in bulk samples. By scRNA-seq, the maximum expression of lncRNAs was similar to that of both mRNAs and housekeeping genes (Figure 2C). Genes with high variance tended to be captured by the single cell analysis rather than by the bulk approach (Online Supplementary Figure S4). Thus, lncRNA expression appeared to be better detected among single cells due to an expression pattern of high cell-to-cell variation and cell-specificity.

We then sought to infer putative functions of defined lncRNAs in hematopoiesis by a comprehensive â&#x20AC;&#x153;guilt by associationâ&#x20AC;? approach (Online Supplementary Methods and Results), correlating expression of lncRNAs with proteincoding genes of known functions.4,15,39-41 Associated protein-coding genes of defined lncRNAs across CD34+ cells were enriched in gene ontology (GO) terms related to myeloid cell differentiation, cell growth, and cellular functions including DNA repair, mRNA splicing, gene expression, and epigenetic regulation (Figure 2D), implicating lncRNAs in the regulation of human hematopoiesis and associated cellular functions.

Stage- and lineage-specific expression of long noncoding RNAs in normal hematopoiesis To obtain a profile of lncRNA expression in normal human hematopoiesis, we assessed lncRNA expression in 391 CD34+ cells from healthy donors. We first studied whether a lncRNA signature separated CD38- and CD38+ cell populations. lncRNAs detected with 20 reads in at least 20 cells were retained, and highly variable lncRNAs were used for stage-specific analysis (Online Supplementary Figure S5A). The method of t-distributed stochastic neighbor embedding (t-SNE) was adopted for non-linear dimension reduction based solely on batch-corrected (by Combat/SVA) lncRNA expression (Online Supplementary Figure S5B). In an unsupervised t-SNE plot, sorted CD38- cells formed a cluster distinct from CD38+ cells, while CD38+ cells were more dispersed (Figure 3A). To determine stage specificity, we performed pair-wise comparison of lncRNA expression in CD38- cells relative to expression in CD38+ cells. lncRNA expression exhibited substantial differences in two stages (Online Supplementary Table S3); heatmaps of differentially expressed mRNAs and lncRNAs of CD38- and CD38+ populations are shown in Figure 3B. We previously assigned single CD34+ cells to a cell type according to their protein-coding transcriptome profiles, based on gene expression data from flow cytometricallysorted cell populations.42 The cell types to which the single cells were assigned included HSC, multilymphoid progenitor (MLP), megakaryocyte-erythroid progenitor (MEP), granulocyte-monocyte progenitor (GMP), pro-B cell (ProB), and earliest thymic progenitor (ETP).31 We applied weighted gene co-expression network analysis43 to assess the potential functions of lncRNAs in CD38+ and CD38- cells. When protein-coding and lncRNA-encoding genes were simultaneously analyzed, they clustered into seven unsupervised modules (Online Supplementary Table S4), and genes in individual modules were analyzed for GO term enrichment (Figure 3C). Genes in module 1 showed high enrichment of lymphocyte activation pathway genes, and their expression levels were higher in ProB and ETP than in other cell types. Genes in module 6 were enriched in the heme metabolic process, and they showed higher expression in MEP. These data suggest 898

roles of lncRNAs in hematopoiesis and lineage specificity of lncRNA expression.

By t-SNE, cells tended to cluster according to cell types (Figure 4A, right) and were coincident with the pattern of hematopoietic differentiation based on mRNA expression in pseudotime ordering (Figure 4A, middle).31 Thus lncRNAs appeared as powerful as their protein-coding counterparts in resolving subtypes of CD34+ cells. We then analyzed cell-type specificity of gene expression by celltype variance (Figure 4B) and assessed a Jensen-Shannon score8 (JScore) (Figure 4C). lncRNA expression showed higher cell-type specificity than did mRNA expression (JScore, P=1x10-16). There was more cell-to-cell variation in lncRNA expression than in mRNA expression, even within the same cell type (Online Supplementary Figure S6). We investigated our dataset for lncRNA signatures in various lineages, using difference in expression in a lineage, relative to expression in all other subsets, by pairwise comparisons, at a threshold P<0.05 (Figure 4E and Online Supplementary Table S5). Heatmaps revealed that MLP had signatures of both mRNAs and lncRNAs similar to those of HSC, in contrast to distinctive gene expression patterns in other lineages. These data were congruent with those of earlier studies,31,42,44 and indicated that HSC and MLP defined by a transcriptome signature were enriched in a phenotypically characterized CD34+CD38- population, while the other lineages comprised the more heterogeneous CD34+CD38+ population. We examined overlap of lncRNA and mRNA expression among lineages: 94.8% of mRNAs were shared by at least five out of six lineages, but only 62.2% of lncRNAs were so widely expressed (Figure 4D, top panel); conversely, 81.4% of lineage-signature mRNAs were specific to only one lineage, while 92.2% of lncRNAs were equivalently specific (Figure 4D, bottom panel). Again, lncRNA expression appeared more lineage-restricted than did the counterpart coding gene expression. In summary, we found lncRNA expression to be highly stage- and lineage-specific during early hematopoiesis.

To confirm our findings of potential novel lncRNAs and lineage-specific expression patterns of lncRNAs, we compared our results with a publicly available dataset.44 This scRNA-seq study was conducted with human HSPCs sorted based on cell surface antigens (GSE75478). Lineage-specific lncRNAs (and mRNAs) defined in the current study were also detected and showed consistent lineage-specific expression in the two datasets (Online Supplementary Results and Online Supplementary Figures S7 and S8). We then assessed 39 lncRNAs and 14 mRNAs by quantitative RT-PCR of aliquots of whole transcriptome amplification from those 391 single CD34+ cells and another set of flow cytometry-sorted bulk samples (Online Supplementary Methods and Results; Online Supplementary Table S6). All 39 signature lncRNAs, including 20 novel lncRNAs, were detectable in single cells and bulk samples by quantitative RT-PCR, indicating expression in human CD34+ cells. We confirmed cell type assignment of single cells by expression of well-recognized mRNAs (Online Supplementary Figure S9C) and confirmed lineage-specific expression for 35 out of 39 lineage signature lncRNAs in single cells.

Moreover, their lineage-specific expression patterns in single cells were reproducible in independent sorted bulk samples (Online Supplementary Figure S9A,B). Expression of these lineage-specific lncRNAs in hematopoietic differentiation, by scRNA-seq and quantitative RT-PCR, is illustrated in Figure 4F. haematologica | 2019; 104(5)

Single cell lncRNAs in hematopoiesis

A

B

C

Figure 3. Long noncoding RNA expression exhibited a cell stage-specific pattern. (A) Single cell RNA sequencing (scRNA-seq) data of 391 cells with merely long noncoding RNA (lncRNA) expression were clustered using t-SNE in the Seurate package to obtain nonlinear dimension reduction and visualization in two dimensions (tSNE1 and t-SNE2). scRNA-seq data of two different cell stages (CD34+CD38- and CD34+CD38+) sorted by FACS were plotted in red and blue, respectively. (B) Heatmaps of messenger RNA (mRNA) (left) and lncRNA (right) expression in CD34+CD38- and CD34+CD38+ cells. (C) Modules of protein-coding and lncRNA-encoding gene expression across single cells identified through weighted gene co-expression network analysis. Gene co-expression modules including both lncRNAs and mRNAs based on expression quantity and seven unsupervised modules are distinguished by colors (top panel); gene ontology (GO) terms for each module of genes identified in the co-expression matrix (middle panel); expression levels of individual modules of genes in different cell types (bottom panel). Detailed information on individual gene modules is presented in Online Supplementary Table S4.

haematologica | 2019; 104(5)

899

Z. Wu et al.

A

B

C

D

E

F

Figure 4. Long noncoding RNA expression exhibited cell lineage-specific patterns. (A) The same t-SNE plot as in Figure 3A, highlighted single cells with a CD38 expression level (left); cell types were assigned to single cells using messenger RNA (mRNA) expression information, followed by differentiation tree reconstruction using a pseudotime ordering method (middle); single cells are colored according to their corresponding cell types, and gray circles indicate clustering of the same cell type (right). (B) Variance of long noncoding RNA (lncRNA) versus. mRNA expression among different lineages. x axis, Log(TPM+1); y axis, variance. (C) JScore to assess lineage specificity of lncRNA and mRNA expression. x axis, JScore; y axis, cumulative distribution function (CDF). (D) Percentages of mRNAs and lncRNAs defined (top) or preferentially expressed (bottom) in various numbers of cell types. HSC: hematopoietic stem cell; MLP: multilymphoid progenitor; MEP: megakaryocyte-erythroid progenitor; GMP: granulocyte-monocyte progenitor, ProB: pro-B cell; ETP: earliest thymic progenitor. (E) Heatmaps of mRNA (left) and lncRNA (right) expression in different lineages. (F) Expression of a group of lineage-specific lncRNAs for HSC/MLP, MEP, ProB, and ETP along the differentiation tree, measured by single cell RNA sequencing (scRNA-seq) (left) and quantitative reverse transcriptase polymerase chain reaction (RT-PCR) analysis (right). Expression (shown as a mean expression level) is presented as a relative quantity in one lineage relative to expression in all the others.

900

haematologica | 2019; 104(5)

Single cell lncRNAs in hematopoiesis

Coordinated activation and suppression of signature messenger RNAs and long noncoding RNAs during hematopoiesis To systematically assess expression of lncRNAs that might be activated or suppressed during hematopoiesis, we focused on dynamic changes of the mRNA and lncRNA transcriptomes along differentiation trajectories defined by pseudotime ordering of HSC/MLP into MEP and GM/L (granulocyte/monocyte/lymphocyte progenitors) (Figure 5 and Online Supplementary Tables S7 and S8).

Sequentially upregulated/downregulated mRNAs and lncRNAs along the two trajectories were analyzed and gene expression was visualized in heatmaps (MEP trajectory in Figure 5A and GM/L trajectory in Figure 5B).

Common downregulated mRNAs in MEP and GM/L trajectories (Figure 5C) were involved in signaling pathways related to stemness, including NRF2, AP-1, ATF-2, CMYB, HIF-1, and IL-6 signaling. Downregulated genes specifically in the MEP differentiation pathway were mostly enriched in T cells and for broad immune response; enrichment in the EPO signaling pathway was observed only among GM/L downregulated genes. Frequently upregulated genes were involved in DNA replication, cell cycle, and cell proliferation; genes specifically upregulated in GM/L were enriched in B- and T-cell signaling and immune response (Figure 5D, right); hemoglobin synthesis and androgen receptors were enriched only among MEP upregulated genes (Figure 5D, left). lncRNA expression along the two differentiation trajectories was synchronously coordinated with lineage-specific coding genes and interrelated in functional pathways of stemness, megakaryocyte/erythrocyte development, and granulocyte/monocyte/lymphocyte development. Collectively, these data suggest the ordered expression of lncRNAs in hematopoietic differentiation and involvement in the regulation of hematopoiesis.

Long noncoding RNAs are bound by lineage-specific transcription factors and might be regulated by epigenetic mechanisms Transcription factors are critical in cell fate decisions and thus in the regulation of lineage-specific gene expression.

Given the observation of highly ordered expression patterns of lncRNAs during hematopoiesis and co-expression with lineage-specific transcription factors, we investigated roles of lineage-specific transcription factors in regulating lncRNA expression during hematopoiesis. The transcription factor GATA1 regulates erythrocyte and megakaryocyte differentiation,45,46 and indeed its expression was sequentially increased as HSC differentiate into MEP (Figure 5A). Using data obtained by chromatin immunoprecipitation sequencing (ChIP-seq) for GATA1 binding (Encode Ref# ENCSR000EFT), we found that GATA1 binding to promoters was higher in lncRNA-encoding (Figure 6A, top) as well as protein-coding genes (Figure 6A, bottom) preferentially expressed in MEP than for other cell types. lncRNA-encoding genes preferentially expressed in MEP, such as SNHG3 and RP11-620J15.3 (Figure 6B), bound to GATA1 and had high read coverage of active histone marks (H3K27Ac, H3K79me2, and H3K4me2) and low coverage of repressive histone marks (H3K27me3) in erythroid cells. Our analysis, together with published data,8,13,14,16,18,39,47 indicated that cell fate decisions were controlled by critical lineage-specific transcription factors, as evidenced by expression of both lineagehaematologica | 2019; 104(5)

specific mRNAs and lncRNAs bound and regulated by corresponding transcription factors, probably involving epigenetic modification.

Long noncoding RNAs exhibit aberrant expression in aneuploid cells from patients with myelodysplastic syndromes Gene expression of 588 single CD34+ cells from five MDS patients was compared with that of cells from four healthy donors. lncRNAs were differentially expressed in MDS cells compared with those from healthy donors (P<0.05): 372 and 590 lncRNAs were upregulated and downregulated, respectively (Figure 7A and Online Supplementary Table S10). By guilt-by-association, downregulated lncRNAs were associated with gene sets involved in immune response, cellular response, and gene expression and DNA damage response; upregulated lncRNAs were involved in cell metabolism and cell signaling (Figure 7B,C).

We adopted three bioinformatics methods to distinguish cells with abnormal karyotypes from diploid cells.31 We observed that 200 and 56 lncRNAs were downregulated and upregulated, respectively, in monosomy 7 cells, compared to diploid cells (P<0.05) (Figure 7D and Online Supplementary Table S11). By guilt-by-association, downregulated lncRNAs were associated with genes involved in immune response, cell apoptosis and cell death, and DNA modification; upregulated lncRNAs displayed involvement in Ras signaling, Wnt signaling, and interleukin-8 production (Figure 7E,F).

Discussion In the current study, we profiled the repertoire of lncRNAs in human bone marrow-derived CD34+ cells, with the goal of understanding lncRNA biology in early human hematopoiesis. The majority of the human genome is transcribed but only a small proportion of transcripts encode proteins,4,48,49 and thus the number of lncRNA genes is predicted to be very large. Deep RNA sequencing followed by de novo transcriptome reconstruction was adopted for genome-wide annotation and functional characterization of novel lncRNAs.12-14,16-18 Moreover, by scRNA-seq, we and others observed higher cell-to-cell variation of lncRNA expression compared to mRNA expression.26,28,30,50 The validation of defined lncRNAs, including potential novel ones, with quantitative RT-PCR in single cells and a new set of sorted bulk samples proved the validity of scRNA-seq and bioinformatic analysis in defining lncRNAs in the current study. Our strategy of single cell deep sequencing in combination with de novo transcript assembly could be adopted to further facilitate annotation of the complete lncRNA repertoire.

The very large number of both annotated and novel lncRNAs presents a challenge to functional validation.

Based on earlier studies,4,15,39-41 we adopted a systematic, computational guilt-by-association method, from which we could confirm defined lncRNAs in human HSPCs to be likely involved in hematopoietic differentiation and anticipated cell functions. Conventional functional validation of the many hundreds of known and new lncRNAs would not only be prohibitively costly and time-consuming, but the choice of assays and conditions of testing is not obvious, nor is there an established statistic by which to judge 901

Z. Wu et al.

A

B

C

D

Figure 5. Dynamically expressed long noncoding RNAs in differentiation. Expression of sequentially upregulated/downregulated messenger RNAs (mRNAs) (left) and long noncoding RNAs (lncRNAs) (right) from HSC to MEP (A), and to GMP/ProB/ETP (B). MEP downregulated genes (red), MEP upregulated genes (pink), GM/L downregulated genes (orange), and GM/L upregulated genes (blue). (C) A network of commonly downregulated mRNAs and lncRNAs in NRF2, IL-6, HIF1, ATF2, and AP1 signaling pathways. (D) A network of mRNAs and lncRNAs specifically upregulated in MEP in hemoglobin synthesis and androgen signaling pathways (left); and a network of mRNAs and lncRNAs specifically upregulated in GM/L in B-cell, T-cell, and integrin signaling pathways (right). HSC: hematopoietic stem cell; MLP: multilymphoid progenitor; MEP: megakaryocyte-erythroid progenitor; GMP: granulocyte-monocyte progenitor, ProB: pro-B cell; ETP: earliest thymic progenitor; GM/L: granulocyte/monocyte/lymphocyte progenitor.

902

haematologica | 2019; 104(5)

Single cell lncRNAs in hematopoiesis

A

B

Figure 6. Transcription factor occupancy and epigenetic modification of long noncoding RNAs during hematopoietic differentiation. (A) Cumulative distribution of genes [long noncoding RNA (lncRNA)-encoding, up; protein-coding, down] with or without GATA1 binding at promoters. x axis, log10(P value), indicating the significance of gene expression in MEP versus non-MEP cells; y axis, a cumulative distribution function (CDF) of lncRNAs (%) or messenger RNAs (mRNAs) (%). For both lncRNAs and mRNAs, the lower log10(P value), which means the higher significance of preferential gene expression in MEP cells versus non-MEP cells, indicated the higher GATA1 binding CDF. (B) Distribution of single cell RNA sequencing (scRNA-seq) reads across two MEP-specific lncRNAs (SHG3 and RP11-620J15.3) in MEP and other cell types, and the histone modification marks in the same region. Top tracks are images from the IGV Browser depicting scRNA-seq signals as the density of mapped scRNA-seq reads, and chromatin immunoprecipitation sequencing (ChIP-seq) signals as the density of processed signal enrichment of GATA1. Track 2 shows a lncRNA transcript model. Tracks 3 to 7 represent scRNA-seq signals of two MEP-specific lncRNAs (SHG3 and RP11-620J15.3) in two single cells of each cell type including MEP and others. Tracks 8 to 11 depict the ChIP-sequ signal for active histone modification marks (H3K79me2, H3K27Ac, and H3K4me2) and repressive histone modification mark H3K27me3 in a human erythroleukemia cell line, K562. MEP: megakaryocyte-erythroid progenitor; ETP: earliest thymic progenitor; GMP: granulocyte-monocyte progenitor, ProB: pro-B cell; MLP: multilymphoid progenitor.

correlation. We attempted to computationally distinguish lncRNA roles as primary and possibly regulatory from secondary and “epiphenomonal”. To this end, we first determined whether lncRNAs were preferentially expressed in specific cell types; if so, their functions were postulated to relate to lineage-specific protein-coding genes. We then applied pseudotime ordering to reconstruct hematopoietic differentiation in order to examine dynamic gene expression. HSCs are assumed to lose “stemness” and to progressively gain restricted lineage commitment gene expression during differentiation. Indeed, we observed repression of stemness genes and activation of the cell proliferation/metabolism gene program, accompanied by activation of specific-lineage genes and repression of alternative pathway of differentiation genes. By this analysis, we defined lncRNAs that are coordinately expressed in those gene modules and thus have a greater probability of regulatory roles in lineage specification. Our data should assist in narrowing the scope of future efforts including in vitro perturbation and in vivo experiments to study functions of individual lncRNAs in hematopoiesis.

The highly ordered expression pattern of lncRNAs during hematopoiesis implies regulatory constraint. Our analysis and earlier studies8,39,47 indicated that lncRNAs are likely regulated by cell-type specific transcription factors.13,14,16 The observation that lncRNAs exhibited higher expression variability than did mRNAs in the same regulatory program suggests more diverse and active expreshaematologica | 2019; 104(5)

sion of lncRNAs. lncRNAs exert regulatory roles transcriptionally and post-transcriptionally by a variety of mechanisms.1-6 These features of lncRNAs would make them more dynamic participants in cell states and biological processes, facilitating prompt adaptive responses to stimuli or perturbations, and add another layer of complexity in gene expression regulation and cell fate decision.

Our data indicated considerable stage- and lineagespecificity of lncRNAs in human HSPCs and potential engagement in early priming of cell fate, consistent with tissue- and cell type-specificity observed in previous studies.5,7-9, 13-18 This conclusion was confirmed by extension to an external independent scRNA-seq study of 1,034 sorted single human HSPCs,45 and the reproducible lineage-specificity of 35 lncRNAs in both single cells and sorted bulk samples by quantitative RT-PCR. lncRNAs often form secondary structures and there are sensitive, rapid, low-cost methods readily available for lncRNA quantification, all of which make lncRNAs promising biomarkers for disease detection, diagnosis, and prognosis. One study based on microarray assay of bone marrow mononuclear cells from 176 adult patients with MDS established a four-lncRNA risk-scoring system that correlated with distinctive clinical features, and was an independent prognostic factor for survival and leukemia transformation.51 We also found lncRNAs to be dysregulated in MDS cells, but due to the limited number of patients, lncRNA signatures of MDS patients in the current study should be interpreted with 903

Z. Wu et al.

A

B

C

D

E

Figure 7. Long noncoding RNAs are differentially expressed in myelodysplastic syndrome cells and aneuploid cells. (A) A heatmap of long noncoding RNAs (lncRNAs) differentially expressed in myelodysplastic syndrome (MDS) and healthy cells. (B) Pathway analysis of downregulated and upregulated lncRNAs. (C) A network of downregulated lncRNAs with associated messenger RNAs (mRNAs) in different pathways. (D) A heatmap of lncRNAs differentially expressed in aneuploid cells compared with diploid cells. (E) Pathway analysis of downregulated and upregulated lncRNAs in aneuploid cells. (F) A network of downregulated lncRNAs with associated mRNAs in immune-related and DNA damage response pathways. Mono7: monosomy 7.

904

haematologica | 2019; 104(5)

Single cell lncRNAs in hematopoiesis

caution. Nevertheless, our results were in agreement with reported microarray data from 183 MDS patients, which related abnormal lncRNAs with gene expression, cancer, and malignancy.52 Also, differentially expressed lncRNAs in monosomy 7 cells were involved in similar pathways as their mRNA counterparts in our previous study.31 Our results are not a complete profile of lncRNAs due to several limitations, especially the use of only polyAenriched RNAs,8 and the limited cell numbers from a few individuals due to the high cost of scRNA-seq.

Additionally, annotation of novel lncRNAs is context dependent. We adopted commonly used pipelines,12-14,16-18 but annotation might vary using different algorithms.

Nevertheless, our work creates a model for future profiling of the repertoire of lncRNAs in other cell types.

Lineage signatures of lncRNAs are comparison-based, and thus may vary when such comparisons are made among different subsets. Others have categorized HSCs versus cells of specific lineages and among differentiated cells or distinct subsets.12-18 In contrast, we defined lncRNA signatures by making comparisons among subsets within a relatively homogeneous HSPC population, which may compromise our power to detect differences. Furthermore, pseudotime ordering reconstructs the hematopoietic hierarchy based on bioinformatic analysis of transcriptome similarity, and it has demonstrated high agreement with purified cell compartments;44 however, dynamic gene expression in hematopoiesis might be preferably assessed in purified cell populations obtained after physical sorting based on membrane proteins, including after induction of

References 1. Alvarez-Dominguez JR, Lodish HF. Emerging mechanisms of long noncoding RNA function during normal and malignant hematopoiesis. Blood. 2017;130(18):19651975.

2. Satpathy AT, Chang HY. Long noncoding RNA in hematopoiesis and immunity.

Immunity. 2015;42(5):792-804.

3. Engreitz JM, Ollikainen N, Guttman M. Long non-coding RNAs: spatial amplifiers that control nuclear structure and gene expression. Nat Rev Mol Cell Biol.

2016;17(12):756-770.

4. Derrien T, Johnson R, Bussotti G, et al. The GENCODE v7 catalog of human long noncoding RNAs: analysis of their gene structure, evolution, and expression. Genome Res. 2012;22(9):1775-1789.

5. Cabili MN, Dunagin MC, McClanahan PD, et al. Localization and abundance analysis of human lncRNAs at single-cell and singlemolecule resolution. Genome Biol. 2015; 16:20.

6. Djebali S, Davis CA, Merkel A, et al.

Landscape of transcription in human cells.

Nature. 2012;489(7414):101-108.

7. Ulitsky I, Bartel DP. lincRNAs: genomics, evolution, and mechanisms. Cell. 2013;154 (1):26-46.

8. Cabili MN, Trapnell C, Goff L, et al.

Integrative annotation of human large intergenic noncoding RNAs reveals global properties and specific subclasses. Genes Dev. 2011;25(18):1915-1927.

9. Washietl S, Kellis M, Garber M.

haematologica | 2019; 104(5)

10.

11.

12.

13.

14.

15.

16.

17.

differentiation or other in vitro perturbations. Given the high cell-type specificity of lncRNAs, signature lncRNAs may be superior to mRNAs in discriminating and differentiating cell subsets or new cell types that cannot be easily distinguished based on cell surface markers. We did not compare the efficacy of lncRNAs and mRNAs in defining cell types due to a lack of detailed surface marker information for single cells. Future studies with larger cell numbers, complete surface marker characterization, and whole transcriptome expression data should be of great interest in defining new cells/subtypes.

Rapid evolution and low species conservation are features of lncRNAs,10,11 making a human catalog a prerequisite to successful, clinically relevant lncRNA studies.

Based on next-generation sequencing and single cell technology, we provide a global database that should be foundational for future studies of lncRNA biology in human HSPCs. Acknowledgments The authors acknowledge the support of the Trans-NIH Center for Human Immunology, Autoimmunity, and Inflammation (National Institute of Allergy and Infectious Diseases, NIH, Bethesda, MD, USA). We thank patients and healthy volunteers who donated bone marrow. Sequencing and technical support were provided by the DNA Sequencing and Genomics Core of NHLBI. FACS sorting was performed by Keyvan Keyvanfar and the Flow Cytometry Core of NHLBI. This research was supported by an Intramural Research Program of the National Heart, Lung, and Blood Institute.

Evolutionary dynamics and tissue specificity of human long noncoding RNAs in six mammals. Genome Res. 2014;24(4):616628.

Pang KC, Frith MC, Mattick JS. Rapid evolution of noncoding RNAs: lack of conservation does not mean lack of function. Trends Genet. 2006;22(1):1-5.

Wang J, Zhang J, Zheng H, et al. Mouse transcriptome: neutral evolution of 'non-coding' complementary DNAs. Nature. 2004;431 (7010):1 p following 757; discussion following 757.

Luo M, Jeong M, Sun D, et al. Long non-coding RNAs control hematopoietic stem cell function. Cell Stem Cell. 2015;16(4):426438.

Alvarez-Dominguez JR, Hu W, Yuan B, et al.

Global discovery of erythroid long noncoding RNAs reveals novel regulators of red cell maturation. Blood. 2014;123(4):570-581.

Paralkar VR, Mishra T, Luan J, et al. Lineage and species-specific long noncoding RNAs during erythro-megakaryocytic development. Blood. 2014;123(12):1927-1937.

Schwarzer A, Emmrich S, Schmidt F, et al.

The non-coding RNA landscape of human hematopoiesis and leukemia. Nat Commun.

2017;8(1):218.

Hu G, Tang Q, Sharma S, et al. Expression and regulation of intergenic long noncoding RNAs during T cell development and differentiation. Nat Immunol. 2013;14(11):11901198.

Ranzani V, Rossetti G, Panzeri I, et al. The long intergenic noncoding RNA landscape of human lymphocytes highlights the regulation of T cell differentiation by linc-MAF-4.

Nat Immunol. 2015;16(3):318-325.

18. BrazĂŁo TF, Johnson JS, MĂźller J, et al. Long noncoding RNAs in B-cell development and activation. Blood. 2016;128(7):e10-19.

19. Collier SP, Collins PL, Williams CL, et al.

Cutting edge: influence of Tmevpg1, a long intergenic noncoding RNA, on the expression of Ifng by Th1 cells. J Immunol.

2012;189(5):2084-2088.

20. Vigneau S, Rohrlich PS, Brahic M, et al.

Tmevpg1, a candidate gene for the control of Theiler's virus persistence, could be implicated in the regulation of gamma interferon.

J Virol. 2003;77(10):5632-5638.

21. Iyer MK, Niknafs YS, Malik R, et al. The landscape of long noncoding RNAs in the human transcriptome. Nat Genet. 2015;47 (3):199-208.

22. Lei L, Xia S, Liu D, Li X, et al. Genome-wide characterization of lncRNAs in acute myeloid leukemia. Brief Bioinform. 2018.

19(4):627-635.

23. Heward JA, Lindsay MA. Long non-coding RNAs in the regulation of the immune response. Trends Immunol. 2014;35(9):408419.

24. Ulitsky I, Bartel DP. lincRNAs: genomics, evolution, and mechanisms. Cell. 2013;154 (1):26-46.

25. Cabili MN, Trapnell C, Goff L, et al.

Integrative annotation of human large intergenic noncoding RNAs reveals global properties and specific subclasses. Genes Dev. 2011;25(18):1915-1927.

26. Wang J, Roy B. Single-cell RNA-seq reveals lincRNA expression differences in Hela-S3 cells. Biotechnol Lett. 2017;39(3):359-366.

27. Kim DH, Marinov GK, Pepke S, et al. Single-

905

Z. Wu et al.

28.

29.

30.

31.

32.

33.

34.

35.

906

cell transcriptome analysis reveals dynamic changes in lncRNA expression during reprogramming. Cell Stem Cell. 2015;16(1):88-101.

Liu SJ, Nowakowski TJ, Pollen AA, et al.

Single-cell analysis of long non-coding RNAs in the developing human neocortex.

Genome Biol. 2016;17:67.

Gawronski KAB, Kim J. Single cell transcriptomics of noncoding RNAs and their cellspecificity. Wiley Interdiscip Rev RNA. 2017;8(6).

Hu W, Wang T, Yang Y, et al. Tumor heterogeneity uncovered by dynamic expression of long noncoding RNA at single-cell resolution. Cancer Genet. 2015;208(12):581-586.

Zhao X, Gao S, Wu Z, et al. Single-cell RNAseq reveals a distinct transcriptome signature of aneuploid hematopoietic cells. Blood.

2017;130(25):2762-2773.

Liao Y, Smyth GK, Shi W, et al. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features.

Bioinformatics.

2014;30(7):923-930.

Trapnell C, Roberts A, Goff L, et al.

Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc.

2012;7(3):562-578.

Wang L, Park HJ, Dasari S, et al. CPAT: Coding-Potential Assessment Tool using an alignment-free logistic regression model.

Nucleic Acids Res. 2013;41(6):e74.

Rice P, Longden I, Bleasby A. et al. EMBOSS: the European Molecular Biology Open Software Suite. Trends Genet. 2000;16(6): 276-277.

36. Harrow J, Frankish A, Gonzalez JM, et al.

GENCODE: the reference human genome annotation for The ENCODE Project.

Genome Res. 2012;22(9):1760-1774.

37. Hon CC, Ramilowski JA, Harshbarger J, et al. An atlas of human long non-coding RNAs with accurate 5' ends. Nature. 2017;543:199204.

38. Zhang K, Huang K, Luo Y, et al.

Identification and functional analysis of long non-coding RNAs in mouse cleavage stage embryonic development based on single cell transcriptome data. BMC Genomics.

2014;15:845.

39. Guttman M, Amit I, Garber M, et al.

Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals. Nature. 2009;458(7235): 223-227.

40. Huarte M, Guttman M, Feldser D, et al. A large intergenic noncoding RNA induced by p53 mediates global gene repression in the p53 response. Cell. 2010;142(3):409-419.

41. Yan X, Hu Z, Feng Y, et al. Comprehensive genomic characterization of long non-coding RNAs across human cancers. Cancer Cell. 2015;28(4):529-540.

42. Laurenti E, Doulatov S, Zandi S, et al. The transcriptional architecture of early human hematopoiesis identifies multilevel control of lymphoid commitment. Nat Immunol.

2013;14(7):756-763.

43. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.

44. Velten L, Haas SF, Raffel S, et al. Human haematopoietic stem cell lineage commit-

45.

46.

47.

48.

49.

50.

51.

52.

ment is a continuous process. Nat Cell Biol.

2017;19(4):271-281.

Stachura DL, Chou ST, Weiss MJ. Early block to erythromegakaryocytic development conferred by loss of transcription factor GATA-1. Blood. 2006;107(1):87-97.

Shivdasani RA, Fujiwara Y, McDevitt MA, et al. A lineage-selective knockout establishes the critical role of transcription factor GATA-1 in megakaryocyte growth and platelet development. EMBO J. 1997;16 (13):3965-3973.

Guttman M, Donaghey J, Carey BW, et al. lincRNAs act in the circuitry controlling pluripotency and differentiation. Nature.

2011;477(7364):295-300.

Ezkurdia I, Juan D, Rodriguez JM, et al.

Multiple evidence strands suggest that there may be as few as 19,000 human proteincoding genes. Hum Mol Genet. 2014;23(22): 5866-5878.

Mattick JS, Rinn JL. Discovery and annotation of long noncoding RNAs. Nat Struct Mol Biol. 2015;22(1):5-7.

Zhou F, Li X, Wang W, et al. Tracing haematopietic stem cell formation at singlecell resolution. Nature. 2016;533(7604):487492.

Yao CY, Chen CH, Huang HH, et al. A 4lncRNA scoring system for prognostication of adult myelodysplastic syndromes. Blood Adv. 2017;1(19):1505-1516.

Liu K, Beck D, Thoms JAI, et al. Annotating function to differentially expressed LincRNAs in myelodysplastic syndrome using a network-based method.

Bioinformatics. 2017;33(17):2622-2630.

haematologica | 2019; 104(5)

ARTICLE

Red Cell Biology & its Disorders

Finely-tuned regulation of AMP-activated protein kinase is crucial for human adult erythropoiesis

Ferrata Storti Foundation

Meriem Ladli,1,2,3,4 Cyrielle Richard1,2,3,4, Lilia Cantero Aguilar,1,2,3,4 Sarah Ducamp,1,2,3,4 Sabrina Bondu,1,2,3,4 Pierre Sujobert,1,2,3 Jérôme Tamburini,1,2,3 Catherine Lacombe,1,2,3,4 Nabih Azar,5 Marc Foretz,1,2,3,4 Yael Zermati,1,2,3,4 Patrick Mayeux,1,2,3,4 Benoit Viollet1,2,3,4 and Frédérique Verdier1,2,3,4

Institut Cochin, INSERM U1016; 2CNRS UMR 8104, Paris; 3Université Paris Descartes, Sorbonne Paris Cité; 4Labex GREX and 5Service d’Hémobiologie, Hôpital La Pitié Salpétrière, Paris, France 1

Haematologica 2019 Volume 104(5):907-918

ABSTRACT

A

MP-activated protein kinase (AMPK) is a heterotrimeric complex containing a, β, and γ subunits involved in maintaining integrity and survival of murine red blood cells. Indeed, Ampk a1-/-, Ampk -/β1 and Ampk γ1-/- mice develop hemolytic anemia and the plasma membrane of their red blood cells shows elasticity defects. The membrane composition evolves continuously along erythropoiesis and during red blood cell maturation; defects due to the absence of Ampk could be initiated during erythropoiesis. We, therefore, studied the role of AMPK during human erythropoiesis. Our data show that AMPK activation had two distinct phases in primary erythroblasts. The phosphorylation of AMPK (Thr172) and its target acetyl CoA carboxylase (Ser79) was elevated in immature erythroblasts (glycophorin Alow), then decreased conjointly with erythroid differentiation. In erythroblasts, knockdown of the a1 catalytic subunit by short hairpin RNA led to a decrease in cell proliferation and alterations in the expression of membrane proteins (band 3 and glycophorin A) associated with an increase in phosphorylation of adducin (Ser726). AMPK activation in mature erythroblasts (glycophorin Ahigh), achieved through the use of direct activators (GSK621 and compound 991), induced cell cycle arrest in the S phase, the induction of autophagy and caspase-dependent apoptosis, whereas no such effects were observed in similarly treated immature erythroblasts. Thus, our work suggests that AMPK activation during the final stages of erythropoiesis is deleterious. As the use of direct AMPK activators is being considered as a treatment in several pathologies (diabetes, acute myeloid leukemia), this observation is pivotal. Our data highlighted the importance of the finely-tuned regulation of AMPK during human erythropoiesis.

Introduction Mammalian AMP-activated protein kinase (AMPK) is a highly conserved eukaryotic serine/threonine protein kinase and a heterotrimeric complex consisting of a single catalytic (a) and two regulatory (β and γ) subunits, encoded by different genes (a1, a2, β1, β2, γ1, γ2, and γ3). In the case of energy depletion, a decrease in the cellular ATP-to-AMP ratio leads to allosteric AMPK activation by AMP but also by the phosphorylation of Thr172 within the activation loop segment of the a subunit by an upstream AMPK kinase, liver kinase B1 (LKB1).

Another ‘‘canonical’’ mechanism of activation involves the phosphorylation of Thr172 by calcium/calmodulin-dependent kinase kinase β (CaMKKβ) in response to a rise in intracellular Ca2+.1 Once activated, AMPK phosphorylates metabolic targets, leading to a decrease in ATP consumption and an increase in ATP production. In particular, AMPK inhibits fatty acid synthesis via phosphorylation and inactivation of acetyl-CoA-carboxylase (ACC) or induces autophagy via the phosphorylation of Unc-51 like autophagy activating kinase 1 (ULK1).2 Thus, AMPK is haematologica | 2019; 104(5)

Correspondence: FRÉDÉRIQUE VERDIER frederique.verdier@inserm.fr Received: February 18, 2018.

Accepted: October 3, 2018.

Pre-published: October 11, 2018. doi:10.3324/haematol.2018.191403 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/907 ©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.

907

M. Ladli et al.

a major sensor of energy status that maintains cellular energy homeostasis but also exerts non-metabolic functions such as the maintenance of cell survival, cell polarity and regulation of the cell cycle.3,4 Erythropoiesis is a tightly regulated process that permits the production of around two million red cells each second throughout a human life, while the total cell number has to be kept within a narrow margin. This extremely dynamic process is also very flexible, since it must increase rapidly in response to blood loss and hypoxia. Furthermore, maintaining homeostasis is crucial and an imbalance in erythropoiesis can lead to the development of erythroid pathologies such as polycythemias and anemia.

We and other groups have previously demonstrated that AMPK plays a crucial role in the integrity and survival of red blood cells. We showed that mice that are globally deficient in the catalytic subunit, Ampka1 but not in those lacking the isoform Ampka2, as well as those globally deficient in the regulatory subunits Ampkβ1 and Ampkγ1, develop regenerative hemolytic anemia caused by increased sequestration of abnormal erythrocytes.

Ampka1-/-, Ampkβ1-/- and Ampkγ1-/- mice develop splenomegaly and iron accumulation due to a compensatory response through extramedullary erythropoiesis in the spleen and enhanced erythrophagocytosis. The lifespan of erythrocytes from Ampka1-/- and Ampkγ1-/- mice was shorter than that of wild-type littermates. Moreover, Ampka1-/- and Ampkγ1-/- erythrocytes were highly resistant to osmotic stress and poorly deformable in response to increasing shear stress, which is consistent with a loss of membrane elasticity.5-8 The defects in Ampk-deficient erythrocytes suggested that alterations might occur early during terminal erythroid maturation but no data were available on the importance of AMPK in human erythropoiesis. We, therefore, decided to investigate whether AMPK could be implicated in regulating the proliferation, survival and differentiation of human erythroid precursors.

In the present study, we analyzed the expression and activation of AMPK along human erythroid differentiation. Our experiments show that AMPK is highly activated in immature erythroblasts and weakly active in mature erythroblasts. We studied the impact of knocking down AMPK and of AMPK activation by direct activators. In erythroblasts, the knockdown of the AMPK α1 catalytic subunit expression by short hairpin (sh) RNA induced a decrease in cell proliferation and alterations in the expression or phosphorylation of membrane proteins whereas no defect in hemoglobin synthesis or erythroid maturation was observed. The activation of AMPK is necessary in immature erythroblasts but maintaining the activation in mature erythroblasts is deleterious, demonstrating that AMPK activation has to be tightly regulated during human terminal erythroid differentiation.

adducin antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Actin (A5441) antibodies, dexamethasone and chloroquine were purchased from SigmaAldrich (Lyon, France). Anti-ankyrin antibodies were obtained from Neuromab (Davis, CA, USA) (#75-380) and anti β-spectrin antibodies from Abcam (Cambridge, UK). Compound GSK-621 was purchased from Selleckchem (Houston, TX, USA) and 991 (5-[[6-chloro-5-(1-methylindol-5-yl)-1H-benzimidazol-2yl]oxy]-2-methyl-benzoic acid) was synthesized by Spirochem (Basel, Switzerland).

Cell lines and cell culture CD34+ cells were obtained from human donors who gave informed consent in accordance with the Declaration of Helsinki and the study was approved by the French ministry of higher education and research review board. Granulocyte colony-stimulating factor-mobilized CD34+ cells were purified from peripheral blood after cytapheresis. CD34+ cells were isolated by positive selection using an immunomagnetic procedure (MACS CD34 isolation kit; Miltenyi Biotech (Paris, France). CD34+ cells were cultured in 5% CO2 at 37°C for 7 days in IMDM medium (Life Technologies, Waltham MA, USA) containing 1% glutamine, 15% BIT 9500 (Stem Cell Technologies), 100 ng/mL stem cell factor, 10 ng/mL interleukin-6 and 10 ng/mL interleukin-3 (Miltenyi Biotech). After 7 days of culture, CD36+ cells corresponding to a highly purified population of human erythroid progenitors were obtained by positive selection on CD36 immunomagnetic beads (CD36 unlabeled antibodies purchased from Beckman Coulter, Villepinte, France) coupled to antimouse IgG1 microbeads purchased from Miltenyi Biotech) .

CD36+ cells were then cultured with 2 U/mL erythropoietin, 100 ng/mL stem cell factor and 10 ng/mL interleukin-3 for up to 14 days for erythroid differentiation. GSK621 or compound 991 was added from day 0 after CD36+ selection; cells were counted daily and diluted to a final concentration of 0.8 x106 cells/mL by the addition of fresh medium containing the indicated concentration of AMPK activator. Because of interindividual variability, the kinetics of erythroid differentiation varies between different human samples. Thus, the days of culture corresponding to the same stage of differentiation have been grouped.

Lentiviral constructs, lentiviral production and cell infection Lentiviral constructs for control and AMPKa1 shRNA [(SHC002 and SHCLNG-NM006251 (TRCN00000000859), respectively)] were purchased from Sigma (Lyon, France). To obtain recombinant lentiviruses, 293T cells were transiently transfected by calcium phosphate precipitation with three different plasmids: pCMV-G (VSVG envelope coding sequence), pCMV-gag-pol and a recombinant pLKO.1 vector encoding either a control or AMPKa1 shRNA. Supernatants containing infectious lentiviral particles were concentrated by ultracentrifugation. Infections of human erythroblasts were performed at day 1 and at day 4 after CD36 cell sorting and culture in the presence of interleukin-3, stem cell factor and erythropoietin, as described above.

Methods Flow cytometry Materials

AMPK a1 and a2 antibodies were obtained from Graham Hardie (University of Dundee, UK);9,10 antibodies against the AMPK β1 and γ1 isoforms, phospho-Thr 172 AMPK, phosphoSer 79 ACC, phospho-Ser 555 ULK1, LKB1, LC3B and cleaved caspase 3 were from Cell Signaling Technology (Danvers, USA) and anti-HSC70, anti-a spectrin, anti-band 3 and anti-P-Ser 726 908

Cells were labeled as previously described.11 Briefly, PC7-conjugated anti-glycophorin A (GPA), APC-conjugated anti-cd49d (α4 integrin) or an appropriate isotype control were purchased from Beckman Coulter; anti-BRIC6 (anti-band 3) was from the NHSBT International Blood Group Reference Laboratory (Bristol, UK). FITC-conjugated annexin V was used to measure the percentage of cell apoptosis. haematologica | 2019; 104(5)

Role of AMPK in human erythropoiesis

Cell proliferation Cell proliferation was determined by trypan blue exclusion dye.

Statistics Results are expressed as means ± standard deviation (SD). A Student t test was used to determine statistical significance. P values <0.05 were considered statistically significant.

Results AMPK a1 activation is tightly regulated during human erythroid differentiation Because AMPK occurs as a heterotrimeric complex containing catalytic a subunits and regulatory β and γ subunits, we aimed to identify which isoforms are expressed in human erythroblasts and to study their variation during human erythroid differentiation. Human primary erythroid progenitors were maintained for up to 12 days in culture and were able to differentiate from the pro-erythroblastic stage (day 2) to the reticulocyte stage (day 12) (Figure 1A and Online Supplementary Figure S1 for cell morphology).

During maturation, erythroblasts progressively acquired cell surface-specific markers such as GPA (from the pro-erythroblastic stage), band 3 (from the basophilic-erythroblastic stage) and decreased expression of a4 integrin at the orthochromatic erythroblast stage. They also started to synthesize hemoglobin from the basophilic stage. Western blot analysis demonstrated that the a1 catalytic subunit was expressed while the a2 isoform was not detectable and that the expression of a1 was constant along erythroid differentiation (Figure 1B). The regulatory subunits, β1 and γ1, were expressed throughout erythroid differentiation. We previously determined the copy number of individual proteins for each stage of erythroid differentiation by an absolute quantitative proteomics analysis.11 We confirmed the expression of a1, β1 and γ1, while a2, β2 and γ2 isoforms were not detectable by mass spectrometry analysis (Online Supplementary Figure S2). Overall, our results suggest that a1/β1/γ1 is the heterotrimeric complex that is predominantly present in human erythroblasts.

Despite the global constant expression of AMPK, the activation of this protein might be modulated during differentiation. We, therefore, studied AMPK activation by detection of phosphorylation of the a1 catalytic subunit at Thr172 and phosphorylation of one of its substrates, ACC, at Ser79 (Figure 1C). From day 2 to day 6, the phosphorylation of AMPK and ACC was clearly detectable, but concomitantly decreased from day 8 to the end of differentiation. LKB1 was expressed throughout erythroid differentiation. These data showed the biphasic pattern of activation of AMPK during erythroid differentiation with clear activation in immature erythroblasts from the pro-erythroblast stage until day 6 when the cells were GPAmed/band 3low (basophilic erythroblasts) and reduced activation in mature GPAhigh/band 3med erythroblasts from day 8 to day 12, corresponding to stages from polychromatic erythroblasts to reticulocytes.

The absence of AMPK induces decreased proliferation and alterations in the expression of membrane proteins of human erythroblasts To decipher the role of AMPK in erythroblasts, we inhibited AMPK expression by a specific shRNA targeting the a1 haematologica | 2019; 104(5)

catalytic subunit. Cells were infected by a lentivirus coding for either shAMPKa1 or shControl (shCtrl) at day 1 and then at day 4 after CD36 cell sorting. The decrease in a1 AMPK expression resulted in an expected decrease of the phosphorylation of AMPK and its substrates ACC and ULK1 (Figure 2A). No compensatory expression of the a2 catalytic subunit was observed in response to inhibition of the a1 isoform (Online Supplementary Figure S3). The inhibition of AMPK a1 expression did not significantly modify the differentiation of the cells. Indeed at days 6, 8 and 10 of culture, the cell population was mainly composed of basophilic erythroblasts, polychromatic erythroblasts and orthochromatic erythroblasts, respectively, and this pattern was not different when AMPKa1 was knocked down (Figure 2B). The morphology of the shCtrl and shAMPKa1 cells was very similar (Online Supplementary Figure S4). The percentages of hemoglobinized cells estimated at the indicated stages of differentiation were identical between the shCtrl and shAMPKa1 cells (Figure 2C). shAMPKa1 erythroblasts showed a reduced ability to proliferate compared to shCtrl erythroblasts (Figure 2D). The inhibition of AMPKa1 expression did not significantly affect the viability of the cells measured by the trypan blue exclusion assay (Figure 2E) or by annexin V flow cytometry analysis (data not shown).

Because red cells from Ampka1-/- and Ampkγ1-/- mice are highly resistant to osmotic stress and poorly deformable, the expression of membrane proteins involved in these processes was studied (Figure 3A). In shAMPKa1 polychromatic and orthochromatic erythroblasts, western blot experiments showed that the phosphorylation of adducin on Ser726 was increased while the expression of spectrins and ankyrin was not affected (Figure 3A). Western blots demonstrated that the global expression of band 3 was significantly decreased in the shAMPKa1 cells while its expression at the cell surface, measured by FACS, was abnormally increased (Figure 3B). FACS analyses also showed a decrease of the cell surface expression of GPA at each stage of differentiation for the shAMPKa1 cells in comparison to ShCtrl (39.9% at day 6/basophilic erythroblasts, 50% at day 8/polychromatlic erythroblasts and 58.8% at day 10/orthochromatic erythroblast versus 100% for shCtrl) (Figure 3C). Overall, the decrease in AMPK expression induced major abnormalities in the expression of the membrane proteins band 3 and GPA and, as in murine Ampk knockout mice, led to enhanced phosphorylation of adducin. Furthermore, AMPKα1 knockdown provoked a decrease in cell proliferation without affecting cell viability and erythroblast maturation.

Proliferation and survival of mature GPAhigh erythroblasts are specifically and drastically diminished by GSK621-mediated AMPK activation To further investigate the role of AMPK in erythroid cells, the activation of AMPK was enhanced by GSK621, a direct, potent, novel activator of AMPK.12-15 In primary erythroblasts, the activation of AMPK by GSK621 was dose-dependent, with increased phosphorylation of T172 AMPKa, and also gradual stimulation of the phosphorylation of the substrates of AMPK, ACC at Ser79 and ULK1 at Ser 555, from 5 to 20 mM (Figure 4A). The latter dose was then used in the experiments. GSK621 was added to the medium from day 0 after CD36 cell sorting until the indicated days. GSK621 induced the phosphorylation of AMPK and its substrates at each stage of erythroid maturation (Online Supplementary 909

M. Ladli et al.

A

B

C

Figure 1. Expression of AMPK isoforms and AMPK activation along terminal erythroid differentiation. (A) Representative experiment of one ex vivo culture of erythroblasts derived from CD34+ cells. Cells were analyzed on days 2, 4, 6, 8, 10 and 12 after CD36+ selection. Expression of cell surface markers GPA, band 3 and a4β1 integrin was studied by flow cytometry along terminal erythroid differentiation. The percentage of hemoglobinized cells was determined by benzidine staining.

A minimum of 200 cells were counted and the percentage of blue-stained cells among the total cell count was determined. Cell morphology was examined following staining with May-Grünwald-Giemsa; the percentage of each cell population was determined. (B) AMPK isoforms during erythropoiesis were determined by western blot. Protein extracts from human primary erythroblasts from day 2 to day 12 of culture were analyzed by western blot using anti-a1, -a2, -β1, and -γ1 antibodies.

Anti-β-actin was used as a loading control and mouse liver protein extracts were used as a positive control for the expression of AMPKa2 (Ctrl). The a1, a2, and γ1 isoforms were analyzed on the same blot, the β1 isoform on a different one. (C) AMPK activation during erythroid differentiation. Anti-pT172 AMPK, anti-p S79 ACC, anti AMPKa1 and LKB1 were used. Anti-β-actin or anti-HSC70 antibodies were used as loading controls. The upper panel shows a representative experiment of three independent ones. Quantification of western blots and determination of the ratio between pAMPK/AMPK, pACC/AMPK and LKB1/AMPKa1 at the indicated days are presented as the mean of three independent experiments ± SD; ns: non-significant, *P<0.05 (lower panel). d: day; GPA: glycophorin A; AMPK: AMP-activated protein kinase; ACC: acetyl-CoA-carboxylase; LKB1: liver kinase B1; HSC70: heat shock 70kDa protein; E: erythroblasts.

910

haematologica | 2019; 104(5)

Role of AMPK in human erythropoiesis

A

B

D

C

E

Figure 2. AMPKa1 knockdown reduces erythroblast proliferation without affecting erythroblast maturation. (A) Efficiency of AMPKa1 knockdown: Erythroblasts were infected twice, on days 1 and 4 after CD36 cell sorting, with either a lentivirus coding a scrambled shRNA (sh Ctrl) or an AMPKa1 shRNA (shAMPK a1). The efficiency of AMPKa1 knockdown in human primary erythroblasts was determined by anti-a1 AMPK, anti-PT172 AMPK, anti-PS79 ACC and anti-PS555 ULK1 western blots on day 8 or day 10 with the same results. A representative experiment on day 8 is presented. Western blot quantification was performed with anti-β−actin or anti-HSC70 antibodies; values are the mean of three independent experiments ± SD; ns: non-significant, *P<0.05. (B) Effect of shAMPKa1 on erythroid cell maturation. Cell composition for each culture at days 6, 8 and 10. The mean composition ± SD was calculated based on erythroblast morphology determined by MayGrünwald-Giemsa staining in three independent experiments. (C) Absence of AMPKa1 does not modify the percentage of hemoglobinized cells. The percentage of hemoglobinized cells was determined by benzidine staining at the indicated days of culture and corresponding stage of differentiation determined as in Figure 2A. (D) Proliferation of shCtrl versus shAMPKa1 erythroblasts. Cumulative cell number was determined by counting cells with the trypan blue exclusion method at day 6 (basophilic erythroblasts), day 8 (polychromatic erythroblasts) and day 10 (orthochromatic erythroblasts) of culture. Results are expressed as the mean ± SD of three independent experiments; ns: non-significant, *P<0.05. (E) Cell death in shAMPKa1 versus shCtrl cells. The proportion of dead cells was determined by trypan blue exclusion dye at days 6, 8 and 10 of culture. Results are expressed as the mean ± SD of three independent experiments; ns: non-significant, *P<0.05. d: day; GPA: glycophorin A; AMPK: AMP-activated protein kinase; ACC: acetyl-CoA-carboxylase; ULK1: Unc-51 like autophagy activating kinase 1; HSC70: heat shock 70kDa protein; Pro-E: pro-erythroblasts; Baso-E: basophilic erythroblasts; Poly-E: polychromatic erythroblasts; Ortho-E: orthochromatic erythroblasts.

haematologica | 2019; 104(5)

911

M. Ladli et al.

A

B

C

Figure 3. AMPK a1 knockdown disturbs glycophorin A, band 3 expression and phosphorylation of aduccin. (A) Analysis of expression of membrane proteins. Protein extracts from human primary erythroblasts were analyzed by western blot using anti-AMPKa1 antibodies to confirm the knockdown and anti-spectrin a and β, ankyrin, band 3, p-aduccin S726 antibodies. Anti-HSC70 was used as a loading control. AMPKa1, anti-spectrin a, ankyrin, band 3, p-adducin S726 were analyzed on the same western blot, spectrin β on a distinct one (representative western blot at day 8/polychromatic erythroblasts, left panel), the same result was obtained at day 10/polychromatic erythroblasts. The blots were quantified using anti-HSC70 antibodies; values are the mean ± SD of three independent experiments at day 8 and day 10; ns: non-significant, *P<0.05; (right panel). (B) Representative cytometry profile for band 3/a4 integrin gated on GPA-positive cells (left panel) The percentages of band 3lowa4 integrinhigh and band 3meda4 integrinmed in the shCtrl and shAMPKa1 cell populations are plotted in histograms (right panel). Results are expressed as the mean ± SD of three independent experiments; ns: non-significant, *P<0.05; **P<0.01. (C) Expression of GPA determined by FACS. Representative cytometry profiles are shown for GPA at day 6/basophilic erythroblasts, at day 8/polychromatic erythroblasts and at day 10/orthochromatic erythroblasts. Control isotype: unfilled black for shCtrl, unfilled gray for shAMPKa1; GPA-labeled: filled black for shCtrl, filled gray for shAMPKa1) (left panel). The histogram shows the mean percentage of the GPA fluorescence intensity on the cell surface; 100% represents the fluorescence intensity for shCtrl cells (right panel). Results are expressed as the mean ± SD of three independent experiments; ns: non-significant, *P<0.05; **P<0.01. AMPK: AMP-activated protein kinase; HSC70: heat shock 70kDa protein; d: day; B3: band 3; Baso-E: basophilic erythroblasts; Poly-E: polychromatic erythroblasts; Ortho-E: orthochromatic erythroblasts; GPA: glycophorin A.

912

haematologica | 2019; 104(5)

Role of AMPK in human erythropoiesis

A

C

B

D

Figure 4. GSK621-mediated AMPK activation decreases the proliferation of mature erythroblasts. (A) GSK621 dose-dependent activation of AMPK in erythroblasts. Western blot analysis of PT172 AMPKa1, PS79 ACC and PS555 ULK1 in primary erythroblasts incubated for 3 h with increasing doses of GSK621. AntiHSC70 was used as a loading control. (B) GSK621 induces massive cell death from days 5-7. The percentage of dead cells was determined by trypan blue exclusion dye. In the three experiments, the days of culture were grouped at the same stage of differentiation for the vehicle-treated cells (see Methods section).

(C) Inhibition of erythroblast proliferation by GSK621. Erythroid cells were incubated in the absence (vehicle) or presence of 20 mM GSK621 from day 0.

Cumulative cell number was determined by counting cells with the trypan blue exclusion method at day 0, days 3-4, days 5-7 and days 8-9 in three independent cultures. (D) GSK621 induces an accumulation of cells in the S phase. Erythroid cells were incubated in the absence (vehicle) or presence of 20 mM GSK621 from day 0 to the indicated days. The propidium Iodide incorporation assay was performed. A representative experiment is shown; four independent experiments were performed and the results from day 9 are expressed as the mean Âą SD; ns: non-significant, *P<0.05. AMPK: AMP-activated protein kinase; ACC: acetyl-CoAcarboxylase; ULK1: Unc-51 like autophagy activating kinase 1; HSC70: heat shock 70kDa protein; d: day.

haematologica | 2019; 104(5)

913

M. Ladli et al.

Figure S5). GSK621 provoked 32% cell death at days 5-7 and 70% at days 8-9 (Figure 4B). GSK621 dramatically reduced the proliferation of cells with a more drastic impact on the most mature erythroblasts (Figure 4C).

To gain further insight into the inhibition of mature erythroblast proliferation by GSK621, the cell cycle was analyzed by quantification of DNA content with propidium iodide. At days 8 and 9, GSK921 induced a reduction in the number of cells in the G2/M phase of the cell cycle and an increase in cells in the early S phase, demonstrating blockage in the S phase (Figure 4D). The protein level of AMPK substrates involved in the cell cycle, P5316 and a target gene of P53, P21 was determined. GSK621-mediated cell cycle arrest was not due to the phosphorylation and consequent stabilization of P53 since there was no variation in their expression, which is in agreement with the absence of defects in G1/S transition (Online Supplementary Figure S6).

We then studied in more detail whether GSK621-mediated AMPK activation could affect erythroblast differentiation. In this set of experiments, at days 3-4 cells were immature and did not synthesize hemoglobin, while at days 8-9, more than 80% of the cells were hemoglobinized (Figure 5A). In the presence of GSK621, at days 8-9, the percentage of cells that synthesized hemoglobin was very low.

To decipher more precisely the stage at which the GSK621-mediated activation of AMPK induced cell death, erythroblasts were analyzed for GPA and annexin V by flow cytometry (Figure 5B). GSK621 did not affect immature GPAlow cells at days 3-4, but induced massive death of GPAhigh erythroblasts at days 5-7 (46% annexin V-positive cells with GSK621 versus 16% in control cells) and at days 8-9 (75% versus 16%, respectively). Indeed, after AMPK activation at days 5-7, only 3.5% of erythroblasts were GPAhigh, in contrast to 46.4% in control cells. Furthermore, at day 7 only 7.5% of GSK621-treated cells were band 3high compared to 42% of control cells. At day 9, in the control conditions, erythroid cells continued to differentiate, which is in contrast to immature GSK621-treated cells.

Furthermore, morphological studies after staining with May-Grünwald-Giemsa confirmed the blockage in maturation. Indeed, in vehicle-treated cultures, at day 9, the population was mainly constituted of polychromatic erythroblasts, and at day 14, orthochromatic erythroblasts and reticulocytes, whereas in the GSK621-treated culture, at days 9 and 14, cells were very immature with large nuclei and uncondensed chromatin; no mature cells were detected at day 14 (Figure 5C). The same results (decreased proliferation and survival, differentiation blockage) were obtained with another direct activator, compound 991 (Online Supplementary Figure S7). Overall, our results show that activation of AMPK by direct activators induced a blockage in the cell cycle, proliferation arrest and death of mature erythroblasts after the basophilic stage.

To reinforce our data, we took advantage of the fact that erythroid progenitors and early precursors can proliferate with delayed differentiation in response to erythropoietin, stem cell factor and dexamethasone.17 We maintained the cells for the indicated number of days in culture medium with vehicle, GSK621, dexamethasone + vehicle or dexamethasone + GSK621 (Figure 6A). As expected, the presence of dexamethasone delayed erythroid differentiation, as demonstrated by the absence of a GPAhigh population after 7 days and even 9 days of culture. After 7 days, GSK621 induced cell death, as previously described (Figures 4B and 5B), with 54.6% of cells being positive for annexin V and 914

35.3% being stained by trypan blue; however, immature erythroblasts treated with dexamethasone + GSK621 were resistant to GSK621-induced cell death. Indeed, only 23.4% of cells were annexin V-positive and 16.6% were stained by trypan blue (Figure 6A,B).

To confirm that the activation of AMPK in mature erythroblasts provoked cell death, GSK621 was added for 24 and 48 h on day 9 when the erythroblastic population already contained 35% of mature GPAhigh cells (Figure 6C).

The GPA/annexin V staining clearly demonstrated that GSK621 induced massive death in mature GPAhigh cells within less than 48 h. With GSK621, 52% of the total cells were positive for annexin V and 43% were stained by trypan blue versus 16% and 10%, respectively, of the control cells.

Overall, our results demonstrated that the activation of AMPK was deleterious for mature GPAhigh cells, specifically in contrast to immature erythroblasts (from the progenitor stage to the basophilic stage), which were not affected.

AMPK activation induced autophagy and apoptotic death of mature erythroblasts In erythroblasts, AMPK activation leads to ULK1 phosphorylation at S555 (Figure 4A), which is well known to be important for the induction of autophagy in several types of cells.18 In GSK621-treated erythroblasts, LC3B-II accumulation was clearly detected by immunoblotting (Figure 7A, left panel). The induction of autophagy was confirmed by the use of chloroquine, which blocks the degradation of autophagosomes.19 Indeed, in addition to GSK621, chloroquine treatment further increased LC3B-II, showing that the activation of AMPK by GSK621 in mature erythroblasts induced autophagy (Figure 7A, right panel).

We, therefore, wondered whether GSK621 provoked caspase-dependent apoptotic cell death and treated cells with a pan-caspase inhibitor Q-VD-OPh (QVD) in addition to GSK621. When caspase activity was blocked by QVD for 48 h (as demonstrated by the anti-cleaved-caspase 3 immunoblot), mature erythroblasts were protected from GSK621-induced cell death, showing that AMPK activation induced caspase-dependent apoptotic cell death (Figure 7B).

Discussion Ampk a1-/-, Ampk β1-/- and Ampk γ1-/- mice develop hemolytic anemia, and the plasma membrane of their red blood cells shows elasticity defects.5–8 The membrane composition evolves continuously throughout erythropoiesis and during red blood cell maturation; the defects due to the absence of Ampk are most likely initiated during erythropoiesis. We, therefore, studied the role of AMPK during human erythropoiesis.

As in murine red blood cells,5 a1 is the only catalytic subunit expressed in erythroblasts, a2 is not detected and we showed that the heterotrimer a1/β1/γ1 is predominant from erythroid progenitors to orthochromatic erythroblasts.

During the earliest stages of terminal differentiation, from progenitors to the basophilic stage, AMPK is activated and then its activation is drastically reduced to the reticulocyte stage. Several kinases and phosphatases regulate AMPK activation.1,20 AMPK is activated by phosphorylation at T172 by three upstream kinases: LKB1, which seems to be constitutively active, CaMKK2 which is activated by an haematologica | 2019; 104(5)

Role of AMPK in human erythropoiesis

A

B

C

Figure 5. GSK621-mediated AMPK activation specifically and drastically decreases the survival of mature GPAhigh erythroblasts. (A) GSK621-dependent AMPK activation reduces the percentage of hemoglobinized cells. The percentage of hemoglobinized cells was determined by benzidine staining after incubation in the absence (vehicle) or presence of 20 mM GSK621 from day 0 to the indicated days. (B) GSK621-mediated AMPK activation blocked erythroblast differentiation. Erythroblasts were incubated in the absence (vehicle) or presence of 20 mM GSK621 from day 0 to the indicated days. Representative cytometry profiles for GPA/annexin V and band 3/a4 integrin are shown. For the GPA/annexin V profiles, annexin V-positive cells are shown in the upper part of the quadrant and GPAhigh cells in the right part.

The percentage of GPAhigh cells and percentage of annexin V-positive cells are represented. Results are expressed as the mean Âą SD of three independent experiments; ns: non-significant, *P<0.05. (C) Morphology was analyzed by May-GrĂźnwald-Giemsa staining. Representative experiments at day 4, day 9 and day 14 are shown. d: day; GPA: glycophorin A.

haematologica | 2019; 104(5)

915

M. Ladli et al.

A

C

B

D

Figure 6. AMPK activation by GSK621 provokes the death of mature GPAhigh erythroblasts. (A) GSK621 does not increase the percentage of annexin V-positive cells in an immature cell population. Erythroid cells were incubated in the absence (vehicle) or presence of 20 mM GSK621 and 2x10-7 M dexamethasone (Dexa)/vehicle or Dexa/GSK621 from day 0 to the indicated days. Cells were labeled with anti-GPA antibodies and annexin V before analysis by flow cytometry. A representative experiment and the percentage of annexin V-positive cells from three experiments are shown. (B) GSK621 does not induce cell death when the population of erythroblasts is immature. The proportion of dead cells was determined by trypan blue exclusion dye assay. (C) GSK621-mediated AMPK activation specifically induces the death of mature day 9 erythroblasts. Mature erythroblasts were incubated in the absence (vehicle) or presence of 20 mM GSK621 for 24 or 48 h. Cells were incubated with anti-GPA antibodies and annexin V before analysis by flow cytometry. A representative experiment and the percentage of annexin V-positive cells from three experiments are shown (lower panel). (D) The proportion of dead cells was determined by a trypan blue exclusion dye assay. Results are expressed as the mean Âą SD. *P<0.05, **P<0.01, ***P<0.001. d: day; Dexa: dexamethasone; GPA: glycophorin A; h: hours.

916

haematologica | 2019; 104(5)

Role of AMPK in human erythropoiesis

increase in cytosolic Ca2+, and possibly TAK1, which is activated by cytokines. The phosphatases PP1, PP2A and PP2C dephosphorylate T172 and the kinases GSK3, PKA, PKB, and PKC inhibit AMPK activation; they may contribute to the reduced AMPK activation in mature erythroblasts. Our data from the quantitative mass spectrometry analysis of human erythropoiesis11 did not allow us to quantify these kinases and phosphatases because of their absence, their very weak expression or the scarcity of peptides generated.

Nevertheless, our western blot studies showed the expression of LKB1 along erythroid differentiation and suggested that LKB1 could be the upstream activating kinase for AMPKa1. Further studies are needed to understand the kinetics of AMPK activation and its upstream regulators during erythroid terminal differentiation.

Our results in human erythroblasts show that knockdown of the expression of the a1 subunit by shRNA induces a decrease in cell proliferation and does not inhibit cell survival or erythroid maturation. Red blood cells from Ampk a1-/- mice have defects in membrane elasticity leading to hemolytic anemia. The absence of the a1 subunit in human erythroblasts induces important changes in the expression of membrane proteins and could potentially affect the expression of membrane proteins involved later in erythrocyte membrane elasticity. As we previously showed, phosphorylation of adducin at Ser724 is increased in red blood cells from Ampk a1-/- and Ampk γ1-/- mice. Our data demonstrate that this modification is also present earlier in human erythroblasts. Interestingly, in sickle cell disease, the reduction of red blood cell deformability is associ-

A

B

Figure 7. GSK621-mediated AMPK activation leads to autophagy and apoptosis. (A) GSK621-mediated AMPK activation induces autophagy. Erythroid cells were incubated in the absence (vehicle) or presence of 20 mM GSK621 from day 0 to the indicated days of culture (left panel). Chloroquine (10 mM) was added for 4 h before harvesting the cells (right panel). LC3BII was detected by western blot experiments using specific antibodies. Anti-β actin or anti-HSC70 antibodies were used as loading controls. (B) GSK621 provokes the caspase-dependent apoptosis of mature day 8 erythroblasts. Mature erythroblasts were incubated in the absence (vehicle) or presence of 20 mM GSK621 or 20 µM GSK621 and 10 mM QVD for 48 h; cells were labeled with anti-GPA and for annexin V binding and analyzed by flow cytometry. A representative experiment and the percentage of annexin V-positive cells from three experiments are shown. Results are expressed as the mean ± SD. *P<0.05, **P<0.01. Efficiency of QVD to block caspase activity was determined by western blot using an anti-caspase 3 antibody that specifically detects the cleaved isoform of caspase 3. β-actin was used as a loading control. d: day; CQ: chloroquine; GPA: glycophorin A; QVD: Q-VD-OPh.

haematologica | 2019; 104(5)

917

M. Ladli et al.

ated with increased phosphorylation of adducin at Ser 726.21 Thus, it would be interesting to determine whether AMPK plays a role in this pathology. Our results suggest that expression of band 3 is regulated by AMPKa1.

Through in vitro studies, Thali et al. identified band 3 as a potential direct substrate for AMPK.22 An attractive hypothesis would be that AMPK induces band 3 phosphorylation resulting in an increase in its global expression but a less efficient expression at the cell surface of erythroblasts.

To activate AMPK specifically, we used direct activators of AMPK because these molecules bind directly to the β subunit and do not affect the AMP/ATP ratio as metformin does. Several recent studies have demonstrated the involvement and specificity of GSK621 in activating AMPK.12-15 In hematopoietic cells, GSK621 has been reported to be more potent in primary acute myeloid leukemia cells and cell lines than the direct activator A-769662.12 In the present study, we demonstrated, through the use of direct activators (GSK621 and compound 991), that AMPK activation in mature erythroblasts (GPAhigh) (polychromatic to reticulocytes) induced apoptotic cell death, whereas no such effect was observed in similarly treated immature erythroblasts. Furthermore, the fact that GSK621 induced the apoptosis of mature erythroblasts after only 48 h of treatment but did not affect erythroblasts that were maintained in an immature state (by dexamethasone) after 9 days of GSK621 excludes a potential toxic effect due to the accumulation of compounds. We propose that maintaining AMPK activation after the basophilic stage, when AMPK is not normally activated, induces cell cycle arrest followed by the induction of autophagy and caspase-dependent apoptosis. Thus, our

References 1. Hardie DG. AMPK: positive and negative regulation, and its role in whole-body energy homeostasis. Curr Opin Cell Biol.

2015;331–337.

2. Hardie DG. AMP-activated protein kinase: an energy sensor that regulates all aspects of cell function. Genes Dev. 2011;25(18):1895–1908.

3. Steinberg GR, Kemp BE. AMPK in health and disease. Physiol Rev. 2009;89(3):1025–1078.

4. Williams T, Brenman JE. LKB1 and AMPK in cell polarity and division. Trends Cell Biol.

2008;18(4):193–198.

5. Foretz M, Guihard S, Leclerc J, et al.

Maintenance of red blood cell integrity by AMP-activated protein kinase a1 catalytic subunit. FEBS Lett. 2010;584(16):3667–3671.

6. Foretz M, Hébrard S, Guihard S, et al. The AMPK 1 subunit plays an essential role in erythrocyte membrane elasticity, and its genetic inactivation induces splenomegaly and anemia. FASEB J. 2011;25(1):337–347.

7. Wang S, Dale GL, Song P, Viollet B, Zou MH. AMPKalpha1 deletion shortens erythrocyte life span in mice: role of oxidative stress. J Biol Chem. 2010;285(26):19976– 19985.

8. Cambridge EL, McIntyre Z, Clare S, et al.

The AMP-activated protein kinase beta 1 subunit modulates erythrocyte integrity.

Exp Hematol. 2017;45:64-68.e5.

9. Vara-Ciruelos D, Dandapani M, Gray A, Egbani EO, Evans AM, Hardie DG. Genotoxic damage activates the AMPK-a1

918

10.

11.

12.

13.

14.

15.

16.

17.

work suggests that AMPK activation during the final stages of erythropoiesis is deleterious.

The present work highlights the role of AMPK in erythropoiesis and adds further support to the involvement of AMPK in the regulation of hematopoiesis. In hematopoietic stem cells, AMPK deficiency partially phenocopies the mitochondrial defects observed in Lkb1-/- mice without affecting hematopoietic stem cell maintenance,23 Obba et al. recently demonstrated that the activation of AMPK is crucial for CSF-1-induced autophagy and human monocyte differentiation into macrophages.24 Our results demonstrate the importance of the finely tuned regulation of AMPK during adult human erythropoiesis. This observation is of significant value since deciphering the molecular mechanisms regulating proliferation, survival and differentiation of erythroblasts is necessary to better understand how erythroid progenitors and precursors can physiologically give rise to red blood cells. The use of direct AMPK activators is being considered as a therapeutic treatment in several chronic metabolic diseases. Phase I and II trials investigating the use of the activators PXL770 (clinical trial NCT03395470) and compound 0304 (betagenon.se) are in progress in patients with non-alcoholic hepatic steatosis or type 2 diabetes.

These activators could induce the apoptosis of mature erythroblasts in the bone marrow so it will be necessary to analyze hematologic parameters to prevent potential anemia.

Acknowledgments ML was funded by the Ministère de l’Enseignement Supérieur et de la Recherche and the Labex GRex. This work was supported by the Laboratory of Excellence Labex GRex.

isoform in the nucleus via Ca2+/CaMKK2 signaling to enhance tumor cell survival.

Mol Cancer Res. 2018;16(2):345–357.

Fogarty S, Ross FA, Vara Ciruelos D, Gray A, Gowans GJ, Hardie DG. AMPK causes cell cycle arrest in LKB1-deficient cells via activation of CAMKK2. Mol Cancer Res. 2016;14(8):683–695.

Gautier EF, Ducamp S, Leduc M, et al.

Comprehensive proteomic analysis of human erythropoiesis. Cell Rep. 2016;16(5): 1470–1484.

Sujobert P, Poulain L, Paubelle E, et al. Coactivation of AMPK and mTORC1 induces cytotoxicity in acute myeloid leukemia. Cell Rep. 2015;11(9):1446–1457.

Liu W, Mao L, Ji F, et al. Targeted activation of AMPK by GSK621 ameliorates H 2 O 2 induced damages in osteoblasts.

Oncotarget. 2017;8(6):10543–10552.

Jiang H, Liu W, Zhan S-K, et al. GSK621 targets glioma cells via activating AMP-activated protein kinase signalings. PLoS One. 2016;11(8):e0161017.

Chen L, Chen Q, Deng G, et al. AMPK activation by GSK621 inhibits human melanoma cells in vitro and in vivo.

Biochem Biophys Res Commun.

2016;480(4):515–521.

Jones RG, Plas DR, Kubek S, et al. AMP-activated protein kinase induces a p53-dependent metabolic checkpoint. Mol Cell.

2005;18(3):283–293. von Lindern M, Zauner W, Mellitzer G, et al.

The glucocorticoid receptor cooperates with the erythropoietin receptor and c-Kit to

18.

19.

20.

21.

22.

23.

24.

enhance and sustain proliferation of erythroid progenitors in vitro. Blood. 1999;94 (2):550–559.

Kim J, Kundu M, Viollet B, Guan K-L. AMPK and mTOR regulate autophagy through direct phosphorylation of Ulk1. Nat Cell Biol. 2011;13(2):132–141.

Mizushima N, Yoshimorim T, Levine B. Methods in mammalian autophagy research. Cell. 2010;140(3):313–326.

Carling D, Thornton C, Woods A, Sanders MJ. AMP-activated protein kinase: new regulation, new roles?

Biochem J. 2012;445(1):11–27.

George A, Pushkaran S, Li L, et al. Altered phosphorylation of cytoskeleton proteins in sickle red blood cells: the role of protein kinase C, Rac GTPases, and reactive oxygen species. Blood Cells Mol Dis. 2010;45(1):41– 45.

Thali RF, Tuerk RD, Scholz R, Yoho-Auchli Y, Brunisholz RA, Neumann D. Novel candidate substrates of AMP-activated protein kinase identified in red blood cell lysates.

Biochem Biophys Res Commun.

2010;398(2):296–301.

Nakada D, Saunders TL, Morrison SJ. Lkb1 regulates cell cycle and energy metabolism in haematopoietic stem cells. Nature.

2010;468(7324):653–658.

Obba S, Hizir Z, Boyer L, et al. The PRKAA1/AMPK 1 pathway triggers autophagy during CSF1-induced human monocyte differentiation and is a potential target in CMML. Autophagy. 2015;11(7): 1114–1129.

haematologica | 2019; 104(5)

ARTICLE

Red Cell Biology & its Disorders

Factor H interferes with the adhesion of sickle red cells to vascular endothelium: a novel disease-modulating molecule

Elisabetta Lombardi,1* Alessandro Matte,2* Antonio M. Risitano,3 Daniel Ricklin,4 John D. Lambris,5 Denise De Zanet,1,6 Sakari T. Jokiranta,7 Nicola Martinelli,2 Cinzia Scambi2, Gianluca Salvagno,8 Zeno Bisoffi,9,10 Chiara Colato,10 Angela Siciliano,2 Oscar Bortolami,11 Mario Mazzuccato,1 Francesco Zorzi,2 Luigi De Marco1,12# and Lucia De Franceschi2

Department of Translational Research, National Cancer Center, Aviano, Italy; Department of Medicine, University of Verona-AOUI Verona; Italy; 3Hematology, Department of Clinical Medicine and Surgery, Federico II University, Naples, Italy; 4 Molecular Pharmacy Group, Department of Pharmaceutical Sciences, University of Basel, Switzerland; 5Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; USA; 6Polytechnic Department of Engineering and Architecture, University of Udine, Italy; 7Research Programs Unit, Immunobiology, University of Helsinki and United Medix Laboratories, Helsinki, Finland; 8 Laboratory of Clinical Biochemistry, Department of Life and Reproduction Sciences, University of Verona, Italy; 9Centre of Tropical Diseases, Sacro Cuore-Don Calabria Hospital Negrar, Verona, Italy; 10Department of Diagnostics and Public Health, University of VeronaAOUI Verona, Italy; 11Unit of Epidemiology and Medical Statistics, Department of Diagnostic & Public Health, University of Verona and 12Department of Molecular Medicine, The Scripps Research Institute, La Jolla, CA, USA 1 2

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):919-928

*EL and AM contributed equally to this work.

LDM are co-last author.

#

ABSTRACT

S

ickle cell disease is an autosomal recessive genetic red cell disorder with a worldwide distribution. Growing evidence suggests a possible involvement of complement activation in the severity of clinical complications of sickle cell disease. In this study we found activation of the alternative complement pathway with microvascular deposition of C5b-9 on skin biopsies from patients with sickle cell disease. There was also deposition of C3b on sickle red cell membranes, which is promoted locally by the exposure of phosphatidylserine. In addition, we showed for the first time a peculiar “stop-and-go” motion of sickle cell red blood cells on tumor factor-a−activated vascular endothelial surfaces. Using the C3b/iC3b binding plasma protein factor H as an inhibitor of C3b cell-cell interactions, we found that factor H and its domains 19-20 prevent the adhesion of sickle red cells to the endothelium, normalizing speed transition times of red cells. We documented that factor H acts by preventing the adhesion of sickle red cells to P-selectin and/or the Mac-1 receptor (CD11b/CD18), supporting the activation of the alternative pathway of complement as an additional mechanism in the pathogenesis of acute sickle cell related vaso-occlusive crises.

Our data provide a rationale for further investigation of the potential contribution of factor H and other modulators of the alternative complement pathway with potential implications for the treatment of sickle cell disease.

Introduction Sickle cell disease (SCD; OMIM # 603903) is an autosomal recessive genetic red blood cell (RBC) disorder with a worldwide distribution. SCD results from a point mutation (βS, 6V) in codon 6 of the β-globin gene where the insertion of valine in place of glutamic acid leads to the production of a defective form of hemoglobin, termed hemoglobin S (HbS).1-3 Pathophysiological studies have shown that intravascular sickling in capillaries and small vessels leads to vaso-occlusion and haematologica | 2019; 104(5)

Correspondence: LUCIA DE FRANCESCHI lucia.defranceschi@univr.it Received: May 27, 2018.

Accepted: January 8, 2019.

Pre-published: January 10, 2019. doi:10.3324/haematol.2018.198622 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/919 ©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.

919

E. Lombardi et al.

impaired blood flow. Vaso-occlusive events in the microcirculation result from a complex and only partially understood scenario involving interactions between different cell types. These cells include dense, dehydrated sickle cells, reticulocytes, abnormally activated endothelial cells, leukocytes and platelets.1-4 Plasma factors such as coagulation system cytokines and oxidized pro-inflammatory lipids may also be involved. In addition, cyclic polymerization-depolymerization promotes RBC membrane oxidation and reduces RBC survival in the peripheral circulation.1,5,6 The resulting increase in free hemoglobin and free heme, a consequence of the saturation of the physiological system and local reduction of nitric oxide bioavailability, leads to a pro-coagulant state with increased risk of thrombotic events.2,3,7-10 All this evidence indicates that sickle cell vasculopathy is a crucial player in RBC adhesion and in the development of acute vaso-occlusion in SCD patients.

Although progress has been made in recent decades in understanding the pathogenesis of SCD, the molecular events involved in these processes are still only partially delineated. Whereas a key role for complement activation has been highlighted in chronic inflammatory processes characterized by hemolysis and inflammatory vasculopathy such as atypical hemolytic uremic syndromes and paroxysmal nocturnal hemoglobinuria11-14 the involvement of complement in SCD has been less extensively explored.

Previous studies have revealed: (i) an activation of the alternative complement pathway (AP) in SCD patients; (ii) a reduction in the activating proteases factors B and D, modulating complement activation; (iii) a decrease in the plasma levels of factor H (FH), the major soluble regulator of AP activation; and (iv) increased deposition of the complement opsonin C3b on RBC exposing phosphatidylserine.15-22 Preliminary data from a mouse model of SCD suggest a possible role for complement activation in the generation of vaso-occlusive crises, as an additional disease mechanism contributing to the severity of acute clinical manifestations related to SCD.23,24 Because of its potential detrimental effects on host cells, the AP is finely regulated by membrane-bound and soluble regulators. Circulating FH plays a particularly important role, since this regulator not only binds to C3b and prevents the formation of C3b convertases, but it is also able to recognize self-associated molecular patterns such as sialic acid and glycosaminoglycans present on the membranes of most healthy cells.25-27 Any interference with this recognition process, resulting from either polymorphisms or blocking antibodies against FH, may have severe pathological consequences as described for atypical hemolytic uremic syndromes and other complementmediated disorders.28 Here, we found that sickle RBC are characterized by membrane deposition of C3b, which acts as a marker for the activation of the AP on sickle RBC. We sought to determine whether C3b deposition on RBC might possibly stimulate vaso-occlusive crises by favoring cell-cell interactions. Indeed, we now demonstrate for the first time a peculiar ex vivo motion profile (“stop-and-go” behavior) of SCD red cells during their transit on vascular endothelial surfaces, a motion that prolongs their transit on the vascular endothelial surface and promotes the adhesion of sickle RBC. We show that FH and its 19-20 domain,29,30 which primarily targets C3b, prevent the adhesion of sickle RBC to the endothelium. We further 920

document that FH acts by preventing the adhesion of sickle RBC to P-selectin and/or the receptor Mac-1 (CD11b/CD18). Our data provide a rationale for further investigation of FH and other modulators of the AP as novel disease-modifying molecules with potential implications for the treatment of the clinical manifestations of SCD.

Methods Study design

We studied SCD subjects (n=29; 26 SS and 3 Sβ0) referred to the Department of Medicine, University of Verona and Azienda Ospedaliera Integrata of Verona (Italy) between January 2012 and January 2017. SCD patients were evaluated at steady state, and none of them had been on either hydroxyurea or a transfusion regimen during the 6 months immediately prior to our analysis.

Healthy controls were matched by age, sex and ethnic background. The study was approved by the Ethical Committee of the Azienda Ospedaliera Integrata of Verona (Italy) and informed consent was obtained from patients and healthy controls (ethical approval FGRF13IT). Table 1 shows the demographic characteristics of both the healthy subjects and SCD patients studied.

Biochemical and hematologic parameters as well as plasma levels of C3 and C4 were determined according to clinical and laboratory standards at the Laboratory of Medicine, University of Verona and Azienda Ospedaliera Integrata of Verona (Italy). Plasma C5a (EIA Quidel Corp., San Diego, CA, USA), plasma vascular cell adhesion molecule-1 (Invitrogen, Carlsbad, CA, USA) and serum FH (Hycult Biotech, Uden, the Netherlands) were determined by enzyme-linked immunosorbent assays, according to the manufacturers’ protocols.31

Evaluation of C5b-9 complement deposition on fixed skin biopsies Skin punch biopsies were carried out on the volar surface of the left arm on apparently normal skin. The samples were paraffinembedded and stained with hematoxylin and eosin before examination; they were also used to estimate the presence of C5b-9 complement deposition within the microvasculature.32 Details on the immunofluorescent and immunohistochemical staining assays are reported in the Online Supplementary Methods.32-34

Measurements of phosphatidylserine-positive and C3d-positive red blood cells Phosphatidylserine-positive cells were detected as previously reported.35-37 Details are given in the Online Supplementary Methods.

Red blood cell adhesion assay The real-time adhesion of RBC collected from healthy and SCD subjects on inactive or activated endothelium [in the absence or presence of tumor necrosis factor-alpha (TNF-a), respectively] was determined as previously reported38-40 with or without FH or the FH19-20 or FH68 domains.41 Details are provided in the Online Supplementary Methods.

Development of an algorithm to determine red blood cell transit and flux trajectory The algorithm to determine RBC transit and flux trajectory is described in the Online Supplementary Methods.

P-selectin and Mac-1 expression in vitro in vascular endothelial cells Details on immunoblotting42 and the cytofluorimetric analysis haematologica | 2019; 104(5)

Complement is involved in sickle red cell adhesion

to determine P-selectin and Mac-1 expression are given in the Online Supplementary Methods.

Table 1. Demographic, hematologic and biochemical data for healthy subjects and patients with sickle cell disease.

Parameters

Statistical analysis All calculations were performed using the IBM SPSS 20.0 statistical package (IBM Inc., Armonk, NY, USA). The level of aggregation was expressed as median values with the minimum-maximum range and shown by means of either dot or box plots. Data were analyzed with non-parametric tests, the Mann-Whitney U test for unpaired samples and the Wilcoxon signed-rank test for paired samples. A P value <0.05 was considered statistically significant.

Results Patients with sickle cell disease show activation of the alternative complement pathway and increased C3d-positive red blood cells Untreated SCD subjects at steady state were studied. As shown in Table 1, reduced hemoglobin, increased reticulocyte counts and raised levels of plasma lactate dehydrogenase, all signs of chronic hemolytic anemia, were observed. In SCD patients, we also found significant increases of the levels of plasma C-reactive protein and surface vascular cell adhesion molecule-1, indicating the presence of a chronic inflammatory vasculopathy in agreement with previous reports.2,3,9,43 Serum C3 and C4 levels were similar in healthy and SCD subjects (data not shown); whereas levels of the complement activation fragment C5a were significantly elevated in SCD patients when compared to the levels in healthy individuals (Figure 1A); statistical analyses were performed to exclude the possible contribution of confounding factors such as gender or smoking status. Our finding is consistent with previous reports, and confirms the substantial complement activation in SCD patients, likely via the AP.15-22 Since studies in atypical hemolytic uremic syndromes have shown that skin biopsy might be a feasible method for documenting AP activation by C5b-9 vascular deposition,32 we obtained skin biopsies from SCD patients at steady state. We considered skin to be an interesting window of observation in SCD, since (i) it might be involved in the clinical manifestations of SCD, such as leg ulcers; and (ii) it has been widely used in SCD mouse models to functionally characterize the microvasculature.44-47 As shown in Figure 1B, we observed microvascular intense, focal granular deposition of C5b-9 in small vessels throughout the dermis of SCD patients. This pattern is similar to that reported in skin biopsies from patients with atypical hemolytic uremic syndrome, which is a thrombotic microangiopathy related to complement dysfunction.32 No deposition of C5b9 was observed in skin from healthy controls (Online Supplementary Figure S1A,B). Colocalization of C59b deposits and CD31+ skin vessels was confirmed by immunohistochemical staining only in the skin biopsies from SCD patients (Figure 1C, Online Supplementary Figure S2A,B).

Taken together, our data indicate an activation of the AP in SCD, with possible involvement of complement in SCD vasculopathy and in related cellular adhesion events.

To understand whether AP activation occurs directly on sickle RBC, we measured the amounts of circulating RBC carrying C3-derived opsonins by detecting the presence of the common C3d fragment.11,12 As shown in Figure 1D, haematologica | 2019; 104(5)

Healthy subjects (n=29)

SCD patients (n=29)

Age (years) 30.0 (25.0-42.5) 20.0 (17.0-47.0)* Gender (male/female) (n) 11/18 10/19 Smokers (%) (n) 10.3% (3) 6.8% (2) Systolic blood pressure (mmHg) 120 (110-130) 120 (112-130) Diastolic blood pressure (mmHg) 65 (58-77) 64 (60-76) Hemoglobin (g/dL) 13.4 (12.5) 8.5 (8-9.5)* HbF (%) 2 (1.8-2.1) 4.5 (2.5-7.6)* Reticulocytes (cells * 103/mL) 44.6 ±12 250 ±27* 9 White blood cells (10 cells/mL) 4.5 (3.8-5.2) 10.2 (8.8-11.9)* Creatinine (mg/dL) 0.8 (0.7-1.0) 0.7 (0.3-1.5) Lactate dehydrogenase (U/L) 325.0 (261.5-413.0) 484.0 (310.2-1104.2)* Albumin (g/L) 42.3 (39.2-46.7) 44.3 (35.6-49.7) C-reactive protein (mg/L) 0.9 (0.2-12.4) 3.0 (1.0-15.8)* sVCAM-1 (pg/mL) 280 ±14 820±36* HbF: fetal hemoglobin; sVCAM-1: serum vascular adhesion molecule-1. Ranges are shown in parentheses. Bold * P<0.05 compared to healthy subjects.

higher numbers of C3d+ RBC were found in SCD patients in the steady state than in healthy subjects (representative scatter-plots are shown in Online Supplementary Figure S3).

This fraction further increased in a subgroup of SCD patients during acute pain crises (SS steady state 2.5±0.8% versus SS pain crisis: 6.1±0.7% P<0.02; n=10), in agreement with an observation by Mold et al.18 Since a previous report linked activation of the AP with deposition of C3 opsonins to the exposure of phosphatidylserine on SCD RBC surfaces, we evaluated the percentage of phosphatidylserine-positive RBC in our SCD patients17 and did indeed find higher percentages of phosphatidylserine-positive RBC in SCD patients than in healthy subjects in agreement with previous studies35,36 (Figure 1E). In contrast to paroxysmal nocturnal hemoglobinuria, in which opsonization leads to rapid lysis of affected RBC.11,12 the presence of functional membrane regulators on SCD erythrocytes allowed a low level of C3-fragment deposition, which might also be considered as a marker of complement activation on sickle RBC. We, therefore, reasoned that the presence of C3b, iC3b or C3dg may function as a site of adhesion of SCD RBC to activated vascular endothelial surfaces, which may carry ligands for C3 fragments, such as Mac-1 and P-selectin.48,49 To study this possibility, we used the C3b and iC3b binding plasma protein FH as an inhibitor25-27 and saw no significant differences in FH serum levels between healthy subjects and SCD patients [AA median: 841 (range 616-1528) mg/mL versus SCD median: 980 (741-1301) mg/mL; P=NS].

Factor H prevents the adhesion of sickle red blood cells to tumor necrosis factor-a-activated vascular endothelium We performed an ex vivo adhesion assay using TNF-aactivated vascular endothelial cells and monitored, in real time, the adhesion of RBC collected from healthy and SCD subjects to inactive (no TNF-a) or activated endothelium (plus TNF-a) as previously described.38-40 Consistent with the literature, we observed significantly increased 921

E. Lombardi et al.

adhesion of SCD RBC to TNF-a-activated endothelium when compared to normal RBC (Figure 2A, Online Supplementary Video S1). We next investigated the effect of FH on the adhesion of healthy or SCD RBC to activated vascular endothelium. The dose-response curve with FH showed a significant reduction in RBC adhesion at concentrations ≥9 nM (Figure 2B). This was more pronounced for SCD RBC than for healthy RBC (Figure 2B). We, therefore, expanded the tests to include a larger number of SCD patients using FH concentrations of 9 or 18 nM; as expected, reduced adhesion of SCD RBC to the vascular endothelium was observed in the presence of either 9 nM or 18 nM FH (Figure 3A). Previous studies have identified two major regions on FH (i.e., domains 6-8 and 19-20) as the putative sites of interaction with cell surface con-

A

stituents such as glycosaminoglycans; furthermore, domains 19-20 are also responsible for binding to sialic acids or C3d-containing deposits.29 We, therefore, evaluated whether any of these segments had activity similar to that of full-length FH. Perfusion experiments performed on TNF-a-activated endothelium demonstrated that, like intact FH, FH19-20 (18 nM) strongly prevented the adhesion of sickle RBC to the surface of TNF-a-activated vascular endothelium.

Conversely, FH6-8 showed a trend, which did not reach statistical significance, toward a reduction in RBC adhesion (Figure 3B). Our data indicate that FH prevents sickle cell adhesion to the activated endothelium through its interaction with cell-surface sialic acids and C3b/iC3b found on the surface of pathological RBC (Figure 3B).

B

C

D

E

Figure 1. The adhesion of C3d-positive sickle red blood cells is prevented by factor H. (A) Plasma samples from healthy controls (AA) and patients with sickle cell disease (SCD) were tested by enzyme-linked immunosorbent assay (see Methods section); *P<0.05 AA versus SCD; n=10 in each group. (B) Deposition of C5b-9 (orange fluorescence) assessed by immunofluorescent staining involving the abluminal aspect of the microvasculature in apparently normal skin of a patient with SCD (direct immunofluorescence; original magnification x100). Inset. Detail of the vessels showing intense granular deposition of C5b-9; direct immunofluorescence; original magnification: x400. Nuclei were stained with Prolong Gold antifade reagent with DAPI (blue fluorescence). The image shown is one representative image of 16 others with similar results. Lower panel. The percentage of vessels positive (+) for C5b9 granular deposition in skin biopsies from healthy subjects (AA) and sickle cell patients (SCD). Data are shown as means ± SD **P<0.002 AA versus SCD. (C) Left panel. Representative immunohistochemical image of a normal skin biopsy from a SCD patient showing a small vessel in the superficial dermis (arrow) characterized by co-expression of C5b9 (brown) and CD31 (red); for comparison a C5b9negative vessel (circle) only decorated with CD31 staining (red) is highlighted. One representative image is shown; all 16 gave similar results (n=16). Right panel.

Quantification of C5b9-positive vessels in skin biopsies from healthy subjects (see Online Supplementary Figure S2) and SCD patients. Data are shown as means ± SD; **P<0.01 AA versus SCD. (D) Percentages of C3d+ red blood cells (RBC) in healthy donors (AA) and in sickle cell subjects (n=16 AA; n=16 SCD). The dashed line indicates the threshold of normality, corresponding to C3d+ <0.5% of RBC; *P<0.05 AA versus SCD; **P<0.01 AA versus SCD. (E) Percentages of phosphatidylserine-positive (PS+) RBC in healthy donors (AA) and in sickle cell subjects (n=20 AA; n=32 SCD). The dashed line indicates the threshold of normality, corresponding to PS+ <0.5 % of RBC; **P<0.01 AA versus SCD.

922

haematologica | 2019; 104(5)

Complement is involved in sickle red cell adhesion

Factor H and its 19-20 domain normalize the sickle red blood cell trajectory and transverse velocity on tumor necrosis factor-α-activated vascular endothelium

during their transit in the flow-chamber. The trajectory of each sickle RBC appeared irregular in space and was not uniform in time, especially when compared to that of healthy RBC (Figure 4A,B). This observation was based on the high transverse displacement and the presence of frequent stop-and-go motion that characterized the sickle RBC transit on the activated vascular endothelium. The

We then developed a new algorithm for RBC to analyze their trajectory and transverse velocity, the velocity component perpendicular to the direction of flow of the RBC

A

B

Figure 2. Adhesion of red blood cells to endothelium activated or not by tumor necrosis factor-a. (A) Adhesion of healthy (AA) or sickle cell disease (SCD) red blood cells (RBC) on immortalized endothelium (EA926.hy) treated or not with tumor necrosis factor-a (TNF-a) under flow conditions (data are expressed as cells/mm2).

The data were obtained from six separate comparable experiments. All calculations were performed using the IBM SPSS 20.0 statistical package (IBM Inc., Armonk, NY, USA). The results of the adhesion tests are expressed as median values with the minimum-maximum range and are illustrated by box plots. The data were analyzed with non-parametric tests, the Mann-Whitney U test for unpaired samples and the Wilcoxon signed-rank test for paired samples. A value of P<0.05 is considered statistically significant. (B) Dose-response curve for factor H (FH) in adhesion assays for healthy (AA) or sickle (SCD) RBC. Data were obtained at 6 min flux on endothelium treated with either vehicle or TNF-a. The curves are representative of six separate and independent experiments with similar results.

A

B

Figure 3. Factor H and its 19-20 segment normalized the transit of sickle red blood cells on the tumor necrosis factor-α-activated vascular endothelial surface. (A) Sickle cell adhesion after 6 min of perfusion on activated or non-activated endothelium (±TNF-a) in the presence of FH 9 nM or 18 nM final concentration. The data shown are representative of six other independent assays with similar results. Wilcoxon test: *indicates the corresponding significance. A P value <0.05 is considered statistically significant. Statistical analysis as in Figure 2A. (B) Sickle cell adhesion after 6 min of perfusion on activated or non-activated endothelium (± TNF-a) in the presence of FH and its fragments 19-20 and 6-8 (18 nM final concentration). Data shown are representative of six other independent assays with similar results.

A P value <0.05 is considered statistically significant. Statistical analysis as in Figure 2A.

haematologica | 2019; 104(5)

923

E. Lombardi et al.

transverse velocity is a parameter that is correlated with the disturbed movement of a particle. The path of healthy RBC appeared to be more regular, uniform, and parallel to the flow (Figure 4A, Online Supplementary Video S2). Thus, healthy RBC crossed the field of view more rapidly than did sickle RBC (0.8-1.2 s versus 1.7-2 s, respectively; P<0.05). In the presence of FH, the trajectory of sickle RBC was regularized (Figure 4C), reducing their transverse velocity to 7.71 ± 6.7 mm/s (average decrease of 81.13% versus vehicle-treated sickle RBC). The same behavior appeared in sickle RBC treated only with the FH 19-20 domain (Figure 4D), with the transverse speed being 6.56 ± 5.6 mm/s (an average decrease of 83.94% versus vehicletreated sickle RBC). When the particle kinematics were quantified (Figure 3E), we found that the absolute value of the instantaneous transverse speed of sickle RBC was

40.86 ± 27.6 mm/s, whereas that of healthy cells was only 5.64 ± 4.9 mm/s (a difference of 86.20%). When healthy cells were compared to either FH- or FH19-20-treated sickle RBC, the absolute values for instantaneous transverse speed decreased significantly, reaching values similar to those observed for healthy RBC.

Anti-P-selectin and anti-Mac-1 antibodies prevent adhesion of sickle red blood cells to tumor necrosis factor-a-activated vascular endothelial surface To better understand the binding proteins that may be important for adhesion of C3b/iC3b+ SCD red cells, we then evaluated the effects of anti-P-selectin or anti-Mac-1 antibodies on the adhesion of SCD to TNF-a-activated vascular endothelium. We chose these two molecules for the following reasons: (i) they are both modulated in

A

B

C

D

E

Figure 4. Factor H and its 19-20 fragment normalized the “stop-and-go” motion of sickle red blood cells. (A) Trajectory of three representative healthy (AA) red blood cells (RBC) in the field of view: each coordinate indicates their centroid at every consecutive frame (flow direction: x axis). (B) Trajectory of three representative sickle (SCD) RBC, showing the “stop-and-go” motion. (C) Trajectory of three representative sickle RBC treated with fator H (FH) (18 nM). (D) Trajectory of three representative sickle RBC treated with FH 19-20 segment (18 nM). (E) Absolute values for instantaneous transverse speed expressed as mean ± standard deviation (**P<0.01 versus healthy RBC).

924

haematologica | 2019; 104(5)

Complement is involved in sickle red cell adhesion

endothelial cells:2,50-55 (ii) P-selectin has been reported to bind C3b present on the membranes of circulating cells such as platelets;56 and (iii) Mac-1 is a well-known receptor for iC3b, and potentially C3dg associated with the cell membrane.25 In addition, P-selectin has been reported to bind RBC, targeting red cell plasma-membrane sialic acid, whereas the presence of a Mac-1 binding site on RBC membranes is still under investigation.29,48,51,57 In our model, the expression of P-selectin and Mac-1 was significantly increased on TNF-a-activated vascular endothelium when compared to the expression on vehicle-treated cells (Online Supplementary Figure S4A). Both anti-P-selectin and anti-Mac1 antibodies prevented the adhesion of sickle RBC to the activated vascular endothelium (Figure 5A).

Whereas the effect of the anti-P-selectin antibody on RBC adhesion may be mediated through interference with two different targets (i.e., iC3b and/or sialic acid), the interplay between Mac-1 and iC3b deposited on sickle RBC is considered to be more selective.25,58 The anti-adhesive effect of the anti-Mac-1 antibody was more pronounced than that of the anti-P-selectin antibody, but the difference was not statistically significant. No effect on the adhesion of sickle RBC was documented in the presence of a control antibody (Online Supplementary Figure S4B). Collectively, these findings indicate that C3b/iC3b is deposited on sickle RBC membranes as a new ligand, bridging sickle RBC to the activated vascular endothelial surface through binding to the pro-adhesive molecules P-selectin and/or Mac-1.

Discussion In this study, we confirmed activation of the AP in SCD and demonstrated, for the first time, cutaneous vascular deposition of C5b-9, supporting the involvement of complement in the microvascular injury associated with SCD.

A

We then showed that activation of the AP results in C3 split-fragments being bound to the sickle RBC surface, thereby contributing to the adhesion of RBC to the inflammatory activated vascular endothelial surface.

Notably, decoration of erythrocytes with C3 opsonins was more evident in patients undergoing vaso-occlusive crises, further supporting the involvement of complement in microvascular injury in SCD. Since sickle RBC, in contrast to paroxysmal nocturnal hemoglobinuria RBC, still contain the functional complement regulators CD55 and CD59, phosphatidylserine-mediated complement activation may allow for a certain degree of opsonization without inducing hemolysis. This situation would lead to the observed accumulation of C3d-containing opsonins (i.e., C3b, iC3b, C3dg), which have all been associated with cell interactions and signaling functions.59 Thus, we propose that C3 split-fragments on RBC might favor cell-cell interactions, supporting previous observations of reduced RBC-cell interactions in mice genetically lacking the C3 complement fraction.51,60 We also demonstrated here that FH prevents the adhesion of sickle RBC and normalizes their trajectory and transverse velocity on TNF-a-activated vascular endothelial surfaces through a mechanism involving P-selectin and/or Mac-1 as pro-adhesion molecule(s). FH binds to C3b/iC3b molecules present on cell surfaces and inhibits the AP. The algorithm developed in the present study allowed us to go further in describing RBC adhesion to the endothelial surface. Indeed, we showed that SCD RBC adhere to endothelial cells with a peculiar dynamic behavior not shown by healthy RBC. The stop-and-go motion of sickle RBC contributes to the irregular trajectory of these cells during their transit on the vascular endothelial surface. This slow and irregular movement clearly affects the speed transition time of sickle RBC when compared to healthy control RBC, possibly contributing to the reduc-

B

Figure 5. The anti-adhesive effect of factor H involves P-selectin and Mac-1 pro-adhesive molecules. (A) Adhesion of sickle cell disease red blood cells (SCD RBC) after 6 min of perfusion on activated or non-activated endothelium (Âą TNF-a) pre-coated with either anti-P-selectin antibody or anti-Mac1 (CD11b/CD18) antibody.

Data shown are representative of six other independent assays with similar results. Wilcoxon test: *indicates the corresponding significance. Adhesion is expressed as median values with a minimum-maximum range and illustrated by box plots. A value of P<0.05 is considered statistically significant. Statistical analysis as in Figure 2A. (B) Schematic model of the beneficial action of factor H in reducing adhesion of C3-derived opsonins on sickle RBC to the TNF-a-activated vascular endothelium. C3 split-fragments on erythrocytes might favor cell-cell interactions through P-selectin and Mac-1. P-selectin might bind RBC through two different targets, iC3b and/or sialic acid; in contrast, Mac-1 targets only iC3b deposits on sickle RBC as a more selective interaction. FH and FH19-20 segment normalized the transit of sickle RBC across the TNF-a-activated vascular endothelial surface, abolishing the â&#x20AC;&#x153;stop-and-goâ&#x20AC;? behavior of the sickle RBC. This effect positively affected (shortened) the transit time of sickle RBC, thereby reducing the likelihood of the RBC sickling during their transit through the microcirculation AP: alternative complement pathway; SCD: sickle cell disease; PS: phosphatidylserine; FH: factor H.

haematologica | 2019; 104(5)

925

E. Lombardi et al.

tion in blood flow in the microcirculation that generally characterizes the early phase of an acute vaso-occlusive crisis.1,5,61-63 It is thus conceivable that our flow-based methodology could have a positive impact on monitoring therapy for SCD. Both FH and its FH19-20 segment normalized the transit of sickle RBC across the TNF-a-activated vascular endothelial surface, abolishing the â&#x20AC;&#x153;stopand-goâ&#x20AC;? behavior of sickle RBC. This effect is of great importance because the transition time of sickle RBC in the microcirculation is critically related to the HbS polymerization time and the generation of dense, dehydrated RBC, which contribute to the development of the acute clinical manifestations of SCD.1,10 It is important to note that FH appears to exert a non-canonical function in preventing cell adhesion. This complement regulator typically inhibits complement activation via the AP by accelerating the decay of C3 convertases and by mediating the degradation of C3b to iC3b and C3dg via the plasma protease factor I. Since our experiments were performed using purified cells in the absence of plasma or serum, factor I and the components involved in convertase formation cannot have been involved in the observed effects.

Rather, FH appears to interfere directly with cell-cell interaction events between sickle RBC and endothelial cells.

To further elucidate the mechanism of complement opsonin-mediated adhesion and the role of FH as an antiadhesive molecule for SCD RBC, we pre-coated vascular endothelial cells with either anti-P-selectin or anti-Mac-1 antibodies. Both surface molecules are expressed on activated vascular endothelial cells and have been associated with opsonin interactions. Mac-1 is well-established as a functionally important receptor for iC3b which contributes to phagocytosis and cell activation. Whereas, the interplay of complement with P-selectin is less well described, several studies have shown interactions between C3b and P-selectin.56,64,65 We found that the adhesion of sickle RBC could be prevented by either anti-Pselectin or anti-Mac-1 antibody, indicating that both molecules contribute to the adhesion of sickle RBC to the vascular endothelium. In our studies, blocking Mac-1 had a more pronounced effect on adhesion than did impairing Pselectin activity. The beneficial impact of interfering with Mac-1 in SCD has been supported by the reduction of RBC-neutrophil interactions in SCD mice treated with anti-Mac-1 antibody51 or in Mac-1-deficient SCD mice.57 Thus, our data indicate that complement is involved in the interaction between sickle RBC and the endothelium, pointing to a new additional mechanism contributing to the biocomplexity of acute events in SCD. Our study therefore has potential implications for the

References 1. De Franceschi L, Cappellini MD, Olivieri O. Thrombosis and sickle cell disease. Semin Thromb Hemost. 2011;37(3):226-236.

2. Kato GJ, Hebbel RP, Steinberg MH, Gladwin MT. Vasculopathy in sickle cell disease: biology, pathophysiology, genetics, translational medicine, and new research directions. Am J Hematol. 2009;84(9):618-625.

3. Hebbel RP. The systems biology-based argu-

926

clinical management of SCD. Current treatments in development are focused on the role of selectins in the pathogenesis of SCD. For example, an anti-P-selectin antibody (crizanlizumab) that has been used to treat SCD has been reported to affect the period of time between acute pain crises in SCD patients.52,66,67 Similarly, the small moleculebased pan-selectin inhibitor, rivipansel, currently undergoing phase III clinical trials, is able to reduce the time required to resolve vaso-occlusive crises with a reduction in opioid treatment.68 Our findings suggest that targeting complement opsonization and/or opsonin-mediated cell adhesion could provide an alternative strategy. Whereas the use of exogenous full-length FH as a therapeutic tool is associated with some challenges, several smaller variants of the regulator have shown promise in preclinical trials for complement-mediated diseases such as paroxysmal nocturnal hemoglobinuria. Given the importance of FH domains 19-20 in interfering with RBC adhesion, mini-FH constructs containing this domain pair may be considered, since they may affect both AP activity and the adhesive function of existing opsonins.69,70 Alternatively, blocking opsonization itself at the level of C3 activation is also expected to impair complement-mediated adhesion.

In conclusion, we have first shown that complement activation on sickle RBC participates in the adhesion of sickle erythrocytes to the TNF-a-activated vascular endothelium (Figure 5B). We then further demonstrated that the FH19-20 segment is as efficient as FH in preventing the adhesion of sickle RBC, and results in normalization of sickle RBC transit across the vascular endothelial surface. We suggest that chronic hemolysis may require high levels of FH to prevent RBC adhesion and entrapment in the microcirculation. Finally, our data indicate that FH might act as a multimodal molecule, preventing the opsonization of sickle RBC with C3 opsonins and targeting the interaction of sickle RBC with the endothelium through the adhesion molecules P-selectin and Mac-1 (Figure 5B). Our findings provide a rationale for considering FH-based inhibitors and other modulators of the AP as potential new therapeutic options in SCD. Acknowledgments This work was supported in part by grants FUR 2016-2017 to LDF and by NIH grant AI068730 to JDL We thank Dr. Sara Ugolini, Dr. Monica Battiston and Dr. Francesco Agostini for technical support, Ing. Leoardo Buscemi for LB software writing and Ing. Vincenzo Insalaca for video editing. We also thank Dr. Letizia Delmonte and Dr. Elisa Vencato for their contribution in preliminary experiments and Dr. Deborah McClellan for editing the manuscript.

ment for taking a bold step in chemoprophylaxis of sickle vasculopathy. Am J Hematol. 2009;84(9):543-545.

4. Parise LV, Telen MJ. Erythrocyte adhesion in sickle cell disease. Curr Hematol Rep. 2003;2(2):102-108.

5. De Franceschi L, Corrocher R. Established and experimental treatments for sickle cell disease. Haematologica. 2004;89(3):348-356.

6. Sabaa N, de Franceschi L, Bonnin P, et al.

Endothelin receptor antagonism prevents

hypoxia-induced mortality and morbidity in a mouse model of sickle-cell disease. J Clin Invest. 2008;118(5):1924-1933.

7. Vinchi F, De Franceschi L, Ghigo A, et al.

Hemopexin therapy improves cardiovascular function by preventing heme-induced endothelial toxicity in mouse models of hemolytic diseases. Circulation. 2013;127 (12):1317-1329.

8. Schaer DJ, Buehler PW, Alayash AI, Belcher JD, Vercellotti GM. Hemolysis and free

haematologica | 2019; 104(5)

Complement is involved in sickle red cell adhesion

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

hemoglobin revisited: exploring hemoglobin and hemin scavengers as a novel class of therapeutic proteins. Blood. 2013;121(8): 1276-1284.

Hebbel RP. Adhesion of sickle red cells to endothelium: myths and future directions.

Transfus Clin Biol. 2008;15(1-2):14-18.

Telen MJ. Beyond hydroxyurea: new and old drugs in the pipeline for sickle cell disease. Blood. 2016;127(7):810-819.

Risitano AM, Notaro R, Marando L, et al.

Complement fraction 3 binding on erythrocytes as additional mechanism of disease in paroxysmal nocturnal hemoglobinuria patients treated by eculizumab. Blood.

2009;113(17):4094-4100.

Risitano AM, Notaro R, Pascariello C, et al.

The complement receptor 2/factor H fusion protein TT30 protects paroxysmal nocturnal hemoglobinuria erythrocytes from complement-mediated hemolysis and C3 fragment.

Blood. 2012;119(26):6307-6316.

Frimat M, Tabarin F, Dimitrov JD, et al.

Complement activation by heme as a secondary hit for atypical hemolytic uremic syndrome. Blood. 2013;122(2):282-292.

Roumenina LT, Loirat C, Dragon-Durey MA, et al. Alternative complement pathway assessment in patients with atypical HUS. J Immunol Methods. 2011;365(1-2):8-26.

Test ST, Woolworth VS. Defective regulation of complement by the sickle erythrocyte: evidence for a defect in control of membrane attack complex formation.

Blood. 1994;83(3):842-852.

Chudwin DS, Papierniak C, Lint TF, Korenblit AD. Activation of the alternative complement pathway by red blood cells from patients with sickle cell disease. Clin Immunol Immunopathol. 1994;71(2):199202.

Wang RH, Phillips G Jr, Medof ME, Mold C. Activation of the alternative complement pathway by exposure of phosphatidylethanolamine and phosphatidylserine on erythrocytes from sickle cell disease patients. J Clin Invest. 1993;92(3):1326-1335.

Mold C, Tamerius JD, Phillips G Jr. Complement activation during painful crisis in sickle cell anemia. Clin Immunol Immunopathol. 1995;76(3 Pt 1):314-320.

Gavriilaki E, Mainou M, Christodoulou I, et al. In vitro evidence of complement activation in patients with sickle cell disease.

Haematologica. 2017;102(12):e481-e482.

Koethe SM, Casper JT, Rodey GE. Alternative complement pathway activity in sera from patients with sickle cell disease.

Clin Exp Immunol. 1976;23(1):56-60.

Strauss RG, Asbrock T, Forristal J, West CD. Alternative pathway of complement in sickle cell disease. Pediatr Res. 1977;11(4):285289. de Ciutiis A, Polley MJ, Metakis LJ, Peterson CM. Immunologic defect of the alternate pathway-of-complement activation postsplenectomy: a possible relation between splenectomy and infection. J Natl Med Assoc. 1978;70(9):667-670.

Schaid TR, Nguyen J, Chen C, et al.

Complement activation in a murine model of sickle cell disease: inhibition of vasoocclusion by blocking C5 activation. Blood.

2016;128(22):158.

Merle NS, Grunenwald A, Rajaratnam H, et al. Intravascular hemolysis activates complement via cell-free heme and heme-loaded microvesicles. JCI Insight. 2018;3(12).

Zipfel PF, Skerka C. Complement regulators and inhibitory proteins. Nat Rev Immunol.

haematologica | 2019; 104(5)

2009;9(10):729-740.

26. de Cordoba SR, de Jorge EG. Translational mini-review series on complement factor H: genetics and disease associations of human complement factor H. Clin Exp Immunol.

2008;151(1):1-13.

27. Kajander T, Lehtinen MJ, Hyvarinen S, et al.

Dual interaction of factor H with C3d and glycosaminoglycans in host-nonhost discrimination by complement. Proc Natl Acad Sci U S A. 2011;108(7):2897-2902.

28. Ricklin D, Reis ES, Lambris JD. Complement in disease: a defence system turning offensive. Nat Rev Nephrol. 2016;12(7):383-401.

29. Wu J, Wu YQ, Ricklin D, et al. Structure of complement fragment C3b-factor H and implications for host protection by complement regulators. Nat Immunol. 2009;10(7): 728-733.

30. Morgan HP, Schmidt CQ, Guariento M, et al. Structural basis for engagement by complement factor H of C3b on a self surface.

Nat Struct Mol Biol. 2011;18(4):463-470.

31. Dworkis DA, Klings ES, Solovieff N, et al.

Severe sickle cell anemia is associated with increased plasma levels of TNF-R1 and VCAM-1. Am J Hematol. 2011;86(2):220223.

32. Magro CM, Momtahen S, Mulvey JJ, et al.

Role of the skin biopsy in the diagnosis of atypical hemolytic uremic syndrome. Am J Dermatopathol. 2015;37(5):349-356; quiz 357-349.

33. Scambi C, Ugolini S, Jokiranta TS, et al. The local complement activation on vascular bed of patients with systemic sclerosis: a hypothesis-generating study. PLoS One. 2015;10(2):e0114856.

34. Vianello A, Vencato E, Cantini M, et al.

Improvement of maternal and fetal outcomes in women with sickle cell disease treated with early prophylactic erythrocytapheresis. Transfusion. 2018;58(9):21922201.

35. de Jong K, Emerson RK, Butler J, et al. Short survival of phosphatidylserine-exposing red blood cells in murine sickle cell anemia.

Blood. 2001;98(5):1577-1584.

36. de Jong K, Larkin SK, Styles LA, Bookchin RM, Kuypers FA. Characterization of the phosphatidylserine-exposing subpopulation of sickle cells. Blood. 2001;98(3):860-867.

37. de Franceschi L, Turrini F, Honczarenko M, et al. In vivo reduction of erythrocyte oxidant stress in a murine model of beta-thalassemia. Haematologica. 2004;89(11):12871298.

38. Hebbel RP. Adhesive interactions of sickle erythrocytes with endothelium. J Clin Invest. 1997;100(11 Suppl):S83-86.

39. Montes RA, Eckman JR, Hsu LL, Wick TM. Sickle erythrocyte adherence to endothelium at low shear: role of shear stress in propagation of vaso-occlusion. Am J Hematol.

2002;70(3):216-227.

40. Barabino GA, McIntire LV, Eskin SG, Sears DA, Udden M. Endothelial cell interactions with sickle cell, sickle trait, mechanically injured, and normal erythrocytes under controlled flow. Blood. 1987;70(1):152-157.

41. Jokiranta TS, Hellwage J, Koistinen V, Zipfel PF, Meri S. Each of the three binding sites on complement factor H interacts with a distinct site on C3b. J Biol Chem.

2000;275(36):27657-27662.

42. Matte A, De Falco L, Federti E, et al.

Peroxiredoxin-2: a novel regulator of iron homeostasis in ineffective erythropoiesis.

Antioxid Redox Signal. 2018;28(1):1-14.

43. Elmariah H, Garrett ME, De Castro LM, et

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

al. Factors associated with survival in a contemporary adult sickle cell disease cohort.

Am J Hematol. 2014;89(5):530-535.

Kalambur VS, Mahaseth H, Bischof JC, et al.

Microvascular blood flow and stasis in transgenic sickle mice: utility of a dorsal skin fold chamber for intravital microscopy. Am J Hematol. 2004;77(2):117-125.

Vercellotti GM, Zhang P, Nguyen J, et al.

Hepatic overexpression of hemopexin inhibits inflammation and vascular stasis in murine models of sickle cell disease. Mol Med. 2016;22.

Serjeant GR. Leg ulceration in sickle cell anemia. Arch Intern Med. 1974;133(4):690-694.

Yawn BP, Buchanan GR, Afenyi-Annan AN, et al. Management of sickle cell disease: summary of the 2014 evidence-based report by expert panel members. JAMA. 2014;312(10):1033-1048.

Matsui NM, Borsig L, Rosen SD, et al. Pselectin mediates the adhesion of sickle erythrocytes to the endothelium. Blood.

2001;98(6):1955-1962.

Manwani D, Frenette PS. Vaso-occlusion in sickle cell disease: pathophysiology and novel targeted therapies.

Blood.

2013;122(24):3892-3898.

Wood K, Russell J, Hebbel RP, Granger DN. Differential expression of E- and P-selectin in the microvasculature of sickle cell transgenic mice. Microcirculation. 2004;11(4): 377-385.

Hidalgo A, Chang J, Jang JE, et al.

Heterotypic interactions enabled by polarized neutrophil microdomains mediate thromboinflammatory injury. Nat Med. 2009;15(4):384-391.

Ataga KI, Kutlar A, Kanter J. Crizanlizumab in sickle cell disease. N Engl J Med. 2017;376(18):1796.

Cherqui S, Kurian SM, Schussler O, et al.

Isolation and angiogenesis by endothelial progenitors in the fetal liver. Stem Cells.

2006;24(1):44-54.

Aranguren XL, Agirre X, Beerens M, et al.

Unraveling a novel transcription factor code determining the human arterial-specific endothelial cell signature.

Blood.

2013;122(24):3982-3992.

Lotzer K, Dopping S, Connert S, et al.

Mouse aorta smooth muscle cells differentiate into lymphoid tissue organizer-like cells on combined tumor necrosis factor receptor1/lymphotoxin beta-receptor NF-kappaB signaling. Arterioscler Thromb Vasc Biol.

2010;30(3):395-402.

Del Conde I, Cruz MA, Zhang H, Lopez JA, Afshar-Kharghan V. Platelet activation leads to activation and propagation of the complement system. J Exp Med. 2005;201(6):871879.

Chen G, Chang J, Zhang D, et al. Targeting Mac-1-mediated leukocyte-RBC interactions uncouples the benefits for acute vasoocclusion and chronic organ damage. Exp Hematol. 2016;44(10):940-946.

Kassebaum NJ, Jasrasaria R, Naghavi M, et al. A systematic analysis of global anemia burden from 1990 to 2010. Blood.

2014;123(5):615-624.

Ricklin D, Reis ES, Mastellos DC, Gros P, Lambris JD. Complement component C3 the "Swiss Army Knife" of innate immunity and host defense. Immunol Rev. 2016;274(1):33-58.

Beller DI, Springer TA, Schreiber RD. AntiMac-1 selectively inhibits the mouse and human type three complement receptor. J Exp Med. 1982;156(4):1000-1009.

927

E. Lombardi et al.

61. De Franceschi L, Franco RS, Bertoldi M, et al.

Pharmacological inhibition of calpain-1 prevents red cell dehydration and reduces Gardos channel activity in a mouse model of sickle cell disease. FASEB J. 2013;27(2):750759.

62. Stocker JW, De Franceschi L, McNaughtonSmith GA, et al. ICA-17043, a novel Gardos channel blocker, prevents sickled red blood cell dehydration in vitro and in vivo in SAD mice. Blood. 2003;101(6):2412-2418.

63. de Franceschi L, Baron A, Scarpa A, et al.

Inhaled nitric oxide protects transgenic SAD mice from sickle cell disease-specific lung injury induced by hypoxia/reoxygenation.

Blood. 2003;102(3):1087-1096.

928

64. Saggu G, Cortes C, Emch HN, et al.

Identification of a novel mode of complement activation on stimulated platelets mediated by properdin and C3(H2O). J Immunol. 2013;190(12):6457-6467.

65. Atkinson C, Zhu H, Qiao F, et al.

Complement-dependent P-selectin expression and injury following ischemic stroke. J Immunol. 2006;177(10):7266-7274.

66. Ataga KI, Kutlar A, Kanter J, et al.

Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439.

67. Slomski A. Crizanlizumab prevents sickle cell pain crises. JAMA. 2017;317(8):798.

68. Telen MJ, Wun T, McCavit TL, et al.

Randomized phase 2 study of GMI-1070 in SCD: reduction in time to resolution of vaso-occlusive events and decreased opioid use. Blood. 2015;125(17):2656-2664.

69. Harder MJ, Anliker M, Hochsmann B, et al.

Comparative analysis of novel complementtargetedinhibitors, miniFH, and the natural regulators factor H and factor H-like protein 1 reveal functional determinants of complement regulation. J Immunol. 2016;196(2): 866-876.

70. Nichols EM, Barbour TD, Pappworth IY, et al. An extended mini-complement factor H molecule ameliorates experimental C3 glomerulopathy. Kidney Int. 2015;88(6): 1314-1322.

haematologica | 2019; 104(5)

ARTICLE

Myelodysplastic Syndromes

Optimized EBMT transplant-specific risk score in myelodysplastic syndromes after allogeneic stem-cell transplantation Nico Gagelmann,1 Diderik-Jan Eikema,2 Matthias Stelljes,3 Dietrich Beelen,4 Liesbeth de Wreede,2 Ghulam Mufti5, Nina Simone Knelange,6 Dietger Niederwieser,7 Lone S. Friis,8 Gerhard Ehninger,9 Arnon Nagler,10 Ibrahim Yakoub-Agha,11 Ellen Meijer,12 Per Ljungman,13 Johan Maertens,14 Lothar Kanz,15 Lucia Lopez-Corral,16 Arne Brecht,17 Charles Craddock,18 Jürgen Finke,19 Jan J. Cornelissen,20 Paolo Bernasconi,21 Patrice Chevallier,22 Jorge Sierra,23 Marie Robin24 and Nicolaus Kröger1

1 University Medical Center Hamburg-Eppendorf, Hamburg, Germany; 2EBMT Statistics, Leiden, the Netherlands; 3University of Münster, Germany; 4Department of Bone Marrow Transplantation, West German Cancer Center, University Hospital of Essen, Germany; 5 GKT School of Medicine, London, UK; 6EBMT Data Office, Leiden, the Netherlands; 7 University Hospital Leipzig, Germany; 8Rigshospitalet, Copenhagen, Denmark; 9 Universitätsklinikum Dresden, Germany; 10Chaim Sheba Medical Center, Tel-Hashomer, Israel; 11CHU de Lille, LIRIC, INSERM U995, Université Lille2, France; 12VU University Medical Center, Amsterdam, the Netherlands; 13Karolinska University Hospital and Karolinska Institutet, Stockholm, Sweden; 14University Hospital Gasthuisberg, Leuven, Belgium; 15Universität Tübingen, Germany; 16Hospital Clínico, Salamanca, Spain; 17 Deutsche Klinik für Diagnostik, Wiesbaden, Germany; 18Centre for Clinical Haematology, Birmingham, UK; 19University of Freiburg, Germany; 20Erasmus MC Cancer Institute, Erasmus University Medical Center, Rotterdam, the Netherlands; 21Fondazione IRCCS Policlinico San Matteo, Pavia, Italy; 22CHU Nantes, France; 23Hospital Santa Creu i Sant Pau, Jose Carreras Leukemia Research Institute, Barcelona, Spain and 24Hopital St. Louis, Paris, France

ABSTRACT

T

he aim of this study was to develop and validate a clinical and transplant-specific prognostic score using data from a large cohort of patients with myelodysplastic syndromes reported to the European Society for Blood and Marrow Transplantation registry. A Cox model was fitted to detect clinical and transplant-related variables prognostic of outcome. Then, cross-validation was performed to evaluate the validity and consistency of the model. Seven independent risk factors for survival were identified: age ≥50 years, matched unrelated donor, Karnofsky Performance Status <90%, very poor cytogenetics or monosomal karyotype, positive cytomegalovirus status of the recipient, blood blasts >1%, and platelet count ≤50 x 109/L prior to transplantation. Incorporating these factors into a four-level risk score yielded hazard ratios for death, with low-risk (score of 0-1) as reference, of 2.02 (95% CI: 1.41-2.90) for the intermediate-risk group (score of 2-3), 3.49 (95% CI: 2.45-4.97) for the high-risk group (score of 4-5), and 5.90 (95% CI: 4.01-8.67) for the very high-risk group (score of >5). The score was predictive of survival, relapse-free survival, relapse, and non-relapse mortality (P<0.001, respectively). Cross-validation yielded significant and reproducible improvement in prognostic ability with C-statistics being 0.609 (95% CI: 0.588-0.629) versus 0.555 for the Gruppo Italiano Trapianto di Midollo Osseo registry and 0.579 for the Center for Blood and Marrow Transplant Research registry. Prediction was even further augmented after applying a nomogram using age and platelets as continuous variables showing C-statistics of 0.628 (95% CI: 0.616-0.637). In conclusion, compared to existing prognostic systems, this proposed transplantspecific risk score offers improved performance with respect to post-transplant risk stratification in myelodysplastic syndromes. haematologica | 2019; 104(5)

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):929-936

Correspondence: NICOLAUS KRÖGER nkroeger@uke.uni-hamburg.de Received: July 2, 2018.

Accepted: January 9, 2019.

Pre-published: January 17, 2019. doi:10.3324/haematol.2018.200808 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/929 ©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.

929

N. Gagelmann et al.

Introduction Myelodysplastic syndromes (MDS) are a heterogeneous group of clonal hematopoietic disorders that are characterized by abnormal cellular maturation resulting in cytopenia and a variable risk of progression to acute leukemia.1 Allogeneic stem-cell transplantation is still the only curative treatment option.2-4 In order to recommend MDS patients for transplantation besides considering only disease-specific factors, such as those proposed by the International Prognostic Scoring System (IPSS) and its revision (IPSS-R),5,6 a valid and readily reproducible transplant-specific scoring system is needed.7,8 Recently, two groups have developed prognostic systems incorporating patient-specific as well as transplantspecific factors. The Gruppo Italiano Trapianto di Midollo Osseo (GITMO) registry developed a transplantation risk index consisting of the following factors: age (<50 or ≥50 years), IPSS-R score (low, intermediate, high, or very high), monosomal karyotype (yes or no), refractoriness to chemotherapy (yes or no), and Hematopoietic Cell Transplantatiom Comorbidity Index score (HCT-CI).9 The resulting risk index could clearly distinguish four different groups (low, intermediate, high, and very high) with corresponding overall survival rates at 5 years of 76%, 48%, 18%, and 5%. The other prognostic score, by the Center for Blood and Marrow Transplant Research (CIBMTR) registry, included the following criteria: age (18-29, 30-49, or ≥50 years), Karnofsky status (90-100 or <90%), cytogenetics (very good to intermediate, poor, very poor or monosomal karyotype), blood blasts before transplantation (>3 or ≤3%) and platelet count before transplantation (>50 or ≤50 x 109/L). This system also discriminated four risk groups (low, intermediate, high, and very high) and showed corresponding overall survival rates at 3 years of 71%, 49%, 41%, and 25%.10 Both systems resulted in relevant re-classification of patients in comparison with the IPSS-R while providing only modest improvement in predictability. While neither score has been externally validated nor investigated regarding prognostic power in large and independent cohorts, we hypothesized that both scores not only vary in design and follow-up but would also vary in prognostic ability when applied to the same cohort.7,8 We, therefore, aimed to develop and validate a clinical and transplant-specific prognostic score using data from a large cohort of MDS patients reported to the European Society for Blood and Marrow Transplantation (EBMT) registry and to validate and compare currently existing systems with respect to the resulting EBMT transplantspecific risk score.

Methods Data source The EBMT is a voluntary organization that comprises more than 500 transplant centers, mainly from Europe. Membership requires submission of the minimal essential data form A for all consecutive patients to a central registry from which patients can be identified by diagnosis of underlying disease and type of transplantation. The information in the minimal essential data form A data is updated annually. Informed consent to transplantation was obtained and data were collected locally according to

930

regulations that were applicable at the time of transplantation.

All transplantation centers were required to obtain written informed consent before data registration with the EBMT in accordance with the 1975 Helsinki Declaration.

Patients Adult patients (≥18 years) with MDS who underwent transplantation from an HLA-identical sibling or matched unrelated donor between 2000 to 2014 were included. Patients were eligible if there was full information on their: diagnosis, donor data, cytogenetic risk, platelet count and blood blasts at transplantation. Cytogenetic risk was stratified based on previously established systems.6,11 The prognostic subgroups were the following: del(11q) and -Y (very good); del(5q), del(12p), del(20q), and normal karyotype (good); del(7q), +8, i(17q), +19, and other independent clones (intermediate); complex karyotype (three abnormalities), inv(3), del(3q), and translocations involving 3q (poor), and very complex karyotype with more than three abnormalities (very poor). Monosomal karyotype was defined as monosomy of two or more chromosomes or one single autosomal monosomy in the presence of other structural abnormalities.12 The IPSS-R was calculated prior to transplantation. In total, 1059 patients met the criteria and were included in the EBMT cohort.

To evaluate possible selection bias, outcomes were compared between the final cohort and remaining patients not included in the analysis due to missing data in the registry (n=5122). Within the EBMT cohort, 519 and 876 patients had full data on all factors included in either the GITMO or CIBMTR score.

Score development The development of the transplant-specific risk score consisted of two steps. First, a Cox proportional hazards model using backward and forward selection was used to identify significant covariates for overall survival.13 Then, the hazard ratios (HR) obtained were classified as large (HR >1.59), intermediate (1.25< HR <1.60) and small (HR <1.25). Subsequently, a scoring rule was defined in which large effects were assigned two points, intermediate effects were assigned one point and small effects were assigned zero points. Scores were grouped based on associated hazard ratios into low-, intermediate-, high-, and very high–risk groups, providing group-based risk predictions for MDS. A second score was developed, based on the β coefficients derived from the model defined above, to provide individualized/patient-specific risk predictions. Second, both developed scores were validated and then compared to existing systems by assessing each score’s prognostic performance.

Statistical analysis Overall survival and relapse-free survival were estimated using the Kaplan-Meier method and compared with the log-rank test in univariable analysis. Non-relapse mortality and relapse were analyzed in a competing risks framework by using the cumulative incidence estimator and the Gray test for univariable analysis.14 Cox proportional hazards regression of complete data was used to develop the two new scores. Maximum likelihood from the Cox model was used to establish cutoffs for continuous variables.

Score performance was analyzed using the concordance index (C): the probability that a patient who experienced an event had a higher risk score than a patient who did not (C >0.5 suggesting predictive ability).15,16 Each system was validated using 5-fold cross-validation with 100 repetitions. P values <0.05 were considered statistically significant. Analyses were performed using SPSS for Windows version 24 (SPSS, Chicago, IL, USA) and R package version 3.4.3 (The R Foundation, Vienna, Austria).

haematologica | 2019; 104(5)

EBMT transplant-specific risk score for MDS

Results Patients The characteristics of the patients and transplants of the total EBMT cohort (n=1059) are listed in Table 1. The median follow-up was 69 months (95% CI: 62-76) and the 5-year overall survival was 38.6% (95% CI: 35.4-41.7) while the relapse-free survival was 33.5% (95% CI: 30.436.6).

Transplant-specific risk score Table 2 summarizes the seven variables identified as independent predictors of overall survival in the multivariable analysis: age ≥50 years, matched unrelated donor, Karnofsky Performance Status <90%, very poor cytogenetics or monosomal karyotype, positive cytomegalovirus (CMV) status of the recipient, blood blasts >1%, and platelet count ≤50 x 109/L at the time of transplantation. A weighted score of two was assigned to older age (≥50 years) and very poor cytogenetics or monosomal karyotype, whereas matched unrelated donor, Karnofsky Performance Status <90%, positive CMV status of the recipient, blood

blasts >1%, and platelet count ≤50 x 109/L prior to transplantation were assigned a score of one.

The overall score ranged from zero to nine, with increasing scores indicating higher risk of death. Based on the score, four risk groups were delineated: low (0-1), intermediate (2-3), high (4-5), and very high (>5). The hazard ratio for death (with the low-risk group as the reference) was 2.02 (95% CI: 1.41-2.90) for the intermediate-risk group, 3.49 (95% CI: 2.45-4.97) for the high-risk group, and 5.90 (95% CI: 4.01-8.67) for the very high–risk group.

Corresponding survival rates were 68.7% for the low-risk group, 43.2% for the intermediate-risk group, 26.6% for the high-risk group, and 9.5% for the very high-risk group.

Overall, the EBMT transplant-specific risk score was predictive of overall survival (P<0.001) (Figure 1).

Secondary endpoints The EBMT cohort was also used to apply the developed score to all secondary objectives. The developed score (overall and in all risk groups) was associated with all secondary endpoints (P=0.001) (Table 3 and Figure 1B-D).

The 5-year relapse-free survival rate was 68.4% (95% CI:

Table 1. Patient and transplantation characteristics of 1059 patients with myelodysplastic syndromes in the total EBMT cohort.

Characteristics Number of patients Age at transplant, years median (range) Patient’s sex female male Secondary disorder no yes unknown Karnofsky index, % 90 to 100 < 90 unknown Comorbidity index, % 0 to 2 ≥3 unknown Cytogenetic risk very good good intermediate poor very poor monosomal karyotype Marrow blasts at transplant, % ≤2 2 to 5 5 to 10 > 10 unknown Blood blasts at transplant, % ≤1 >1

Total cohort, n (%) 1059 56 (18 to 73) 446 (42) 613 (58) 776 (74) 201 (19) 92 (7) 630 (60) 252 (23) 177(17) 462 (44) 190 (18) 407 (38) 26 (3) 132 (13) 357 (34) 299 (28) 45 (4) 200 (19) 333 (31) 258 (24) 170 (16) 114 (12) 210 (17) 929 (88) 130 (12)

continued from the previous coloum

Platelets at transplant, x 109/L ≤ 50 > 50 Revised IPSS category very low low intermediate high very high unknown Conditioning intensity myeloablative reduced non-myeloablative Pretreatment chemotherapy hypomethylating agents both none Graft source bone marrow peripheral blood Time to transplant, months median (range) Donor type HLA-identical sibling matched unrelated Cytomegalovirus serostatus of recipient negative positive unknown Anti-thymocyte globulin Ex vivo T-cell depletion

382 (38) 677 (62) 32 (3) 259 (24) 317 (32) 205 (20) 86 (8) 150 (13) 336 (32) 620 (58) 103 (10) 235 (22) 214 (20) 42 (4) 568 (54) 147 (14) 912 (86) 7 (0.2 to 328) 622 (59) 437 (41) 416 (39) 614 (58) 29 (3) 491 (46) 43 (4)

n: number; IPSS: International Prognosis Scoring System.

continued in the next coloum

haematologica | 2019; 104(5)

931

N. Gagelmann et al.

58.4-78.4) for the low-risk group, 43.3% (95% CI: 37.249.4) for the intermediate-risk group, 26.0% (95% CI: 20.7-31.3) for the high-risk group, and 11.5% (95% CI: 5.4-17.6) for the very high-risk group (P<0.001) (Figure 2B). The cumulative incidence of relapse and non-relapse mortality at 5 years according to risk group were, respectively: 16.9% and 12.8% (low-risk group), 28.2% and 27.7% (intermediate-risk group), 36.0% and 36.9% (highrisk group), and 46.0% and 44.3% (very high-risk group) (Figure 2C,D).

Nomogram We aimed to refine the transplant-specific risk score further by using age and platelet count at transplantation as continuous variables while the other variables remained in their categorical structure. We provide a discrete/continuous nomogram to interpolate the final score and assess an individual patient’s risk in an easy manner (Figure 2). For each of the seven prognostic factors, individual points are assigned, which are subsequently summed to a total point scale. The final score is then translated into predicted survival rates at different time points for each patient. For instance, 55-year old patients presenting with a platelet count of 150 x 109/L, blasts ≤1%, good cytogenetics, a Karnofsky Performance Status of 90% who are seropositive for CMV and for whom an identical sibling donor is available show estimated survival rates at 3 and 5 years of 57% (95% CI: 51-63) and 47% (95% CI: 41-54), respectively.

Validation of existing systems The overall score from the CIBMTR ranged from zero to seven and was used to define four risk categories: low risk (0-1, n=46), intermediate risk (2-3, n=434), high risk (4-5, n=365), and very high risk (>5, n=31). The 5-year overall survival rates were 55.8% (95% CI: 39.9-71.7) for the low-risk group, 42.3% (95% CI: 37.2-47.4) for the intermediate-risk group, and 27.2% (95% CI: 21.9-32.5) for the high-risk group. Rates were not estimable in the very high-risk group because the median follow-up was 42.6 months in this group while median overall survival was 11.8 months (95% CI: 3.1-20.4) (Figure 3A). The GITMO score values ranged from zero to eight. The actual score delineated four risk-groups: low (0-1, n=81), intermediate (2-3, n=200), high (4, n=123), and very high (>4, n=115). The overall survival rates at 5 years were 62.7% (95% CI: 51.7-73.7) for the low-risk group, 41.4% (95% CI: 34.0-48.8) for the intermediate-risk group, 24.1% (95% CI: 15.1-33.1) for the high-risk group, and 15.8% (95% CI: 8.2-23.4) for the very high-risk group (Figure 3B).

Overall, both scores could be validated (P<0.001).

Comparison of prognostic systems Both existing scores from GITMO and CIBMTR were then compared with the developed EBMT transplant-specific risk score with respect to their performance regarding overall survival. The CIBMTR and GITMO scores showed modest performance, with C-statistics after cross-validation being 0.555 (95% CI: 0.524-0.586) and 0.579 (95% CI: 0.570-0.588), whereas the IPSS-R resulted in C-statistics of 0.551 (95% CI: 0.530-0.566). The developed categorized EBMT transplant-specific risk score showed C-statistics of 0.609 (95% CI: 0.588-0.629) indicating an improvement in prognostic performance, which was further improved using age and platelet count as con932

tinuous variables as indicated by the C-statistics of 0.628 (95% CI: 0.616-0.637).

Discussion The major findings of this analysis can be summarized as follows. First, seven independent risk factors (age, cytogenetics, thrombocytopenia and increasing blood blasts at transplantation, Karnofsky Performance Status, donor relation, and CMV status of the recipient) could be successfully incorporated into a transplant-specific risk score.

Second, this EBMT transplant-specific risk score enabled significantly improved prediction of outcome in comparison with currently existing systems.

The EBMT transplant-specific risk score presented here considered conventional clinically derived risk factors at transplantation and can thus be readily calculated. The primary objective in developing this score was to improve our ability to predict survival of MDS patients after transplantation; furthermore, the score was predictive of relapse-free survival, relapse and non-relapse mortality.

Moreover, sub-analyses of all MDS patients in the EBMT registry (n=6181), of whom 5122 were not included in the present study because of missing data, revealed a better overall survival rate at 5 years for excluded patients (48%) compared with that of the patients used to develop the score (39%) while non-relapse mortality was 29%, respectively. Most patients had to be excluded because of missing information regarding cytogenetics. Multiple imputation of these patients revealed no difference in score performance. Collectively, the prognostic ability could be improved using the proposed EBMT transplant-specific

Table 2. Multivariate analysis on overall survival showing seven independent risk factors after stepwise selection using a Cox proportional hazards model.

Factor Age at transplant, years < 50 ≥ 50 Blood blasts at transplant, % ≤1 >1 Platelets at transplant, x109/L > 50 ≤ 50 Donor type HLA-identical sibling matched unrelated Cytogenetic risk very good to poor very poor/monosomal karyotype Cytomegalovirus serostatus of recipient negative positive Karnofsky index, % 90 to 100 < 90

HR (95% CI)

P

Score value

<0.001 reference 1.71 (1.39-2.09)

2 0.03

reference 1.39 (1.03-1.86)

1 0.001

reference 1.46 (1.17-1.82)

1 <0.001

reference 1.39 (1.13-1.71)

1 <0.001

reference 1.71 (1.43-2.06)

2 <0.001

reference 1.39 (1.16-1.65)

1 <0.001

reference 1.44 (1.20-1.72)

1

HR: hazard ratio; CI: confidence interval.

haematologica | 2019; 104(5)

EBMT transplant-specific risk score for MDS

risk score while C-statistics still showed moderate power; thus, prognostic systems in MDS may be validated and updated offering improved risk stratification if more data become available in the registries in the future.

While another study by the EBMT assessed the prognostic utility of the IPSS-R in patients following transplantation, finding modest applicability of this disease-specific approach,17 we evaluated two systems that used transplant- and patient-related approaches. The CIBMTR score showed limited to modest performance (0.555) but resulted in an even better prognostic ability (with a concordance index up to 0.582 after cross-validation) in our cohort than originally reported (0.575) while the GITMO score showed better performance (0.579). We acknowledge the limitation of the lack of information regarding the HCT-CI

in a sufficient number of patients, which could therefore not be included in the final multivariable model for the development of the EBMT transplant-specific risk score.

The Karnofsky Performance Status was used instead.

However, the actual performance status of a patient may vary according to clinicians or at different times during the transplantation evaluation. Other tools evaluating patient fitness, including the HCT-CI may be additionally used as they become available in larger populations.

To investigate to what extent a transplant-specific approach will be generally feasible, we evaluated the prognostic power of the IPSS-R resulting in C-statistics of 0.551 (95% CI: 0.530-0.566), confirming that transplantspecific risk stratification may enable optimized counseling of patients. However, although the proposed EBMT

A

B

C

D

Figure 1. Kaplan-Meier analysis of survival following allogeneic stem cell transplantation in patients with myelodysplastic syndrome stratified according to each risk group of the EBMT transplant-specific risk score. (A) Overall survival. (B) Relapse-free survival. (C) Cumulative incidence of relapse. (D) Non-relapse mortality.

haematologica | 2019; 104(5)

933

N. Gagelmann et al.

transplant-specific risk score demonstrated improved prognostic capacity over the IPSS-R, the magnitude of this benefit was still moderate, suggesting that the combined use of prognostic systems may provide the most appropriate prognostication, until systems with significantly better performance become available.

None of the existing scores investigated the possible impact of CMV on outcome. CMV is an important cause of morbidity and mortality after allogeneic stem-cell transplantation.18 During recent years, major advances have been achieved regarding antiviral prophylactic strategies, and new sensitive diagnostic techniques have

been developed.19-21 A recent evidence synthesis of the efficacy and safety of different prophylactic strategies for CMV highlighted inconclusive results in terms of survival while CMV disease and infection could be significantly reduced using antiviral agents.20 Furthermore, it is unclear whether different prophylactic agents for graftversus-host disease increase the risk for CMV infection or disease after transplantation. Aggregated evidence did not show an increased risk for CMV reactivation in randomized trials on antithymocyte globulin, which was given to 46% of our EBMT cohort, in comparison with standard prophylaxis using cyclosporine and methotrex-

Table 3. Transplant-specific MDS risk score prediction of relapse-free survival, non-relapse mortality and incidence of relapse.

Risk group Score overall low intermediate high very high

Relapse-free survival HR (95% CI) P

Non-relapse mortality HR (95% CI) P

<0.001 reference 2.03 (1.38-2.98) 3.47 (2.39-5.06) 5.77 (3.85-8.66)

<0.001 <0.001 <0.001

Incidence of relapse HR (95% CI) P

<0.001 reference 2.08 (1.16-3.75) 2.99 (1.68-5.30) 4.18 (2.25-7.76)

0.01 <0.001 <0.001

<0.001 reference 1.80 (1.04-3.10) 2.68 (1.57-4.57) 3.70 (2.09-6.55)

0.03 <0.001 <0.001

HR: hazard ratio; CI: confidence interval.

Figure 2. Nomogram of the EBMT transplant-specific risk score: for each of the seven prognostic factors. Corresponding points are assigned, which are subsequently summed to make a total point scale. This final score is then translated into predicted survival rates at different time points for each patient. CMV: cytomegalovirus; Tx: transplantation; MK: monosomal karyotype; 12m: 12-month; 36m: 36-month; 60m: 60-month.

934

haematologica | 2019; 104(5)

EBMT transplant-specific risk score for MDS

A

B

Figure 3. Kaplan-Meier analysis of survival following allogeneic stem-cell transplantation in patients with myelodysplastic syndrome stratified according to their risk group. (A) Center for International Blood and Marrow Transplant Research (CIBMTR) registry. (B) Gruppo Italiano Trapianto di Midollo Osseo (GITMO) registry.

ate or tacrolimus.22 Further analyses, however, showed no significant impact of T-cell depletion or the presence of acute graft-versus-host disease at baseline on outcome of antiviral prophylaxis in terms of CMV infection, survival or safety.20 Hence, the CMV serological status of the transplant recipient still has a strong influence on outcome.23 In our analysis, a positive CMV serostatus of the recipient was associated with a 39% increased risk of death, as well as higher rates of non-relapse mortality.

The significant impact of CMV on outcome in a multivariable model in this large cohort of MDS patients supports further evaluation of antiviral agents, such as letermovir, affecting not only CMV infection and disease but also mortality.20,21 The role of molecular genetics after transplantation has been investigated recently. Most studies24-26 found a negative impact on outcome in patients with p53 mutations while one study26 suggested that p53 as well as RAS-pathway mutations were mainly seen in patients carrying complex karyotypes. Although molecular genetics have not been included in any existing system and were not available in a suitable number of patients in this analysis, incorporation of genomic aberrations may refine systems in the future.

References 1. Adès L, Itzykson R, Fenaux P. Myelodysplastic syndromes.

Lancet.

2014;383(9936):2239-2252.

2. Fenaux P, Mufti GJ, Hellstrom-Lindberg 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

haematologica | 2019; 104(5)

Regarding conditioning regimens, the introduction of reduced-intensity conditioning has resulted in a significant reduction of transplant-related toxicity and mortality.27 Our analysis, in line with the CIBMTR study, found at least similar effects of reduced-intensity and myeloablative regimens. As with any retrospective analysis, these results are prone to bias. Patients perceived as being at greater risk might have been favorably treated using myeloablative conditioning. In our study, patients receiving reducedintensity conditioning were even older and showed a worse performance status than patients given myeloablative conditioning. This observation is supported by a recent prospective study by the EBMT,28 in which it was found that the administration of reduced-intensity conditioning before transplantation in MDS patients resulted in at least an equivalent survival trend for a better overall survival at 2 years, whereas non-relapse mortality appeared to be higher after using myeloablative conditioning.

In conclusion, this EBMT transplant-specific risk score could improve prediction of outcome for MDS patients undergoing allogeneic stem-cell transplantation. This readily available score enables optimized clinical decisionmaking with respect to allogeneic stem-cell transplantation in patients with MDS.

III study. Lancet Oncol. 2009;10(3): 223-232.

3. 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.

4. KrĂśger N. Allogeneic stem cell transplantation for elderly patients with myelodysplastic syndrome. Blood. 2012;119(24):56325639.

5. Greenberg P, Cox C, LeBeau MM, et al.

International scoring system for evaluating prognosis in myelodysplastic syndromes.

Blood. 1997;89(6):2079-2088.

6. Greenberg PL, Tuechler H, Schanz J, et al.

Revised international prognostic scoring system for myelodysplastic syndromes. Blood.

2012;120(12):2454-2465.

7. KrĂśger N. Maximizing the benefit of allogeneic stem cell transplantation in myelodysplastic syndromes. Semin Hematol. 2017;54 (3):154-158.

935

N. Gagelmann et al.

8. Malcovati L, Hellström-Lindberg E, Bowen D, et al. Diagnosis and treatment of primary myelodysplastic syndromes in adults: recommendations from the European LeukemiaNet. Blood. 2013;122(17):29432964.

9. Della Porta MG, Alessandrino EP, Bacigalupo A, et al. Predictive factors for the outcome of allogeneic transplantation in patients with MDS stratified according to the revised IPSS-R. Blood. 2014;123(15): 2333-2342.

10. Shaffer BC, Ahn KW, Hu ZH, et al. Scoring system prognostic of outcome in patients undergoing allogeneic hematopoietic cell transplantation for myelodysplastic syndrome. J Clin Oncol. 2016;34(16):1864-1871.

11. Schanz J, Tüchler H, Solé F, et al. New comprehensive cytogenetic scoring system for primary myelodysplastic syndromes (MDS) and oligoblastic acute myeloid leukemia after MDS derived from an international database merge. J Clin Oncol. 2012;30(17): 820-829.

12. Breems DA, Van Putten WL, De Greef GE, et al. Monosomal karyotype in acute myeloid leukemia: a better indicator of poor prognosis than a complex karyotype. J Clin Oncol. 2008;26(29):4791-4797.

13. Cox DR. Regression models and life tables. J R Stat Soc B. 1972;34:187-220.

14. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(9):496509.

15. Gónen M, Heller G. Concordance probability and discriminatory power in proportional hazards regression. Biometrika. 2005;92(4): 965-970.

16. Raykar VC, Steck H, Krishnapuram B,

936

17.

18.

19.

20.

21.

22.

Dehing-Oberije C, Lambin P. On ranking in survival analysis: bounds on the concordance index. Presented at the 21 Annual Conference on Neural Information Processing Systems, Vancouver, British Columbia, Canada, December 3-6, 2007.

Koenecke C, Göhring G, de Wreede LC, et al. Impact of the revised International Prognostic Scoring System, cytogenetics and monosomal karyotype on outcome after allogeneic stem cell transplantation for myelodysplastic syndromes and secondary acute myeloid leukemia evolving from myelodysplastic syndromes: a retrospective multi-center study of the European Society of Blood and Marrow Transplantation.

Haematologica. 2015;100(3):400-408.

Teira P, Battiwalla M, Ramanathan M, et al.

Early cytomegalovirus reactivation remains associated with increased transplant-related mortality in the current era: a CIBMTR analysis. Blood. 2016;127(20):2427-2438.

Goodrich J, Bowden R, Fisher L, Keller C, Schoch G, Meyers J. Ganciclovir prophylaxis to prevent cytomegalovirus disease after allogeneic marrow transplant. Ann Intern Med. 1993;118(3):173-178.

Gagelmann N, Ljungman P, Styczynski J, Kröger N. Comparative efficacy and safety of different antiviral agents for cytomegalovirus prophylaxis in allogeneic hematopoietic cell transplantation: a systematic review and meta-analysis. Biol Blood Marrow Transplant. 2018;24(10): 2101-2109.

Marty FM, Ljungman P, Chemaly RF, et al.

Letermovir prophylaxis for cytomegalovirus in hematopoietic-cell transplantation. N Engl J Med. 2017;377(25):2433-2444.

Gagelmann N, Ayuk F, Wolschke C, Kröger

23.

24.

25.

26.

27.

28.

N. Comparison of different rabbit anti-thymocyte globulin formulations in allogeneic stem cell transplantation: systematic literature review and network meta-analysis. Biol Blood Marrow Transplant. 2017;23(12): 2184-2219.

Ljungman P, Brand R, Hoek J, et al. Donor cytomegalovirus status influences the outcome of allogeneic stem cell transplant: a study by the European Group for Blood and Marrow Transplantation. Clin Infect Dis. 2014;59(4):473-481.

Della Porta MG, Gallì A, Bacigalupo A, et al.

Clinical effects of driver somatic mutations on the outcomes of patients with myelodysplastic syndromes treated with allogeneic hematopoietic stem-cell transplantation. J Clin Oncol. 2016;34(30):3627-3637.

Lindsley RC, Saber W, Mar BG, et al.

Prognostic mutations in myelodysplastic syndrome after stem-cell transplantation. N Engl J Med. 2017;376(6):536-547.

Yoshizato T, Nannya Y, Atsuta Y, et al.

Genetic abnormalities in myelodysplasia and secondary acute myeloid leukemia: impact on outcome after stem cell transplantation. Blood. 2017;129(17):2347-2358.

Sorror ML, Sandmaier BM, Storer BE, et al.

Long-term outcomes among older patients following nonmyeloablative conditioning and allogeneic hematopoietic cell transplantation for advanced hematologic malignancies. JAMA. 2011;306(17):1874-1883.

Kröger N, Iacobelli S, Franke GN, et al.

Dose-reduced versus standard conditioning followed by allogeneic stem-cell transplantation for patients with myelodysplastic syndrome: a prospective randomized phase III study of the EBMT (RICMAC trial). J Clin Oncol. 2017;35(19):2157-2164.

haematologica | 2019; 104(5)

ARTICLE

Myeloproliferative Neoplasms

Ruxolitinib in combination with prednisone and nilotinib exhibit synergistic effects in human cells lines and primary cells from myeloproliferative neoplasms

Ferrata Storti Foundation

Alicia Arenas Cortés,1 Rosa Ayala Diaz,1 Pilar Hernández-Campo,2 Julián Gorrochategui,2 Daniel Primo,2 Alicia Robles,2 María Luz Morales,1,3 Joan Ballesteros,2 Inmaculada Rapado,1 Miguel Gallardo,*3 María Linares*1,3,4 and Joaquín Martínez-López*1,3,4 *MG, ML and JM-L contributed equally to this work

Hematology Service, Hospital Universitario 12 de Octubre; 2Vivia Biotech, Tres Cantos; H12O-CNIO Haematological Malignancies Clinical Research Unit, Centro Nacional de Investigaciones Oncológicas (CNIO) and 4Universidad Complutense de Madrid, Madrid, Spain.

1 3

Haematologica 2019 Volume 104(5):937-946

ABSTRACT

R

uxolitinib is the front-line non-palliative treatment for myelofibrosis (MF). However, a significant number of patients lose or present suboptimal response, are resistant or have unacceptable toxicity.

In an attempt to improve response and avoid the adverse effects of this drug, we evaluated the combination of 17 drugs with ruxolitinib in ex vivo models of peripheral blood mononuclear cells from MF patients and cell lines. We found that the combination ruxolitinib and nilotinib had a synergistic effect against MF cells (ΔEC50 nilotinib, -21.6%).

Moreover, the addition of prednisone to combined ruxolitinib/nilotinib improved the synergistic effect in all MF samples studied. We evaluated the molecular mechanisms of combined ruxolitinib/nilotinib/prednisone and observed inhibition of JAK/STAT (STAT5, 69.2+11.8% inhibition) and MAPK (ERK, 29.4+4.5% inhibition) signaling pathways.

Furthermore, we found that the triple therapy combination inhibited collagen protein and COL1A1 gene expression in human bone marrow mesenchymal cells. Taken together, we provide evidence that combined ruxolitinib/nilotinib/prednisone is a potential therapy for MF, possibly through the anti-fibrotic effect of nilotinib, the immunomodulatory effect of ruxolitinib and prednisone, and the anti-proliferative effect of ruxolitinib. This combination will be further investigated in a phase Ib/II clinical trial in MF.

Introduction Myelofibrosis (MF) is a Philadelphia chromosome-negative chronic myeloproliferative neoplasm (MPN) clinically characterized by stem cell-derived clonal myeloproliferation and a reactive cytokine-driven increase in bone marrow (BM) fibrosis.1,2 Patients with MF have a poor prognosis and a median survival of 5.8 years.1 Dysregulation of JAK/STAT signaling is the main cause of MPN and, accordingly, inhibitors of the JAK/STAT signal transduction pathway are currently the best clinical approach to treat this disease. Discovered in 2005,3-7 a mutation in the JAK2 gene resulting in a substitution of valine for phenylalanine (V617F) was found in approximately 90% of patients with polycythemia vera (PV), and in 50-60% of patients with essential thrombocythemia (ET) and primary myelofibrosis (PMF).8 In addition, mutations in the MPL gene, which encodes the thrombopoietin receptor, were found in approximately 1% of patients with MPN,9 and 12% of patients with MPN (35-50% of MF) have mutations in calreticulin (CALR).10-12 Interestingly, the mutated forms of CALR acquire the ability to activate the thrombopoietin haematologica | 2019; 104(5)

Correspondence: MIGUEL GALLARDO miguelgallardodelgado@gmail.com Received: July 10, 2018.

Accepted: December 10, 2018.

Pre-published: December 13, 2018. doi:10.3324/haematol.2018.201038 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/937 ©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.

937

A.A. Cortés et al.

receptor and, therefore, constitutively activate the JAK/STAT pathway.13,14 Ruxolitinib, a JAK1/JAK2 inhibitor, is the first and only drug approved by the European Medicines Agency for the treatment of PMF, post-PV MF, and post-ET MF15 and is the first-line treatment for MF. Results of the COMFORTI and II clinical trials showed that ruxolitinib produced a reduction in spleen volume, improved MF-related symptoms, and was associated with prolonged overall survival of patients compared with controls.15 Despite the beneficial effect of ruxolitinib, a high percentage of patients lose their response at some point during treatment, and others are refractory or present a suboptimal response. Because of this, the use of drug combinations might increase the effectiveness of the treatment and response time, and overcome treatment resistance.

Indeed, numerous studies have used this premise, and a number of combinations have been tested in clinical trials with varying success. For instance, whereas the combination of ruxolitinib with simtuzumab (clinicaltrials.gov identifier: 01369498) produced no clinical benefit,16 and the combination with lenalidomide (clinicaltrials.gov identifier: 01375140) had to be terminated early because the efficacy objectives were not achieved,17 the combination with danazol (clinicaltrials.gov identifier: 01732445) achieved a hematologic stabilization but did not increase the response to ruxolitinib.18 Other combinations including ruxolitinib with buparlisib (clinicaltrials.gov identifier: 01730248) or with panobinostat (clinicaltrials.gov identifiers: 01693601 and 01433445) are currently under evaluation in clinical trials. In this scenario, the objective of the present study was to develop a drug combination that enhances the effect of ruxolitinib in the treatment of MF. The proposed combination in this work, ruxolitinib, nilotinib and prednisone, is the result of testing 17 drugs with ruxolitinib to evaluate the best therapy for MF. We hypothesized that this combination would be synergic through a decrease in the proinflammatory status by ruxolitinib and prednisone19 and the known antifibrogenic effect of nilotinib,20 and would result in a better histological response.

Table 1. Drugs used in the screening to search for the best combination with ruxolitinib.

Drug Ruxolitinib Nilotinib Bosutinib Ponatinib Midostaurin Sorafenib Buparlisib Dactolisib Everolimus Sonidegib SB 431542 LCL161 Bortezomib Panobinostat HSP990 Prednisone Anagrelide Danazol

Target Jak 1/2 inhibitor PDGF-R. c-kit and BCR/ABL inhibitor Src/Abl kinase inhibitor BCR/ABL inhibitor FLT3 inhibitor Multikinase inhibitor PI3K inhibitor PI3K/Akt/mTOR inhibitor mTOR inhibitor SMO inhibitor Inhibitor of TGF-β receptor SMAC mimetic Proteosome inhibitor Deacetylase histone inhibitor HSP90 inhibitor Immunosuppressant PDE3 immunosuppressant Antigonadotropic and anti-estrogenic activity

mutation (BA/F3 V617F JAK2), and WEHI cell lines were kindly provided by Dr Quintás-Cardama (MD Anderson Cancer Center, Houston, TX, USA). The WEHI cell line, which produces IL-3, was cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (Biowest, Nuaillé, France) with 10% heat-inactivated fetal bovine serum (FBS). BA/F3 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Biowest, Nuaillé, Francia) with 10% FBS plus 10% conditioned medium from WEHI cells. The SET2 cell line (DSMZ, Braunschweig, Germany), which harbors the JAK2-V617F mutation, was cultured in RPMI 1640 with 20% FBS. The HS27a human BM mesenchymal cell line (DSMZ) was cultured in DMEM with 10% FBS.

Dose response and synergy analysis Methods Primary samples and cell lines Peripheral blood (PB) samples were collected from MF patients and from healthy donors after obtaining informed consent in accordance with the guidelines of the 12 Octubre Hospital ethics committee and the Declaration of Helsinki. The diagnosis of MF was based on 2016 World Health Organization criteria.21 PB mononuclear cells (PBMCs) were isolated from 6-10 mL of PB by density gradient centrifugation (Ficoll-Paque™ PLUS, GE Healthcare, Little Chalfont, UK) and were cultured (0.4x106 cells/mL) in Methocult TM GF_H4535 supplemented with 20 ng/mL IL-3 and 50 ng/mL Stem Cell Factor (both from StemCell Technologies, Vancouver, Canada) at 37°C in a humidified atmosphere containing 5% CO2. For the drug screening study, samples from 9 patients were used; age range was 49-83 years, there were 5 males and 4 females, and 6 of them harbored the JAK2-V617F mutation (Online Supplementary Table S1). For synergy studies, all patients (aged 66-83 years) had the JAK2-V617F mutation (3 males and 2 females). No patient had been treated previously (Online Supplementary Table S2).

The BA/F3 wild-type (BA/F3 WT), BA/F3 with JAK2-V617F 938

A total of 10,000-20,000 cells of the different cell lines were seeded per well in 96-well plates in the presence of the drugs alone (Table 1 and Online Supplementary Table S3) or in combination with ruxolitinib. Dimethyl sulfoxide (DMSO) was used as vehicle.

After 48-72 hours (h), cell viability was measured by flow cytometry with Annexin V-phycoerythrin (Biolegend, San Diego, CA, USA) or the metabolic WST-8 assay (Cell Counting Kit - 8 BioChemika; Sigma-Aldrich). Drugs were purchased from SigmaAldrich (St. Louis, MO, USA), Tocris (Bristol, UK) or kindly donated by Novartis (Basel, Switzerland).

Peripheral blood mononuclear cells were treated as follows: a) directly delivering the drugs to methylcellulose solid culture at the outset of the experiment (ex vivo, model A); or b) after 14 days of methylcellulose culture (ex vivo, model B). In model B, colonyforming cells were collected, washed with phosphate buffer saline (PBS) and cultured in RPMI with 10% FBS at 15,000 cells per well in 96-well plates in the presence of the drugs alone (Table 1) or in combination with ruxolitinib, for 72 h. DMSO was used as vehicle at a maximum concentration of 0.5%. Flow cytometry to measure myeloid cell viability was performed with monoclonal antibodies against CD45-allophycocyanin-Cy7, CD13-allophycocyanin and Annexin V-phycoerythrin (all from Biolegend, San Diego, CA, USA) using the ExviTech automated flow cytometry plataform. haematologica | 2019; 104(5)

Ruxolitinib, nilotinib and prednisone for myelofibrosis Table 2. Results of the dose-response curves of drugs in monotherapy in patients' samples after 72 hours of incubation with drugs: median. First (Q1) and third (Q3) quartiles.

Patientsâ&#x20AC;&#x2122; samples Drugs

Median

EC50 (ÂľM) Q1

Q3

Median

Emax (% Survival) Q1

Q3

Panobinostat Bortezomib Prednisolone HSP990 BKM120 Ponatinib Pomalidomide Anagrelide Bosutinib Nilotinib Danazol Sorafenib Everolimus SB431542

0.008 0.033 0.144 0.041 0.893 1.716 1.961 3.518 4.517 5.996 9.220 10.272 22.606 27.334

0.002 0.002 0.079 0.029 0.866 1.317 1.961 2.528 2.674 4.804 8.026 10.164 17.676 16.362

0.024 0.056 0.286 0.048 0.964 3.341 1.961 6.787 9.740 9.651 10.414 10.897 25.872 38.307

3.3 6.4 26.0 39.9 28.3 0.0 72.1 45.8 0.0 2.5 45.4 0.0 0.0 0.0

0.0 1.9 24.6 16.6 25.8 0.0 72.1 38.3 0.0 0.0 39.6 0.0 0.0 0.0

7.5 11.0 28.6 43.7 34.4 0.0 72.1 51.7 0.0 10.3 51.3 0.3 0.0 0.0

Collagen I expression study

Hs27a cells were treated with 100 nM ruxolitinib, 1 mM nilotinib or 1 mM prednisone or their combinations for 1 h.

Subsequently, 2 ng/mL of TGF-β (R&D Systems, Minneapolis, MN, USA) was added and cells were incubated for a further 24 h.

Immunocytochemistry and quantitative polymerase chain reaction (qPCR) analysis was used to measure collagen I expression.

inactivate endogenous peroxidase. After incubation with a peroxidase-conjugated secondary antibody for 1 h, signals were revealed with 3,3 diaminobenzidine (Abcam). Counterstaining was performed with Carazzi's hematoxylin (AppliChem Panreac, Darmstadt, Germany). Images were visualized on the Eclipse 80i (Nikon) microscope equipped with a DS-Fi1 camera (Nikon, Minato, Tokyo, Japan). Stained areas were calculated with ImageJ (Rasband, W.S., ImageJ, NIH, Bethesda, MD, USA).

Protein array and western blotting

The effects of 32 nM ruxolitinib, 1.6 mM nilotinib, 0.8 mM prednisone and their combinations, on protein phosphorylation were analyzed using the Human Phospho-kinase Array (Proteome ProfilerTM, R&D Systems) and by western blotting. Antibodies against phosphorylated or non-phosphorylated STAT5 and ERK 1/2 proteins (Cell Signaling Technology, Danvers, MA, USA) were used in western blotting analysis. Tubulin was used as a loading control and was purchased from Abcam (Cambridge Science Park, Cambridge, UK). Proteins were visualized with the ChemiDoc MP imaging system (BioRad laboratories, Hercules, CA, USA), quantified, corrected for housekeeping expression, and normalized to control samples using the ImageLab software program (v.5.1, BioRad).

Quantitative polymerase chain reaction Total RNA was prepared with the AllPrepTM DNA/RNA Micro Kit (Qiagen, Hilden, Germany). Reverse transcription reaction was carried out using the High Capacity cDNA Reverse Transcription Kit system (Life Technologies, Carlsbad, MA, USA). Real-time PCR was performed with Taqman Gene Expression Master Mix and gene-specific Taqman probe COL1A1 (Hs00164004_m1) using the 7900HT Fast Real-Time PCR Systems platform (all from Life Technologies). Normalized gene expression levels were calculated using GAPDH mRNA expression as a housekeeping gene.

Immunocytochemistry HS27a cells were fixed with 4% paraformaldehyde (Merck Millipore, Billerica, MA, USA) permeabilized and blocked with 0.25% Triton X-100 plus 1% BSA in PBS for 30 minutes (min).

Slides were incubated with antibodies against collagen I (Abcam) for one hour, followed by a 5-min incubation with 3% H2O2 to haematologica | 2019; 104(5)

Statistical analysis The analysis of drug dose-response was performed using the non-linear regression model (Equation 1): E=E0+Emaxâ&#x2C6;&#x2019;E01+ 10(logEC50â&#x2C6;&#x2019;C) where C is the drug concentration; E is the drug effect; Emax, the maximum drug efficacy in terms of survival; E0, survival when only DMSO is applied; EC50, drug concentration in which 50% of the total drug action is achieved; and Îł the slope of the curve. The area under curve (AUC) of dose response curves was also calculated.

The study of the behavior of drugs in combination was performed using Î&#x201D;EC50, the percentage of difference between EC50 of each drug in combination with ruxolitinib minus their EC50 in monotherapy.

Synergy analysis was performed using Calcusyn v.2.0 (Biosoft, Ferguson, MO, USA). The calculations performed by the program are based on the median-effect equation formulated by Chou.22 The combination index (CI) is the parameter by which the synergy or antagonism of two drugs were quantified (Equation 2):

đ??śđ??ź= đ??śđ??ˇđ??śđ??ˇ|đ?&#x2018;&#x2026;+đ??śđ?&#x2018;&#x2026;đ??śđ?&#x2018;&#x2026;|đ??ˇ where CD is the concentration of each drug, CR is the ruxolitinib concentration, and is the concentration of a drug in the presence of another drug that causes a certain effect. A CI <0.8 indicates synergism.

Shapiro-Wilk and Leveneâ&#x20AC;&#x2122;s robust test statistic were applied to evaluate normality and homoscedasticity, respectively. Linear 939

A.A. Cortés et al.

A

B

Figure 1. Effect of each drug in combination with 100 nM ruxolitinib in BA/F3 JAK2-V617F3 cell line (A) or in patients' peripheral blood mononuclear cells (B). y-axis: the increment between the EC50 for each drug in the presence of ruxolitinib minus its EC50 in monotherapy. Results are expressed as the mean±Standard Deviation (SD) of 2 independent experiments in cell lines (A) and median and interquartile range in patients' samples (B).

regression was performed for the correlation of time of response with ex vivo activity of ruxolitinib. For the statistical analysis of phospho-kinase array, an ANOVA test was performed. A t-test was used to assess whether the CI of each combination was significantly synergistic. For collagen expression assays and phosphoproteomics studies, Student t-test was used when the populations were normal and the non-parametric Wilcocox t-test when they were not. P<0.05 was considered statistically significant. Statistical analyses were performed with GraphPad Prism v.6.00 for Windows (GraphPad Software, La Jolla, CA, USA) or STATA v.13 (StataCorp., College Station, TX, USA).

Results Ruxolitinib activity in cell lines and patients’ samples We first evaluated the activity of ruxolitinib in JAK2-mutated cell lines. Ruxolitinib efficiently inhibited the viability of BA/F3 and SET2 V617F JAK2 cells with an EC50 of 35 nM and 25 nM, respectively (Online Supplementary Figure S1A). The EC50 for ruxolitinib in BA/F3 WT cells was 212 nM, indicating the importance of the JAK2-V617F mutation for the activity of ruxolitinib.

Nonetheless, when we compared the activity of ruxolitinib in patients' PBMCs with or without a JAK2 mutation, using ex vivo model A, we found that its activity was not significantly different, with an EC50 of 55 nM. For this reason, subgroups based on the mutation in JAK2 were not studied further.

To determine the best cell model to screen drugs in combination with ruxolititinb, its activity was tested in the two different ex vivo culture models. The only method that provided a sufficient number of cells for screening was model B, although the EC50 for ruxolititinb using this model was 0.747 µM. Greater ruxolitinib activity was found when PBMCs were seeded in methylcellulose in the presence of ruxolitinib (model A: EC50 = 43 nM) (Online Supplementary Table S4 and Online Supplementary Figure S1B). Moreover, if ex vivo activity of ruxolitinib was com940

pared with the time of response to ruxolitinib of each patient sample, it was found that both models A and B distinguished patients' samples with responses >6 months (Online Supplementary Figure S1C).

BCR/ABL or ABL kinase inhibitors and PDGFR and TGFβR inhibitors are effective combinations with ruxolitinib in cell lines and patients' samples To examine the best combination with ruxolitinib, dose-response curves of all tested drugs in monotherapy or in combination with ruxolitinib were first analyzed in BA/F3 V617F JAK2 cells using an automated flow cytometry platform. Drugs exhibiting the best behavior in the presence of ruxolitinib were then selected to perform the same assay using PBMCs of MF patients in ex vivo model B. Drugs with more activity in the screening were also included in dose-response assays in monotherapy with patients' samples.

Results showed that the BCR/ABL or SRC/ABL tyrosine kinase inhibitors (TKI) nilotinib and bosutinib, respectively, together with danazol, a synthetic androgen reported to reverse anemia,23 and SB432542, an inhibitor of the TGF-β receptor related to the fibrogenic processes, were among the four best combinations in BA/F3 JAK2 V617F cells (Figure 1A). Accordingly, they presented the lowest increments between their EC50 in the presence or absence of ruxolitinib (ΔEC50 nilotinib = -92.4%; ΔEC50 bosutinib = -87.7%; ΔEC50 danazol = -80.1%; ΔEC50 SB432542 = -77.1%). When tested in patients' samples, of the two BCR/ABL inhibitors, only nilotinib showed a lower EC50 in the presence of ruxolitinib than in its absence (ΔEC50 nilotinib = -21.6%), together with SB432542 (ΔEC50 = -11.7%) (Figure 1B).

Online Supplementary Table S5 shows the drugs listed by their Emax and EC50. The most active drugs in BA/F3 JAK2 V617F cells were the histone deacetylase HDAC6 inhibitor panobinostat (EC50 = 0.041mM), the proteasome inhibitor bortezomib (EC50 = 0.041 mM), and the immunomodulatory HSP90 inhibitor HSP990 (EC50 = haematologica | 2019; 104(5)

Ruxolitinib, nilotinib and prednisone for myelofibrosis

Figure 2. Screening for signaling proteins affected by 32 nM ruxolitinib (R), 1.6 µM nilotinib (N), 0.8 mM prednisone (P), and their combinations, using a phosphoprotein array. C: Control. The SET2 cell line was incubated with R, N, P or their combination for 48 hours and phosphoprotein analysis was performed.

0.045 mM). Interestingly, these inhibitors were also the most active drugs in patients' samples (Table 2), showing an EC50 of 0.008 mM, 0.033 mM and 0.041 mM, respectively. Furthermore, prednisone, an immunosuppressant used to control symptoms of MF by decreasing the levels of cytokines and growth factors such as TGF-β,24 showed an EC50 of 0.144 mM. Other active drugs in the low micromolar range included the signaling pathway inhibitors BKM120 (EC50 = 3.331 mM) and ponatinib (EC50 = 5.530 mM) (Online Supplementary Table S5). Patients' PBMCs were even more sensitive to these drugs: EC50 BKM120 = 0.893 mM, EC50 ponatinib = 1.716 mM (Table 2).

Curiously, drugs used in treatment of MF such as danazol or prednisone did not have any effect on BA/F3 JAK2 V617F viability (Online Supplementary Table S5), but prednisone was effective in combination with ruxolitinib (ΔEC50 = -13.9%) (Figure 1A). Patients' samples showed the same response (ΔEC50 = -18.0%) (Figure 1B).

Ruxolitinib/nilotinb/prednisone combination has a synergistic effect on myeloid cell lines and patient peripheral blood mononuclear cells Given the synergistic effect of nilotinib with ruxolitinib in BA/F3 JAK2 V617F cells, we next studied the effect of adding prednisone to this combination in myeloid cell lines and patients' PBMCs. As well as being a potent BCR/ABL TKI, nilotinib has been reported to inhibit the PDGF receptor, which is involved in fibrogenesis.20 A synergistic effect with all combinations of the three haematologica | 2019; 104(5)

drugs tested was found in BA/F3 JAK2 V617F and SET2 cells (Table 3). By contrast, only four of the eight doses tested were synergistic in BA/F3 wild-type (WT) cells, again suggesting an important role for the V617F mutation in the activity of the drug combination. We repeated this assay using monotherapy or combination regimens in model A cultures (PBMCs) to test the effect of the combinations in myeloid progenitors. Myelofibrosis patients' samples were sensitive to nilotinib and prednisone with an EC50 of 6.6 mM and 13.1 mM, respectively. Moreover, all combinations tested (ruxolitinib/nilotinib and ruxolitinib/niloitnib/prednisone) exhibited synergic behavior in at least two of the five patients' samples tested. In addition, the combination of 160 nM ruxolitinib, 8 mM nilotinib and 0.8 mM prednisone was synergistic in all the patients' samples tested (Table 4).

Ruxolitinib/nilotinb/prednisone combination blocks JAK/STAT and MAPK signaling The signaling pathways affected by the ruxolitinib/nilotinib/prednisone combination or monotherapy were next characterized in SET2 cells using the Proteome ProfilerTM phosphoprotein array after treatment for 30 min. Results showed that, among others, phosphorylation of STAT5 was significantly inhibited 88.94%, TOR 76.55%, ERK 87.55%, and SRC 76.88% (Figure 2) by drug combination (ANOVA ***P<0.005). Western blotting confirmed that the phosphorylation of STAT5 and ERK was inhibited by ruxolitinib in monotherapy by 69.8±8.1% and 87.7±2.3%, 941

A.A. Cortés et al.

respectively. In combination with nilotinib and prednisone, STAT5 and ERK were inhibited by 69.2±11.8% and 29.4±4.5%, respectively (Figure 3A and B). By contrast, phosphorylation of AKT was not inhibited in any case (data not shown).

gen synthesis was reduced, especially in monotherapy (79.13±20.5%) and in combination with ruxolitinib (79.20±2.3%) (Figure 4B and Online Supplementary Figure S2).

Discussion

Ruxolitinib/nilotinib/prednisone combination decreases the synthesis of collagen I in bone marrow mesenchymal cells Myelofibrosis is characterized by an increase in collagen deposition, among other fibrillar proteins, in BM, which prevents its proper functioning. To study the effect of ruxolitinib, nilotinib and prednisone (in monotherapy and in combination), on collagen expression, we utilized the HS27a mesenchymal cell line, together with TGF-β as an inductor of collagen expression. TGF-β increased the expression of COL1A1 200.9% over untreated cells (Figure 4A). Nilotinib decreased the mRNA expression of COL1A1 in HS27a cells by 60.2±0.9% in monotherapy, by 51.9±2.9% in combination with ruxolitinib and 62.2±1.9% in combination with ruxolitinib and prednisone. As a complementary test, we measured collagen expression by immunocytochemistry, finding that colla-

The management of MF remains challenging, even in the era of TKIs and personalized medicine. The discovery of the V617F mutation in JAK2 as a physiopathogenic mechanism of MPN3-7 prompted the development of JAK2 inhibitors and represented a revolution in the treatment of MF. Currently, the only approved JAK2 inhibitor for the treatment of MF and PV in the second-line is ruxolitinib,25 which has been shown to be effective in reducing hepatosplenomegaly, resolving disease-related symptoms, and producing a significant increase in overall survival when compared with conventional therapies.26 Nevertheless, there are some limitations to the use of ruxolitinib, including hematologic toxicity (anemia and thrombocytopenia) and a failure to achieve histopathological and molecular complete responses.15 Accordingly, the

Table 3. Combination Index of ruxolitinib (R), nilotinib (N) and prednisone (P) in samples of cell lines. SET2: BA/F3 JAK2 wt or V617F cell lines were incubated with R, N, P or their combination for 48 hours and then Wst8 was performed.

Ruxolitinib (nM) 6.4 6.4 32 32 6.4 6.4 6.4 6.4 32 32 32 32

Ruxolitinib (nM) 32 32 160 160 32 32 32 32 160 160 160 160

Prednisone (mM)

CI Mean

BAF3 JAK2 V617F SEM

320 1.6 0.32 1.6 0.32 0.32 1.6 1.6 0.32 0.32 1.6 1.6

0.16 0.8 0.16 0.8 0.16 0.8 0.16 0.8

>2 >2 >2 >2 0.067 0.198 0.082 0.095 0.049 0.040 0.051 0.063

Nilotinib (mM)

Prednisone (mM)

0.16 0.8 0.16 0.8 0.16 0.8 0.16 0.8

Nilotinib (mM)

1.6 8 1.6 8 1.6 1.6 8 8 1.6 1.6 8 8

P

CI Mean

9.47 3.35 12.49 187.51 0.02 0.12 0.03 0.03 0.02 0.01 0.01 0.01

* ** *** *** *** *** *** ***

NA NA NA NA NA NA 0.311 0.172 NA 0.198 NA 0.248

CI Mean

SET2 SEM

P

0.815 0.596 1.374 0.857 0.538 0.448 0.396 0.337 0.155 0.158 0.173 0.153

0.261 0.189 0.144 0.109 0.162 0.057 0.089 0.040 0.015 0.010 0.011 0.011

* * *** *** *** ***

BAF3 JAK2 wt CI SEM

P

0.192 0.090 0.086

**

0.060

**

NA: drugs combination with an effect less than of 20% compared to control. Combination Index (CI) < 0.9 indicates synergy (Italic); CI > 1.1 indicates antagonism (Italic Bold); CI from 0.9 to 1.1 indicates additivity (Bold). *P<0.05; **P<0.01; ***P<0.001.

942

haematologica | 2019; 104(5)

Ruxolitinib, nilotinib and prednisone for myelofibrosis

unmet clinical need to increase the effectiveness of the treatment and decrease its toxicity guides the search for combination treatments with ruxolitinib.

The main objective of this work was to evaluate the best combination of drugs for the treatment of MF, maintaining ruxolitinib as a therapeutic base and reducing its toxicity while maintaining its efficacy. Since MF is not characterized by large amounts of pathological cells, as in acute myeloid leukemia,27 it is challenging to develop

an ex vivo model to screen 17 combinations.

Consequently, we elected to utilize myeloid cells obtained from 14-day old cultures of PBMCs in methylcellulose (ex vivo model B), which produce a sufficient supply of cells for screening. Interestingly, the most active drugs in monotherapy coincided with the most active drugs in preclinical trials, including bortezomib,28 panobinostat29 and HSP990.30 Therefore, the search for combination treatments with ruxolitinib is in response to

A

B

Figure 3. Effect of 32 nM ruxolitinib (R), 1.6 ÂľM nilotinib (N), 0.8 mM prednisone (P), and their combinations, on the phosphorylation of ERK1/2 and STAT5 in cell lines. Results are expressed as the meanÂąStandard Error of Mean. Data are representative of at least 2 separate experiments. *P<0.05; **P<0.01; ***P<0.001.

C: control.

Table 4. Combination Index (CI) of ruxolitinib (R), nilotinib (N) and prednisone (P) in samples of myelofibrosis (MF) patients and healthy donors.

Mononuclear cells from peripheral blood of MF patients and healthy donors were seeded at 200,000-500,000 cell/mL in Metocult supplemented with SCF and IL3 in presence of drugs for two weeks. Then flow cytometry analysis was performed.

Drugs Ruxolitinib (nM) 32 32 160 160 32 32 160 160

Nilotinib (mM) 1.6 8 1.6 8 1.6 1.6 8 8

Prednisone (mM)

P20

P33

MF P27

P19

P34

0.8 4 0.8 4

>2 1.44 1.83 1.55 >2 >2 0.31 0.39

0.50 0.18 1.05 0.4 >2 >2 0.64 >2

1.09 0.30 0.32 0.22 0.36 0.15 0.20 0.13

0.40 0.34 0.20 0.20 0.43 0.24 0.10 0.18

0.22 0.55 0.16 0.05 0.10 0.02 -

Healthy donors C1 0.24 0.12 0.30 0.21 >2 0.53 0.48 0.54

C3 0.20 0.13 >2 0.13 >2 >2 >2 >2

CI < 0.9 indicates sinergy (Italic); CI > n 1.1 indicates antagonism (Italic Bold). P<0.05 is considered significative.

haematologica | 2019; 104(5)

943

A.A. Cortés et al.

A

B

Figure 4. Effect of ruxolitinib (R), nilotinib (N), prednisone (P), and their combination, on collagen I mRNA expression using quantitative polymerase chain reaction or protein expression by immunocytochemistry. HS27a cell line was incubated with R, N, P or their combination for 1 hour (h) and then for 24 h with 2 ng/mL TGF-β. Results were expressed as the mean±Standard Error of Mean.

Data are representative of at least 2 independent experiments. *P<0.05; **P<0.01.

the need to increase the effectiveness of the treatment and decrease its toxicity.

We found that the best combinations were those with BCR/ABL inhibitors, nilotinib and bosutinib, used for the treatment of chronic myeloid leukemia (CML).31 Nevertheless, TKIs such as perifosine or BKM120, corticosteroids such as prednisone, androgens such as danazol, and the TGF-β receptor inhibitor SB431542, also decreased the EC50 in combination with ruxolitinib. It is interesting to note that some of these combinations have already been tested in clinical trials, such as ruxolitinib plus danazol (clinicaltrials.gov identifier: 01732445);18 this study showed that while there was no improvement in hematologic response, stabilization of the patients was achieved.

Regarding ruxolitinib plus BKM120 (clinicaltrials.gov identifier: 01730248), although no results have yet been reported, preliminary analyses (ASH 2015) are not encouraging.

Interestingly, it has previously been described that the combination nilotinib plus ruxolitinib can eliminate CD34+ leukemic progenitors in CML32 and Philadelphia chromo944

some positive acute lymphoblastic leukemia;33 however, it was not known whether this also holds for MF. We show here synergistic behavior of nilotinib plus ruxolitinib, which blocks colony formation in clonogenic assays with patients' PMBCs, and inhibits the phosphorylation of both STAT5 and ERK 1/2. Furthermore, the inclusion of prednisone is key to achieve synergy in the survival assays of cell lines and increases the power of synergy in patients' samples. In addition, this combination inhibited the synthesis of collagen in BM mesenchymal cells, which is important to achieve histopathological response. The antifibrogenic effect of nilotinib has been previously described in skin cells,20,34 liver35 and muscle,36 via its ability to inhibit the PDGF receptor, which is directly involved in the induction of collagen synthesis. In addition, ruxolitinib is able to stabilize, or even ameliorate, fibrosis through its ability to reduce the proinflamatory state, which is typical in MF.37 Similar results are seen with corticosteroids like prednisone, which decreases the levels of cytokines and proinflammatory growth factors including TGF-β.24 haematologica | 2019; 104(5)

Ruxolitinib, nilotinib and prednisone for myelofibrosis

Our results indicate that the combination ruxolitinib, nilotinib and prednisone would eliminate more efficiently pathological cells, stopping and/or reverting MF. It must be remembered, however, that this evidence was obtained in vitro and the impact on the clinical situation still needs to be proved. With this is mind, and given that all the drugs are approved for clinical practice, we have recently initiated a clinical trial (the RuNic Trial; clinicaltrials.gov identifier: 02973711) which does not require an in vivo analysis in animal models.

In summary, MF is a complex disease in which alterations in tyrosine kinase-related signaling participates in the amplification of hematopoietic clones and in the increased production of cytokines and growth factors by pathological cell clones, which stimulates MF.38 It is often advantageous to use combinations of drugs that target different pathways involved in the pathophysiology of a disease, shown here with the objective of decreasing fibrosis of the BM, and not only eradicating the tumor clone, as previously attempted.15,30,39,40 Our results lead us to hypothesize that the combination therapy ruxolitinib and prednisone might provide a dampened proinflammatory environment and nilotinib would block fibrosis. Accordingly,

References 1. Tefferi A, Vardiman JW. Classification and diagnosis of myeloproliferative neoplasms: The 2008 World Health Organization criteria and point-of-care diagnostic algorithms.

Leukemia. 2008;22(1):14-22.

2. Tefferi A. How I treat myelofibrosis. Blood.

2011;117(13):3494-3504.

3. James C, Ugo V, Le Couédic J-P, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera.

Nature.

2005; 434(7037):1144-1148.

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. Zhao R, Xing S, Li Z, et al. Identification of an acquired JAK2 mutation in polycythemia vera.

J Biol Chem.

2005;280(24):22788-22792.

7. 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.

8. Vainchenker W, Kralovics R. Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms. Blood.

2017;129(6):667-679.

9. Ma W, Zhang X, Wang X, et al. MPL mutation profile in JAK2 mutation-negative patients with myeloproliferative disorders.

Diagn Mol Pathol. 2011;20(1):34-39.

10. Klampfl T, Gisslinger H, Harutyunyan AS, et al. Somatic mutations of calreticulin in myeloproliferative neoplasms. N Engl J Med. 2013;369(25):2379-2390.

11. Nangalia J, Massie CE, Baxter EJ, et al.

Somatic CALR mutations in myeloproliferative neoplasms with nonmutated JAK2. N Engl J Med. 2013;369(25):2391-2405.

haematologica | 2019; 104(5)

the combination ruxolitinib/nilotinib/prednisone is configured as a therapeutic strategy against MF that aims to enhance the effect of ruxolitinib, and promote the reduction of fibrosis in the BM and reduce inflammation. As mentioned above, this combination will be studied in a phase Ib/II clinical trial in MF (the RuNic Trial; clinicaltrials.gov identifier: 02973711).

Further information is available in the Online Supplementary Appendix.

Funding This study was supported by the Subdirección General de Investigación Sanitaria (Instituto de Salud Carlos III, Spain) grants PI13/02387 and PI16/01530, and the CRIS against Cancer foundation grant 2014/0120. M.L. holds a postdoctoral fellowship of the Spanish Ministry of Economy and Competitiveness (FPDI- 2013-16409).

Acknowledgments The authors would like to thank to Carmen Delgado, from H12O Hematology Department for the support with the samples, Kenneth McCreath for language support, and to all patients who participated in the study.

12. Ha J-S, Kim Y-K. Calreticulin exon 9 mutations in myeloproliferative neoplasms. Ann Lab Med. 2015;35(1):22-27.

13. Marty C, Pecquet C, Nivarthi H, et al.

Calreticulin mutants in mice induce an MPL-dependent thrombocytosis with frequent progression to myelofibrosis. Blood.

2016;127(10):1317-1324.

14. Chachoua I, Pecquet C, El-Khoury M, et al.

Thrombopoietin receptor activation by myeloproliferative neoplasm associated calreticulin mutants. Blood. 2016; 127(10):1325-1335.

15. Verstovsek S, Gotlib J, Mesa RA, et al.

Long-term survival in patients treated with ruxolitinib for myelofibrosis: COMFORT-I and -II pooled analyses. J Hematol Oncol.

2017;10(1):156.

16. Verstovsek S, Savona MR, Mesa RA, et al.

A phase 2 study of simtuzumab in patients with primary, post-polycythaemia vera or post-essential thrombocythaemia myelofibrosis. Br J Haematol. 2017;176(6):939-949.

17. Daver N, Cortes J, Newberry K, et al.

Ruxolitinib in combination with Lenalidomide as therapy for patients with myelofibrosis.

Haematologica 2015; 100(8):1058-1063.

18. Gowin K, Kosiorek H, Dueck A, et al.

Multicenter phase 2 study of combination therapy with ruxolitinib and danazol in patients with myelofibrosis. Leuk Res 2017;6031-6035.

19. Mascarenhas J, Hoffman R. A comprehensive review and analysis of the effect of ruxolitinib therapy on the survival of patients with myelofibrosis. Blood. 2013; 121(24):4832-4837.

20. Akhmetshina A, Dees C, Pileckyte M, et al.

Dual inhibition of c-abl and PDGF receptor signaling by dasatinib and nilotinib for the treatment of dermal fibrosis. FASEB J. 2008;22(7):2214-2222.

21. 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.

22. Chou T-C. Theoretical Basis, Experimental Design, and Computerized Simulation of Synergism and Antagonism in Drug Combination Studies. Pharmacol Rev. 2006;58(3):621-681.

23. Cervantes F, Isola IM, Alvarez-Larrán A, Hernández-Boluda J-C, Correa J-G, Pereira A. Danazol therapy for the anemia of myelofibrosis: assessment of efficacy with current criteria of response and long-term results. Ann Hematol. 2015;94(11):17911796.

24. Yu W, Guo F, Song X. Effects and mechanisms of pirfenidone, prednisone and acetylcysteine on pulmonary fibrosis in rat idiopathic pulmonary fibrosis models.

Pharm Biol. 2017;55(1):450-455.

25. Vainchenker W, Leroy E, Gilles L, Marty C, Plo I, Constantinescu SN. JAK inhibitors for the treatment of myeloproliferative neoplasms and other disorders. F1000Res. 2018;7:82.

26. Vannucchi AM, Kantarjian HM, Kiladjian JJ, et al. A pooled analysis of overall survival in COMFORT-I and COMFORT-II, 2 randomized phase III trials of ruxolitinib for the treatment of myelofibrosis.

Haematologica. 2015;100(9):1139-1145.

27. Bennett TA, Montesinos P, Moscardo F, et al. Pharmacological Profiles of Acute Myeloid Leukemia Treatments in Patient Samples by Automated Flow Cytometry: A Bridge to Individualized Medicine. Clin Lymphoma Myeloma Leuk. 2014; 14(4):305-318.

28. Wagner-Ballon O, Pisani DF, Gastinne T, et al. Proteasome inhibitor bortezomib impairs both myelofibrosis and osteosclerosis induced by high thrombopoietin levels in mice. Blood. 2007;110(1):345-353.

29. Evrot E, Ebel N, Romanet V, et al. JAK1/2 and Pan-deacetylase inhibitor combination therapy yields improved efficacy in preclinical mouse models of JAK2V617F-driven disease. Clin Cancer Res. 2013; 19(22):6230-6241.

30. Marubayashi S, Koppikar P, Taldone T, et al. HSP90 is a therapeutic target in JAK2-

945

A.A. Cortés et al.

31.

32.

33.

34.

946

dependent myeloproliferative neoplasms in mice and humans. J Clin Invest. 2010; 120(10):3578-3593.

Green MR, Newton MD, Fancher KM. Off-Target Effects of BCR-ABL and JAK2 Inhibitors. Am J Clin Oncol. 2016;39(1): 76–84.

Gallipoli P, Cook A, Rhodes S, et al.

JAK2/STAT5 inhibition by nilotinib with ruxolitinib contributes to the elimination of CML CD34+ cells in vitro and in vivo.

Blood.2014;124(9):1492-1501.

Kong Y, Wu Y-L, Song Y, et al.

Ruxolitinib/nilotinib cotreatment inhibits leukemia-propagating cells in Philadelphia chromosome-positive ALL. J Transl Med. 2017;15(1):184.

Luchetti MM, Moroncini G, Jose Escamez

M, et al. Induction of Scleroderma Fibrosis in Skin-Humanized Mice by Administration of Anti-Platelet-Derived Growth Factor Receptor Agonistic Autoantibodies. Arthritis Rheumatol. 2016; 68(9):2263-2273.

35. Liu Y, Wang Z, Kwong SQ, et al. Inhibition of PDGF, TGF-β, and Abl signaling and reduction of liver fibrosis by the small molecule Bcr-Abl tyrosine kinase antagonist Nilotinib. J Hepatol. 2011;55(3):612-625.

36. Lemos DR, Babaeijandaghi F, Low M, et al.

Nilotinib reduces muscle fibrosis in chronic muscle injury by promoting TNF-mediated apoptosis of fibro/adipogenic progenitors.

Nat Med. 2015;21(7):786-794.

37. Massaro F, Molica M, Breccia M. How ruxolitinib modified the outcome in myelofi-

brosis: focus on overall survival, allele burden reduction and fibrosis changes. Expert Rev Hematol. 2017;10(2):155-159.

38. Lataillade J-J, Pierre-Louis O, Hasselbalch HC, et al. Does primary myelofibrosis involve a defective stem cell niche? From concept to evidence. Blood. 2008;112(8): 3026-3035.

39. Barosi G, Gattoni E, Guglielmelli P, et al.

Phase I/II study of single-agent bortezomib for the treatment of patients with myelofibrosis. Clinical and biological effects of proteasome inhibition. Am J Hematol.

2010;85(8):616-619.

40. Tefferi A, Lasho TL, Begna KH, et al. A Pilot Study of the Telomerase Inhibitor Imetelstat for Myelofibrosis. N Engl J Med. 2015;373(10):908-919.

haematologica | 2019; 104(5)

ARTICLE

Myeloproliferative Neoplasms

EXPAND, a dose-finding study of ruxolitinib in patients with myelofibrosis and low platelet counts: 48-week follow-up analysis

Ferrata Storti Foundation

Alessandro M. Vannucchi,1 Peter A. W. te Boekhorst,2 Claire N. Harrison,3 Guangsheng He,4 Marianna Caramella,5 Dietger Niederwieser,6 Françoise Boyer-Perrard,7 Minghui Duan,8 Nathalie Francillard,9 Betty Molloy,10 Monika Wroclawska10 and Heinz Gisslinger11

Center for Research and Innovation of Myeloproliferative Neoplasms, Azienda Ospedaliero Universitaria Careggi, University of Florence, Italy; 2Erasmus Medical Center, Rotterdam, the Netherlands; 3Guy’s and St Thomas’ NHS Foundation Trust, Guy’s Hospital, London, UK; 4No. 1 Hospital of Nanjing Medical University, China; 5ASST Grande Ospedale Metropolitano, Milano, Italy; 6Department of Hematology and Medical Oncology, University of Leipzig, Germany; 7Centre Hospitalier Universitaire d’Angers, France; 8Peking Union Medical College Hospital, Beijing, China; 9Novartis Pharma S.A.S, Rueil Malmaison, France; 10Novartis AG, Basel, Switzerland and 11Department of Medicine I, Division of Hematology and Hemostaseology, Medical University of Vienna, Austria 1

Haematologica 2019 Volume 104(5):947-954

ABSTRACT

E

EXPAND (phase Ib, dose-finding study) evaluated the starting dose of ruxolitinib in patients with myelofibrosis with baseline platelet counts of 50-99×109/L. The study consisted of dose-escalation and safety-expansion phases. Based on the baseline platelet counts, patients were assigned to stratum 1 (75-99x109/L) or stratum 2 (50-74x109/L), with the primary objective of determining the maximum safe starting dose (MSSD); key secondary objectives included safety and efficacy. At week 48 data cutoff (stratum 1, n=44; stratum 2, n=25), 24.6% (17 out of 69) of patients were still receiving treatment. The MSSD was established as ruxolitinib 10 mg twice daily in both strata. Thrombocytopenia [grade 4 (stratum 1, n=1; stratum 2, n=2)] was the only reported dose-limiting toxicity (study drug related) at 10 mg twice daily. In the MSSD cohort (stratum 1, n=20; stratum 2, n=18), adverse events (regardless of study drug relationship) led to treatment discontinuation in 15.0% and 33.3% of patients in stratum 1 and stratum 2, respectively, and dose adjustment/interruption in 45.0% and 66.7% of patients in stratum 1 and stratum 2, respectively. Three cases of on-treatment deaths were reported at the MSSD. Spleen response was achieved at week 48 in 33.3% and 30.0% of patients in stratum 1 and stratum 2, respectively.

Improvements in the Total Symptom Score were also observed. In this study, ruxolitinib demonstrated acceptable tolerability in both the strata at the MSSD of 10 mg twice daily. (Registered at: clinicaltrials.gov identifier: 01317875).

Introduction Myelofibrosis (MF) is a rare, chronic, Philadelphia chromosome-negative myeloproliferative neoplasm caused by clonal proliferation of pluripotent hematopoietic stem cells.1,2 The common clinical presentations associated with MF include splenomegaly due to extramedullary hematopoiesis, progressive bone marrow fibrosis with cytopenias, and debilitating constitutional symptoms (e.g., fatigue, night sweats, and fever), which substantially diminish the quality of life.3-6 The dysregulated activation of the Janus kinase (JAK)/signal transducer and activator of transcription pathway is the hallmark of MF and can result from mutations in JAK2, in the cytokine receptor, or in other components of the signaling pathway.7,8 Ruxolitinib, a potent and selective oral JAK1/JAK2 inhibitor, was approved for the treatment of intermediate- and high-risk patients with MF based on two randomized, phase III studies: COMFORT-I (n=309; clinicaltrials.gov identifier: 00952289) and haematologica | 2019; 104(5)

Correspondence: ALESSANDRO M. VANNUCCHI amvannucchi@unifi.it Received: August 16, 2018.

Accepted: November 13, 2018.

Pre-published: November 15, 2018. doi:10.3324/haematol.2018.204602 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/947 ©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.

947

A.M. Vannucchi et al.

COMFORT-II (n=219; clinicaltrials.gov identifier: 00934544).9-12 In both COMFORT studies, ruxolitinib demonstrated marked and sustained clinical benefits in spleen size and improvement in symptom burden, and was generally well tolerated.9-12 Patients with MF may present with thrombocytopenia (platelet counts, <100x109/L) owing to the nature of the disease.13,14 Across various studies, approximately 16-26% of patients with MF were found to be thrombocytopenic at diagnosis.15-19 In addition, patients with MF may also experience treatment-emergent thrombocytopenia while on ruxolitinib therapy (owing to its mechanism of action).9,12,20 The Study 251 (n=153; clinicaltrials.gov identifier: 00509899) identified thrombocytopenia as a dose-limiting toxicity (DLT) for ruxolitinib.21 Patients in the COMFORT studies had baseline platelet counts ≥100x109/L, limiting the safety and efficacy data in patients with lower platelet counts.14,18,21 To date, only a few treatment options have been evaluated for patients with MF and thrombocytopenia.

The safety and efficacy of JAK inhibitors in thrombocytopenic patients with MF have also not been adequately explored.15,22 Evidence from clinical trials evaluating the use of ruxolitinib in patients with MF with baseline thrombocytopenia (platelet counts <100x109/L) is limited.14,23-25 The Study 258 (n=50; clinicaltrials.gov identifier: 01348490) evaluated the efficacy and safety of low-dose ruxolitinib [5 mg twice daily (bid)] with subsequent dose escalation in patients with low platelet counts (50 to <100x109/L).14 Ruxolitinib was generally well tolerated and provided efficacy benefits, suggesting that a starting dose of 5 mg bid with escalation to 10 mg bid may be suitable for the low platelet count population.14 The JUMP study (n=2233; clinicaltrials.gov.identifier: 01493414), a phase IIIb expandedaccess study, was amended to enroll patients with baseline platelet counts ≥50x109/L to gather additional safety and efficacy data in patients with low platelet counts.23-25 In the JUMP study, the safety profile of ruxolitinib in the low-platelet patient cohort was consistent with that observed in patients with platelet counts ≥100x109/L. Spleen and symptom responses achieved with low-dose ruxolitinib (5 mg bid) were within the expected range based on the COMFORT studies.23 The recommended starting dose of ruxolitinib (prescribing information) is based on the platelet count.26 The maximum recommended starting dose in patients with platelet counts between 50x109/L and 100x109/L is 5 mg bid, and the dose should be titrated with caution.26,27 However, the findings from the COMFORT-I and the Study 258 demonstrated that the final titrated doses of ≥10 mg bid resulted in larger improvements in spleen volume and MF-related symptoms compared to titrated doses of ≤5 mg bid.14,28 The purpose of the EXPAND study (clinicaltrials.gov identifier: 01317875: open-label, phase Ib, dose-finding study) was to establish the maximum safe starting dose (MSSD) of ruxolitinib in patients with MF with baseline platelet counts between 50x109/L and 100x109/L. The study also intended to assess the safety and tolerability of ruxolitinib in this patient population.

The preliminary findings from the dose-escalation and safety-expansion phases of EXPAND at the preplanned interim analysis [day 168 (week 24)] were previously 948

reported.29 Guided by the occurrence of protocol-defined DLTs during the first cycle of treatment (28 days), 15 mg bid was initially declared as the MSSD for patients enrolled in stratum 1 (S1; platelet counts: 75-99×109/L) of the study, whereas 10 mg bid was declared as the MSSD for patients in stratum 2 (S2; platelet counts: 50-74×109/L).

However, based on the safety and efficacy findings from the interim analysis, the MSSD for S1 was subsequently lowered to 10 mg bid (as per the protocol amendment).

Here, we present the results from the 48-week follow up of EXPAND for the MSSD cohorts.

Methods Patient population

Eligible patients: i) were aged ≥18 years; ii) had been diagnosed with intermediate-1, intermediate-2, or high-risk MF (primary MF, post-polycythemia vera MF, or post-essential thrombocythemia MF);30 iii) had a palpable spleen (≥5 cm from the costal margin); and iv) fulfilled the platelet count criteria at screening or study day 1 (S1: <100x109/L and ≥75x109/L; S2: <75x109/L and ≥50x109/L).

An Eastern Oncology Cooperative Group performance status of ≤2 was required at screening. The key exclusion criteria included: i) patients with any history of platelet counts <45x109/L within 30 days prior to screening; ii) platelet transfusion within 14 days prior to screening; iii) history or predisposition to clinically significant bleeding; iv) history of platelet dysfunction and/or bleeding diathesis; and v) regular use of drugs inhibiting platelet function.

Study design EXPAND was a phase Ib, open-label, multicenter, dose-finding study of ruxolitinib in patients with intermediate- or high-risk primary MF, post-polycythemia vera MF, or post-essential thrombocythemia MF who had baseline platelet counts between ≥50x109/L and <100x109/L. The study design is shown in Figure 1.

The study period consisted of 2 phases: dose escalation and safety expansion. The successive cohorts of newly enrolled patients received increasing doses of ruxolitinib until the MSSD was determined in the dose-escalation phase. The MSSD was defined as the dose level most closely associated with a posterior probability of DLT between 16% and 33% that did not also have >25% probability of excessive toxicity. A DLT was defined as the occurrence of any treatment-related toxicity occurring through study day 28 (Online Supplementary Table S1). A preplanned interim analysis was conducted when the last patient enrolled in the dose-escalation phase completed week 24.29 An adaptive Bayesian logistic regression model guided by escalation with overdose control was used to allocate patients into each cohort (5 dose levels) in the dose-escalation phase (Treatment Dose Levels) (Online Supplementary Appendix). The patients in the dose-determining set (DDS) enrolled in the dose-finding part of the study were assessed to determine the MSSD. The DDS consisted of all patients from the safety set who met the minimum exposure criterion and had sufficient safety evaluations or who experienced a DLT. The safety set consisted of all patients who received at least 1 dose of ruxolitinib. The DDS definition, minimum exposure criterion, and planned enrollment are provided in the Online Supplementary Appendix.

The safety-expansion phase was conducted after determination of the MSSD to further evaluate the safety and tolerability of the MSSD, and establish that the dose was suitable for use in patients with MF with low platelet counts. Per protocol amendment, 10 mg bid was evaluated as the starting dose for all new patients haematologica | 2019; 104(5)

EXPAND: 48-week follow-up analysis

Figure 1. Study design. Dark arrows represent escalation from a given dose level to the following one, only if both that dose level and the previous one have been deemed safe. Dotted arrows represent each dose level in stratum 2, which will open to patients only if both that dose level and the following one have been deemed safe in stratum 1. Per protocol amendment, new patients enrolled in stratum 1 in the safety-expansion phase will be given the 10 mg twice-daily (bid) dose instead of the 15 mg bid dose level previously evaluated as the maximum safe starting dose (MSSD). The MSSD cohort (10 mg bid) in stratum 1 included 3 patients from the dose-escalation and 17 patients from the safety-expansion phases. In stratum 2, the MSSD cohort included 8 patients from the dose-escalation and 10 patients from the safety-expansion phases.

enrolled in S1 in the safety-expansion phase (Online Supplementary Appendix). Patients who were already receiving the 15 mg bid dose continued to take their assigned dose.

The end of the study will occur after all study patients complete their last assessment as per protocol (follow-up visit 30 days after the end of the treatment visit) (Online Supplementary Appendix).

Details of the statistical analyses are presented in the Online Supplementary Appendix.

The study was approved by the institutional review boards of the respective institutions prior to patient enrollment and was conducted in accordance with the principles of the Declaration of Helsinki. All patients provided written informed consent. The trial is registered at clinicaltrials.gov identifier: 01317875.

Assessments In the dose-escalation phase, the primary objective was to determine the MSSD (incidence rate of DLTs) of ruxolitinib. The key secondary objectives included safety [frequency, duration, and severity of adverse events (AEs) and serious AEs] and efficacy (spleen response: proportion of patients achieving ≥50% of reduction in palpable measurement of spleen length at week 48 data cutoff relative to day 1). The key exploratory objectives included patient-reported outcomes [change in the Total Symptom Score (TSS) as assessed by the modified Myelofibrosis Symptom Assessment Form (MFSAF) v.2.0 diary].31-33

Overall study cohort At week 48 data cutoff (December 7, 2017), the final enrollment for EXPAND included 69 patients (S1, n=44; S2, n=25) (Online Supplementary Table S2). Overall, 31.8% (14 out of 44) of patients in S1 and 12.0% (3 out of 25) of patients in S2 were still receiving ruxolitinib treatment (Online Supplementary Table S3). The median exposure to ruxolitinib was 51.4 weeks (range, 0.9-210.0 weeks) in S1 and 67.4 weeks (range, 4.4-161.1 weeks) in S2.

The AEs (in ≥20% of patients in either stratum, regardless of study drug relationship) reported in the overall cohort are presented in Online Supplementary Table S4. Reasons for on-treatment death included acute myeloid leukemia (1 patient), cardiac arrest (1 patient), and unknown (1 patient, not suspected to be related to study drug) in S1 and complications following gastrointestinal ulcer (1 patient) and multiple organ failure (1 patient) in S2 (Online Supplementary Table S5). Hemoglobin levels and platelet counts over time are presented in Figure 2. An initial decrease in the blood count parameters was observed in the first few weeks; however, the parameters stabilized with time. Spleen response at week 48 was achieved in 7 out of 22 patients [31.8% (95%CI: 13.9, 54.9)] in S1 and 5 out of 14 patients [35.7% (95%CI: 12.8, 64.9)] in S2. A spleen response at any time point was observed in 22 out of 43 patients [51.2% (95%CI: 35.5, 66.7)] in S1 and 17 out of 25 patients [68.0% (95%CI: 46.5, 85.1)] in S2 (Online Supplementary Figure S1).

Maximum safe starting dose cohort Results Results from the interim analysis (week 24 data cutoff: January 20, 2015) of the study have been presented at the 2015 American Society of Hematology meeting.29 At that data cutoff, 46 patients (S1, n=27; S2, n=19) had received treatment. haematologica | 2019; 104(5)

Patients’ characteristics. Baseline patients’ characteristics (S1, n=20; S2, n=18) were indicative of an advanced disease stage (Online Supplementary Table S6). In the MSSD cohort, 70.0% (14 out of 20) of patients in S1 and 16.7% (3 out of 18) of patients in S2 were still receiving ruxolitinib treatment (Table 1). The primary reasons for the end of treatment included AEs [S1, n=1 (5.0%); S2, 949

A.M. Vannucchi et al.

A

B

Figure 2. Blood parameters over time. (A) Hemoglobin levels. (B) platelet counts. BL: baseline.

n=4 (22.2%)], treatment duration completed [S1, n=0; S2, n=3 (16.7%]), physician decision [S1, n=1 (5.0%); S2, n=3 (16.7%)], disease progression [S1, n=3 (15.0%); S2, n=1 (5.6%)], and death [S1, n=0; S2, n=2 (11.1%)].

Dosing and exposure. The median exposure to ruxolitinib was 54.8 weeks (range, 4.3-210.0 weeks) in S1 and 83.2 weeks (range, 4.4-161.1 weeks) in S2. Overall, 45.0% (9 out of 20) of patients in S1 and 88.9% (16 out of 18) of patients in S2 had at least 1 dose reduction/interruption (Online Supplementary Table S7). The mean total daily dose over time plot by stratum at MSSD (10 mg bid in both S1 and S2) is shown in Figure 3. The mean dose intensity was 17.96 mg/day [standard deviation (SD)=3.055] in S1 and 13.27 mg/day (SD=5.030) in S2.

Dose modifications were observed during the first 12 weeks of treatment in some patients; 30.0% (6 out of 20) of patients in S1 and 61.1% (11 out of 18) of patients in S2 had at least 1 dose reduction/interruption (Online Supplementary Table S8). Among these patients with a dose down-titration, 3 of 6 patients in S1 and 10 of 11 patients in S2 did not resume the initial 10 mg bid dose.

Thrombocytopenia was the most frequent AE leading to an early dose titration.

Safety in the maximum safe starting dose cohort

Adverse events (in â&#x2030;Ľ15% of patients in either stratum, regardless of study drug relationship) in the MSSD cohort are presented in Table 2. As observed in the interim analysis, anemia and thrombocytopenia (all grades) were the most common hematologic AEs in both strata [anemia: S1, n=9 (45.0%); S2, n=8 (44.4%) and thrombocytopenia: S1, n=8 (40.0%); S2, n=14 (77.8%)].

Grade 3 or 4 AEs were reported in 70.0% (14 out of 20) of patients in S1 and 88.9% (16 out of 18) of patients in S2. The AEs [regardless of study drug relationship; any system organ class (SOC)] led to treatment discontinuation in 15.0% (3 out of 20) of patients in S1 and 33.3% (6 out of 18) of patients in S2, with thrombocytopenia being reported as the most common reason for treatment discontinuation [S1, n=1 (5.0%); S2, n=3 (16.7%)]. Dose adjustment or study drug interruption due to AEs [regardless of study drug 950

Table 1. Patient disposition (week 48 analysis; maximum safe starting dose cohort).

Patient disposition

Stratum 1 (N=20) Stratum 2 (N=18) N (%) N (%)

Patients treated End of treatment Treatment ongoinga Primary reason for end of treatment AE Completed Death Other Physician decision Progressive disease Withdrawal by patient

6 (30.0) 14 (70.0)

15 (83.3) 3 (16.7)

1 (5.0) 0 0 1 (5.0) 1 (5.0) 3 (15.0) 0

4 (22.2) 3 (16.7) 2 (11.1) 0 3 (16.7) 1 (5.6) 2 (11.1)

AE: adverse event. aPatients under ongoing treatment at the time of data cutoff (December 7, 2017).

relationship (any SOC)] was observed in 45.0% (9 out of 20) of patients in S1 and 66.7% (12 out of 18) of patients in S2.

Thrombocytopenia was the primary reason for dose adjustment/interruption in both strata [S1, n=4 (20.0%); S2, n=12 (66.7%)]. Overall, 25% (5 out of 20) of patients in S1 and 38.9% (7 out of 18) of patients in S2 experienced a serious AE [regardless of study drug relationship (any SOC)]. Thrombocytopenia (grade 4, related to study drug) was the only DLT reported in both strata at 10 mg bid [S1, n=1; S2, n=2 (1 DLT in S2 was reported at the interim analysis)]. Grade 4 worsening from baseline in platelet count was observed in 1 patient (5.0%) in S1 and 7 patients (38.9%) in S2 (Table 3). No grade 4 worsening from baseline was reported in either stratum for hemoglobin levels. Reasons for ontreatment death (in the MSSD cohort) included acute myeloid leukemia (1 patient) and cardiac arrest (1 patient) in S1, and multiple organ failure (1 patient) in S2 (Online Supplementary Table S5). haematologica | 2019; 104(5)

EXPAND: 48-week follow-up analysis

Efficacy in the maximum safe starting dose cohort Spleen response. At week 48, spleen response was achieved in 5 out of 15 patients [33.3% (95%CI: 11.8, 61.6)] in S1 and 3 out of 10 patients [30.0% (95%CI: 6.7, 65.2)] in S2. A spleen response at any time point was observed in 8 out of 20 patients [40.0% (95%CI: 19.1, 63.9)] in S1 and 12 out of 18 patients [66.7%

(95%CI: 41.0, 86.7)] in S2 (Figure 4).

The waterfall plot for the best response in spleen length for patients treated at the MSSD, with or without dose titration, within the first 12 weeks is presented in Online Supplementary Figure S2. A decrease in best percentage change from baseline in spleen length was evident in 100.0% of patients with dose down-

Table 2. All grade adverse events, regardless of study drug relationship, in â&#x2030;Ľ15% of patients in either stratum (week 48 analysis; maximum safe starting dose cohort).

Preferred term

Stratum 1 (N=20) All grades Grade 3 or 4 N (%) N (%)

Stratum 2 (N=18) All grades Grade 3 or 4 N (%) N (%)

Anemia Thrombocytopenia Platelet count decreased Pyrexia Abdominal pain Diarrhea Ecchymosis Epistaxis White blood cell count decreased Back pain Blood bilirubin increased Alanine aminotransferase increased Aspartate aminotransferase increased Asthenia Fatigue Neutrophil count decreased Cough Hypocalcemia Nasopharyngitis Headache Hypertension Nausea Leukocytosis Peripheral edema Pain in extremity Vomiting

9 (45.0) 8 (40.0) 6 (30.0) 6 (30.0) 5 (25.0) 5 (25.0) 5 (25.0) 5 (25.0) 5 (25.0) 4 (20.0) 4 (20.0) 3 (15.0) 3 (15.0) 3 (15.0) 3 (15.0) 3 (15.0) 0 2 (10.0) 2 (10.0) 1 (5.0) 1 (5.0) 1 (5.0) 1 (5.0) 1 (5.0) 0 1 (5.0)

8 (44.4) 14 (77.8) 0 4 (22.2) 4 (22.2) 5 (27.8) 2 (11.1) 0 1 (5.6) 2 (11.1) 1 (5.6) 1 (5.6) 1 (5.6) 5 (27.8) 3 (16.7) 0 6 (33.3) 5 (27.8) 5 (27.8) 4 (22.2) 4 (22.2) 4 (22.2) 3 (16.7) 3 (16.7) 3 (16.7) 3 (16.7)

4 (20.0) 7 (35.0) 5 (25.0) 0 0 1 (5.0) 0 0 2 (10.0) 0 2 (10.0) 1 (5.0) 1 (5.0) 1 (5.0) 1 (5.0) 1 (5.0) 0 0 0 0 1 (5.0) 0 1 (5.0) 1 (5.0) 0 0

3 (16.7) 14 (77.8) 0 1 (5.6) 0 0 0 0 0 0 0 1 (5.6) 0 2 (11.1) 0 0 0 0 0 0 0 0 1 (5.6) 0 0 0

AE: adverse event; MSSD: maximum safe starting dose.

Figure 3. Mean daily dose over time by stratum at maximum safe starting dose. bid: twice daily.

haematologica | 2019; 104(5)

951

A.M. Vannucchi et al.

titration in both S1 (n=3) and S2 (n=10). In patients without a dose down-titration, a decrease in best percentage change from baseline in spleen length was seen in 94.1% (16 out of 17) of patients in S1 and 87.5% (7 out of 8) of patients in S2.

Symptom response. An improvement (i.e. decrease) in TSS was observed at the MSSD in both strata. The mean change in TSS from baseline at week 24 was –7.7 (SD=9.70) in S1 and –3.9 (SD=11.36) in S2. Compared to the baseline symptom score, the mean individual symptom scores decreased (improved) for all categories except bone/muscle pain (slight worsening) at week 24 (Figure 5). A trend in symptom improvement was also observed in patients with early dose titration (Online Supplementary Table S9 and Online Supplementary Figure S3).

Discussion Only a few clinical trials are currently evaluating the treatment options for patients with MF with thrombocytopenia (platelet counts <100x109/L), highlighting the need for conducting this analysis. The JAK inhibitors that have been evaluated in this setting include ruxolitinib, pacritinib, momelotinib, and fedratinib;34-36 however, only ruxolitinib is currently approved for the treatment of patients with MF. Ruxolitinib was approved for the treatment of intermediate- and high-risk patients with MF

based on the COMFORT studies in patients with normal platelet counts (≥100x109/L).9-12,28 Findings from a post hoc analysis of COMFORT-I showed that patients with cytopenias at baseline could be effectively managed with ruxolitinib dose adjustments and that doses of ≥10 mg bid yielded clinically meaningful reductions in spleen volume and symptom improvement.28 These findings support the use of ruxolitinib as a therapeutic option for patients with MF with low baseline platelet counts (<100x109/L).

The preliminary observations based on toxicity during the first cycle of treatment indicated ruxolitinib 15 mg and 10 mg bid as MSSDs for S1 and S2, respectively. However, observations from the interim analysis showed that the majority of patients receiving the 15 mg bid MSSD dose in S1 experienced thrombocytopenia, thus requiring dose reductions after the first cycle. These patients subsequently continued study treatment at the 10 mg bid or lower dose. Clinical benefit was observed across all starting dose levels, including in those patients who started treatment at the 10 mg bid dose level. Based on these observations, the initially stated MSSD of 15 mg bid for S1 was revised to 10 mg bid as per protocol amendment.

Based on the results from the interim and 48-week analyses of EXPAND, 10 mg bid was established as the MSSD for both strata (S1: platelet count=75-99x109/L; S2: platelet count=50-74x109/L).

Table 3. New or worsened hematologic abnormalities (week 48 analysis; maximum safe starting dose cohort).

Parameter

Worsening from baseline to the following

Platelets (×109/L)

Hemoglobin (g/dL)

Grade 1 Grade 2 Grade 3 Grade 4 Grade 1 Grade 2 Grade 3 Grade 4

Stratum 1 (N=20) Total N (%) 2 15 19 20 5 13 18 20

0 5 (25.0) 9 (45.0) 1 (5.0) 2 (10.0) 3 (15.0) 5 (25.0) 0

Stratum 2 (N=18) Total

N (%)

0 1 18 18 1 9 12 18

0 1 (5.6) 7 (38.9) 7 (38.9) 0 6 (33.3) 1 (5.6) 0

MSSD: maximum safe starting dose.

A

B

Figure 4. Waterfall plot of best response in spleen length by stratum at maximum safe starting dose.

952

haematologica | 2019; 104(5)

EXPAND: 48-week follow-up analysis

Figure 5. Change in total symptom score and individual symptom scores of Myelofibrosis Symptom Assessment Form diary from baseline to week 24 by stratum at maximum safe starting dose.

Ruxolitinib was generally well tolerated at all dose levels, including the MSSDs, and no new safety signal was observed. However, as expected, a higher frequency of thrombocytopenia was observed, which was managed by dose reduction/interruption. Evidently, the 10 mg bid dose was better tolerated in S1 versus S2; 3 out of 6 patients in S1 versus 10 out of 11 patients in S2 did not resume the 10 mg bid dose after the initial dose reduction (first 12 weeks). A medically meaningful spleen size response was observed with ruxolitinib treatment at the 10 mg bid dose.

In the MSSD cohort, at least 50% of reduction in the spleen length was observed in 33.3% of patients in S1 and 30.0% of patients in S2 at week 48, whereas a spleen response at any time point was achieved by 40.0% of patients in S1 and 66.7% of patients in S2. A decrease in the best percentage change from baseline was observed at the MSSD in 95.0% of patients in S1 and 94.4% of patients in S2. An improvement in symptom response (decrease in the MFSAF-TSS) was also observed at the MSSD in both strata.

During the first 12 weeks of treatment, the ruxolitinib dose was reduced below 10 mg bid in some patients, mostly due to ruxolitinib-associated hematologic toxicity.

However, spleen and symptom benefit was observed in these patients despite the early dose titration, and the treatment was continued at the reduced dose, suggesting that a starting dose of 10 mg bid may still be effective in the long-term (as was also observed in the COMFORT-I study).28 In the subgroup with dose down-titration in S1 (3 out of 6 patients), a decrease in the best percentage change from baseline in spleen length was observed; however, none of these patients achieved an at least 50% reduction in spleen length. All patients in S2 (n=10) who had a dose down-titration achieved a decrease in the best percentage change from baseline in spleen length; 7 out of 10 patients achieved an at least 50% reduction in spleen length (Online Supplementary Figure S2). A trend in symptom improvement was also observed in patients with early haematologica | 2019; 104(5)

dose titration at the MSSD; this effect was more pronounced in S1 than in S2 (Online Supplementary Table S9 and Online Supplementary Figure S3).

The observations from the EXPAND study are consistent with the findings from other clinical trials evaluating the use of ruxolitinib in patients with MF with baseline thrombocytopenia. In the Study 258, the median percentage change from baseline in spleen length at week 24 in the 30 evaluable patients was –29.7% (range, –100.0% to 58.3%).

In the EXPAND study, the median percentage change from baseline in spleen length at week 24 for the overall population (n=69) was –36.9% (range, –100.0% to 55.6%). As expected (owing to baseline patients’ characteristics and the mechanism of action of ruxolitinib), thrombocytopenia was frequent in both the Study 258 and the EXPAND study (64.0% vs. 68.1%, respectively). Low-dose ruxolitinib was shown to be generally well tolerated and efficacious in patients with MF with low platelet counts in JUMP.23 The findings to date from the 48-week follow-up analysis of the EXPAND study provide evidence to support a starting dose of ruxolitinib at 10 mg bid for patients with MF with low baseline platelet counts of 75-99x109/L (S1) but are less conclusive for baseline platelet counts of 50-74x109/L (S2). The reported AEs were consistent with the known safety profile of ruxolitinib, with the exception of thrombocytopenia in S2, which was expected.

Ruxolitinib treatment was generally well tolerated and provided spleen size reduction and symptom response benefit. The tolerability of ruxolitinib in this previously unstudied patient population with MF with low platelet counts at baseline was acceptable at doses of 10 mg bid in both strata. The study is ongoing, and further evaluations will be performed at the end of the study to confirm the safety and efficacy of ruxolitinib in the study cohorts.

Acknowledgments The authors would like to thank Archana Rai and Ambrin Fatima, PhD (Novartis Healthcare Pvt Ltd) for providing medical writing assistance.

953

A.M. Vannucchi et al.

References 1. Campbell PJ, Green AR. The myeloproliferative disorders. N Engl J Med. 2006;355(23): 2452-2466.

2. Le Bousse-Kerdiles MC. Primary myelofibrosis and the "bad seeds in bad soil" concept. Fibrogenesis Tissue Repair. 2012; 5(Suppl 1):S20.

3. Abdel-Wahab OI, Levine RL. Primary myelofibrosis: update on definition, pathogenesis, and treatment. Annu Rev Med. 2009;60:233-245.

4. Barosi G. Myelofibrosis with myeloid metaplasia: diagnostic definition and prognostic classification for clinical studies and treatment guidelines. J Clin Oncol. 1999; 17(9):2954-2970.

5. Mesa RA, Schwager S, Radia D, et al. The Myelofibrosis Symptom Assessment Form (MFSAF): an evidence-based brief inventory to measure quality of life and symptomatic response to treatment in myelofibrosis. Leuk Res. 2009;33(9):1199-1203.

6. Scherber R, Dueck AC, Johansson P, et al.

The Myeloproliferative Neoplasm Symptom Assessment Form (MPN-SAF): international prospective validation and reliability trial in 402 patients. Blood.

2011;118(2):401-408.

7. Ihle JN, Gilliland DG. Jak2: normal function and role in hematopoietic disorders. Curr Opin Genet Dev. 2007;17(1):8-14.

8. Parganas E, Wang D, Stravopodis D, et al.

Jak2 is essential for signaling through a variety of cytokine receptors.

Cell.

1998;93(3):385-395.

9. 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.

10. Harrison CN, Vannucchi AM, Kiladjian JJ, et al. Long-term findings from COMFORT-II, a phase 3 study of ruxolitinib vs best available therapy for myelofibrosis. Leukemia.

2016;30(8):1701-1707.

11. Verstovsek S, Mesa RA, Gotlib J, et al. Longterm treatment with ruxolitinib for patients with myelofibrosis: 5-year update from the randomized, double-blind, placebo-controlled, phase 3 COMFORT-I trial. J Hematol Oncol. 2017;10(1):55.

12. 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.

13. Balduini A, Badalucco S, Pugliano MT, et al.

In vitro megakaryocyte differentiation and proplatelet formation in Ph-negative classical myeloproliferative neoplasms: distinct patterns in the different clinical phenotypes.

PLoS One. 2011;6(6):e21015.

954

14. Talpaz M, Paquette R, Afrin L, et al. Interim analysis of safety and efficacy of ruxolitinib in patients with myelofibrosis and low platelet counts. J Hematol Oncol.

2013;6(1):81.

15. Al-Ali HK, Vannucchi AM. Managing patients with myelofibrosis and low platelet counts. Ann Hematol. 2017; 96(4):537-548.

16. Emanuel RM, Dueck AC, Geyer HL, et al.

Myeloproliferative neoplasm (MPN) symptom assessment form total symptom score: prospective international assessment of an abbreviated symptom burden scoring system among patients with MPNs. J Clin Oncol. 2012;30(33):4098-4103.

17. Gangat N, Caramazza D, Vaidya R, et al.

DIPSS plus: a refined Dynamic International Prognostic Scoring System for primary myelofibrosis that incorporates prognostic information from karyotype, platelet count, and transfusion status. J Clin Oncol.

2011;29(4):392-397.

18. Mesa RA, Niblack J, Wadleigh M, et al. The burden of fatigue and quality of life in myeloproliferative disorders (MPDs): an international Internet-based survey of 1179 MPD patients. Cancer. 2007;109(1):68-76.

19. Tefferi A, Lasho TL, Jimma T, et al. One thousand patients with primary myelofibrosis: the Mayo Clinic experience. Mayo Clin Proc. 2012;87(1):25-33.

20. Verstovsek S, Mesa RA, Gotlib J, et al.

Efficacy, safety and survival with ruxolitinib in patients with myelofibrosis: results of a median 2-year follow-up of COMFORT-I. Haematologica. 2013;98(12):1865-1871.

21. Verstovsek S, Kantarjian H, Mesa RA, et al.

Safety and efficacy of INCB018424, a JAK1 and JAK2 inhibitor, in myelofibrosis. N Engl J Med. 2010;363(12):1117-1127.

22. Harrison CN, Gisslinger H, Miller CB, et al.

EXPAND: A phase 1b, open-label, dosefinding study of ruxolitinib in patients with myelofibrosis and baseline platelet counts between 50×109/L and 99×109/L. Blood.

2012;120(21):177.

23. Griesshammer M, Vannucchi AM, le Coutre P, et al. Safety and efficacy of ruxolitinib in patients with low platelets enrolled in a phase 3b expanded-access study in myelofibrosis (MF). Blood 2014; 124(21):1859.

24. Tavares R, Palumbo GA, Le Coutre P, et al.

Safety and efficacy of ruxolitinib in an 1869patient cohort of JUMP: an open-label, multicenter, single-arm, expanded-access study in patients with myelofibrosis. Blood.

2015;126(23):2799.

25. Al-Ali HK, Griesshammer M, le Coutre P, et al. Safety and efficacy of ruxolitinib in an open-label, multicenter, single-arm phase 3b expanded-access study in patients with myelofibrosis: a snapshot of 1144 patients in

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

the JUMP trial. Haematologica. 2016; 101(9):1065-1073.

JAKAFI® (ruxolitinib) prescribing information USFDA. https:// www. accessdata.fda.gov/drugsatfda_docs/label/2 017/202192s015lbl.pdf. Accessed March 21, 2018.

JAKAVI® (ruxolitinib) Summary of product characteristics. http://www.ema. europa.eu/ docs/ en_GB/ document_library/EPAR__Product_Information/human/002464/WC5 00133223.pdf. Accessed March 21, 2018.

Verstovsek S, Gotlib J, Gupta V, et al.

Management of cytopenias in patients with myelofibrosis treated with ruxolitinib and effect of dose modifications on efficacy outcomes. Onco Targets Ther. 2013;7:13-21.

Vannucchi AM, Gisslinger H, Harrison CN, et al. EXPAND: A phase 1b, open-label, dose-finding study of ruxolitinib in patients with myelofibrosis (MF) and low platelet counts (50×109/L to 99×109/L) at baseline.

Blood. 2015;126(23):2817.

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.

Mukuria C, Rowen D, Brazier JE, et al.

Deriving a preference-based measure for myelofibrosis from the EORTC QLQ-C30 and the MF-SAF. Value Health.

2015;18(6):846-855.

Mesa RA, Schwager S, Radia D, et al. The Myelofibrosis Symptom Assessment Form (MFSAF): an evidence-based brief inventory to measure quality of life and symptomatic response to treatment in myelofibrosis.

Leukemia Res. 2009;33(9):1199-1203.

Mesa RA, Kantarjian H, Tefferi A, et al.

Evaluating the serial use of the Myelofibrosis Symptom Assessment Form for measuring symptomatic improvement: performance in 87 myelofibrosis patients on a JAK1 and JAK2 inhibitor (INCB018424) clinical trial. Cancer. 2011; 117(21):48694877.

Harrison CN, Vannucchi AM, Platzbecker U, et al. Momelotinib versus best available therapy in patients with myelofibrosis previously treated with ruxolitinib (SIMPLIFY 2): a randomised, open-label, phase 3 trial.

Lancet Haematol. 2018;5(2):e73-e81.

Pardanani A, Harrison C, Cortes JE, et al.

Safety and efficacy of fedratinib in patients with primary or secondary myelofibrosis: a randomized clinical trial. JAMA Oncol.

2015;1(5):643-651.

Verstovsek S, Komrokji RS. A comprehensive review of pacritinib in myelofibrosis.

Future Oncol. 2015;11(20):2819-2830.

haematologica | 2019; 104(5)

ARTICLE

Chronic Myeloid Leukemia

Imatinib dose reduction in major molecular response of chronic myeloid leukemia: results from the German Chronic Myeloid Leukemia-Study IV

Christian Michel,1* Andreas Burchert,1,* Andreas Hochhaus,2 Susanne Saussele,3 Andreas Neubauer,1 Michael Lauseker,4 Stefan W. Krause,5 Hans-Jochem Kolb,6 Dieter Kurt Hossfeld,7 Christoph Nerl,8 Gabriela M. Baerlocher,9 Dominik Heim,10 Tim H Brümmendorf,11 Alice Fabarius,3 Claudia Haferlach,12 Brigitte Schlegelberger,13 Leopold Balleisen,14 Maria-Elisabeth Goebeler,15 Mathias Hänel,16 Anthony Ho,17 Jolanta Dengler,18 Christiane Falge,19 Robert Möhle,20 Stephan Kremers,21 Michael Kneba,22 Frank Stegelmann,23 Claus-Henning Köhne,24 Hans-Walter Lindemann,25 Cornelius F. Waller,26 Karsten Spiekermann,6 Wolfgang E. Berdel,27 Lothar Müller,28 Matthias Edinger,29 Jiri Mayer,30 Dietrich W. Beelen,31 Martin Bentz,32 Hartmut Link,33 Bernd Hertenstein,34 Roland Fuchs,11 Martin Wernli,35 Frank Schlegel,36 Rudolf Schlag,37 Maike de Wit,38 Lorenz Trümper,39 Holger Hebart,40 Markus Hahn,41 Jörg Thomalla,42 Christof Scheid,43 Philippe Schafhausen,7 Walter Verbeek,44 Michael J. Eckart,45 Winfried Gassmann,46 Michael Schenk,47 Peter Brossart,48 Thomas Wündisch,1 Thomas Geer,49 Stephan Bildat,50 Erhardt Schäfer,51 Joerg Hasford,4 Rüdiger Hehlmann3 and Markus Pfirrmann4

Universitätsklinikum Gießen und Marburg, Campus Marburg, Klinik für Hämatologie, r Innere Onkologie und Immunologie, Philipps Universität Marburg, Germany; 2Klinik füm Medizin II, Hämatologie und Internistische Onkologie, Jena, Germany; 3III. Medizinische Klinik, Medizinische Fakultät Mannheim, University Heidelberg, Mannheim, Germany; 4 Institut für Medizinische Informationsverarbeitung, Biometrie und Epidemiologie, Ludwig-Maximilians-Universität München, Munich, Germany; 5Medizinische Klinik 5, Universitätsklinikum, Erlangen, Germany; 6Medizinische Klinik III, Universität München, Germany; 7Medizinische Klinik, Universitätsklinikum Eppendorf, Hamburg, Germany; 8 Klinikum Schwabing, Munich, Germany; 9Inselspital, Bern, Switzerland; 10 Universitätsspital, Basel, Switzerland; 11RWTH, Aachen, Germany; 12MLL, Munich, Germany; 13Institut für Humangenetik, MHH, Hannover, Germany; 14Ev. Krankenhaus, Hamm, Germany; 15Comprehensive Cancer Center Mainfranken und Medizinische Klinik II, Zentrum für Innere Medizin, Würzburg, Germany; 16Klinik für Innere Medizin 3, Chemnitz, Germany; 17Medizinische Klinik V, Universität Heidelberg, Germany; 18 Onkologische Schwerpunktpraxis, Heilbronn, Germany; 19Medizinische Klinik 5, Klinikum Nürnberg-Nord, Germany; 20Medizinische Abteilung 2, Universitätsklinikum, Tübingen, Germany; 21Caritas Krankenhaus, Lebach, Germany; 222. Medizinische Klinik und Poliklinik, Universitätsklinikum Schleswig-Holstein, Kiel, Germany; 23Klinik für Innere Medizin 3, Universitätsklinikum, Ulm, Germany; 24Universitätsklinik für Onkologie und Hämatologie, Oldenburg, Germany; 25St Marien-Hospital, Hagen, Germany; 26Innere Medizin 1, Universitätsklinikum, Freiburg, Germany; 27Medizinische Klinik A, Universitätsklinikum, Münster, Germany; 28Onkologie Leer Unter Ems, Leer, Germany; 29 Klinik und Poliklinik für Innere Medizin 3, Universitätsklinikum, Regensburg, Germany; 30 Masaryk University Hospital, Brno, Czech Republic; 31Klinik für Knochenmarktransplantation, Essen, Germany; 32Medizinische Klinik 3, Städtisches Klinikum, Karlsruhe, Germany; 33Hematology, Medical Oncology, Kaiserslautern, Germany; 341. Medizinische Klinik, Klinikum Bremen Mitte, Bremen, Germany; 35 Kantonsspital, Aarau, Switzerland; 36St Antonius-Hospital, Eschweiler, Germany; 37 Hämatologische-Onkologische Schwerpunktpraxis, Würzburg, Germany; 38Klinik für Innere Medizin II, Hämatologie, Onkologie und Palliativmedizin, Vivantes Klinikum Neukölln, Berlin, Germany; 39Klinik für Hämatologie und medizinische Onkologie, Universitätsmedizin, Göttingen, Germany; 40Stauferklinikum Schwäbisch Gmünd, Mutlangen, Germany; 41Onkologie Zentrum, Ansbach, Germany; 42Praxisklinik für Hämatologie und Onkologie, Koblenz, Germany; 43Klinik 1 für Innere Medizin, Universitätsklinikum, Köln, Germany; 44Ambulante Hämatologie und Onkologie, Bonn, Germany; 45Internistische Schwerpunktpraxis, Erlangen, Germany; 46St MarienKrankenhaus, Siegen, Germany; 47Barmherzige Brüder, Regensburg, Germany; 48 Medizinische Klinik 3, Universität, Bonn, Germany; 49Diakonie, Schwäbisch Hall, Germany; 50Medizinische Klinik 2, Herford, Germany and 51Onkologische Schwerpunktpraxis, Bielefeld, Germany

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):955-962

1

*CM and AB contributed equally to this work.

haematologica | 2019; 104(5)

Correspondence: ANDREAS BURCHERT burchert@staff.uni-marburg.de Received: September 23, 2018.

Accepted: November 22, 2018.

Pre-published: December 4, 2018. doi:10.3324/haematol.2018.206797 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/955 ©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.

955

C. Michel et al.

ABSTRACT

S

tandard first-line therapy of chronic myeloid leukemia is treatment with imatinib. In the randomized German Chronic Myeloid Leukemia-Study IV, more potent BCR-ABL inhibition with 800 mg (â&#x20AC;&#x2DC;highdoseâ&#x20AC;&#x2122;) imatinib accelerated achievement of a deep molecular remission. However, whether and when a de-escalation of the dose intensity under high-dose imatinib can be safely performed without increasing the risk of losing deep molecular response is unknown. To gain insights into this clinically relevant question, we analyzed the outcome of imatinib dose reductions from 800 mg to 400 mg daily in the Chronic Myeloid Leukemia-Study IV. Of the 422 patients that were randomized to the 800 mg arm, 68 reduced imatinib to 400 mg after they had achieved at least a stable major molecular response. Of these 68 patients, 61 (90%) maintained major molecular remission on imatinib at 400 mg. Five of the seven patients who lost major molecular remission on the imatinib standard dose regained major molecular remission while still on 400 mg imatinib. Only two of 68 patients had to switch to more potent kinase inhibition to regain major molecular remission. Importantly, the lengths of the intervals between imatinib high-dose treatment before and after achieving major molecular remission were associated with the probabilities of maintaining major molecular remission with the standard dose of imatinib. Taken together, the data support the view that a deep molecular remission achieved with high-dose imatinib can be safely maintained with standard dose in most patients. Study protocol registered at clinicaltrials.gov 00055874.

Introduction Approved first-line therapies of chronic myeloid leukemia (CML) are the tyrosine kinase inhibitors (TKIs) imatinib, dasatinib, nilotinib and bosutinib.1-5 Imatinib led to distinctively improved progression-free and overall survival of chronic phase CML patients as compared with previous conventional treatment standards in CML.6,7 Second-generation TKIs, such as nilotinib and dasatinib, but also a higher dose of imatinib (800 mg/day), induce deep molecular response (MR) faster8,9 and in a larger proportion of patients.10,11 As a consequence, deep molecular remission (an essential eligibility criterion for TKI discontinuation) can be achieved earlier and in more patients when compared to imatinib standard dose.12,13 However, the benefit of pursuing highly-potent BCR-ABL-kinase inhibition once deep MR has been achieved is less clear. Moreover, for those patients in deep MR, which (for whatever reason) require long-term treatment, the tolerability and prevention of organ damage through clinically relevant and potentially irreversible side effects, such as pulmonary hypertension, diabetes, hypercholesterinemia, and cardiovascular morbidity become the most important priority.14-17 Thus, if highpotency BCR-ABL inhibition is not needed to sustain remission or improve survival, then the risk of potentially harmful side effects from second- or third-generation TKI must be weighed against the long-term safety of using imatinib,1820 especially when also considering that generic imatinib is more cost effective. By analyzing the outcome of 800 mg to 400 mg imatinib dose reductions performed in at least stable major molecular remission (MMR) within the randomized German CML-Study IV,8,21 we aimed to address the clinically important questions of in which patients and at what time after initiation of strong BCR-ABL inhibition with 800 mg imatinib less potent BCR-ABL inhibition with standard dose imatinib is sufficient to maintain stable MMR.

Methods Patients and Chronic Myeloid Leukemia-Study IV protocol All patients investigated in this study were treated within the randomized German CML-Study IV.8,21 Imatinib monotherapy at 956

800 mg/day was one of the five arms in this trial. The study protocol was registered at clinicaltrials.gov 00055874. Randomization took place from July 2002 through March 2012. During a pilotphase of 3 years, only high-risk patients according to the Euro score22 were randomized to imatinib 800 mg/day. In 2005, imatinib 800 mg/day was started as a full study arm.

To avoid selection bias towards high-risk patients, in this retrospective analysis, only patients randomized from 2005 were evaluated.

Definition of high-dose imatinib treatment Imatinib at a dose of 800 mg/day for at least 6 months was classified as high-dose therapy. Six months was chosen because the presence of MMR after 6 months significantly increased the probabilities of patients going on to achieve deep MR later.8 A highdose treatment interval began with the first dose of 800 mg/day and ended at the time of imatinib dose reduction to 400 mg/day.

An intermittent 600 mg/day interval which directly preceded or followed a high-dose treatment interval with 800 mg/day was still considered high-dose treatment because the effective median dose of imatinib in the 800 mg arm was seen to be only 600 mg in the CML-Study IV.8 The molecular analyses are described in the Online Supplementary Appendix.

Statistical analysis Survival without loss of MMR was defined as the time between the start of reduced imatinib therapy with 400 mg/day either until loss of MMR or until the date of the last evaluation of MR status with the date linkable to the reduction period, as defined in the Online Supplementary Methods. Probabilities of molecular relapsefree survival (RFS) were estimated by the Kaplan-Meier method.

The association between a variable and molecular RFS was assessed by Cox regression.23 For identification of cutoffs, the minimal P-value approach was used while assuming that the smallest group should contain at least 10% of patients.24 Bootstrap resampling and kernel density estimation were carried out to assess the stability of a cutoff.25,26 Point estimates are given together with their 95% confidence intervals (95%CI). In the case of the hazard ratios (HR), for estimation of the 95%CI, the profile likelihood was used and P-values were calculated from the likelihood ratio test. All analyses are descriptive and exploratory. Apart from the minimal P-value haematologica | 2019; 104(5)

Imatinib dose reduction in CML

approach, the significance level of the two-sided P-value was 0.05 for all statistical tests. Analyses were carried out with SAS v.9.4 or R v.3.4.3.

Ethical approval The CML-Study IV was performed in accordance with the Declaration of Helsinki, and was approved by the central ethics committee of the Medizinische Fakultaet Mannheim and the local ethics committees of all participating centers. Written informed consent was obtained from all patients prior to entering the CMLStudy IV.

Results Imatinib dose reduction in the 800 mg cohort of the Chronic Myeloid Leukemia-Study IV Of 1551 patients with newly diagnosed chronic phase CML, 422 were randomly assigned to 800 mg imatinib per day. Of these, two patients violated CML-Study IV inclu-

sion criteria, ten were part of the pilot study, and a further ten were excluded from this analysis due to missing treatment data (see the CONSORT flow diagram in Figure 1).

Of the remaining 400 patients randomized to 800 mg imatinib, 92 patients had never received imatinib 800 mg/day and 163 patients never achieved MMR within the imatinib 800 mg/day interval. Of the remaining 145 patients, 39 had never reduced the 800 mg/day dose or had no observation time after dose reduction. A further 21 had an 800 mg/day interval of less than 6 months (i.e. not high-dose imatinib by our definition). Two patients were excluded because they had more than 6 weeks without any therapy between ending 800 mg/day imatinib and recommencing 400 mg/day. Eight patients were not considered because treatment with 400 mg/day lasted for less than 6 months. After exclusion of a further seven patients who had not been monitored by the central molecular diagnostic laboratory of the CML-Study IV, 68 patients were evaluable for our analyses.

Data entry was closed on July 21, 2015.

Figure 1. Flow diagram: patients of the Chronic Myeloid Leukemia-Study IV considered for final analysis.

haematologica | 2019; 104(5)

957

C. Michel et al.

Table 1. Univariate Cox regression estimating the influence on relapse-free survival after reduction to imatinib at 400 mg.

Variable

n / Loss Estimation of of MMR coefficient β

Standard deviation of estimated β

Hazard ratio

Lower 95%CI limit for hazard ratio

Variables recorded at diagnosis Age (years) 68 / 7 -0.001 0.03 1.00 0.94 Gender 68 / 7 Male 48 / 6 baseline Female 20 / 1 -1.04 1.08 0.36 0.02 Spleen size below costal margin (cm) 68 / 7 0.16 0.06 1.17 1.03 9 White blood cell count (x10 /L) / 100 68 / 7 1.01 0.25 2.76 1.69 Blasts in peripheral blood (%) 68 / 7 0.24 0.11 1.27 0.99 Eosinophils in peripheral blood (%) 68 / 7 0.19 0.07 1.21 1.02 Basophils in peripheral blood (%) 68 / 7 0.09 0.07 1.10 0.93 Platelet count (x109/L) /1000 68 / 7 -2.44 2.35 1.00 0.00 Sokal score 68 / 7 Low risk 29 / 2 baseline Intermediate risk 27 / 3 0.60 0.91 1.82 0.30 High risk 12 / 2 1.05 1.01 2.87 0.34 Euro score 68 / 7 Low risk 30 / 1 baseline Intermediate risk 34 / 4 1.42 1.12 4.13 0.61 High risk 4/2 3.04 1.23 20.98 1.99 EUTOS score 68 / 7 Low risk 62 / 4 baseline High risk 6/3 2.26 0.77 9.56 1.88 ELTS score 68 / 7 Low risk 44 / 1 baseline Intermediate risk 19 / 4 2.30 1.12 9.98 1.48 High risk 5/2 3.17 1.23 23.71 2.25 Variables recorded under treatment and their influence after stopping treatment at 800mg/day Time with 800mg dosage, months 68 / 7 -0.05 0.03 0.95 0.891 Time from start of 800mg until MMR, months 68 / 7 0.13 0.05 1.14 1.02 Time from MMR until end of 800mg, months 68 / 7 -0.15 0.07 0.86 0.72

Upper 95%CI ratio limit for hazard ratio

P

1.06

195.25 513.27

0.97 0.28 0.02 0.0001 0.06 0.03 0.24 0.24 0.57 0.04 0.01 0.01 -

0.998 1.27 0.96

0.04 0.02 0.0006

2.08 1.31 4.73 1.55 1.37 1.25 3.40 13.79 24.15 80.86 454.17 43.50

CI: Confidence Interval; MMR: major molecular remission; EUTOS: European Treatment and Outcome Study; ELTS: EUTOS long-term survival.

Patients' characteristics are shown in Online Supplementary Table S1. Median age was 52 years and 71% of the 68 patients were male.

Treatment course of patients with imatinib dose reduction to 400 mg Twenty-five of the 68 patients on high-dose imatinib (37%) started their primary treatment directly with 800 mg/day (first treatment interval). Forty patients (59%) increased to 800 mg/day after a first period of 400 mg imatinib. Three patients only started the 800 mg imatinib dose later.

Median time on high-dose imatinib therapy was 31 months (range: 6-98 months) for the 68 patients who later reduced imatinib treatment to 400 mg (Online Supplementary Table S1). In this cohort, the median dura958

tion of treatment with 400 mg/day after dose reduction was 34 months (range: 6-78 months). For 53 out of 68 patients (78%), this was the last reported treatment and dose. Five patients (7%) eventually stopped any TKI therapy following imatinib dose reduction to 400 mg. In one patient, no information regarding treatment after dose reduction to 400 mg was available. The remaining 9 out of 68 patients (13%) received a more potent ABL-kinase inhibition: 600 mg imatinib (n=1), 800 mg imatinib (n=5), nilotinib (n=2), or dasatinib (n=1).

Molecular relapse-free survival after imatinib dose reduction to 400 mg Seven of 68 patients experienced a loss of MMR during the reduction interval (Figure 2). This resulted in a 1-year molecular relapse-free survival (RFS) of 90% (95%CI: 81haematologica | 2019; 104(5)

Imatinib dose reduction in CML

Figure 2. Courses of BCR-ABL (IS) in 68 patients with imatinib dose reduction. Results below the horizontal red line represent at least major molecular response (MMR). Sixty-one patients have never lost MMR (courses with black lines). Five patients with loss of MMR regained MMR while continuing with reduced imatinib dose at 400 mg/day (blue lines). Two patients with loss of MMR did not regain MMR while on the lower imatinib dose and were switched to nilotinib or imatinib at 600 mg/day, respectively (orange lines).

96%). With only one MMR loss occurring later than 12 months after dose reduction, the 3-year molecular RFS was 88% (95%CI: 77-94%) (Figure 3). However, MMR loss was only temporary in five of the seven patients; these patients regained MMR while still on the lower 400 mg imatinib dose. Only two patients with MMR loss were switched to more potent ABL-inhibition with nilotinib or 600 mg imatinib to regain MMR (Figure 2). It is worthy of note that, at the time of stopping high-dose treatment, 43 of 68 patients were at least in MR4, 33 of them even at least in MR4.5. Of the 43 patients, 10 lost MR4 at some point; none lost MMR.

Clinical variables and high-dose imatinib treatment durations prior to and after achieving major molecular response were associated with probabilities of relapse-free survival Of the clinical variables evaluated at diagnosis, larger spleen size below costal margin (HR: 1.17, 95%CI: 1.031.31; P=0.02), higher white blood cell count (WBC) (HR: 2.76, 95%CI: 1.69-4.73; P=0.0001), and a higher percentage of eosinophils (HR: 1.21, 95%CI: 1.02-1.37; P=0.03) were significantly associated with probabilities of lower RFS (Table 1). Furthermore, the high-risk groups according to the Euro and the European Treatment and Outcome Study (EUTOS) scores, as well as the intermediate- and high-risk groups according to the EUTOS long-term survival (ELTS) score, suggested significantly worse molecular RFS than their corresponding low-risk groups.

The longer the total treatment time at 800 mg/day, the higher were the probabilities of RFS (HR: 0.95, 95%CI: 0.891-0.998; P=0.04). However, as in the EURO-SKI study, which analyzed TKI discontinuation, the main focus was to investigate treatment time after dividing this time interhaematologica | 2019; 104(5)

val into the time of high-dose treatment before and after achievement of MR.27 Four of the 68 patients had already achieved MMR with 400 mg imatinib/ day prior to the high-dose treatment interval. The median time to achieving MMR was 5 months (range: 0-23 months) (Online Supplementary Table S1). The longer the time with treatment at 800 mg/day until achieving MMR, the lower were the probabilities of RFS (HR: 1.14, 95%CI: 1.02-1.27; P=0.02). Using the minimum P-value approach, with the prerequisite that the smallest group contained at least 10% of the patients, a cutoff of 13 months was observed (Padjusted=0.007). This cutoff was confirmed with bootstrap resampling: in 1000 bootstrap samples, the cutoff of 13 months was most frequently chosen. For the 56 patients who had an MMR within 13 months while on treatment at 800 mg/day, the probability of molecular RFS 12 months after stopping high-dose treatment was 94% (95%CI: 84-98%), whereas it was 74% [95%CI: 39-91%; HR: 7.39 (95%CI: 1.6237.74)] for the 12 patients who had an MMR after more than 13 months of high-dose treatment (Figure 4A).

The median time from achievement of MMR until dose reduction to 400 mg was 23 months (range: 0-93 months) (Online Supplementary Table S1). The longer the time with MMR while on treatment at 800 mg/day, the higher were the probabilities of RFS (HR: 0.86, 95%CI: 0.72-0.96; P=0.0006). Using the minimum P-value approach, with the prerequisite that the smallest group contained at least 10% of the patients, a cutoff of 8.5 months was observed (Padjusted=0.011). This cutoff was confirmed with bootstrap resampling: in 1000 bootstrap samples, the cutoff of 8.5 months was most frequently chosen. For the 53 patients who were more than 9 months on high-dose treatment after achievement of MMR, the probability of molecular 959

C. Michel et al.

Figure 3. Probabilities of molecular relapse-free survival after dose reduction to imatinib at 400 mg/day. At 12 and 36 months, horizontal crossbars indicate the upper and lower limit of the 95% confidence interval (CI) for the estimated probability.

RFS 12 months after stopping high-dose treatment was 98% (95%CI: 87-99%), whereas it was 65% [95%CI: 3484%; HR: 0.096 (95%CI: 0.014-0.449)] for the 15 patients who were only on high-dose treatment for 9 months or less after achieving MMR (Figure 4B).

Discussion The concept of starting CML therapy upfront with more potent BCR-ABL inhibition than is achievable with 400 mg imatinib has been introduced to prevent early disease progression and induce deep MR faster and more effectively.2 However, it is not known in which patients and when after the initiation of a more potent BCR-ABL kinase inhibition (second/third-generation TKI or 800 mg imatinib) a deep MR can be maintained after de-escalation to 400 mg imatinib.

Trials investigating dose reductions are rare. In the DESTINY study, the dose of second-generation TKIs was reduced to half the respective standard dose.28 However, in terms of BCR-ABL inhibitory potency, even reduced second-generation TKI doses such as those used in the DESTINY trial demonstrate significantly more BCR-ABL inhibition than 400 mg imatinib. To our knowledge, a controlled switch from highly potent BCR-ABL kinase inhibition with 800 mg imatinib or second/third-generation TKI to 400 mg/day imatinib has never been performed prospectively.

On the other hand, a reduction of imatinib treatment intensity to 400 mg is frequently required in patients who achieve a deep MR but experience toxicities or acquire/ have worsening comorbidities of a type that prevents the further use of second-generation TKI. Furthermore, those 960

patients with deep MR who relapse after TKI cessation or who do not wish to discontinue their TKI, and therefore require life-long TKI therapy, are all candidates for a dose de-escalation to imatinib 400 mg.

Here, we studied the stability of a deep MR in patients of the German CML-Study IV who had MMR or better response for at least 6 months when they reduced imatinib from 800 mg to 400 mg per day. We wished to gain insight into whether treatment duration with 800 mg imatinib has an impact on subsequent maintenance of deep MR with imatinib at the 400 mg standard dose. We also searched for clinical variables associated with maintenance of at least MMR after dose reduction of imatinib.

We found that, if BCR-ABL-monitoring once every three months is ensured, imatinib dose reduction from 800 mg to 400 mg/day in patients with stable MMR did not compromise efficacy or risk sustained MMR in most patients, as only two of seven patients who had lost MMR on 400 mg imatinib required a rescue treatment with more potent BCR-ABL-kinase inhibitors. This also suggests that if standard dose imatinib treatment and BCR-ABL monitoring are ensured, a single loss of MMR might not require immediate re-intensification of TKI treatment.29 Secondly, despite only 7 events, statistical modeling suggested that achieving an MMR within 13 months under 800 mg imatinib, as well as staying on 800 mg imatinib for at least 9 months after achievement of MMR, are both good predictors of a successful continuous MMR maintenance under the standard imatinib dose of 400 mg.

Interestingly, with the exception of WBC count, all other prognostic markers identified for a successful imatinib dose reduction have previously also been reported as predictors for treatment-free remission (TFR).7,22,27,30 Based on this, it is tempting to speculate that the biology of TFR haematologica | 2019; 104(5)

Imatinib dose reduction in CML

A

Figure 4. Factors influencing the probabilities of molecular relapse-free survival after imatinib dose reduction to 400 mg/day. (A) Impact of time to major molecular response (MMR) and molecular relapsefree survival. (B) Impact of interval between MMR and imatinib dose reduction and molecular relapsefree survival. At 12 months (mo), horizontal crossbars indicate the upper and lower limit of the 95% confidence interval (CI) for the estimated probability.

B

and rapid MR are mechanistically linked. For example, immuno-biological features such as CD86+ plasmacytoid dendritic cell counts were recently shown to be associated with both TFR rate and rapid, deep MR under TKI therapy.31,32 With 68 patients and 7 events only, the cutoffs and our prognostic analyses remain exploratory and should be confirmed independently. In summary, here we show that if MMR was achieved within 13 months on 800 mg imatinib, a reduction of treatment intensity to 400 mg imatinib is feasible with a high probably that MMR will be maintained. From these results, we speculate that a switch from a second-generation TKI to 400 mg imatinib is unlikely to haematologica | 2019; 104(5)

lead to a loss of MMR. This is no trivial speculation because there are no data to support it, and a prospective clinical trial investigating a switch from second-generation TKI to imatinib will probably never be performed.

Funding This work was supported by the “Deutsche Forschungsgemeinschaft, DFG, Klinische Forschergruppe 210 “Genetics of Drug resistance in Cancer”, and the Deutsche José Carreras Leukämiestiftung (AR12/12), the “Anneliese Pohl Stiftung” and the EUTOS 2016 program (Novartis) to AB. The CML-Study IV has been supported by the German Government (BMBF 01GI0270); Deutsche Krebshilfe (Nr. 961

C. Michel et al.

106642); Deutsche José-Carreras Leukämiestiftung (DJCLS H09/f, H06/04v, H03/01, R05/23, AH06.01); European Union (LSHC-CT-2004–503216); Novartis Oncology, Nürnberg (Drs G Gerhard, S Schaffert, A Jacob and U Haus);

References 1. Hehlmann R, Lauseker M, Saussele S, et al.

Assessment of imatinib as first-line treatment of chronic myeloid leukemia: 10-year survival results of the randomized CML study IV and impact of non-CML determinants. Leukemia. 2017;31(11):2398-2406.

2. Hochhaus A, Larson RA, Guilhot F, et al.

Long-Term outcomes of imatinib treatment for chronic myeloid leukemia. N Engl J Med. 2017;376(10):917-927.

3. Cortes JE, Saglio G, Kantarjian HM, et al.

Final 5-year study results of DASISION: The dasatinib versus imatinib study in treatment-naive chronic myeloid leukemia patients trial. J Clin Oncol. 2016; 34(20):2333-2340.

4. Hochhaus A, Saglio G, Hughes TP, et al.

Long-term benefits and risks of frontline nilotinib vs imatinib for chronic myeloid leukemia in chronic phase: 5-year update of the randomized ENESTnd trial. Leukemia.

2016;30(5):1044-1054.

5. Cortes JE, Gambacorti-Passerini C, Deininger MW, et al. Bosutinib versus imatinib for newly diagnosed chronic myeloid leukemia: Results from the randomized BFORE trial. J Clin Oncol. 2018;36(3):231237.

6. Druker BJ, Guilhot F, O'Brien SG, et al.

Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med. 2006;355(23):2408-2417.

7. Pfirrmann M, Baccarani M, Saussele S, et al.

Prognosis of long-term survival considering disease-specific death in patients with chronic myeloid leukemia. Leukemia.

2016;30(1):48-56.

8. Hehlmann R, Muller MC, Lauseker M, et al. Deep molecular response is reached by the majority of patients treated with imatinib, predicts survival, and is achieved more quickly by optimized high-dose imatinib: results from the randomized CMLstudy IV. J Clin Oncol. 2014;32(5):415-423.

9. Cortes JE, Baccarani M, Guilhot F, et al.

Phase III, randomized, open-label study of daily imatinib mesylate 400 mg versus 800 mg in patients with newly diagnosed, previously untreated chronic myeloid leukemia in chronic phase using molecular end points: tyrosine kinase inhibitor optimization and selectivity study. J Clin Oncol. 2010;28(3):424-430.

10. Kantarjian H, Shah NP, Hochhaus A, et al.

Dasatinib versus imatinib in newly diagnosed chronic-phase chronic myeloid leukemia. N Engl J Med. 2010;

962

Roche, Grenzach- Wyhlen; and Essex, Munich, Germany. We thank E Matzat, R Pleil-Lösch, I Stalljann, G Bartsch, C Sodan-Boyer, M Meckesheimer, U Böhm, A Gil and J Hehlmann for assistance.

362(24):2260-2270.

11. Saglio G, Kim DW, Issaragrisil S, et al.

Nilotinib versus imatinib for newly diagnosed chronic myeloid leukemia. N Engl J Med. 2010;362(24):2251-2259.

12. Saussele S, Richter J, Hochhaus A, Mahon FX. The concept of treatment-free remission in chronic myeloid leukemia.

Leukemia. 2016;30(8):1638-1647.

13. Hochhaus A, Masszi T, Giles FJ, et al.

Treatment-free remission following frontline nilotinib in patients with chronic myeloid leukemia in chronic phase: results from the ENESTfreedom study. Leukemia.

2017;31(7):1525-1531.

14. Steegmann JL, Baccarani M, Breccia M, et al. European LeukemiaNet recommendations for the management and avoidance of adverse events of treatment in chronic myeloid leukaemia. Leukemia. 2016; 30(8):1648-1671.

15. Hochhaus A, Baccarani M, Deininger M, et al. Dasatinib induces durable cytogenetic responses in patients with chronic myelogenous leukemia in chronic phase with resistance or intolerance to imatinib.

Leukemia. 2008;22(6):1200-1206.

16. Hughes TP, Lipton JH, Spector N, et al.

Deep molecular responses achieved in patients with CML-CP who are switched to nilotinib after long-term imatinib. Blood.

2014;124(5):729-736.

17. Gambacorti-Passerini C, Brummendorf TH, Kim DW, et al. Bosutinib efficacy and safety in chronic phase chronic myeloid leukemia after imatinib resistance or intolerance: Minimum 24-month follow-up.

Am J Hematol. 2014;89(7):732-742.

18. Efficace F, Baccarani M, Breccia M, et al.

Chronic fatigue is the most important factor limiting health-related quality of life of chronic myeloid leukemia patients treated with imatinib. Leukemia. 2013;27(7):15111519.

19. Guerin A, Chen L, Ionescu-Ittu R, et al.

Impact of low-grade adverse events on health-related quality of life in adult patients receiving imatinib or nilotinib for newly diagnosed Philadelphia chromosome positive chronic myelogenous leukemia in chronic phase. Curr Med Res Opin. 2014;30(11):2317-2328.

20. Flynn KE, Atallah E. Quality of life and long-term therapy in patients with chronic myeloid leukemia. Curr Hematol Malig Rep. 2016;11(2):80-85.

21. Hehlmann R, Lauseker M, Jung-Munkwitz S, et al. Tolerability-adapted imatinib 800 mg/d versus 400 mg/d versus 400 mg/d plus interferon-alpha in newly diagnosed

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

chronic myeloid leukemia. J Clin Oncol.

2011;29(12):1634-1642.

Hasford J, Pfirrmann M, Hehlmann R, et al.

A new prognostic score for survival of patients with chronic myeloid leukemia treated with interferon alfa. Writing Committee for the Collaborative CML Prognostic Factors Project Group. J Natl Cancer Inst. 1998;90(11):850-858.

Therneau TM, Grambsch PM. Modeling survival data: Extending the Cox model.

New York: Springer, 2000.

Altmann DG, Lausen B, Sauerbrei W, Schumacher M. Dangers of using "Optimal" cutpoints in the evaluation of prognostic factors. J Natl Cancer Inst.

1994;86(11):829-835.

Davison AC, Hinkley DV. Bootstrap methods and their application. Cambridge: Cambridge University Press, 1997.

Silverman B. Density estimation for statistics and data analysis. London: Chapman and Hall, 1986.

Saussele S, Richter J, Guilhot J, et al.

Discontinuation of tyrosine kinase inhibitor therapy in chronic myeloid leukaemia (EURO-SKI): a prespecified interim analysis of a prospective, multicentre, non-randomised, trial. Lancet Oncol.

2018;19(6):747-757.

Clark RE, Polydoros F, Apperley JF, et al.

De-escalation of tyrosine kinase inhibitor dose in patients with chronic myeloid leukaemia with stable major molecular response (DESTINY): an interim analysis of a non-randomised, phase 2 trial. Lancet Haematol. 2017;4(7):e310-e316.

Fassoni AC, Baldow C, Roeder I, Glauche I. Reduced tyrosine kinase inhibitor dose is predicted to be as effective as standard dose in chronic myeloid leukemia: A simulation study based on phase 3 trial data.

Haematologica. 2018;103(11):1825-1834.

Hasford J, Baccarani M, Hoffmann V, et al.

Predicting complete cytogenetic response and subsequent progression-free survival in 2060 patients with CML on imatinib treatment: the EUTOS score. Blood. 2011; 118(3):686-692.

Schutz C, Inselmann S, Sausslele S, et al.

Expression of the CTLA-4 ligand CD86 on plasmacytoid dendritic cells (pDC) predicts risk of disease recurrence after treatment discontinuation in CML. Leukemia. 2017; 31(4):829-836.

Inselmann S, Wang Y, Saussele S, et al.

Development, function and clinical significance of plasmacytoid dendritic cells in chronic myeloid leukemia. Cancer Res. 2018;78(21):6223-6234.

haematologica | 2019; 104(5)

ARTICLE

Acute Myeloid Leukemia

The thymidine dideoxynucleoside analog, alovudine, inhibits the mitochondrial DNA polymerase γ, impairs oxidative phosphorylation and promotes monocytic differentiation in acute myeloid leukemia

Dana Yehudai,1,2 Sanduni U. Liyanage,1 Rose Hurren,1 Biljana Rizoska,2 Mark Albertella,1 Marcela Gronda,1 Danny V Jeyaraju,1 Xiaoming Wang,1 Samir H. Barghout,1 Neil MacLean,1 Thirushi P. Siriwardena,1 Yulia Jitkova,1 Paul Targett-Adams1 and Aaron D. Schimmer1

Princess Margaret Cancer Centre, University Health Network, ON, Canada and 2Medivir AB, Huddinge, Sweden

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):963-972

1

ABSTRACT

M

itochondrial DNA encodes 13 proteins that comprise components of the respiratory chain that maintain oxidative phosphorylation. The replication of mitochondrial DNA is performed by the sole mitochondrial DNA polymerase γ. As acute myeloid leukemia (AML) cells and stem cells have an increased reliance on oxidative phosphorylation, we sought to evaluate polymerase γ inhibitors in AML. The thymidine dideoxynucleoside analog, alovudine, is an inhibitor of polymerase γ. In AML cells, alovudine depleted mitochondrial DNA, reduced mitochondrial encoded proteins, decreased basal oxygen consumption, and decreased cell proliferation and viability. To evaluate the effects of polymerase γ inhibition with alovudine in vivo, mice were xenografted with OCI-AML2 cells and then treated with alovudine. Systemic administration of alovudine reduced leukemic growth without evidence of toxicity and decreased levels of mitochondrial DNA in the leukemic cells. We also showed that alovudine increased the monocytic differentiation of AML cells. Genetic knockdown and other chemical inhibitors of polymerase γ also promoted AML differentiation, but the effects on AML differentiation were independent of reductions in oxidative phosphorylation or respiratory chain proteins. Thus, we have identified a novel mechanism by which mitochondria regulate AML fate and differentiation independent of oxidative phosphorylation.

Moreover, we highlight polymerase γ inhibitors, such as alovudine, as novel therapeutic agents for AML.

Introduction Acute myeloid leukemia (AML) is a hematologic malignancy with a poor prognosis, characterized by clonal, pathological and often poorly differentiated hematopoietic cells that infiltrate the bone marrow (BM), blood and extramedullary tissues. In spite of some recent new therapies for AML, the disease is curable in only up to 40% of adults under 60 years of age, and older patients and those with high-risk cytogenetics have a dismal outcome, with a 2-year survival rate of <10%.1,2 Thus, new treatment options for this disease are required.

As previously described by us and others, a subset of AML cells present unique features that make them more vulnerable to impairment of mitochondrial function, such as increased mitochondrial biogenesis, decreased spare reserve capacity, and increased dependence on oxidative phosphorylation compared to normal hematopoietic progenitor cells.3-6 Based on these data, our current study focuses on inhibiting polymerase gamma (POLG), the sole mitochondrial DNA polymerase, as a new therapeutic target for this disease. haematologica | 2019; 104(5)

Correspondence: AARON D. SCHIMMER aaron.schimmer@uhn.ca Received: April 11, 2018.

Accepted: December 17, 2018.

Pre-published: December 20, 2018. doi:10.3324/haematol.2018.195172 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/963 ©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.

963

D. Yehudai et al.

Eukaryotic cells have two separate genomes: nuclear DNA (nuDNA), organized in chromosomes, and the circular mitochondrial DNA (mtDNA), organized within nucleoids located within the mitochondria.

Mitochondrial DNA is comprised of a double-stranded circular genome that is 16.6 kb in length and lacks introns. It encodes two rRNAs, 22 t-RNAs, and 13 of the 90 proteins in the mitochondrial respiratory chain, the cascade of enzymes central to ATP production in the mitochondria via oxidative phosphorylation.2 Synthesis of the 13 proteins encoded by the mtDNA occurs in the mitochondrial matrix on mitochondrial ribosomes using mitochondrial-specific protein synthesis machinery.7,8 The remaining approximately 1300 mitochondrial proteins are encoded by nuclear genes and translated in the cytoplasm.

Contrary to nuDNA, that replicates once during cell division, mtDNA is continuously replicated independently from the cell cycle.9 Mitochondria contain their own specialized machinery for DNA replication, transcription and translation of the mitochondrial genome.

Loss of integrity of mtDNA results in dysfunctional respiratory complexes and negatively affects the production of ATP. In humans, the nuclear-encoded mitochondrial POLG is the sole polymerase responsible for mitochondrial DNA replication. Human POLG is comprised of a catalytic polymerase domain at the C-terminus and an exonuclease domain separated by a linker region at the N-terminus. The holoenzyme consists of the primary subunit POLG and a homodimeric form of its accessory subunit POLG2.10 POLG forms a multi-protein-DNA complex, termed nucleoid, which acts as a hub for mtDNA replication, transcription and translation.11 Mitochondrial and cytoplasmic pathways support mtDNA biosynthesis by supplying it with sufficient nucleotide pools. The former, the mitochondrial nucleotide salvage pathway, converts nucleoside precursors to nucleotides by a cascade of kinases within the mitochondria,12 while in the latter, the cytoplasmic pathway, kinases catalyze the phosphorylation of nucleosides to nucleotides in the cytoplasm. Nucleosides in this cytoplasmic pathway are synthesized from de novo biosynthesis,13,14 and following their phosphorylation to nucleotides they are imported into the mitochondria by specific nucleotide transporters.15,16 In this current study, we investigated the POLG inhibitor, dideoxynucleoside analog of thymidine, alovudine (3'-deoxy-3'-fluorothymidine, FLT), in AML. We evaluated the effects of alovudine on mitochondrial function as well as on stemness and differentiation in AML.

Methods Cell lines and primary samples OCI-AML2, MV4-11 and K562 cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM) augmented with 10% fetal bovine serum (FBS) and antibiotics. TEX cells (a gift from Dr. J. Dick) were cultured in IMDM augmented with 20% FBS, 2 mM L-glutamine, 2 ng/mL human Interleukin-3 (IL-3), 20 ng/mL human stem cell factor (SCF) (R&D Systems) and antibiotics. NB4 cells were cultured in RPMI 1640 medium augmented with 10% FBS. A total of 8227 cells (a gift from Dr. J. Dick) were cultured in X-VIVO 10 supplemented with 20% BIT (StemCell Technologies, 964

Vancouver, Canada), 10 ng/mL hIL-3, 50 ng/mL hSCF, 10 ng/mL G-CSF, and 25 ng/mL TPO (Pepro Tech).

Peripheral blood was collected from consenting patients with AML. Samples with at least 80% leukemic blasts among low-density cells isolated by Ficoll density gradient centrifugation were included in this analysis. Normal hematopoietic cells were derived from healthy volunteers donating peripheral blood stem cells (PBSCs) for allogeneic stem cell transplantation following granulocyte colony-stimulating factor (G-CSF) mobilization. Primary AML and normal hematopoietic mononuclear cells were cultured in Iscove-modified Dulbecco medium (IMDM) supplemented with 20% FBS, 2 mM L-glutamine, 2 ng/mL human IL-3, and 20 ng/mL human SCF. Sample collection and the use of human tissue were approved by the University Health Network institutional review.

All cell lines and primary samples used in our experiments were incubated at 37°C and 5% CO2 in humidified atmosphere.

Xenograft models of human acute myeloid leukemia For in vivo studies, alovudine was supplied by Medivir AB (Huddinge, Sweden). OCI-AML2 leukemia cells (1x106) were injected subcutaneously into the flanks of severe combined immune deficient (SCID) mice (Ontario Cancer Institute, Toronto, ON, Canada). After the appearance of a palpable tumor (9-11 days), the mice were treated orally with alovudine (50 mg/kg) twice daily or vehicle (saline) control (n=10 per group) at a treatment schedule of 5 out of 7 days for a total of 21 days (total number of 27 doses). Tumors were measured 3 times a week based on caliper measurements of tumor length and width (volume=tumor length x width2 x 0.5236). At the end of treatment, mice were sacrificed and tumor volumes and mass were measured from excised tumors. Mitochondrial DNA (mtDNA) assessment was also carried out from excised tumors, using qRT-PCR. To test alovudine efficacy in a primary AML engraftment mouse model, a frozen aliquot of primary AML cells was thawed, counted, and re-suspended in phosphate-buffered saline. Viable trypan blue-negative cells (2.5x106) were injected into the right femur of 10-week old female NOD-SCID mice that were sublethally irradiated (2 Gy), and pretreated with 200 µg of antimouse CD-122. Mice were treated once daily with oral alovudine at 25 mg/kg or vehicle (saline) control (n=10 per group) 5 out of 7 days for 24 days (total number of 17 doses). Mice were then sacrificed, femurs flushed, and primary AML engraftment (CD45+CD33+CD19- cells) in the left femur was determined by flow cytometry.

In vivo studies were performed according to the regulations of the Canadian Council on Animal Care and with the approval of the Ontario Cancer Institute Animal Ethics Review Board.

Results The nucleoside analog alovudine depletes mitochondrial DNA and reduces cell growth and viability in acute myeloid leukemia cells Alovudine (Figure 1A) is a dideoxynucleoside analog of thymidine, originally developed as a reverse transcriptase inhibitor and evaluated in patients for the treatment of HIV.17-19 In addition to inhibiting viral replication, alovudine triphosphate is incorporated into DNA by POLG preferentially over nuclear polymerases, resulting in chain termination and inhibition of its enzymatic activity in cellfree assays.20 However, its anti-leukemic activity has not been previously described. Therefore, we focused our further investigations on this compound. haematologica | 2019; 104(5)

POLG inhibition promotes AML differentiation

Alovudine reduces mtDNA and impairs mitochondrial function in acute myeloid leukemia cells To investigate the effects of alovudine on mitochondrial function, OCI-AML2 and MV4-11 leukemia cells were treated with increasing concentrations of alovudine. Six days after incubation, changes in mtDNA and bioenergetics were measured. Alovudine decreased mtDNA in both OCI-AML2 and MV4-11 cells, although MV4-11 cells were more sensitive with reductions in mtDNA observed at low nM concentrations and >80% reductions in mtDNA observed with 25 nM of the drug (Figure 1B). In

A

contrast to the large reduction in levels of mtDNA, there was only a small reduction in mitochondrial mass after 6 days of alovudine treatment (Online Supplementary Figure S1) and no change in nuclear DNA (Online Supplementary Figure S2). In keeping with reductions in mtDNA, we noted prominent reductions in protein levels of the mtDNA-encoded respiratory chain IV complex subunits, mt-COXI and mt-COX II (Figure 1C). In contrast, no changes were seen in levels of nu-COX IV, a subunit of the same respiratory chain complex IV, but encoded by nuclear DNA (Figure 1C). mt-COX I and mt-COX II

B

Figure 1. Alovudine inhibits mitochondrial DNA biosynthesis and oxidative phosphorylation in acute myeloid leukemia (AML). (A) Alovudine’s chemical structure. (B) OCI-AML2 and MV4-11 cells were treated with increasing concentrations of alovudine for 6 days.

Relative mitochondrial DNA (mtDNA) content was assessed by qRT-PCR as described in the Methods section. Data represent mean+Standard Deviation (SD) mtDNA relative to untreated controls from one of three representative experiments. (C) OCI-AML2 and MV411 were treated with increasing concentrations of alovudine. Three and 6 days after treatment, cells were harvested, lysed and levels of cytochrome C oxidase subunits: mitochondrial COXI (mt-COX1), mitochondrial COXII (mt-COXII), nuclear COX IV (nu-COX IV), and β-tubulin were measured by immunoblotting.

The immunoblot from one of three representative experiments is shown. (D) Basal oxygen consumption rate (OCR) was assessed in OCIAML2 and MV411 cells following 6 days of alovudine treatment, using the Seahorse XF96 Metabolic Flux Assay. Data represent the mean±SD basal OCR from one of three representative experiments. n=6. (E) OCIAML2 and MV411 cells were treated with increasing concentrations of alovudine for 6 days. ATP production was assessed by CellTitre-Glo Luminescent Cell Viability Assay.

Data represent the mean±SD from two independent experiments in triplicate. (F) OCI-AML2 and MV411 cells were treated with increasing concentrations of alovudine. Cell growth and viability was assessed by trypan blue exclusion staining at increasing times after incubation.

Data represent the mean±SEM from one of three representative experiments. For all experiments, ***P<0.001 and ****P<0.0001 using Dunnett’s multiple comparisons test after one-way ANOVA (B, D and E). (F) Two-way ANOVA.

C

D

E

F

haematologica | 2019; 104(5)

965

D. Yehudai et al.

mRNA were also decreased >95%, but little change in the mRNA expression of nu-COX IV was seen (Online Supplementary Figure S3). We also observed reductions in basal oxygen consumption at concentrations associated with reductions in mtDNA and respiratory chain proteins (Figure 1D). Of note, alovudine did not reduce ATP level in treated cells (Figure 1E).

A

Finally, we observed reductions in cell proliferation and viability at concentrations of alovudine that reduced mtDNA and inhibited oxidative phosphorylation. Of the tested AML cell lines (MV4-11, OCI-AML2, TEX, NB4, and K562), MV4-11 was the most sensitive to alovudine (Figure 1F and Online Supplementary Figure S4A-C).

Compared to the other tested cell lines, MV4-11 had high-

Figure 2. Effects of alovudine on primary acute myeloid leukemia (AML) and normal hematopoietic cells. Primary AML samples and normal peripheral blood stem cells (PBSCs) were treated with 2000 nM alovudine for 6 days. (A) Relative mitochondrial DNA (mtDNA) was assessed by qRT-PCR in AML patients (n=2). (B) Primary AML and normal hematopoietic cells (PBSCs) were treated with 2000 nM of alovudine for 6 days. Cell viability was assessed by trypan blue exclusion staining in primary AML cells (n=7) and CellTiter-Fluor for PBSCs (n=3). The dotted line represents cell viability of DMSO control. (C and D) Primary AML and normal hematopoietic cells (PBSCs) were treated with increasing concentrations of alovudine for 6 days. Colony forming abilities were assessed in primary AML cells and PBSCs (n=2 of each) as described above in the Online Supplementary Methods. PBSC samples 1 and 2 were the same samples as used in (B). For all experiments, ns: non-significant; *P<0.5, **P<0.01, ***P<0.001, and ****P<0.0001 using Sidakâ&#x20AC;&#x2122;s (A) or Dunnettâ&#x20AC;&#x2122;s (B-D) multiple comparisons test after one-way ANOVA. BFU-E: primitive erythroid progenitor cells; CFU-GM: granulocyte precursors.

B

C

D

966

haematologica | 2019; 104(5)

POLG inhibition promotes AML differentiation

est expression of POLG (Online Supplementary Figure S4D).

There was no increase in apoptosis as measured by Annexin V staining 6 days after alovudine treatment, a time at which reductions in cell viability and mtDNA biosynthesis were already evident. However, cell death was observed after more prolonged (11-day) incubation with alovudine (Online Supplementary Figure S4E).

Furthermore, analysis of the cell cycle in alovudine-treated AML cells showed little or no change (Online Supplementary Figure S5).

Next, we investigated the effects of alovudine on primary AML cells and normal hematopoietic cells. Primary AML and normal hematopoietic cells were treated for 6 days with increasing concentrations of alovudine. After incubation, we measured cell viability and levels of mtDNA. The number of viable primary cells treated with the vehicle control did not increase during the period of incubation, suggesting that the primary cells were quiescent and did not proliferate. Alovudine reduced levels of mtDNA in primary AML cells at concentrations that also reduced cell viability (Figure 2A and B and Online Supplementary Table S1). In contrast, alovudine did not reduce the viability of the normal hematopoietic cells (Figure 2B). Furthermore, alovudine reduced the clonogenic growth of primary AML cells. We also tested the effects of alovudine on the clonogeneic growth of normal hematopoietic cells. Compared to AML cells, normal hematopoietic cells were more resistant to alovudine.

However, at the highest tested dose (2000 nM), alovudine partially reduced the clonogenic growth of granulocyte precursors (CFU-GM) and completely inhibited the growth of primitive erythroid progenitor cells (BFU-E) (Figure 2C and D and Online Supplementary Table S1).

Thus, alovudine inhibits mtDNA biosynthesis in primary AML cells and selectively targets a subset of AML cells in vitro.

treatment dose, an alovudine-resistant primary AML sample, or a protective effect from the marrow niche.

Alovudine promotes monocytic differentiation in acute myeloid leukemia Recent studies reported that mitochondrial pathways and metabolism can regulate the differentiation of malignant cells.21-24 Therefore, we explored the effects of alovudine on the differentiation of AML cells. OCI-AML2 and MV4-11 cells were treated with concentrations of alovudine that depleted mtDNA for 10 days. Alovudine increased expression of CD11b, a cell surface marker associated with monocytic differentiation (Figure 4A). It also induced morphological changes typical of monocytic dif-

A

B

Alovudine displays efficacy in mouse models of human leukemia Given the anti-leukemic effects of alovudine in vitro, we examined the efficacy and toxicity of alovudine in a mouse model of leukemia. Severe combined immune deficient (SCID) mice xenografted with OCI-AML2 cells were treated with alovudine (50 mg/kg bid) or vehicle control by oral gavage. Alovudine reduced the growth of leukemia in vivo by approximately 70% without evidence of toxicity (Figure 3A). Specifically, doses of alovudine that reduced tumor growth did not alter mouse body weight, behavior, serum chemistries, or organ histology (Figure 3B and Online Supplementary Figure S6). Finally, we conducted correlative studies and measured mtDNA in leukemic cells isolated from mice treated with alovudine. We observed reductions in mtDNA by greater than 75% in OCI-AML2 tumors excised from mice treated with alovudine (Figure 3C).

Next, we tested whether alovudine targets primary AML cells in vivo. Sublethally irradiated NOD-SCID mice preconditioned with anti-CD122 received intrafemural injection of primary AML cells. In this mouse model, the maximum tolerated dose of alovudine was only 25 mg/kg/day and less than the dose tolerated by the SCID mice engrafted with OCI-AML2 cells. Therefore, mice were treated with 25 mg/kg alovudine for 5 of 7 days for 17 days. A small, but statistically significant reduction in primary AML engraftment was observed (Online Supplementary Figure S7), potentially reflecting the reduced haematologica | 2019; 104(5)

C

Figure 3. Alovudine displays efficacy in mouse models of human acute myeloid leukemia. OCI-AML2 cells were injected subcutaneously into the ďŹ&#x201A;ank of SCID mice. Once tumors were palpable, mice were treated with oral (PO) alovudine (50 mg/kg bid) or vehicle control for 14 days. (A) Tumor volume was assessed from excised tumors. The error bars represent Standard Deviation (SD) (n=10 per group). (B) Body weight was assessed every 2-3 days. The error bars represent SD (n=10 per group). (C) Relative mitochondrial DNA (mtDNA) from xenograft tumors excised in (B) (n=5 per group). For all experiments, ***P<0.001 and ****P<0.0001 using two-way ANOVA (A and B) and t-test (C).

967

D. Yehudai et al.

ferentiation by Giemsa stain (Figure 4B). Finally, alovudine decreased global DNA methylation (Figure 4C), in keeping with prior studies showing a reduction in methylation correlates with differentiation.25-27 Next, we evaluated alovudine in the 8227 primary AML culture model; 8227 leukemia cells are patient-derived cells that are organized into a hierarchy of stem and bulk cells with the stem cells residing in the CD34+CD38– com-

partment.28 Treatment of 8227 cells with alovudine decreased the CD34+CD38– stem cells (Figure 4D).

Inhibition of polymerase gamma but not reductions in oxidative phosphorylation or respiratory chain proteins influence acute myeloid leukemia differentiation We then explored the mechanism by which alovudine promoted AML differentiation. We tested whether reduc-

A

B

C

D

968

Figure 4. Alovudine promotes monocytic differentiation in acute myeloid leukemia. (A) OCI-AML2 and MV4-11 cells were treated with increasing concentrations of alovudine for 10 days. CD11b expression was assessed by flow cytometry. (B) MV4-11 cells were treated with alovudine (200 nM) for 9 days.

Morphology was assessed by Giemsa stain (magnification 40X, scale bar=60 µm). (C) OCI-AML2 and MV4-11 cells were treated with alovudine for 6 days.

Methylation was assessed by dot-blot assay. (D) 8227 cells were treated with increasing concentrations of alovudine for 6 days. CD34+/CD38– expression was assessed by flow cytometry. For all experiments, *P<0.5, **P<0.01, ***P<0.001, and ****P<0.0001 using Dunnett's multiple comparisons test after one-way ANOVA. Data represent the mean+Standard Deviaiton (SD) from one of three representative experiments, except (C) which represent the mean+SEM of average of 3 (MV4-11) or 4 experiments (OCI-AML2).

haematologica | 2019; 104(5)

POLG inhibition promotes AML differentiation

A

B

C

E

D

Figure 5. Effects of inhibiting mitochondrial translation and polymerase gamma (POLG) on acute myeloid leukemia growth and differentiation. (A-C) OCI-AML2 cells were treated with increasing concentrations of Tigecycline for 72 hours (h). (A) Cell growth and viability was assessed by Alamar Blue assay [Mean+Standard Deviation (SD)]. (B) Cells were harvested 72 h after treatment, lysed, and levels of cytochrome C oxidase subunits: mitochondrial COXI (mt-COXI), mitochondrial COXII (mt-COXII), nuclear COX IV (nu-COX IV), and β-tubulin measured by immunoblotting. The immunoblot from a representative experiment is shown. (C) CD11b expression was assessed by flow cytometry. (D) OCI-AML2 cells were treated with increasing concentrations of antiviral ddC. Cell growth and viability was assessed by trypan blue exclusion staining at increasing times after incubation. Data represent the mean±Standard Error of Mean (SEM) from one of three representative experiments.

(E) OCI-AML2 cells were treated with increasing concentrations of 2'3'-dideoxycytidine (ddC) for 10 days. CD11b expression was assessed by flow cytometry. For all experiments, *P<0.5, **P<0.01, ***P<0.001, and ****P<0.0001 using Dunnett's multiple comparisons test after one-way ANOVA (C) and two-way ANOVA (D). (E) Unpaired t-test.

ing levels of mitochondrial encoded respiratory chain proteins and basal oxygen consumption without altering levels of mtDNA would be sufficient to mimic the effects of alovudine and induce AML differentiation. We treated OCI-AML2 cells with tigecycline that we previously showed inhibits mitochondrial protein translation by targeting mitochondrial ribosomes. Consistent with our previous studies,4 tigecycline reduced levels of mtDNAencoded proteins but not nuclear-DNA-encoded proteins and reduced the proliferation and viability of AML cells (Figure 5A and B). However, there was no evidence of differentiation (Figure 5C).

Then, we asked whether other chemical or genetic inhibitors of POLG would also promote AML differentiation. We treated OCI-AML2 cells with the antiviral 2'3'-dideoxycytidine (ddC) that we and others previously showed inhibited POLG.6,20 Similar to alovudine, ddC reduced mtDNA and reduced levels of mitochondrial encoded respiratory chain proteins.6 Like alovudine, ddC also reduced AML growth (Figure 5D) and promoted AML differentiation (Figure 5E). As a genetic approach, we knocked down POLG with shRNA in lentiviral vectors. Partial POLG knockdown reduced mtDNA, but the reductions in mtDNA were not sufficient to significantly reduce levels of respiratory chain proteins or impair oxidative phosphorylation. However, partial POLG knockdown mimicked the effects of alovudine and reduced AML proliferation, and induced differentiation of AML cells (Figure 6A-F).

Thus, taken together, our data suggest that inhibition of POLG promotes AML differentiation through effects that haematologica | 2019; 104(5)

are independent of reductions in oxidative phosphorylation or respiratory chain proteins.

Discussion Inhibiting nuclear DNA replication by targeting nuclear DNA polymerases is the primary mechanism of many standard anti-leukemic agents, including cytarabine, clofarabine and cladribine. Some nucleoside analogs also cross react with the mitochondrial polymerase, POLG, and thereby inhibit mitochondrial DNA replication. Given that a subset of AML cells and stem cells have increased mitochondrial mass, mitochondrial DNA, and a greater reliance on oxidative phosphorylation,3,5,29 we sought to identify new nucleoside analogs that deplete mitochondrial DNA. In this current study, we evaluated alovudine as a potent nucleoside analog targeting mtDNA in AML. Alovudine is a dideoxynucleoside analog of thymidine, which inhibits mtDNA polymerase activity through its 5’triphosphate metabolite. First introduced in the early 90s, this drug was originally developed as a reverse transcriptase inhibitor and evaluated in patients for the treatment of HIV.30-32 More recently, it has been labeled with 18F and used as a reagent in PET imaging to visualize malignancy.33-35 We demonstrated that inhibition of POLG with alovudine inhibits the viability and proliferation of AML cells in vitro and in vivo. Human mitochondrial DNA is replicated exclusively by the mitochondrial DNA polymerase gamma. The holoenzyme is a heterotrimer which consists 969

D. Yehudai et al.

of the primary subunit POLG and a homodimeric form of its accessory subunit POLG2.10 The nuclear-encoded primary subunit contains a C-terminal catalytic polymerase domain and N-terminal exonuclease domain separated by a linker/spacer region.36 The polymerase domain is responsible for extension of the DNA strand, and similar to nuclear-DNA polymerase, utilizes deoxynucleotide triphosphates (dNTPs) as its substrate to the growing chain.37 Any alteration of dNTP (i.e. analogs of dNTPs, called ddNTPs) can inhibit POLG function through chain termination. We showed that alovudine decreased the growth of OCI-AML2 cells xenografted into SCID mice.

A

C

E

However, in sublethally irradiated NOD/SCID mice engrafted by intrafemural injection with a primary AML patient sample, alovudine produced only a small reduction in primary AML engraftment. Unfortunately, in this mouse model, the maximum tolerated dose of alovudine was only 25 mg/kg/day. In contrast, in SCID mice engrafted with OCI-AML2 cells, the maximum tolerated dose was over 50 mg/kg/day. We suspect that the difference in the maximal tolerated dose is due to differences in the mouse strains or the sublethal irradiation that the NOD/SCID mice received prior to engraftment with primary AML cells. Alternatively, the reduced efficacy in the

B

D

F

Figure 6. Polymerase gamma (POLG) knockdown promotes acute myeloid leukemia differentiation. OCI-AML2 cells were infected with shRNA targeting POLG or a control sequenced in lentiviral vectors. (A) POLG knockdown at 14 days post transduction was assessed by qRT-PCR. Data represent the mean+Standard Deviation (SD) mtDNA from one of three representative experiments. (B) Relative mitochondrial DNA (mtDNA) content was assessed by qRT-PCR 14 days post transduction with POLG shRNA, as described in the “Methods” section. Data represent the mean+Standard Error of Mean (SEM) from two independent experiments. (C) Cells were harvested 11, 14 and 18 days post POLG knockdown and levels of cytochrome C oxidase subunits: mitochondrial COXI (mt-COX1), nuclear COX IV (nu-COX IV), and β-tubulin in whole-cell extracts were measured by immunoblotting. The immunoblot from one of three representative experiments is shown. (D) Basal oxygen consumption rate (OCR) was assessed 10 days following POLG knockdown, using the Seahorse XF96 Metabolic Flux Assay. Data represent Mean±SEM basal OCR from two independent experiments; n=6. (E) Cell growth and viability were assessed by trypan blue exclusion staining. Data represent the mean±SD from one of three representative experiments. (F) CD11b expression was assessed by flow cytometry. Data represent the mean±SD from one of three representative experiments. For all experiments, *P<0.5, **P<0.01, ***P<0.001, and ****P<0.0001 using Dunnett's multiple comparison following one-way ANOVA (A and F) or two-way ANOVA (B and D). (E) Two-way ANOVA test.

970

haematologica | 2019; 104(5)

POLG inhibition promotes AML differentiation

primary AML model may reflect protection from the BM niche, or that the tested sample was resistant to alovudine.

Further studies will be needed to clarify the basis for this sensitivity and resistance.

In clinical trials with patients with AIDS and HIV, alovudine reduced viral load but its development for HIV did not progress beyond phase II and small scale phase III trials.17,30,36 In these trials, alovudine was associated with dose-dependent reversible leukopenia and anemia.19 Although myelosuppression was not an acceptable toxicity for HIV, it may be a more manageable side effect for an anti-leukemic therapy. Moreover, our data suggest that alovudine can preferentially target AML cells over normal hematopoeitic cells. Thus, given the known toxicology and pharmacology of alovudine, along with the prior experience of its clinical use, it could potentially be rapidly repositioned for the treatment of AML. Using a combination of pharmacological and genetic approaches, we discovered that inhibition of POLG induces the monocytic differentiation of AML cells.

Thus, we identified new mechanisms by which mitochondrial pathways control differentiation in AML and highlight alovudine as a novel potential therapeutic agent for AML. Most therapies currently in use or under investigation for AML are cytotoxic and induce cell death. A less explored therapeutic strategy is to promote the differentiation of AML cells into more mature progeny. Upon differentiation, the leukemic cells will cease to proliferate or die. Targeting the block in differentiation is the standard therapy for acute promyelocytic leukemia (APL)-AML.39 There is growing evidence that targeting mitochondrial pathways can impact cell fate and differentiation of AML beyond the subset of APL.22-24 For example, the cytosolic isocitrate dehydrogenase 1 (IDH1) and its mitochondrial homolog IDH2 encode enzymes that convert isocitrate to a-ketoglutarate (aKG), a co-factor for TET2 that demethylates DNA. Mutations in IDH1 and IDH2 modify the affinity between the encoded

References 1. DĂśhner H, Weisdorf DJ, Bloomfield CD. Acute Myeloid Leukemia. N Engl J Med. 2015;373(12):1136-1152.

2. Lang BF, Gray MW, Burger G. Mitochondrial genome evolution and the origin of eukaryotes. Annu Rev Genet.

1999;33(1):351-397.

3. Lagadinou ED, Sach A, Callahan K, et al.

BCL-2 inhibition targets oxidative phosphorylation and selectively eradicates quiescent human leukemia stem cells. Cell Stem Cell. 2013;12(3):329-341.

4. SkrtiÄ&#x2021; M, Sriskanthadevan S, Jhas B, et al.

Inhibition of mitochondrial translation as a therapeutic strategy for human acute myeloid leukemia. Cancer Cell. 2011; 20(5):674-688.

5. Sriskanthadevan S, Jeyaraju DV, Chung TE, et al. AML cells have low spare reserve capacity in their respiratory chain that renders them susceptible to oxidative metabolic stress. Blood. 2015;125(13):21202130.

6. Liyanage SU, Hurren R, Voisin V, et al.

haematologica | 2019; 104(5)

7.

8.

9.

10.

11.

12.

enzymes and their substrates, resulting in conversion of aKG to R-2-hydroxyglutarate (R-2-HG) which inhibits both TET2 and other aKG-dependent enzymes.40 The result is an increase in DNA methylation, altered gene expression, and a block in differentiation. Inhibiting mutant IDH1/2 restores normal IDH function, decreases DNA methylation, and promotes the differentiation of AML cells in vitro and in vivo.21,41 In this study, we report a new mechanism by which mitochondrial pathways control differentiation. We demonstrated that inhibition of POLG and/or reductions in levels of mitochondrial DNA that were not sufficient to impair oxidative phosphorylation led to increased monocytoid differentiation. Previous studies have reported that mitochondrial stress such as inhibition of the mitochondrial protease ClpP42,43 or oxidative stress44 can result in translocation of mitochondrial proteins to the nucleus and alter gene expression. Potentially similar mechanisms could occur upon inhibition of POLG and lead to translocation of proteins from the mitochondria to the nucleus, impacting gene expression to promote differentiation.

Thus, we have identified a novel mechanism by which mitochondria regulate AML fate and differentiation independently of oxidative phosphorylation. Moreover, we highlight POLG inhibitors such as alovudine as potential therapeutic agents for AML. Acknowledgments We thank Jill Flewelling (Princess Margaret Cancer Center) for administrative assistance.

Funding This work was supported by Medivir AB, the Leukemia and Lymphoma Society, the Canadian Institutes of Health Research, the Princess Margaret Cancer Centre Foundation, and the Ministry of Long Term Health and Planning in the Province of Ontario. ADS holds the Barbara Baker Chair in Leukemia and Related Diseases

Leveraging increased cytoplasmic nucleoside kinase activity to target mtDNA and oxidative phosphorylation in AML. Blood.

2017;129(19):2657-2666.

Gaur R, Grasso D, Datta PP, et al. A single mammalian mitochondrial translation initiation factor functionally replaces two bacterial factors. Mol Cell. 2008;29(2):180-190.

Christian BE, Spremulli LL. Mechanism of protein biosynthesis in mammalian mitochondria.

Biochim Biophys Acta.

2012;1819: 1035-1054.

Birky CW. Eukaryotic Origins. Science.

1994;266(5183):309-310.

Chan SSL, Copeland WC. DNA polymerase gamma and mitochondrial disease: Understanding the consequence of POLG mutations. Biochim Biophys Acta Bioenerg. 2009;1787(5):312-319.

Bogenhagen DF. Mitochondrial DNA nucleoid structure. Biochim Biophys Acta.

2012;1819(9-10):914-920.

Carling PJ, Cree LM, Chinnery PF. The implications of mitochondrial DNA copy number regulation during embryogenesis.

Mitochondrion. 2011;11(5):686-692.

13. Lane AN, Fan TW-M. Regulation of mammalian nucleotide metabolism and biosynthesis. Nucleic Acids Res. 2015;43(4):24662485.

14. Mathews CK. Deoxyribonucleotide metabolism, mutagenesis and cancer. Nat Rev Cancer. 2015;15(9):528-539.

15. Kakuda TN. Pharmacology of nucleoside and nucleotide reverse transcriptase inhibitor-induced mitochondrial toxicity.

Clin Ther. 2000;22(6):685-708.

16. Gandhi VV, Samuels DC. A review comparing deoxyribonucleoside triphosphate (dNTP) concentrations in the mitochondrial and cytoplasmic compartments of normal and transformed cells. Nucleosides Nucleotides Nucleic Acids. 2011;30(5):317339.

17. Katlama C, Ghosn J, Tubiana R, et al. MIV310 reduces HIV viral load in patients failing multiple antiretroviral therapy: results from a 4-week phase II study. AIDS. 2004;18(9):1299-1304.

18. Rusconi S. Alovudine Medivir. Curr Opin Investig Drugs. 2003;4(2):219-223.

19. Flexner C, van der Horst C, Jacobson MA,

971

D. Yehudai et al.

20.

21.

22.

23.

24.

25.

26.

27.

28.

972

et al. Relationship between plasma concentrations of 3’-deoxy-3’-fluorothymidine (alovudine) and antiretroviral activity in two concentration-controlled trials. J Infect Dis. 1994;170(6):1394-1403.

Martin JL, Brown CE, Matthews-Davis N, Reardon JE. Effects of antiviral nucleoside analogs on human DNA polymerases and mitochondrial DNA synthesis. Antimicrob Agents Chemother. 1994;38(12):2743-2749.

Birendra KC, DiNardo CD. Evidence for Clinical Differentiation and Differentiation Syndrome in Patients With Acute Myeloid Leukemia and IDH1 Mutations Treated With the Targeted Mutant IDH1 Inhibitor, AG-120. Clin Lymphoma Myeloma Leuk.

2016;16(8):460-465.

Miller WH, Waxman S. Differentiation induction as a treatment for hematologic malignancies. Oncogene. 2002;21(21):34963506.

Stegmaier K, Corsello SM, Ross KN, Wong JS, Deangelo DJ, Golub TR. Gefitinib induces myeloid differentiation of acute myeloid leukemia. Blood. 2005; 106(8): 2841-2848.

Waxman S. Differentiation therapy in acute myelogenous leukemia (non-APL).

Leukemia. 2000;14:491-496.

Bröske A-M, Vockentanz L, Kharazi S, et al.

DNA methylation protects hematopoietic stem cell multipotency from myeloerythroid restriction. Nat Genet. 2009;41(11): 1207-1215.

Gereige L-M, Mikkola HKA. DNA methylation is a guardian of stem cell self-renewal and multipotency. Nat Genet 2009;41(11): 1164-1166.

Challen GA, Sun D, Jeong M, et al. Dnmt3a is essential for hematopoietic stem cell differentiation. Nat Genet. 2012;44(1):23-31.

Lechman ER, Gentner B, Ng SWK, et al.

29.

30.

31.

32.

33.

34.

35.

miR-126 Regulates Distinct Self-Renewal Outcomes in Normal and Malignant Hematopoietic Stem Cells. Cancer Cell.

2016;29(4):602-606. Škrtić M, Sriskanthadevan S, Jhas B, et al.

Inhibition of Mitochondrial Translation as a Therapeutic Strategy for Human Acute Myeloid Leukemia. Cancer Cell. 2011; 20(5):674-688.

Ghosn J, Quinson A-M, Sabo N, et al.

Antiviral activity of low-dose alovudine in antiretroviral-experienced patients: results from a 4-week randomized, double-blind, placebo-controlled dose-ranging trial. HIV Med. 2007;8(3):142-147.

Otto MJ. New nucleoside reverse transcriptase inhibitors for the treatment of HIV infections. Curr Opin Pharmacol. 2004; 4(5):431-436.

Kim E-Y, Vrang L, Öberg B, Merigan TC. Anti-HIV Type 1 Activity of 3′-Fluoro-3′Deoxythymidine for Several Different Multidrug-Resistant Mutants. AIDS Res Hum Retroviruses. 2001;17(5):401-407.

Shen G, Ma H, Pang F, Ren P, Kuang A. Correlations of 18F-FDG and 18F-FLT uptake on PET with Ki-67 expression in patients with lung cancer: a meta-analysis.

Acta Radiol. 2018;59(2):188-195.

Bahri H, Laurence L, Edeline J, et al. High Prognostic Value of 18F-FDG PET for Metastatic Gastroenteropancreatic Neuroendocrine Tumors: A Long-Term Evaluation. J Nucl Med. 2014;55(11):17861790.

Gerbaudo VH, Killoran JH, Kim CK, et al.

Pilot study of serial FLT and FDG-PET/CT imaging to monitor response to neoadjuvant chemoradiotherapy of esophageal adenocarcinoma: correlation with histopathologic response. Ann Nucl Med. 2018;32(3):165-174.

36. Lee Y-S, Kennedy WD, Yin YW. Structural Insight into Processive Human Mitochondrial DNA Synthesis and DiseaseRelated Polymerase Mutations. Cell.

2009;139(2):312-324.2018;32(3):165-174.

37. Wang TS. Eukaryotic DNA polymerases.

Annu Rev Biochem. 1991;60(1):513-552.

38. Martin JL, Brown CE, Matthews-Davis N, Reardon JE. Effects of antiviral nucleoside analogs on human DNA polymerases and mitochondrial DNA synthesis. Antimicrob Agents Chemother. 1994;38(12):2743-2749.

39. Warrell RP, Frankel SR, Miller WH, et al.

Differentiation therapy of acute promyelocytic leukemia with tretinoin (all-transretinoic acid). N Engl J Med. 1991; 324(20):1385-1393.

40. Figueroa ME, Abdel-Wahab O, Lu C, et al.

Leukemic IDH1 and IDH2 Mutations Result in a Hypermethylation Phenotype, Disrupt TET2 Function, and Impair Hematopoietic Differentiation. Cancer Cell. 2010;18(6):553-567.

41. Stein EM, DiNardo CD, Pollyea DA, et al.

Enasidenib in mutantIDH2relapsed or refractory acute myeloid leukemia. Blood.

2017;130(6):722-731.

42. Fiorese CJ, Schulz AM, Lin Y-F, Rosin N, Pellegrino MW, Haynes CM. The Transcription Factor ATF5 Mediates a Mammalian Mitochondrial UPR. Curr Biol.

2016;26(15):2037-2043.

43. Haynes CM, Yang Y, Blais SP, Neubert TA, Ron D. The matrix peptide exporter HAF-1 signals a mitochondrial UPR by activating the transcription factor ZC376.7 in C. elegans. Mol Cell. 2010;37(4):529-540.

44. Nagaraj R, Sharpley MS, Chi F, et al.

Nuclear Localization of Mitochondrial TCA Cycle Enzymes as a Critical Step in Mammalian Zygotic Genome Activation.

Cell. 2017;168:210-223.e11.

haematologica | 2019; 104(5)

ARTICLE

Acute Myeloid Leukemia

The small-molecule compound AC-73 targeting CD147 inhibits leukemic cell proliferation, induces autophagy and increases the chemotherapeutic sensitivity of acute myeloid leukemia cells

Isabella Spinello,1 Ernestina Saulle,1 Maria Teresa Quaranta,1 Luca Pasquini,2 Elvira Pelosi,3 Germana Castelli,3 Tiziana Ottone,4 Maria Teresa Voso,4 Ugo Testa3 and Catherine Labbaye1

National Center for Drug Research and Evaluation, Istituto Superiore di Sanità, Rome; Core Facilities, Istituto Superiore di Sanità; 3Department of Oncology and Molecular Medicine, Istituto Superiore di Sanità and 4Department of Biomedicine and Prevention, University of Rome “Tor Vergata”, Rome, Italy 1 2

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):973-985

ABSTRACT

C

D147 is a transmembrane glycoprotein with multiple functions in human healthy tissues and diseases, in particular in cancer.

Overexpression of CD147 correlates with biological functions that promote tumor progression and confers resistance to chemotherapeutic drugs. In contrast to solid tumors, the role of CD147 has not been extensively studied in leukemia. Understanding whether CD147 represents a new hematologic target and whether its inhibitor AC-73 may be used in leukemia therapy may reveal an alternative treatment strategy in patients with acute myeloid leukemia (AML). We analyzed CD147 expression and function in hematopoietic progenitor cells from normal cord blood, in several leukemic cell lines and in primary leukemic blasts obtained from patients with AML. We investigated the effects of AC-73, used alone or in combination with arabinosylcytosine (Ara-C) and arsenic trioxide (ATO), on leukemic cell proliferation. We demonstrated that CD147 overexpression promotes leukemic cell proliferation. We showed that AC-73 exhibits a potent growth inhibitory activity in leukemic cells, by inhibiting the ERK/STAT3 activation pathway and activating autophagy. We demonstrated that AC-73 exerts an anti-proliferative effect additive to chemotherapy by enhancing leukemic cell sensitivity to Ara-C-induced cytotoxicity or to ATO-induced autophagy.

We also reported CD147 expression in the fraction of leukemic blasts expressing CD371, a marker of leukemic stem cells. Altogether, our study indicates CD147 as a novel potential target in the treatment of AML and AC-73 as an anti-proliferative drug and an inducer of autophagy in leukemic cells to use in combination with chemotherapeutic agents.

Introduction Targeted therapy for acute myeloid leukemia (AML) represents an ongoing challenge and in this context, cluster of differentiation 147 (CD147) represents an attractive target for therapeutic intervention in AML and in other hematologic neoplasms.1-3 CD147, also known as basigin or extracellular matrix metalloproteinase inducer (EMMPRIN), is a type-I transmembrane glycoprotein that belongs to the immunoglobulin superfamily. Among the numerous studies that have documented the significance of CD147 in various physiological processes, the best characterized function of CD147 is related to its role in tumor metastasis, angiogenesis and chemoresistance processes.3-6 Overexpression of CD147 correlates with a number of biological functions that promote tumor progression (e.g. cellular proliferation, angiogenesis, matrix metalloproteinase production) and confers resistance to chemotherapeutic drugs such as adriamycin,7,8 cisplatin.9 CD147 mediates molecuhaematologica | 2019; 104(5)

Correspondence: CATHERINE LABBAYE catherine.labbaye@iss.it Received: June 11, 2018.

Accepted: November 15, 2018.

Pre-published: November 22, 2018. doi:10.3324/haematol.2018.199661 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/973 ©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.

973

I. Spinello et al.

lar events by interacting with various binding partners, such as tumor- and inflammation-associated molecules including integrins, monocarboxylate transporters (MCTs), cyclophilins, caveolin-1, and E-selectin, explaining its significant role in the pathogenesis of several diseases.3-6,10 CD147 overexpression and more recently its coexpression with MCTs11,12 are regarded as unfavorable prognostic factors in cancers associated with hypoxia, a common feature of solid tumors, but also a major component of the bone marrow (BM) microenvironment, crucial in leukemia progression.13,14 However, in contrast to solid tumors, the function of CD147 remains poorly defined in leukemia.

Recent studies have shown growing interest in the CD147 molecule in AML15,16 and in some hematologic neoplasia, in particular in multiple myeloma (MM), where CD147 expression levels have a prognostic value and are required for the proliferation of MM cells.17-19 Moreover, CD147 is over-expressed in erythroid cells of myelodysplastic syndrome (MDS) with 5q deletion.18 Here, we show that CD147 is expressed in normal CD34+ hematopoietic progenitor cells (HPCs) and down-regulated during monocytic and granulocytic differentiation of HPCs. We then show that CD147 is over-expressed in blasts pertaining to different subtypes of AML and promotes leukemic cell proliferation. Interestingly, we report that CD147 is expressed at the level of CD34+CD371+ AML cells, previously described for their leukemia-initiating properties.20 Recently, the small-molecule AC-73 has been proposed as a specific inhibitor for CD147.21 First, we checked that the response to AC-73 treatment is not involved in an offtarget mechanism in leukemic cells. Then, we analyzed the effects of CD147 inhibition by AC-73 in AML cell lines and in primary leukemic blasts. We found that AC73 inhibits leukemic cell proliferation by suppressing the ERK/STAT3 activation pathway, known to play a role in AML cell proliferation and survival,22 but also by activating autophagy, an essential phenomenon for hematopoietic stem cell (HSC) maintenance, resistance to stress, survival and differentiation, the machinery of which might be disrupted in AMLs.23-25 Next, we analyzed whether AC-73 enhanced the sensitivity of leukemic cells to conventional chemotherapeutic agents. We used arabinosylcytosine (Ara-C), one of the most active cytotoxic agents in myeloid leukemia, and arsenic trioxide (ATO), an active anti-proliferative agent used in the treatment of patients with acute promyelocytic leukemia (APL) (AML-M3)1,2,26 [although with low efficacy in AML lacking the t(15;17) translocation], and also an inducer of autophagy.25,27 We found that AC-73 used in vitro in combination with Ara-C or ATO, increases the effects of these agents.

Altogether, our data suggest that CD147 plays a key role in leukemic cell proliferation and represents a potential therapeutic target in AML patients, and that AC-73 is a new promising inhibitor that could be used in combination with conventional chemotherapeutic agents as a novel treatment strategy in AML.

Methods Cell cultures Human cord blood (CB) was obtained from healthy donors after informed consent. Leukemic blasts were isolated from BM 974

obtained from patients with newly diagnosed AML, using FicollHypaque density gradient. Informed consent was obtained from patients in accordance with the Declaration of Helsinki. This study was approved by the local ethical committees of the Istituto Superiore di Sanità and the University of Tor Vergata, Rome, Italy.

Cord blood CD34+ HPC purification, unilineage monocytic (Mo) and granulocytic (G) differentiation and morphological analyses were performed,28,29 as described in the Online Supplementary Methods.

Human primary AML blasts were maintained in culture in Iscove medium supplemented with 10% FCS, GM-CSF (10 ng/mL), SCF (50 ng/mL), IL-3 (10 ng/mL) (PeproTech Inc., Rocky Hill, NJ, USA), as described.29 Human AML cell lines used in our study were: U937 as a model of AML-M5, NB4 and HL-60 as models of AML-M3 and AMLM4, respectively; NB4-R4 as AML-M3 resistant to all-trans retinoic acid (ATRA) treatment; MV4-11 as AML-M2 mutated for FLT3ITD; Kasumi-1 as AML-M2 with the t(8;21) translocation. All cell lines were grown in RPMI medium supplemented with 10% FCS (Gibco, Carlsbad, CA, USA).

Cell growth, cell cycle profile, viability and apoptosis analysis were performed, as described in the Online Supplementary Methods.

Flow cytometry, western blot and quantitative real-time RTPCR analysis were as described in the Online Supplementary Methods.

Knockdown of CD147 expression by RNA interference, clonogenic assays and colony formation assays were performed, as described30 in the Online Supplementary Methods.

Cyto-ID autophagy detection was performed using the CytoID assay (Enzo Life Sciences ENZ-51031-K200) as described in the Online Supplementary Methods.

AC-73 treatment of leukemic cells used alone or in combination with chemotherapeutic agents AC-73 (3-{2-[([1,1’-biphenyl]-4-ylmethyl) amino]-1-hydroxyethyl}phenol) (Specs ID number AN-465/42834501, Specs, Zoetermeer, the Netherlands) was dissolved in 20% DMSO (Sigma, St. Louis, MO, USA) and diluted in DMEM, with a final DMSO concentration of no more than 0.2% for all in vitro studies.21 In leukemic cell lines, dose-response and time-course analysis were performed using AC-73 at 1.0, 2.5, 5.0 and 10 mM from 1 to 4 days of treatment; results were compared with 0.2% DMSOtreated leukemic cells, indicated as control leukemic cells. AC-73 was added in cultures every 2 days to maintain its activity.

Combinations of treatment were performed using AC-73 (2.5 mM) alone over 24 hours and then by adding Ara-C (0.01; 0.1 and 1.0 µM) or ATO (0.1; 0.01 and 1.0 µM) for another 1 (for NB4 and NB4-R4 cells) or 2 (for U937, HL-60, MV4-11 and Kasumi-1 cells) days. Cell viability assays were performed to evaluate the effect of AC-73 alone or in combination on cell growth and viability of these cells.

Analysis of The Cancer Genome Atlas data Datasets of The Cancer Genome Atlas (TCGA) Research Network 2008, were processed and obtained directly from the public access data portal (http://tcga-data.nci.nih.gov/).

Statistical analysis Student t-test was applied to assess statistical significance of differences between multiple/group of experiments. Data were analyzed using GraphPad Prism software. For univariate survival analysis, Kaplan-Meier plots with a log-rank test were presented using the overall survival data of AML patients from the TGCA. Additional information is provided in the Online Supplementary Appendix. haematologica | 2019; 104(5)

CD147 as potential therapeutic target in AML

Results CD147 is directly involved in hematopoietic progenitor cell proliferation We analyzed CD147 expression at both mRNA and protein level in CD34+ HPCs and during Mo and G proliferation and differentiation of these cells. CD147 is well expressed in CD34+ HPCs and its level of expression decreases during Mo and G differentiation and maturation of these cells (Figure 1A-C and Online Supplementary Figure

Figure 1. CD147 is down-regulated during monocytic (Mo) and granulocytic (G) differentiation of CD34+ hematopoietic progenitor cells (HPCs) and is involved in HPC proliferation. (A) qRT-PCR analysis of CD147 mRNA expression during selective Mo and G proliferation and differentiation of CD34+ HPCs, as compared to U937 and HL-60 leukemic cells. (B) Western blot analysis of CD147 protein expression level in CD34+ cells and Mo and G differentiating HPCs; Jurkat and SKBR3 cells are shown as positive controls of CD147 expression; HG- and LG- are indicated for High- and Low- glycosylated CD147 isoforms; actin is shown as an internal control; molecular weights are indicated (kDa). (C) Flow cytometry analysis of CD147 membrane protein expression during Mo and G differentiation and maturation of HPCs. (D) Flow cytometry analysis of CD147 membrane protein expression in Mo and G differentiating transfected (CD147-siRNA)-HPCs, as compared to transfected (c-siRNA)-HPCs of control, at day 2 of Mo and G cultures corresponding to day 3 post transfection. (E) Cell growth inhibition of (CD147-siRNA)-HPCs, as compared to (c-siRNA)-HPCs, grown under both G and Mo liquid culture conditions. (F) Clonogenic assays performed under G and Mo culture conditions with (CD147siRNA)-HPCs, as compared to (c-siRNA)HPCs. (A, C-F) MeanÂąStandard Error of Mean of three independent experiments is shown. *P<0.05; **P<0.01; ***P<0.001. (B) One representative experiment out of three is shown. AU: arbitrary units; MFI: mean fluorescence intensity.

A

B

C

D

S1A and B). Western blot analysis shows that CD147 protein is highly glycosylated (HG-CD147 40-60 kDa) in CD34+ HPCs and at all stages of Mo and G cell differentiation (Figure 1B), indicating the presence of CD147 protein in a stable and biologically active conformation mainly translocated to plasma membrane,5 as also shown by flow cytometry analysis (Figure 1C and Online Supplementary Figure S1B). To analyze the role of CD147 in HPCs, we performed siRNA-mediated CD147 knockdown experiments in CD34+ HPCs. We transiently trans-

E

F

haematologica | 2019; 104(5)

975

I. Spinello et al.

fected HPCs with specific-CD147 interfering small RNAs (CD147-siRNA) or with a non-targeting control siRNA (csiRNA). Silencing of CD147 mRNA decreased CD147 protein levels in transfected (CD147-siRNA)-HPCs, as compared to (c-siRNA)-HPCs of control (Figure 1D); then, transfected-HPCs were grown under culture conditions, to allow their selective G or Mo proliferation and differentiation. We observed that the decreased expression of CD147 impaired the proliferation of both G and Mo differentiating (CD147-siRNA)-HPCs, as compared to control (c-siRNA)-HPCs (Figure 1E), without any significant modulation of G or Mo specific antigen expression on these cells (data not shown) and without any significant induction of apoptosis (data not shown). Clonogenic G and Mo progenitor assays performed with transfected-HPCs, showed a significant decrease in the number of CFU-GM and CFU-M colonies obtained from (CD147-siRNA)HPCs, compared with the number of colonies obtained from (c-siRNA)-HPCs (Figure 1F). Altogether, our data indicate that CD147 plays a role in HPC proliferation and clonogenic activity.

Over-expressed in AML, CD147 is down-regulated during differentiation of leukemic cells By analyzing CD147 mRNA expression levels in primary leukemic blast cells obtained from 48 patients with different subtypes of AML,29 we found that CD147 is over-expressed in all AML subtypes, in particular in the AML-M3 subtype, as compared to normal CD34+ HPCs (Figure 2A, left panel), in line with the data from AML samples generated by the TCGA Research Network (Figure 2A, right panel).

Then, we also examined CD147 expression in several AML cell lines induced to terminal differentiation. All leukemic cell lines analyzed over-expressed the fully glycosylated mature CD147 protein, biologically active (Figure 2B, HG-CD147), like also its mRNA, as compared to normal CD34+ HPCs (Figure 2C). We found that CD147 expression is down-regulated during vitamin D3-induced Mo differentiation of U937 cells and ATRA-induced G differentiation of HL-60 and NB4 cell lines, at both mRNA and protein levels during terminal differentiation (Figure 2D-F and Online Supplementary Figure S1C-H), similarly to Mo and G terminal differentiation of normal HPCs (Figure 1A-C). However, CD147 expression is not affected by ATRA treatment of the NB4-R4 cell line (Figure 2G and Online Supplementary Figure S1I and L), indicating that induction of cell differentiation is required for CD147 downregulation.

AC-73 inhibits leukemic cell proliferation by blocking ERK/STAT3 signaling and induces autophagy To investigate whether a high level of CD147 can promote cell proliferation, we used AC-73 to inhibit CD147 function in normal and leukemic cells.

First, we examined the effect of AC-73 treatment on G and Mo differentiation of normal HPCs. AC-73 used at 5 ÂľM moderately decreased cell growth without affecting cell cycling or differentiation of HPCs (Figure 3A-C).

Morphology analysis of HPCs at different days of culture supported these observations, showing that AC-73-treated cells regularly progress along G and Mo differentiation and do not show any significant morphology abnormality or delay in differentiation (Figure 3D).

We analyzed the effects of chronic AC-73 administra976

tion on leukemic cell growth, apoptosis and viability of AML cell lines. Leukemic cells were treated from 1 to 3 days, with various AC-73 doses (2.5, 5 and 10 mM) to determine their sensitivity to AC-73 treatment. We observed a time- and dose-dependent effect of AC-73 on leukemic cell growth, with growth inhibition of all leukemic cell lines treated (Figure 4A and Online Supplementary Figure S2B). We could not detect any effect of AC-73 on cell growth and apoptosis of the Chinese Hamster Ovary (CHO) cells (data not shown) used as a CD147-negative cell line (Online Supplementary Figure S2A, left panel).

NB4 and NB4-R4 cell lines were more sensitive to AC73 treatment than U937 and HL-60 cell lines. Indeed, a higher significant decrease in cell growth (Figure 4A), apoptosis (Figure 4B), and viability (Figure 4C and D) was observed in NB4 and NB4-R4 cells treated with low-dose AC-73 (2.5 mM) than in U937 and HL-60 lines (2 days) (Figure 4D). MV4-11, Kasumi-1 cells were the most resistant to AC-73 treatment (Online Supplementary Figure S2BE) when compared to other cell lines (Figure 4A-D).

Significant cell growth inhibition and sensitivity of MV411 and Kasumi-1 cells occurred after treatment with high concentration of AC-73 (5 and 10 mM) (Online Supplementary Figure S2B-D) or when cells were treated for longer (3-4 days) (Online Supplementary Figure S2E).

AC-73 treatment had no significant effect on cell cycle distribution in the leukemic cell lines tested as compared to control cells (data not shown), suggesting that AC-73 decreases the cell growth rate but does not inhibit the progression of cells in the cell cycle. Analysis of CD11b, CD14 and CD15 expression levels in leukemic cell lines treated for 3 days with 5 mM AC-73 showed no significant effect of AC-73 on leukemic cell differentiation (data not shown). By analyzing the effects of AC-73 (5 mM) on clonogenic growth of U937 and NB-4 cells we found that AC73 decreases the in vitro colony formation of both U937 and NB-4 cell lines, indicating an inhibitory effect of AC73 on the clonogenetic capacity of leukemic cells (Online Supplementary Figure S2F).

Altogether, despite the different sensitivities to AC-73 manifested by the different leukemic cell lines, probably related to the specific molecular alteration, such as PML/RARA, FLT3-ITD or RUNX1/RUNX1T1, of these cell lines, AC-73 exhibits potent growth inhibition and cytotoxic activity on leukemic cells only at high doses. Because low doses of AC-73 inhibit leukemic cell proliferation, but do not cause cell death via apoptosis or cell cycle arrest, we investigated the possibility that AC-73 treatment induces autophagy in leukemic cells. First, we assessed the effect of AC-73 on the level of the autophagic indicator LC3 by western blotting in U937 and NB4 cells. Our results showed a dose-dependent effect of AC-73 on the increase of LC3-II/LC3-I ratio in leukemic cells as compared to control (-) cells (Figure 4E), indicating induction of autophagy.31 Then we monitored autophagy flux by flow cytometry analysis in live leukemic cells treated for 72 hours (h) by AC-73 compared to control cells. Our data demonstrated that AC-73 induces dose-dependent autophagy (Figure 4F) in CD147-expressing leukemic cells (Figure 4G), but not in CHO cells (Online Supplementary Figure S2A, right panels), again indicating that the biological effects induced by AC-73 require CD147 expression on target cells. Considering that AC-73 suppresses CD147/ERK1/2/STAT3/MMP-2 pathways in hepatocelluhaematologica | 2019; 104(5)

CD147 as potential therapeutic target in AML

lar carcinoma cells, and that ERK/STAT3 signaling plays a significant role in promoting AML cell proliferation, survival and autophagy,22,23,32,33 we also investigated the effect of AC-73 on ERK1/2 and STAT3 activation in leukemic cells. In line with previous studies,31,32 western blot analysis showed that ERK and STAT3 are constitutively phosphorylated in all leukemic cell lines [Online Supplementary Figure S3, lanes (-)]. We then found that AC-73 (5 mM, for 3 days) notably decreases both pERK and pSTAT3 (S727) levels without affecting the levels of total ERK and STAT3, in leukemic cells (lanes 5 mM) (Online Supplementary Figure S3, lower panels).

A

B

C

D

E

F

G

haematologica | 2019; 104(5)

Altogether, our data demonstrate that AC-73 inhibits leukemic cell proliferation, in part by suppressing the ERK/STAT3 activation pathway in these cells, in part by activating a non-apoptotic, autophagic form of cell death, which may account for the efficacy of AC-73 in leukemic cells.

AC-73 has an additive anti-proliferative effect on chemotherapeutic treatment of leukemic cells and increases ATO-induced autophagy in leukemic cells To evaluate the effect of AC-73 in combination with standard drugs used for AML treatment, AML cell lines Figure 2. Over-expressed in acute myeloid leukemia (AML), CD147 expression is down-regulated during monocytic and granulocytic differentiation of leukemic cells. (A) qRT-PCR analysis of CD147 mRNA expression in primary leukemic cells of AMLs pertaining from M0 to M5 subtypes of the French-American-British (FAB) classification, as compared to normal CD34+ hematopoietic progenitor cells (HPCs) (left panel); CD147 mRNA expression data from AML samples generated by TCGA Research Network (right panel). (B) Western blot analysis of CD147 protein expression level in AML cell lines; densitometry analysis [arbitrary units (AU)] of CD147 protein expression levels compared with actin levels is indicated. (C) qRT-PCR analysis of CD147 mRNA expression in leukemic cell lines, as compared to normal CD34+ HPCs. (D-G) Western blot (left panels) and flow cytometry (right panels) analysis of CD147 total and membrane protein expression levels during vitamin D3-induced monocytic differentiation of U937 cell (D), ATRAinduced granulocytic differentiation of HL60 cells (E), ATRA-induced differentiation of NB4 cells (F), and in NB4-R4 cells, resistant to ATRA treatment (G). (A and C) MeanÂąStandard Error of Mean (SEM) of three independent experiments is shown. *P<0.05; **P<0.01; ***P<0.001. (B, D-G, left panels) One representative western blot experiment out of three is shown. High (HG) and Low (LG) glycosylated CD147 isoforms are indicated. Actin is shown as internal control. (D-G, right panels) MeanÂąSEM of three independent experiments by flow cytometry analysis is shown. *P<0.05; **P<0.01; ***P<0.001. ns: not significant; RPKM: Reads Per Kilobase Million; MFI: mean fluorescence intensity.

977

I. Spinello et al.

and primary AML blasts were initially treated for 24 h with a low concentration (2.5 mM) of AC-73 and then cotreated for 48 h with Ara-C or ATO at different concentrations (0.01-1 mM). Cell viability assays were performed to assess leukemic cell survival and sensitivity to AC-73 treatment used in combination with Ara-C or ATO, as compared to single AC-73, Ara-C, or ATO treatment (Figure 5A and B and Online Supplementary Figure S4). We observed a significant decrease in leukemic cell viability of all leukemic cell lines after AC-73 treatment in combination with Ara-C or ATO, as compared to AC-73, Ara-C or ATO used alone (Figure 5A and B and Online Supplementary Figure S4). Furthermore, we identified three subgroups of leukemic cells according to their Ara-C drugsensitivity: high-sensitivity lines (U937 and HL-60); intermediate-sensitivity lines (NB4 and NB4-R4); low-sensitivity lines (MV4-11 and Kasumi-1). AC-73 potentiates the sensitivity to Ara-C treatment of all leukemic cells, even

those belonging to the low-sensitivity subgroup (Figure 5A and B and Online Supplementary Figure S4). The antiproliferative effect of AC-73 enhances sensitivity to ATO treatment of both M3 leukemic cells, such as NB4 and NB4-R4 (Figure 5B and Online Supplementary Figure S4B), and non-M3 leukemic cells, such as U937, HL-60, Kasumi1 and MV4-11 cells (Figure 5A and Online Supplementary Figure S4A, C and D).

Then, we showed that when used in combination with Ara-C and ATO, AC-73 treatment inhibits ERK and STAT3 activation in leukemic cells (Figure 5C).

Because ATO is also an inducer of autophagy,25 we analyzed the autophagy flux in both M3 and non-M3 leukemic cells treated by ATO or AC-73 used alone or in combination, as compared to control cells. Our data showed that AC-73 and ATO are both inducers of autophagy and the autophagy flux is significantly increased in M3 (NB4) and non-M3 (U937) leukemic cells

A

B

D

978

C

Figure 3. AC-73 has no significant effect on cell growth, differentiation and cell cycle progression in normal granulocytic (G) and monocytic (Mo) differentiating hematopoietic progenitor cells (HPCs). (A) Cell growth analysis during G and Mo differentiation of HPCs, in presence of AC-73 used at 5 mM and added every 2 days in cultures, as compared to control (C) cells. (B) Cell cycle analysis at day 7 of G and Mo differentiation of HPCs treated with AC-73 (5 µM), as compared to control HPCs. (C) Phenotype analysis performed by analyzing CD11b, CD15 and CD14 expression levels in G and Mo differentiating HPCs at day 15, in the presence or not (C) of AC-73.

(A-C) Mean±Standard Error of Mean (SEM) of three independent experiments is shown. *P<0.05; ns: not significant. (D) Morphological analysis at day 7 of the differentiation and maturation of G and Mo differentiating HPCs treated with AC-73, as compared to control (C) HPCs, and stained with May-GrünwaldGiemsa. (D) One representative experiment out of three is shown.

haematologica | 2019; 104(5)

CD147 as potential therapeutic target in AML

treated by ATO and AC-73 used in combination, as compared to single agent treatment or control cells (C) (Figure 5D and E).

Altogether, our data demonstrate that AC-73 is a potent, novel anti-proliferative molecule that enhances the sensitivity of leukemic cells to conventional chemotherapeutic agents, but also increases ATO-induced autophagy in M3 and non-M3 leukemic cells.

AC-73 increases the sensitivity of primary acute myeloid leukemia blasts co-expressing CD147 and CD371 to chemotherapy We evaluated the effects of AC-73 used alone and in combination with Ara-C or ATO on primary AML blasts.

First, we analyzed CD147 protein expression in samples obtained from different subtypes of AML and we controlled the expression of CD34, CD38 and CD371 mark-

A

B

C

D

E

F

G

Figure 4. Effects of AC-73 on cell growth, apoptosis and viability of acute myeloid leukemia (AML) cell lines and autophagy. (A) Dose response analysis of AC-73 treatment on leukemic cell growth, as compared to control cells (C). (B) Analysis of the effects of AC-73 treatment performed for 3 days (d3) on leukemic cell apoptosis, as compared to control leukemic cell (0). Total apoptosis by annexin V/PI (%) detected by using flow cytometric apoptotic assays is indicated. (C and D) Cell viability assays on leukemic cells treated: (C) with AC-73 used at different concentrations for 2 days; (D) at different times with 2.5 ÂľM AC-73, as compared to control cells (day 0). (C and D) Viability is presented as percentage viable cell relative to control. (E) Western blot analysis of the autophagy related protein LC3 and its conversion from LC3-I to LC3-II form in leukemic cells treated with AC-73 used at different concentrations, as compared to control (-) cells. (F) Dose response analysis of AC-73 treatment on the autophagy flux in U937 and NB4 leukemic cells, as compared to control cells (0). (G) Induction of autophagy by AC-73 treatment (+) in several AML cell lines, as compared to control cells (-). (A-D, F and G) MeanÂąStandard Error of Mean is shown. *P<0.05; **P<0.01; ***P<0.001. ns: not significant.

(E) One representative western blot experiment out of three is shown. Actin was used as an internal control.

haematologica | 2019; 104(5)

979

I. Spinello et al. ers previously identified in LSCs.34 We confirmed that CD147 is expressed at high levels in all AML subtypes (M1, M2, M3, M5 and M5a) analyzed (Table 1). We also report for the first time that CD147 is co-expressed with CD34, and particularly with CD371, in these cells (Table 1 and Figure 6A). It is interesting to note that in 8 non-M3 AMLs analyzed for CD147 and CD371 expression, there was a strong positive correlation between CD147 and CD371 expression (P<0.001) (Table 1). These data suggest

the existence of CD34+/CD147+ and CD371+/CD147+ subpopulations in AMLs that may have functional properties of leukemic stem cells (LSCs).

Then, we investigated the effects of AC-73 used alone or combined with Ara-C and ATO at low concentration in primary AML blasts, and particularly in M2, M3 and M5 AML cells. We found that AC-73 used alone has no significant effect on AML cell apoptosis, unless used at high concentration (10 mM) for AML(7)-M3 and AML(8)-M5

A

B

C

D

980

E

Figure 5. AC-73 potentiates the sensitivity of leukemic cells to chemotherapeutic treatment. (A and B) Cell viability assays on U937 and NB4 leukemic cells, treated 24 hours (h) with AC-73 used alone (2.5 mM), then in combination with Ara-C (AC-73 + AraC, left panels) or ATO (AC-73 + ATO, right panels) for 48 h, as compared to treatment with single drug, AC-73 (AC-73 + 0.00 mM Ara-C; AC-73 + 0.00 mM ATO), Ara-C and ATO. (C) Western blot analysis of phospho-STAT3(ser727) (pSTAT3(S727), STAT3, phospho-ERK1/2 (pERK) and ERK1/2 (ERK) protein expression in U937, NB4 and MV4-11 leukemic cell lines. U937, NB4 and MV4-11 cells were treated for 3 days with AC-73 in combination with ATO (AC-73 + ATO 1 mM), or Ara-C (AC-73 + Ara-C 1 mM), as compared to AC-73 (2.5 mM), ATO (1 mM) or Ara-C (1 mM) used alone or to control (C) leukemic cells. Actin was used as an internal control.

Quantification of total and phosphorylated ERK1/2 and STAT3 proteins by densitometry analysis is indicated. (D and E) Analysis of the autophagy flux induces by ATO (1 mM) and AC-73 (5 mM) used alone or in combination in U937 and NB4 leukemic cells compared to control cells (C).

(A, B and E) MeanÂąStandard Error of Mean of three independent experiments is shown. *P<0.05; **P<0.01; ***P<0.001. ns: not significant. (C and D) One representative experiment out of three is shown. (C) A.U.: arbitrary units.

Ratio pSTAT3/STAT3 and pERK/ERK of three independent experiments is shown. kDa: molecular weights.

haematologica | 2019; 104(5)

CD147 as potential therapeutic target in AML

(Figure 6B). However, AC-73 induces autophagy by increasing the LC3-II/LC3-I ratio in leukemic blasts, as compared to control (-) leukemic cells, even at 2.5 or 5 mM (Figure 6C). However, combination of low-dose AC-73 (2.5 mM) and ATO or Ara-C (0.1 and 1 mM) more efficiently decreased leukemic blast viability, compared with single AC-73, ATO or Ara treatment (Figure 6D). Overall, our data indicate that AC-73 activates a non-apoptotic autophagic form of cell death in AML blasts and increases the sensitivity of these cells to Ara-C or ATO agents.

Analysis of CD147 expression in acute myeloid leukemia based on TCGA dataset By providing selected genetic and clinical data from 200 AML patients, useful for prognosis and diagnosis, the TCGA dataset offered a unique opportunity to explore a possible relationship between the recurrent gene mutations observed in AML and the level of CD147 expression. TCGA data set analysis indicates that CD147 level is particularly elevated in AML-M3, those AML subtypes bearing PML-RARA fusion gene1,26 (Figure 7A, RARA), in

A

B

C

D

Figure 6. AC-73 decreases the survival of CD147+ acute myeloid leukemia (AML) cells exposed to chemotherapeutic agents and activates autophagy in AML cells.

(A) CD147 is expressed in the subfraction of CD34+CD371+ positive AML leukemic cells. (B) Dose response analysis of AC-73 treatment on leukemic cell apoptosis of AML-M2, AML-M3 and AML M5 blasts obtained from 5 AML (4, 5, 7, 8 and 10) patients, as compared to respective control (C) leukemic blasts; flow cytometry analysis of total apoptosis by Annexin V/ PI (%) is shown. (C) Western blot analysis of the autophagic-related LC3 marker shows its conversion from the LC3-I to LC3-II form in leukemic blasts treated in vitro with growing concentration of AC-73 as compared to control leukemic blasts (-). (D) Cell viability assays were performed on M2, M3 and M5 leukemic blasts, obtained from 5 AML patients (4), (5), (7), (8) and (10), and treated in vitro by AC-73 (2.5 mM) in combination with ATO or Ara-C (0.1 and 1 mM), as compared to control (C) leukemic blasts and to single treatment with AC-73, ATO and Ara-C. (A) One representative experiment is shown. (B and D) MeanÂąStandard Error of Mean of three independent experiments is shown. *P<0.05; **P<0.01; ***P<0.001; ns: not significant. (C) One representative western blot experiment out of three is shown. Actin was used as an internal control.

haematologica | 2019; 104(5)

981

I. Spinello et al.

line with the data observed from fresh samples obtained from AML-M3 patients shown in Figure 2A, while significantly lower levels were observed in AMLs blasts mutated for NPM1, DNMT3A or FLT3 (Figure 7A). By analyzing CD147 levels in all AMLs that we have stratified into three risk groups according to the European LeukemiaNet (ENL) risk classification, we found that: 1) the lower CD147 levels were observed in poor- and intermediaterisk AMLs; 2) surprisingly, the highest CD147 levels were observed in AMLs with favorable risk profile, mainly due to the presence in this group of AML-M3 (Figure 7B).

However, it is important to point out that approximately 20% of poor- and intermediate-risk AMLs also display elevated CD147 expression (Figure 7B). Because AML-M3 have the best prognosis among all AML subtypes for the efficiency of target therapy based on ATRA and ATO,26 those patients with favorable risk and high CD147 levels may represent a bias for our analysis. Therefore, we evaluated a possible correlation between CD147 expression level and overall survival of AML patients, excluding AML-M3 patients, and according to CD147 levels classified as: low (CD147<70); medium (CD147; range, 70120); high (CD147 >120) (Figure 7C). In these non-M3 patients who died within 60 months after diagnosis, earlier death (within 20 months) was observed more frequently in cases with high CD147 levels, as compared with (non-M3)-AML subgroups with low-medium CD147 levels (P=0.003 and P=0.0003, respectively) (Figure 7D). These three (non-M3)-AML subgroups had comparable age and white blood count (WBC) at diagnosis (Figure 7D, right panels). Our data indicate that, excluding AML-M3, high CD147 levels are correlated with early death of AML patients.

Discussion CD147 represents a prognostic marker in several solid tumors and in hematologic malignancies.3,6-9,15-19 However, the biological function of CD147 and its potential role as a marker in leukemia remain poorly defined. Apart from the erythrocyte lineage, the expression, regulation and function of CD147 in normal and leukemic hematopoietic cells have not been extensively studied.35 CD147, previ-

ously identified as a carrier molecule for the blood group antigen OKa,36 has been involved in the recirculation of mature erythrocytes from the spleen into the general circulation.37 In our study, we show that CD147 is important for the proliferation of normal CD34+ HPCs and its expression is down-regulated during Mo and G differentiation. Notably, we found that CD147 is over-expressed in all leukemic cell lines and in the large majority of primary leukemic blasts analyzed, as compared to normal CD34+ HPCs. However, CD147 expression significantly decreases during Mo and G differentiation of leukemic cells, thus mimicking what occurs in Mo and G lineage cells. By using AC-73 to inhibit CD147 function in leukemic cell lines and in primary AML blasts, we demonstrated that CD147 overexpression promotes leukemic cell proliferation. Interestingly, we observed that AC-73 treatment inhibits leukemic cell proliferation by suppressing the activation of the ERK/STAT3 signaling pathway, in line with previous studies.5,21 As constitutive STAT3 activity has been described to promote AML cell proliferation and survival,22,38 targeting key tyrosine kinases upstream of STAT3 has been proposed as a strategy to treat AML.38-40 However, due to its role in stem cell renewal, the potential risk of systemic STAT3 inhibition could be a deregulation of hematopoiesis.41 Here, we found that CD147 is expressed in normal HPCs and also showed that a low dose of AC-73 induces only a slight inhibition of cell proliferation, without affecting differentiation of CD34+ HPCs. Importantly, we found that AC-73 does not cause cell death via apoptosis or cell cycle arrest, but induces autophagy in leukemic cells, as described in previous studies carried out in AML cells treated with ATO as an inducer of autophagy,23 or with other novel targeted drugs.23,27,4244 In some cases, autophagy accounted for a non-apoptotic decrease in cell viability.42,43 To further validate its potential therapeutic activity, we used AC-73 in combination with conventional antileukemia treatment, and showed that the anti-proliferative effect of AC-73 on leukemic cells enhanced the sensitivity of leukemic cells to chemotherapeutic treatments such as Ara-C or ATO, which could be consequently used at lower concentration. Our findings are in line with recent studies showing that other compounds, such as chidamide that down-regulates the JAK2/STAT3 signaling,45

Table 1. Analysis of cell surface antigen expression in primary acute myeloid leukemia samples.

Patient n.

1 2 3 4 5 6 7 8 9 10

FAB subtype

CD147 (%)

CD34 (%)

CD371 (%)

CD11b (%)

CD14 (%)

M1 M1 M1 M2 M2 M3 M3 M5 M5a M5

63.1 93.7 44.4 95 92.5 63.7 97.9 71.5 96.6 91.3

21.2 92.4 98.6 11.1 39.5 0.2 12.5 5 8.3 73

80.3 88.6 64.5 88.3 94 31 95.9 90.9 93.2 ND

31 ND 28 5 78 4 3.2 60 33.6 69

1 ND 2.3 1 18.5 ND 0 21 8.8 17

Flow cytometry analysis of CD147, CD34, CD371 and CD11b, CD14 cell surface antigen expression in 10 primary acute myeloid leukemia (AML) samples, obtained from patients with AML pertaining to different French-American-British (FAB) classification subtypes. Results are shown as percentage (%) of expression. ND: not determined.

982

haematologica | 2019; 104(5)

CD147 as potential therapeutic target in AML

inhibit the viability of AML cells and synergize with cytarabine in inhibiting AML cell viability.46 Importantly, we also showed that AC-73 increases ATO-induced autophagy in both M3 and non-M3 leukemic cells, indicating a possible synergistic anti-tumor interaction of ATO with an inhibitor of CD147, both agents being able to induce autophagy.

The analysis of CD147 expression in AML subsets presented several interesting findings. First, CD147 is overexpressed in all French-American-British classification AML subtypes, and particularly in AML-M3; a finding confirmed through the analysis of primary AML samples.

This finding was also corroborated by the analysis of the TCGA data set. Moreover, about 20% of non-M3 AMLs displayed elevated CD147 levels. Although the overall long-term survival of these patients (defined as highlyexpressing CD147) was similar to the other two non-M3 AML groups (defined as middle- and low-expressing

A

CD147), death occurs earlier in patients showing high expression of CD147. This observation supports the negative prognostic role of high CD147 expression levels in AMLs, as observed also in other tumors, including some hematologic neoplasia.3,19 Other recurrent AML mutations,1,2 such as NPM1, FLT3-ITD and DNMT3A, were not associated to particularly elevated levels of CD147. Therefore, AC-73 used in combination with AraC or ATO may have clinical potential implications in treatment of AML patients expressing CD147. Future toxicological and pharmacodynamic studies in suitable animal models will be required for a preclinical evaluation of the possible impact of AC-73 as an anti-leukemic drug.

Our study also reports for the first time that CD147 is expressed in a sub-fraction CD34+CD371+ of AML cells that is able to engraft immunodeficient mice,20 CD34 being predominantly regarded as a marker of hematopoietic stem cells (HSC) and HPCs, and CD371 as a marker of

Figure 7. CD147 gene expression level does not predict overall survival in acute myeloid leukemia (AML), but is correlated in early death of (non-M3)-AML patients.

(A) Relationship between the most recurrent gene mutations observed in all AMLs and the level of CD147 mRNA expression, according to the TCGA dataset. (B) CD147 mRNA levels were analyzed in AMLs stratified into 3 risk groups: poor, intermediate and favorable, according to the European LeukemiaNet (ENL) risk classification. (C) Kaplan-Meier survival analysis in a new group of AML patients that excludes AMLM3 patients, based on CD147 gene expression and stratified according to their CD147 mRNA levels in 3 groups: low (CD147<70); medium (CD147, range, 70-120); high (CD147 >120). (D) Kaplan-Meier survival analysis in the new group of (nonM3)-AML patients deceased within 50-60 months years after diagnosis, indicates that the kinetics of death is more rapid (within 20 months) among CD147 high patients, as compared with those with low (P<0.01) or medium (P<0.001) CD147 levels; P-values calculated by log-rank test. The histograms show that white blood count (WBC) number, age at diagnosis and proportion of patients with poor cytogenetics were comparable in the 3 (non-M3)-AML subgroups subdivided according to CD147 expression level.

B

C

D

High vs. medium P=0.0003 High vs. low P=0.003

haematologica | 2019; 104(5)

983

I. Spinello et al.

LSCs and a novel prognostic predictor in AML.20,34,47 Interestingly, in non-M3 AMLs a strong positive correlation was observed between CD147 and CD371 expression. In this context, while a dedicated future study would shed light on the functional properties of LSCs of the fractions of (CD34+CD147+) and (CD34+CD371+CD147+) AML cells, our data suggest that CD147 is expressed in LSCs and may be a potential new prognostic marker in AML and a potential therapeutic target.

Taken together, our study demonstrates a pivotal role for CD147 in leukemic cell proliferation and indicates CD147 as a potential therapeutic target of AC-73 in AML.

References 1. Saultz JN, Garzon R. Acute Myeloid Leukemia: A Concise Review. J Clin Med. 2016;5:5(3).

2. Khaled S, Al Malki M, Marcucci G. Acute myeloid leukemia: biologic, prognostic, and therapeutic insights. Oncology (Williston Park). 2016;30(4):318-329.

3. Xiong L, Edwards CK 3rd, Zhou L. The biological function and clinical utilization of CD147 in human diseases: a review of the current scientific literature. Int J Mol Sci. 2014;15(10):17411-17441.

4. Somno A, Anuchapreeda S, Chruewkamlow N, Pata S, Kasinrerk W, Chiampanichayakul S. Involvement of CD147 on multidrug resistance through the regulation of P-glycoprotein expression in K562/ADR leukemic cell line. Leuk Res Rep. 2016;6:33-38.

5. Bai Y, Huang W, Ma LT, Jiang JL, Chen ZN. Importance of N-glycosylation on CD147 for its biological functions. Int J Mol Sci. 2014;15(4):6356-6377.

6. Grass GD, Dai L, Qin Z, Parsons C, Toole BP. CD147: regulator of hyaluronan signaling in invasiveness and chemoresistance.

Adv Cancer Res. 2014;123:351-373.

7. Tang J, Guo YS, Zhang Y, et al. CD147 induces UPR to inhibit apoptosis and chemosensitivity by increasing the transcription of Bip in hepatocellular carcinoma.

Cell Death Differ. 2012;19(11):1779-1790.

8. Gao H, Jiang Q, Han Y, Peng J, Wang C. shRNA-mediated EMMPRIN silencing inhibits human leukemic monocyte lymphoma U937 cell proliferation and increases chemosensitivity to adriamycin. Cell Biochem Biophys. 2015;71(2):827-835.

9. Huang Z, Wang L, Wang Y, et al.

Overexpression of CD147 contributes to the chemoresistance of head and neck squamous cell carcinoma cells. J Oral Pathol Med. 2013;42(7):541-546.

10. Muramatsu T. Basigin (CD147), a multifunctional transmembrane glycoprotein with various binding partners. J Biochem.

2016;159(5):481-490.

11. Le Floch R, Chiche J, Marchiq I, et al.

CD147 subunit of lactate/H+ symporters MCT1 and hypoxia-inducible MCT4 is critical for energetics and growth of glycolytic tumors. Proc Natl Acad Sci U S A. 2011;108(40):16663-16668.

12. Kirk P, Wilson MC, Heddle C, Brown MH, Barclay AN, Halestrap AP. CD147 is tightly associated with lactate transporters MCT1

984

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Thus, targeting CD147 by using AC-73 or derivate molecules to induce autophagy or to inhibit the growth of leukemic clones, including LSCs, may be an effective option to improve leukemia treatment.

Acknowledgments The authors would like to thank AM Cerio for assistance in cell cultures.

Funding This study was supported by Istituto Superiore di Sanità (Rome, Italy).

and MCT4, and facilitates their cell surface expression. EMBO J. 2000;19(15):38963904.

Eliasson P, Jönsson JI. The hematopoietic stem cell niche: low in oxygen but a nice place to be. J Cell Physiol. 2010;222(1):1722.

Testa U, Labbaye C, Castelli G, Pelosi E. Oxidative stress and hypoxia in normal and leukemic stem cells. Exp Hematol.

2016;44(7):540-560.

Fu J, Fu J, Chen X, Zhang Y, Gu H, Bai Y. CD147 and VEGF co-expression predicts prognosis in patients with acute myeloid leukemia. Jpn J Clin Oncol. 2010;40 (11):1046-1052.

Honnemyr M, Bruserud Ø, Brenner AK. The constitutive protease release by primary human acute myeloid leukemia cells.

J Cancer Res Clin Oncol. 2017; 143(10):1985-1998.

Zhu D, Wang Z, Zhao JJ, et al. The Cyclophilin A-CD147 complex promotes the proliferation and homing of multiple myeloma cells. Nat Med. 2015;21(6):572580.

Eichner R, Heider M, Fernández-Sáiz V, et al. Immunomodulatory drugs disrupt the cereblon-CD147-MCT1 axis to exert antitumor activity and teratogenicity. Nat Med. 2016;22(7):735-743.

Panchabhai S, Schlam I, Sebastian S, Fonseca R. PKM2 and other key regulators of Warburg effect positively correlate with CD147 (EMMPRIN) gene expression and predict survival in multiple myeloma.

Leukemia. 2017;31(4):991-994. van Rhenen A, van Dongen GA, Kelder A, et al. The novel AML stem cell associated antigen CLL-1 aids in discrimination between normal and leukemic stem cells.

Blood. 2007;110(7):2659-2666.

Fu ZG, Wang L, Cui HY, et al. A novel small-molecule compound targeting CD147 inhibits the motility and invasion of hepatocellular carcinoma cells.

Oncotarget. 2016;7(8):9429-9447.

Redell MS, Ruiz MJ, Alonzo TA, Gerbing RB, Tweardy DJ. Stat3 signaling in acute myeloid leukemia: ligand-dependent and independent activation and induction of apoptosis by a novel small-molecule Stat3 inhibitor. Blood. 2011;117(21):5701-5709.

Orsini M, Morceau F, Dicato M, Diederich M. Autophagy as a pharmacological target in hematopoiesis and hematological disorders. Biochem Pharmacol. 2018;152:347361.

24. Folkerts H, Hilgendorf S, Wierenga ATJ, et al. Inhibition of autophagy as a treatment strategy for p53 wild-type acute myeloid leukemia. Cell Death Dis. 2017;8(7):e2927.

25. Jin J, Britschgi A, Schläfli AM, et al. Low Autophagy (ATG) Gene Expression Is Associated with an Immature AML Blast Cell Phenotype and Can Be Restored during AML Differentiation Therapy. Oxid Med Cell Longev. 2018;2018:1482795.

26. Lo-Coco F, Cicconi L, Breccia M. Current standard treatment of adult acute promyelocytic leukaemia. Br J Haematol.

2016;172(6):841-854.

27. Goussetis DJ, Altman JK, Glaser H, McNeer JL, Tallman MS, Platanias LC. Autophagy is a critical mechanism for the induction of the antileukemic effects of arsenic trioxide. J Biol Chem. 2010; 285(39):29989-29997.

28. Lulli V, Romania P, Riccioni R, et al.

Transcriptional silencing of the ETS1 oncogene contributes to human granulocytic differentiation. Haematologica. 2010; 95(10):1633-1641.

29. Spinello I, Quaranta MT, Paolillo R, et al.

Differential hypoxic regulation of the microRNA-146a/CXCR4 pathway in normal and leukemic monocytic cells: impact on response to chemotherapy.

Haematologica. 2015;100(9):1160-1171.

30. Labbaye C, Spinello I, Quaranta MT, et al.

A three-step pathway comprising PLZF/miR-146a/CXCR4 controls megakaryopoiesis.

Nat Cell Biol.

2008;10(7):788-801.

31. Tanida I, Ueno T, Kominami E. LC3 and Autophagy. Methods Mol Biol. 2008; 445:77-88.

32. Benekli M, Xia Z, Donohue KA, et al.

Constitutive activity of signal transducer and activator of transcription 3 protein in acute myeloid leukemia blasts is associated with short disease-free survival. Blood.

2002;99(1):252-257.

33. Lunghi P, Tabilio A, Dall'Aglio PP, et al.

Downmodulation of ERK activity inhibits the proliferation and induces the apoptosis of primary acute myelogenous leukemia blasts. Leukemia. 2003;17(9):1783-1793.

34. Pelosi E, Castelli G, Testa U. Targeting LSCs through membrane antigens selectively or preferentially expressed on these cells. Blood Cells Mol Dis. 2015;55(4):336346.

35. Papayannopoulou T, Brice M. Integrin expression profiles during erythroid differentiation. Blood. 1992;79(7):1686-1694.

haematologica | 2019; 104(5)

CD147 as potential therapeutic target in AML

36. Spring FA, Holmes CH, Simpson KL, et al.

The Oka blood group antigen is a marker for the M6 leukocyte activation antigen, the human homolog of OX-47 antigen, basigin and neurothelin, an immunoglobulin superfamily molecule that is widely expressed in human cells and tissues. Eur J Immunol. 1997;27(4):891-897.

37. Coste I, Gauchat JF, Wilson A, et al.

Unavailability of CD147 leads to selective erythrocyte trapping in the spleen. Blood.

2001;97(12):3984-3988.

38. Xiong A, Yang Z, Shen Y, Zhou J, Shen Q. Transcription Factor STAT3 as a Novel Molecular Target for Cancer Prevention.

Cancers (Basel). 2014;6(2):926-957.

39. Murone M, Radpour R, Attinger A, et al.

The Multi-kinase Inhibitor Debio 0617B Reduces Maintenance and Self-renewal of Primary Human AML CD34+

haematologica | 2019; 104(5)

40.

41.

42.

43.

Stem/Progenitor Cells. Mol Cancer Ther.

2017;16(8):1497-1510.

Xiang M, Kim H, Ho VT, et al. Gene expression-based discovery of atovaquone as a STAT3 inhibitor and anticancer agent.

Blood. 2016;128(14):1845-1853.

Yu H, Pardoll D, Jove R. STATs in cancer inflammation and immunity: a leading role for STAT3. Nat Rev Cancer. 2009; 9(11):798-809.

Evangelisti C, Evangelisti C, Chiarini F, et al. Autophagy in acute leukemias: a double-edged sword with important therapeutic implications. Biochim Biophys Acta.

2015;1853(1):14-26.

Wei Y, Kadia T, Tong W, et al. The combination of a histone deacetylase inhibitor with the Bcl-2 homology domain-3 mimetic GX15-070 has synergistic antileukemia activity by activating both apoptosis and

44.

45.

46.

47.

autophagy. Clin Cancer Res. 2010; 16(15):3923-3932.

Auberger P, Puissant A. Autophagy, a key mechanism of oncogenesis and resistance in leukemia. Blood. 2017;129(5):547-552.

Zhao S, Guo J, Zhao Y, et al. Chidamide, a novel histone deacetylase inhibitor, inhibits the viability of MDS and AML cells by suppressing JAK2/STAT3 signaling. Am J Transl Res. 2016;8(7):3169-3178.

Li X, Yan X, Guo W, et al. Chidamide in FLT3-ITD positive acute myeloid leukemia and the synergistic effect in combination with cytarabine. Biomed Pharmacother.

2017;90:699-704.

Wang YY, Chen WL, Weng XQ, et al. Low CLL-1 Expression Is a Novel Adverse Predictor in 123 Patients with De Novo CD34+ Acute Myeloid Leukemia. Stem Cells Dev. 2017;26(20):1460-1467.

985

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):986-992

Acute Lymphoblastic Leukemia

Replacing cyclophosphamide/cytarabine/ mercaptopurine with cyclophosphamide/ etoposide during consolidation/delayed intensification does not improve outcome for pediatric B-cell acute lymphoblastic leukemia: a report from the COG

Michael J. Burke,1* Wanda L. Salzer,2* Meenakshi Devidas,3 Yunfeng Dai,3 Lia Gore,4 Joanne M. Hilden,4 Eric Larsen,5 Karen R. Rabin,6 Patrick A. ZweidlerMcKay,7 Michael J. Borowitz,8 Brent Wood,9 Nyla A. Heerema,10 Andrew J. Carroll,11 Naomi Winick,12 William L. Carroll,13 Elizabeth A. Raetz,13** Mignon L. Loh14** and Stephen P. Hunger15**

1 Department of Pediatrics, Children’s Hospital of Wisconsin, Milwaukee, WI; 2U.S. Army Medical Research and Materiel Command, Fort Detrick, MD; 3Department of Biostatistics, Colleges of Medicine and Public Health & Health Professions, University of Florida, Gainesville, FL; 4Department of Pediatrics, Center for Cancer and Blood Disorders, Children’s Hospital Colorado and The University of Colorado School of Medicine, Aurora, CO; 5Department of Pediatrics, Maine Children’s Cancer Program, Scarborough, ME; 6 Department of Pediatrics, Baylor College of Medicine, Houston, TX; 7ImmunoGen, Inc, Waltham, MA; 8Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, MD; 9Department of Laboratory Medicine, University of Washington, Seattle, WA; 10 Department of Pathology, The Ohio State University School of Medicine, Columbus, OH; 11 Department of Genetics, University of Alabama at Birmingham, AL; 12Department of Pediatrics, University of Texas Southwestern Medical Center, Dallas, TX; 13Department of Pediatrics, Perlmutter Cancer Center, New York University Langone Health, New York, NY; 14 Department of Pediatrics, Benioff Children’s Hospital and the Helen Diller Family Comprehensive Cancer Center, University of California at San Francisco, CA and 15 Department of Pediatrics, Children’s Hospital of Philadelphia and the Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA

Correspondence: MICHAEL J. BURKE mmburke@mcw.edu

*These authors contributed equally as first authors.

**These authors contributed equally as senior authors.

ABSTRACT Received: August 20, 2018.

Accepted: December 6, 2018.

Pre-published: December 13, 2018. doi:10.3324/haematol.2018.204545 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/986 ©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.

986

W

ith modern chemotherapy, approximately 90% of patients with pediatric acute lymphoblastic leukemia are now cured.

However, subsets of patients can be identified who remain at very high risk of relapse with expected 4-year disease-free survival rates <80%; such patients are appropriate candidates for intensive therapeutic strategies designed to improve survival. The AALL1131 trial was designed to determine, in a randomized fashion, whether substitution with cyclophosphamide/etoposide (experimental arm 1) would improve the 4-year disease-free survival of children, adolescents, and young adults with very high-risk B-cell acute lymphoblastic leukemia compared to a modified Berlin-Frankfurt-Münster regimen (control arm). Patients 130 years of age with newly diagnosed very high-risk B-cell acute lymphoblastic leukemia were randomized after induction in a 1:2 fashion to the control arm or experimental arm 1 in which they were given cyclophosphamide (440 mg/m2 days 1-5)/ etoposide (100 mg/m2 days 15) during part 2 of consolidation and delayed intensification. Prospective interim monitoring rules for efficacy and futility were included where futility would be determined for a one-sided P-value ≥0.7664. The study was stopped for futility as the interim monitoring boundary was crossed [hazard ratio 0.606 (95% confidence interval: 0.297 - 1.237)] and the very high-risk arm of AALL1131 was closed in February 2017. Using data current as of December 31, 2017, 4-year disease-free survival rates were 85.5±6.8% (control arm) versus 72.3±6.3% (experimental arm 1) (P-value = 0.76). There were no significant differences in grade 3/4 adverse events haematologica | 2019; 104(5)

Cyclophosphamide/etoposide in pediatric B-ALL

between the two arms. Substitution of this therapy for very high-risk B-cell acute lymphoblastic leukemia patients on the Children’s Oncology Group AALL1131 trial (NCT02883049) randomized to cyclophosphamide/etoposide during part 2 of consolidation and delayed intensification did not improve disease-free survival.

Introduction With modern chemotherapy regimens, approximately 90% of patients with pediatric B-cell acute lymphoblastic leukemia (B-ALL) are now cured.1,2 However, subsets of patients remain at very high-risk (VHR) of relapse with an expected 4-year disease-free survival (DFS) rate <80%.

Current post-induction intensification strategies, which have focused on optimizing the use of drugs commonly administered in ALL therapy, have delivered sub-optimal results for these VHR B-ALL patients. In the absence of a specific targeted intervention (such as Abl-tyrosine kinase inhibitors in Philadelphia chromosome-positive ALL), intensive chemotherapy continues to be the mainstay of treatment. We hypothesized that further optimization or intensification of the dose and schedule of established agents or combination regimens typically used to treat newly diagnosed ALL patients would probably not improve outcomes further for VHR B-ALL patients, and therefore novel or targeted therapies should be investigated. Given that there was not a molecularly targeted agent available for this population of patients at the time the study was conceived, this trial was designed to test the use of different consolidation strategies, based on drugs not commonly used in frontline ALL trials, including fractionated cyclophosphamide and etoposide.

The Children’s Oncology Group (COG) AALL1131 trial thus aimed to determine, in a randomized fashion, whether replacing cyclophosphamide, cytarabine, and 6-mercaptopurine during consolidation or cyclophosphamide, cytarabine, and 6-thioguanine during delayed intensification with cyclophosphamide and etoposide (experimental arm 1) during the consolidation and reconsolidation phases of COG augmented Berlin-Frankfurt-Münster therapy (control arm)3 would improve the 4-year DFS of children, adolescents, and young adults with VHR B-ALL. The cyclophosphamide/ etoposide combination was well tolerated in prior relapse B-ALL studies4,5 and a similar combination of ifosfamide/etoposide yielded 40% complete remission rates in children with refractory ALL,6 making cyclophosphamide/etoposide an encouraging combination to study.

Methods COG AALL1131 (NCT02883049), a phase III trial for patients aged 1-30 years with newly diagnosed high-risk B-ALL opened to enrollment on February 27, 2012 and the VHR randomization closed on February 15, 2017. Eligibility criteria included: 1-9 years of age inclusive with a presenting white blood cell count ≥50x109/L; ≥10 to <31 years of age with any white blood cell count; >1 to <31 years of age with testicular leukemia, central nervous system leukemia (CNS3; ≥5/mL white blood cells and cytospin positive for blasts in the cerebral spinal fluid and/or clinical signs of CNS leukemia), or steroid pre-treatment in patients <10 years of age for whom no pre-steroid white blood cell count was obtained.7 At the end of induction therapy, patients were furhaematologica | 2019; 104(5)

ther classified as VHR if they had any of the following criteria: ≥13 years of age; CNS3 leukemia at diagnosis; day 29 bone marrow minimal residual disease ≥0.01% determined by flow cytometry;7,8 induction failure [>25% bone marrow blasts (M3) on induction day 29], severe hypodiploidy (DNA index <0.81 and/or <44 chromosomes); intrachromosomal amplification of chromosome 21, or lysine methyltransferase 2A (KMT2A, formerly mixed lineage leukemia, MLL) rearrangement. In addition, patients with National Cancer Institute standard-risk B-ALL, enrolled on the COG study AALL0932 (NCT01190930) for standard-risk B-ALL (≥1 to <10 years of age with a white blood cell count <50x109/L), were classified as VHR following induction if they met any of the above VHR criteria or if they had day 29 bone marrow minimal residual disease ≥0.01% in the absence of favorable cytogenetics (no trisomies of chromosomes 4 and 10 and no ETV6/RUNX1 fusion). Patients with Down syndrome were not eligible for the VHR stratum given the concern of increased toxicity of the regimen. Toxicities were graded using the National Cancer Institute’s Common Terminology Criteria for Adverse Events version 4.0. The study was approved by the National Cancer Institute, the Pediatric Central Institutional Review Board, and institutional review boards at each participating COG institution.

The AALL1131 study was originally designed to investigate the addition of clofarabine to cyclophosphamide/etoposide as experimental arm 2 versus cyclophosphamide/etoposide (experimental arm 1) versus the control arm in a 2:2:1 randomization for patients with VHR B-ALL. The study design was later amended to a 2:1 randomization between experimental arm 1 and the control arm, retaining those patients initially randomized to experimental arm 1 and the control arm, after the clofarabine arm (experimental arm 2) was closed because of unacceptable toxicity (September 2014).7 Patients classified as VHR were subsequently randomized after induction in a 1:2 fashion to cyclophosphamide (1 g/m2 day 29)/cytarabine (75 mg/m2 days 29-33 and 36-40)/6-mercaptopurine (60 mg/m2 days 29-42 consolidation) or thioguanine (60 mg/m2 days 29-42 during part 2 of delayed intensification) (control arm) or cyclophosphamide (440 mg/m2, days 29-33)/etoposide (100 mg/m2, days 29-33) (experimental arm 1) during part 2 of consolidation and delayed intensification. Both arms included the same dose and schedule of pegaspargase (2,500 IU/m2) on day 43 and vincristine (1.5 mg/m2) on days 43 and 50 of consolidation and delayed intensification. The delayed intensification also included intrathecal methotrexate on days 29 and 36 on all arms. Patients with CNS3 leukemia received 1800 cGy of cranial irradiation during the first month of maintenance therapy. Any patient with testicular leukemia at diagnosis that did not resolve by the end of induction received 2400 cGy testicular irradiation during consolidation. The remainder of the VHR therapy was identical between the two arms.7 The complete AALL1131 VHR treatment regimen is shown in Table 1. The study did not capture detailed information on patients who underwent hematopoietic stem cell transplantation off protocol therapy. The VHR randomization was powered (80%) to compare a 4-year DFS of 70% versus 79% (HR=0.661) using a two-sided log-rank test (a=5%). DFS was defined as the time from post-induction randomization to first event (death in remission, relapse, or second malignant neoplasm) or date of last contact for those who remained event-free. Survival 987

M. Burke et al.

Table 1. AALL1131 very high-risk treatment regimen.

Experimental Arm 1

Control Arm Induction (same for Exp1 and CA) IT ARAC D1a VCR (1.5 mg/m2, 2 mg max) D1, 8, 15, 22 DEX (10 mg/m2/d) D1-14 (<10 years of age) PRED (60 mg/m2/d) D1-28 (>10 years of age) DAUN (25 mg/m2) D1, 8, 15, 22 PEG-ASP (2500 IU/m2) D4 IT MTX D8, 29b Consolidation Part 1 (same for Exp1 and CA) CPM (1 gm/m2) D1 ARAC (75 mg/m2) D1-4, D8-11 MP (60 mg/m2) D1-14 VCR (1.5 mg/m2, 2 mg max) D15, 22 PEG-ASP (2500 IU/m2) D15 IT MTX D1, 8, 15, 22 Consolidation Part 2

CPM (1 g/n2) D29 ARAC (75 mg/m2) D29-32, D36-39 MP (60 mg/m2) D29-42 VCR (1.5 mg/m2, 2 mg max) D43, 50 PEG-ASP (2500 IU/m2) D43

CPM (440 mg/m2) D29-33 ETOP (100 mg/m2) D29-33 VCR (1.5 mg/m2, 2 mg max) D43, 50 PEG-ASP (2500 IU/m2) D43

Interim Maintenance 1 (same for Exp1 and CA) IV MTX (5g/m2) followed by leucovorin rescue D1, 15, 29, 43 MP (25 mg/m2) D1-56 VCR (1.5 mg/m2, 2 mg max) D1, 15, 29, 43 IT MTX D1, 29 Delayed Intensification Part 1 (same for Exp1 and CA) VCR (1.5 mg/m2, 2 mg max) D1, 8, 15 DEX (10 mg/m2/d) D1-14, D15-21 DOX (25 mg/m2) D1, 8, 15 PEG-ASP (2500 IU/m2) D4 IT MTX D1 Delayed Intensification Part 2 CPM (1 g/m2) D29 ARAC (75 mg/m2) D29-32, D36-39 TG (60 mg/m2) D29-42 VCR (1.5 mg/m2, 2 mg max) D43, 50 PEG-ASP (2500 IU/m2) D43 IT MTX D29, 36

CPM (440 mg/m2) D29-33 ETOP (100 mg/m2) D29-33 VCR (1.5 mg/m2, 2 mg max) D43, 50 PEG-ASP (2500 IU/m2) D43 IT MTX D29, 36 Interim Maintenance 2 (same for Exp1 and CA) IV MTX (100 mg/m2) escalating D1, 11, 21, 31, 41c VCR (1.5 mg/m2, 2 mg max) D1, 11, 21, 31, 41 PEG-ASP (2500 IU/m2) D2, 22 IT MTX D1, 31

Maintenance (12-week cycles)d (same for Exp1 and CA) MTX (20 mg/m2) D8, 15, 22, 29, 36, 43, 50, 57, 64, 71, 78 MP (75 mg/m2) D1-84 VCR (1.5 mg/m2, 2 mg max) D1, 29, 57 PRED (40 mg/m2) D1-5, 29-33, 57-61 IT MTX D1 (D29 for first 2 cycles for patients who did not receive CNS radiation) ARAC: cytosine arabinoside; CNS: central nervous system; CPM: cyclophosphamide; DAUN: daunorubicin; DEX: dexamethasone; DOX: doxorubicin; ETOP: etoposide; IT,: intrathecal; MP: mercaptopurine; MTX: methotrexate; PRED: prednisone; PEG-ASP: pegaspargase; TG: thioguanine;VCR: vincristine; aIntrathecal cytarabine: 1-1.99 years (30 mg), 2-2.99 years (50 mg), ≥3 years (70 mg); bintrathecal methotrexate: 1-1.99 years (8 mg), 2-2.99 years (10 mg), 3-8.99 years (12 mg), ≥9 years (15 mg); cthe methotrexate dose was escalated as tolerated 50 mg/m2 every 10 days; dthe total duration of treatment was 2 years for females and 3 years for males from the start of interim maintenance-1 therapy.

988

Cyclophosphamide/etoposide in pediatric B-ALL

rates were estimated using the method of Kaplan-Meier with standard errors of Peto et al.9,10 Interim monitoring for efficacy utilized an at2 spending function and futility monitoring was based on the method of Anderson and High,11 with the first interim analysis scheduled for when 20% of the expected DFS events had been observed. Prospective interim monitoring rules for efficacy and futility were included where futility would be determined for a one-sided P-value ≥0.7664. Cumulative incidence rates were computed using the cumulative incidence function for competing risks, and comparisons were made using the K-sample test.12 Proportions between the two arms were compared using a c2 test or Fisher exact test. A P value <0.05 was considered statistically significant for all comparisons. All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC, USA). Graphics were generated with R version 2.13.1 (http://www.r-project.org).

Results A total of 732 eligible, evaluable patients were enrolled in the VHR part of AALL1131 and randomized to either the control arm (n=242) or experimental arm 1 (n=490) as of the data freeze for this report (December /31, 2017). The Consolidated Standards of Reporting Trials (CONSORT) diagram for the study is shown in Online Supplementary Figure S1. Two patients on the control arm and seven on experimental arm 1 had induction failures and were excluded from analyses, resulting in 240 and 483 patients, respectively, on the two arms included in this report. There were no significant differences in patients’ characteristics between the two arms (Table 2), including no difference in the proportion of patients with minimal residual disease <0.01% at the end of consolidation between those in the control arm (87.4%) and those in experimental arm 1 (87.2%). As of the data cutoff date of December 31, 2016, 20% (n=41) of expected DFS events had occurred, triggering a scheduled interim monitoring for efficacy and futility.

Figure 1. Outcomes based on information in the database at December 31, 2017 with additional follow-up, Four-year disease-free survival rates were 85.5±6.8% in the control arm (Contr) versus 72.3±6.3% in experimental arm 1 (Exp 1) (P=0.76).

haematologica | 2019; 104(5)

The 3-year DFS rates were 88.3±6.3% (control arm) versus 81.2±5.6% (experimental arm 1) (P=0.92). The study was stopped for futility as the interim monitoring boundary was crossed, indicating non-superiority of experimental arm 1 [hazard ratio 0.606 (95% confidence interval: 0.297-1.237)].

As a result, the VHR sub-study of AALL1131 was permanently closed in February 2017.

As of December 31, 2017, the date of freezing the data, the 4-year DFS rates were 85.5±6.8% (control arm) versus 72.3±6.3% (experimental arm 1) (P=0.76) (Figure 1). Table 3 gives the distribution of DFS events by arm. The 4-year cumulative incidence rates for each type of relapse are summarized in Table 4, by regimen. The cumulative incidence of isolated bone marrow relapses was significantly different between the control arm and experimental arm 1 (2.5±1.1% versus 14.5±3.3%, P=0.025). Grade 5 toxicity rates were 5.0% on the control arm (n=12) and 2.9% on experimental arm 1 (n=14). Of the 12 grade 5 toxicities reported among patients on the control arm, six occurred on therapy. These six deaths were attributed to infection

Table 2. Patients’ characteristics.

Characteristic Age <10 years ≥10 years Gender Male Female Race American Indian or Alaska Native Asian Native Hawaiian or other Pacific Islander Black or African American White Multiple races Unknown Ethnicity Hispanic or Latino Not Hispanic or Latino Unknown White blood cell count < 50x109/L ≥ 50x109/L National Cancer Institute risk Standard risk High risk Central nervous system status CNS 1 CNS 2 CNS 3 Day 29 bone marrow M1 M2 M3 (were excluded in DFS analysis) End of induction MRD MRD < 0.01% MRD ≥ 0.01%

VHR CA N (%), N=242

VHR Exp1 N (%) N=490

97 (40.1%) 145 (59.9%)

196 (40.0%) 294 (60.0%)

142 (58.7%) 100 (41.3%)

270 (55.1%) 220 (44.9%)

4 (1.7%)

8 (1.6%)

9 (3.7%) 5 (2.1%)

20 (4.1%) 2 (0.4%)

16 (6.6%) 176 (72.7%) 1 (.04%) 31 (12.8%)

24 (4.9%) 355 (72.5%) 6 (1.2%) 75 (15.3%)

78 (32.2%) 157 (64.9%) 7 (2.9%)

133 (27.1%) 334 (68.2%) 23 (4.7%)

184 (76.0%) 58 (34.0%)

389 (79.4%) 101 (20.6%)

66 (27.3%) 176 (72.7%)

143 (29.2%) 347 (70.8%)

191 (80.3%) 35 (14.7%) 12 (5.0%)

410 (84.0%) 56 (11.5%) 22 (4.5%)

232 (96.3%) 7 (2.9%) 2 (0.8%)

466 (95.3%) 16 (3.3%) 7 (1.4%)

114 (47.1%) 128 (52.9%)

207 (42.2%) 283 (57.8%)

VHR: very high-risk; CA: control arm; Exp1: experimental arm 1; CNS: central nervous system; MRD: minimal residual disease.

989

M. Burke et al.

(n=4) and multi-organ failure (n=2). Of the 14 deaths reported among patients on experimental arm 1, five occurred while on therapy and were attributed to infection (n=3), thrombosis (n=1) and multi-organ failure (n=1). A total of 235 subjects on the control arm and 477 on experimental arm 1 completed part 2 of consolidation and had toxicity data submitted. There were no significant differences in grade 3/4 adverse events or delays in starting the interim maintenance-1 phase of therapy after consolidation between the control arm and experimental arm 1 (Table 5A). In addition, a total of 193 subjects on the control arm and 389 on experimental arm 1 completed part 2 of delayed intensification and had toxicity data submitted. There were no significant differences in grade 3/4 adverse events or delays in starting the interim maintenance-1 phase of ther-

Table 3. Summary of disease-free survival events by randomization arm.

DFS event

CA (N=240)

Exp 1 (N=483)

None Relapse Isolated BM Isolated CNS Combined BM + CNS Combined BM + CNS + Other Other Secondary malignancy Death Death on therapy Death in Follow-up

212 (88.3%) 16 (6.7%) 7 4 2 0 3 0 (0.0%) 12 (5.0%) 6 6

421 (87.2%) 47 (9.7%) 32 7 2 1 5 1 (0.2%)* 14 (2.9%) 5 9

CA: control arm; Exp 1: experimental arm 1; BM: bone marrow; CNS: central nervous system; *acute myeloid leukemia.

apy after delayed intensification between the control arm and experimental arm 1 (Table 5B).

One-hundred and seven patients went off therapy during consolidation or interim maintenance-1, with the reason stated being “physician determines it is in the best interest of the patient” (n=86; 32 in the control arm and 54 in experimental arm 1) or “refusal of further protocol therapy by patient/parent/guardian” (n=21; 3 in the control arm and 18 in experimental arm 1): Some of these patients may have proceeded to hematopoietic stem cell transplantation.

Long-term study outcomes may be published in the future when follow-up data are more mature.

Discussion The 5-year event-free survival rate for patients with highrisk B-ALL enrolled in the COG AALL0232 trial (2004 – 2011) randomized to high-dose methotrexate during interim maintenance-1 was 79.6% compared to 75.2% for the patients in the control arm (Capizzi methotrexate) (P=0.008) of that study.3 Patients identified as having VHR B-ALL are predicted to fare worse than high-risk patients overall and, depending on the specific VHR risk factors, their 4-year DFS rates can range anywhere from 40 to 80%.8,13-22 Based on the relatively poor outcomes for these VHR patients, they are candidates for investigation of novel, more intensive, yet potentially more toxic, therapeutic strategies designed to improve DFS. Based on this hypothesis, and without mature data from the AALL0232 study available at the time of study development, the COG high-risk B-ALL study AALL1131 was designed to further intensify cytotoxic chemotherapy during the consolidation

Table 4. Cumulative incidence rates for types of relapse by randomization arm.

Relapse type

VHR CA 4-year cumulative incidence Rate ±SE 95% CI

VHR Exp 1 4-year cumulative incidence Rate ±SE 95% CI

Isolated BM Isolated CNS Combined BM + CNS Other

2.5±1.1% 2.3±1.2% 0.9±0.6% 1.5±0.9%

14.5±3.3% 3.9±1.7% 2.8±2.2% 1.4±0.7%

(0.9%, 5.5%) (0.7%, 5.6%) (0.2%, 3.0%) (0.4%, 4.1%)

(8.8%, 21.6%) (1.5%, 8.3%) (0.4%, 9.6%) (0.5%, 3.2%)

P-value

0.025 0.880 0.525 0.817

VHR: very high risk; CA: control arm; Exp 1: experimental arm 1; SE: standard error; 95% CI: 95% confidence interval; BM: bone marrow; CNS: central nervous system.

Table 5A. Toxicities in patients completing part 2 of consolidation.

Targeted toxicity Grade (Gr) Gr 4 infection Gr 3/4 AST/ALT* Gr 4 lipase/amylase* Gr 3/4 bilirubin* Gr 3/4 pancreatitis Gr 3/4 acute kidney injury Gr 3/4 non-hematologic* Delays >14 days in starting interim maintenance

CA (N=235) 6 (2.6%)

Exp 1 (N=477) 12 (2.5%)

P-value 0.98

5 (2.1%)

8 (1.7%)

0.68

5 (2.1%) 0 5 (2.1%) 64 (27.2%)

6 (1.3%) 0 20 (4.2%) 102 (21.4%)

0.21 NA 0.12 0.09

Grade 3 and 4 targeted toxicities (CTCAE v4.0). *Do not return to grade ≤2 by the time day 43 vincristine and asparaginase are scheduled to be administered during consolidation. CA, control arm; Exp 1, experimental arm 1.

990

Table 5B. Toxicities in patients completing part 2 of the delayed intensification.

Targeted toxicity Grade (Gr)

CA (N=193)

Gr 4 infection 10 (5.2%) Gr 3/4 AST/ALT* Gr 4 lipase/amylase* 5 (2.6%) Gr 3/4 bilirubin* Gr 3/4 pancreatitis 2 (1.0%) Gr 3/4 acute kidney injury 1 (0.5%) Gr 3/4 non-hematologic* 5 (2.6%) Delays >14 days in 49 (25.4%) starting interim maintenance

Exp 1 (N=389)

P-value

20 (5.1%)

0.98

4 (1.0%)

0.21

9 (2.3%) 1 (0.3%) 15 (3.9%) 84 (21.6%)

0.23 0.65 0.40 0.16

Grade 3 and 4 targeted toxicities (CTCAE v4.0). *Do not return to grade ≤2 by the time day 43 vincristine and asparaginase are scheduled to be administered during delayed intensification. CA, control arm; Exp 1, experimental arm 1.

haematologica | 2019; 104(5)

Cyclophosphamide/etoposide in pediatric B-ALL

and delayed intensification phases of therapy to improve the DFS in the VHR subgroup. Two intensification strategies were tested in AALL1131 and compared to the control arm. The first intensification strategy included the combination of clofarabine with cyclophosphamide/etoposide, a promising combination in relapsed ALL23,24 (experimental arm 2) and cyclophosphamide/etoposide without clofarabine (experimental arm 1). This began as a 1:2:2 randomization between the control arm and experimental arms 1 and 2, respectively. Experimental arm 2, testing clofarabine, was found to be too toxic and not feasible when given in this combination to newly diagnosed patients with VHR B-ALL, and this arm of AALL1131 was, therefore, closed to further accrual in September 2014.7 AALL1131 thus continued as a two-arm study comparing the control arm with experimental arm 1 in a 1:2 randomized fashion. This randomization was later stopped for futility when the interim monitoring boundary was crossed, identifying non-superiority of DFS when consolidation and delayed intensification included cyclophosphamide/etoposide compared to standard VHR therapy (modified augmented Berlin-Frankfurt-Münster regimen).

With additional follow-up after closure of the randomization, there was even stronger evidence that experimental arm 1 would never be superior to the control arm with the reported DFS being 85.5±6.8% for the control arm compared to 72.3±6.3% for experimental arm 1 (P=0.76).

The 4-year DFS of 85.5±6.8% reported for the control arm of this study was higher than the 70% we originally predicted based on data available for patients with VHR features treated in the preceding B-ALL studies for standard-risk (AALL0331) and high-risk (AALL0232) patients.

Many patients in these earlier studies did not receive high-dose methotrexate during interim maintenance-1 which may have resulted in the differences in the DFS rates we report. Additionally, the definitions of VHR were expanded in AALL1131 to include groups of patients at least 13 years of age as well as lower minimal

References 1. Hunger SP, Lu X, Devidas M, et al. Improved survival for children and adolescents with acute lymphoblastic leukemia between 1990 and 2005: a report from the children's oncology group. J Clin Oncol. 2012;30(14): 1663-1669.

2. Pui CH, Yang JJ, Hunger SP, et al. Childhood acute lymphoblastic leukemia: progress through collaboration. J Clin Oncol.

2015;33(27):2938-2948.

3. Larsen EC, Devidas M, Chen S, et al.

Dexamethasone and high-dose methotrexate improve outcome for children and young adults with high-risk B-acute lymphoblastic leukemia: a report from children's oncology group study AALL0232. J Clin Oncol. 2016;34(20):2380-2388.

4. Raetz EA, Borowitz MJ, Devidas M, et al.

Reinduction platform for children with first marrow relapse of acute lymphoblastic leukemia: a Children's Oncology Group Study [corrected]. J Clin Oncol. 2008;26 (28):3971-3978.

5. Raetz EA, Cairo MS, Borowitz MJ, et al. Reinduction chemoimmunotherapy with epratuzumab in relapsed acute lymphoblas-

haematologica | 2019; 104(5)

6.

7.

8.

9.

10.

residual disease thresholds which may also have contributed to differences in the DFS we observed.

In summary, intensification of cytotoxic chemotherapy by substituting either clofarabine/cyclophosphamide/ etoposide or cyclophosphamide/etoposide for cyclophosphamide/cytarabine/6-mercaptopurine (or 6-thioguanine) during part 2 of consolidation and delayed intensification did not improve DFS compared to that of patients receiving standard COG VHR therapy in this study. Future therapeutic studies for pediatric patients with VHR B-ALL could pursue immunologically and/or molecularly targeted therapies which may have more potential to improve outcomes than further intensification of cytotoxic chemotherapy.25-27 In this regard, the COG is currently investigating the tyrosine kinase inhibitor dasatinib for newly diagnosed high-risk patients with Philadelphia-like B-ALL harboring ABL-class lesions (AALL1131, NCT02883049) and ruxolitinib for newly diagnosed NCI-HR patients with Philadelphia-like BALL harboring CRLF2 rearranged and/or JAK pathway mutated B-ALL (AALL1521, NCT02723994). Additionally, the COG plans to bring both inotuzumab ozogamicin (humanized monoclonal antibody against CD22) and blinatumomab (anti-CD19/CD3 bispecific T-cell engager antibody) into the next generation of upfront studies for highrisk and standard-risk B-ALL, respectively, anticipated to open in 2019.

Acknowledgments This trial was supported by grants U10 CA98543, U10 CA98413, U10 CA180886, U10 CA180899 from the National Institutes of Health and supported by St. Baldrick’s Foundation.

LG is the Ergen Family Chair in Pediatric Oncology at the Children’s Hospital of Colorado. EAR is a KiDS of NYU Foundation Professor at NYU Langone Health. MLL is the UCSF Benioff Chair of Children’s Health and Deborah and Arthur Ablin Endowed Chair in Pediatric Molecular Oncology. SPH is the Jeffrey E. Perelman Distinguished Chair in the Department of Pediatrics at The Children's Hospital of Philadelphia.

tic leukemia (ALL): phase II results from Children's Oncology Group (COG) study ADVL04P2. Pediatr Blood Cancer.

2015;62(7):1171-1175.

Crooks GM, Sato JK. Ifosfamide and etoposide in recurrent childhood acute lymphoblastic leukemia. J Pediatr Hematol Oncol. 1995;17(1):34-38.

Salzer WL, Burke MJ, Devidas M, et al.

Toxicity associated with intensive postinduction therapy incorporating clofarabine in the very high-risk stratum of patients with newly diagnosed high-risk B-lymphoblastic leukemia: a report from the Children's Oncology Group study AALL1131. Cancer.

2018;124(6):1150-1159.

Borowitz MJ, Wood BL, Devidas M, et al.

Prognostic significance of minimal residual disease in high risk B-ALL: a report from Children's Oncology Group study AALL0232. Blood. 2015;126(8):964-971.

Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. analysis and examples. Br J Cancer. 1977;35(1):1-39.

Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53:457-481.

11. Anderson JR, High R. Alternatives to the standard Fleming, Harrington, and O'Brien futility boundary. Clin Trials. 2011;8(3): 270-276.

12. Gray R. A Class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16:1141-1154.

13. Schultz KR, Pullen DJ, Sather HN, et al.

Risk- and response-based classification of childhood B-precursor acute lymphoblastic leukemia: a combined analysis of prognostic markers from the Pediatric Oncology Group (POG) and Children's Cancer Group (CCG). Blood. 2007;109(3): 926-935.

14. Pui CH, Crist WM, Look AT. Biology and clinical significance of cytogenetic abnormalities in childhood acute lymphoblastic leukemia. Blood. 1990;76(8):1449-1463.

15. Pui CH, Gaynon PS, Boyett JM, et al.

Outcome of treatment in childhood acute lymphoblastic leukaemia with rearrangements of the 11q23 chromosomal region.

Lancet. 2002;359(9321):1909-1915.

16. Pui CH, Carroll AJ, Raimondi SC, et al.

Clinical presentation, karyotypic characterization, and treatment outcome of childhood acute lymphoblastic leukemia with a nearhaploid or hypodiploid less than 45 line.

991

M. Burke et al.

Blood. 1990;75(5):1170-1177.

17. Schultz KR, Bowman WP, Aledo A, et al.

Improved early event-free survival with imatinib in Philadelphia chromosome-positive acute lymphoblastic leukemia: a Children's Oncology Group study. J Clin Oncol. 2009;27(31):5175-5181.

18. Pui CH, Chessells JM, Camitta B, et al.

Clinical heterogeneity in childhood acute lymphoblastic leukemia with 11q23 rearrangements. Leukemia. 2003;17(4):700706.

19. Heerema NA, Nachman JB, Sather HN, et al.

Hypodiploidy with less than 45 chromosomes confers adverse risk in childhood acute lymphoblastic leukemia: a report from the Children's Cancer Group. Blood. 1999; 94(12):4036-4045.

20. Silverman LB, Gelber RD, Young ML, Dalton VK, Barr RD, Sallan SE. Induction

992

21.

22.

23.

24.

failure in acute lymphoblastic leukemia of childhood. Cancer. 1999;85(6):1395-1404. van Dongen JJ, Seriu T, Panzer-Grumayer ER, et al. Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet. 1998;352 (9142):1731-1738.

Dworzak MN, Froschl G, Printz D, et al.

Prognostic significance and modalities of flow cytometric minimal residual disease detection in childhood acute lymphoblastic leukemia. Blood. 2002;99(6):1952-1958.

Hijiya N, Thomson B, Isakoff MS, et al.

Phase 2 trial of clofarabine in combination with etoposide and cyclophosphamide in pediatric patients with refractory or relapsed acute lymphoblastic leukemia. Blood.

2011;118(23):6043-6049.

Locatelli F, Testi AM, Bernardo ME, et al.

Clofarabine, cyclophosphamide and etopo-

side as single-course re-induction therapy for children with refractory/multiple relapsed acute lymphoblastic leukaemia. Br J Haematol. 2009;147(3):371-378.

25. Maude SL, Frey N, Shaw PA, et al. Chimeric antigen receptor T cells for sustained remissions in leukemia. N Engl J Med. 2014;371(16):1507-1517.

26. Gokbuget N, Zugmaier G, Klinger M, et al.

Long-term relapse-free survival in a phase 2 study of blinatumomab for the treatment of patients with minimal residual disease in Blineage acute lymphoblastic leukemia.

Haematologica. 2017;102(4):e132-e135.

27. Weston BW, Hayden MA, Roberts KG, et al.

Tyrosine kinase inhibitor therapy induces remission in a patient with refractory EBF1PDGFRB-positive acute lymphoblastic leukemia. J Clin Oncol. 2013;31(25):e413416.

haematologica | 2019; 104(5)

ARTICLE

Non-Hodgkin Lymphoma

Mogamulizumab versus investigator’s choice of chemotherapy regimen in relapsed/ refractory adult T-cell leukemia/lymphoma Adrienne A. Phillips,1 Paul A. Fields,2 Olivier Hermine,3 Juan C. Ramos,4 Brady E. Beltran,5 Juliana Pereira,6 Farooq Wandroo,7 Tatyana Feldman,8 Graham P. Taylor,9 Ahmed Sawas,10 Jeffrey Humphrey,11 Michael Kurman,11 Junji Moriya,11 Karen Dwyer,11 Mollie Leoni,11 Kevin Conlon,12 Lucy Cook,13 Jason Gonsky14 and Steven M. Horwitz;15 on behalf of the 0761-009 Study Group

Division of Hematology and Medical Oncology, Weill Cornell Medical College, New York Presbyterian Hospital, New York, NY, USA; 2Department of Haematology Guy's and St Thomas' Hospitals NHS Trust Hospital, London, UK; 3Department of Hematology, Necker University Hospital, Paris, France; 4Division of Hematology/Oncology, University of Miami Miller School of Medicine, Sylvester Comprehensive Cancer Center, FL, USA; 5Hospital Nacional Edgardo Rebagliati Martins and Centro de Investigación de Medicina de Precision, Universidad de San Martin de Porres, Lima, Peru; 6Department of Hematology, University of São Paulo, Brazil; 7Sandwell and West Birmingham Hospitals NHS Trust, West Bromwich, and University of Birmingham, UK; 8John Theurer Cancer Center, Hackensack UMC, NJ, USA; 9National Centre for Human Retrovirology, St Mary's Hospital, Imperial College Healthcare NHS Trust, London, UK; 10Center for Lymphoid Malignancies, Columbia University Irving Medical Center, New York, NY, USA; 11Kyowa Kirin, Princeton, NJ, USA; 12Warren Grant Magnuson Clinical Center, National Cancer Institute, Bethesda, MD, USA; 13Department of Haematology and National Centre for Human Retrovirology, Imperial College Healthcare NHS Trust, London, UK; 14Division of Hematology/Oncology, Department of Medicine, New York City Health + Hospitals/Kings County and SUNY Downstate Medical Center, Brooklyn, NY, USA and 15Hematology/ Oncology, Memorial Sloan-Kettering Cancer Center, New York, NY, USA 1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):993-1003

Correspondence: ABSTRACT

M

ogamulizumab, a humanized defucosylated anti-C-C chemokine receptor 4 monoclonal antibody, has been approved in Japan for the treatment of C-C chemokine receptor 4-positive adult T-cell leukemia/lymphoma (ATL). This phase II study evaluated efficacy and safety of mogamulizumab in ATL patients with acute, lymphoma, and chronic subtypes with relapsed/refractory, aggressive disease in the US, Europe, and Latin America. With stratification by subtype, patients were randomized 2:1 to intravenous mogamulizumab 1.0 mg/kg once weekly for 4 weeks and biweekly thereafter (n=47) or investigator’s choice of chemotherapy (n=24). The primary end point was confirmed overall response rate (cORR) confirmed on a subsequent assessment at 8 weeks by blinded independent review. ORR was 11% (95%CI: 4-23%) and 0% (95%CI: 0-14%) in the mogamulizumab and chemotherapy arms, respectively. Best response was 28% and 8% in the respective arms. The observed hazard ratio for progression-free survival was 0.71 (95%CI: 0.41-1.21) and, after post hoc adjustment for performance status imbalance, 0.57 (95%CI: 0.3370.983). The most frequent treatment-related adverse (grade ≥3) events with mogamulizumab were infusion-related reaction and thrombocytopenia (each 9%). Relapsed/refractory ATL is an aggressive, poor prognosis disease with a high unmet need. Investigator’s choice chemotherapy did not result in tumor response in this trial; however, mogamulizumab treatment resulted in 11% cORR, with a tolerable safety profile. Trial registered at clinicaltrials.gov identifier: 01626664. haematologica | 2019; 104(5)

ADRIENNE A. PHILLIPS adp9002@med.cornell.edu Received: August 22, 2018.

Accepted: December 18, 2018.

Pre-published: December 20 2018. doi:10.3324/haematol.2018.205096 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/993 ©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.

993

Adrienne A. Phillips et al.

Introduction Adult T-cell leukemia/lymphoma (ATL) is an aggressive, rare, peripheral T-cell lymphoma (PTCL) caused by human T-cell lymphotropic virus type I (HTLV-1).1-5 Approximately 2-7% of people infected with HTLV-1 develop ATL, often after decades of infection.6 HTLV-1 is endemic in Southern Japan, the Caribbean, Central and South America, Central and South Africa, parts of the Middle East and Melanesia, and aboriginal regions of Australia.7 In non-endemic areas such as North America and Europe, HTLV-1 infection and ATL have been linked to immigration from endemic areas.8-10 It is estimated that ATL accounts for 0.2% of lymphomas in the US but as many as 37% in Kyushu, Japan.11 Compared to other subtypes of PTCL, ATL has the worst prognosis with 5-year overall survival (OS) of 14%.5 Adult T-cell leukemia/lymphoma is classified as smoldering, chronic, lymphoma, and acute subtypes.12 In Japan, aggressive subtypes of ATL (acute and lymphoma) have a poor prognosis with median OS of around 12 months, even with intensive chemotherapy regimens.13 A long-term retrospective study has shown that even the indolent subtypes of ATL (smoldering, chronic) have a poorer than expected prognosis with median survival of only 4.1 years.14 Outside Japan, there is no approved treatment for ATL. In a retrospective series of 89 ATL patients at three New York City medical centers, median OS across subtypes was approximately 6 months.15 Allogeneic stem cell transplantation (allo-SCT) can significantly prolong survival, but there are few appropriate candidates because of age, availability of a stem cell source, lack of adequate response to primary therapy, and/or absence of effective agents in the relapsed/refractory setting.16-18 Almost all patients (≥90%) with ATL over-express C-C chemokine receptor 4 (CCR4) on tumor cells.19,20 Mogamulizumab is a first-in-class defucosylated humanized IgG1 kappa monoclonal antibody that selectively binds to CCR4 and has enhanced antibody-dependent cellular cytotoxicity (ADCC) activity.21 Mogamulizumab is approved in Japan for the treatment of relapsed/refractory CCR4+ ATL on the basis of a phase II trial showing a 50% overall response rate (ORR) in a relapsed population.22 It was subsequently approved for chemotherapy-naïve CCR4+ ATL on the basis of a randomized phase II trial in combination with the mLSG15 regimen.23 In order to study mogamulizumab outside Japan, we conducted a phase II randomized trial of mogamulizumab monotherapy compared to investigator’s choice of chemotherapy in patients with relapsed/refractory ATL and, herein, report the results.

Methods Patients

Patients ≥18 years of age with a confirmed diagnosis of ATL (HTLV-1 antibody positive) who met criteria for the acute, lymphoma, or chronic ATL subtypes12 and who were refractory or relapsed after at least one prior systemic therapy were eligible to enroll [chronic patients were retrospectively designated favorable or unfavorable based on serum BUN, lactate dehydrogenase (LDH) and albumin levels]. Disease had to be evident in at least one compartment: lymph nodes, extranodal masses, spleen, liver, 994

skin, peripheral blood, or bone marrow. Patients were required to be Eastern Cooperative Oncology Group (ECOG) performance status ≤2, with adequate hematologic, hepatic, and renal function.

Patients were excluded if they had a history of allo-SCT, active concurrent cancers, or central nervous system (CNS) involvement.

Patients randomized to the investigator’s choice arm could not receive a regimen that they had previously received or to which they had a contraindication.

Because the disease is aggressive, refractory patients enrolled early in the study had difficulty completing the first treatment cycle. To enroll a population able to receive adequate drug exposure and more likely to be able to benefit from treatment, the protocol was amended to exclude patients with acute or lymphoma subtypes who had received >2 prior systemic therapy regimens and had not achieved a response or maintained stable disease ≥12 weeks on immediate prior therapy.

The trial was conducted in accordance with the Declaration of Helsinki, the International Conference on Harmonization consolidated good clinical practice guideline, and any applicable national and local laws and regulations. The protocol was reviewed and approved by institutional review boards or independent ethics committees at each site. All patients provided written informed consent.

Study design This was an international, multicenter, open-label, randomized study conducted at 22 centers (see Online Supplementary Appendix) in Belgium, Brazil, France, Martinique, Peru, the UK, and the US; 18 centers screened and 17 randomized patients. A Steering Committee selected investigator’s choice regimens: pralatrexate, GemOx (gemcitabine and oxaliplatin), or DHAP (dexamethasone, cisplatin, and cytarabine) which were appropriate for a relapsed/refractory population. Eligible patients were randomized 2:1 to mogamulizumab or investigator’s choice arms with stratification by ATL subtype (acute, chronic, or lymphoma). Patients who progressed in the investigator’s choice arm were permitted to cross over to mogamulizumab.

The primary objective of the study was to determine the ORR of mogamulizumab that persisted and was confirmed at a subsequent response evaluation, 8 weeks after initial response (confirmed ORR, cORR). Secondary objectives were to compare cORR, progression-free survival (PFS), OS, time to response, and duration of response (DoR) between the treatment arms and to assess safety.

Drug administration Mogamulizumab 1.0 mg/kg was administered by intravenous (IV) infusion over ≥1 hour (h) once weekly during the first cycle (days 1, 8, 15 and 22 of the first 28-day cycle) and on days 1 and 15 of subsequent cycles without dose modification. Pralatrexate 30 mg/m2 was administered IV over 3-5 minutes (min) once weekly for 3 weeks followed by 1 week without.24 GemOx comprised IV gemcitabine 1000 mg/m2 over 30 min followed by IV oxaliplatin 100 mg/m2 over 2 h every 2 weeks. DHAP comprised IV dexamethasone 40 mg over 5-15 min on days 1 to 4 and IV cisplatin 100 mg/m2 over 24 h on day 1 followed by IV cytarabine 2000 mg/m2 over 3 h immediately after cisplatin and again 12 h later on day 2 every 4 weeks. For investigator’s choice regimens, dose modifications were permitted and applicable treatment recommendations followed according to local prescribing information. Treatment continued until progressive disease (PD), unacceptable toxicity, or withdrawal of consent.

Assessments Efficacy was determined by an independent, blinded review haematologica | 2019; 104(5)

Mogamulizumab vs. investigator's choice in ATL

(Independent Review) and by the investigators (Investigator Assessment). Response, stringently and globally evaluated in six potential disease compartments (blood, skin, lymph nodes, extranodal masses, liver/spleen, and bone marrow), was determined according to published response criteria for ATL25 that assessed skin via modified Severity Weighted Assessment Tool; lymph nodes, extranodal masses, liver, and spleen by PET and/or CT; and bone marrow by biopsy at baseline and to confirm PD or CR; central flow cytometry rather than morphology was used for blood evaluation. Response was determined at the end of the first treatment cycle and every 8 weeks thereafter. cORR required confirmation and maintenance of response at the next successive evaluation. Best response included all responses at any time point. PFS, OS, time to response, and DoR were defined according to standard methods.

Adverse events (AEs) were coded by Medical Dictionary for Regulatory Activities, v.15.0 and graded using the National Cancer Institute Common Terminology Criteria v.4.0. For patients receiving mogamulizumab who developed a grade 2 or greater skin rash, treatment was to be interrupted and the rash treated with topical steroids.

Validated electrochemiluminscence immunoassays were used to determine anti-mogamulizumab and neutralizing anti-mogamulizumab antibodies. Correlative studies were done to study mogamulizumab pharmacokinetics and neutralizing antibody.

CCR4 expression status was determined by flow cytometry in patients with blood disease (CD45+CD4+CD25+CCR4+CD7– ≥5% considered positive) or by immunohistochemistry (positive value defined as ≥10%) in those without blood involvement.

Statistical analysis The sample size was estimated based on a feasible accrual of approximately 70 patients, which was predicted to require 3 years. The primary end point, cORR by Independent Review, was estimated using an exact 95% confidence interval (CI). The mogamulizumab arm sample size (n=47) was chosen to yield a maximum width of a 95%CI on ORR to be <30%. This does not assume a target value for ORR due to the rarity of ATL and the lack of published efficacy data for the investigator’s choice options in the relapsed and refractory setting. With 2:1 randomization approximately 23 patients would be enrolled in the investigator’s choice arm.

All analyses were performed using SAS v.9.3 (SAS Institute, Cary, NC, USA). Comparison of cORR between treatment arms was performed using an exact 95% unconditional confidence for the risk difference. Survival estimates were calculated using the Kaplan-Meier method. PFS and OS were analyzed using Cox proportional hazards models, and, if data warranted, for PFS using a multivariate Cox proportional model adjusting for selected potential prognostic factors. Other results are shown descriptively.

Results Patients A total of 71 patients were enrolled to the mogamulizumab (n=47) and investigator’s choice (n=24) arms between August 2012 and May 2015 (Figure 1). Two patients, both in the mogamulizumab arm, remained on treatment at time of efficacy data cut-off on March 31, 2016. A final data cut-off for survival data was made on December 31, 2017. Investigator’s choice regimens were GemOx (n=21), pralatrexate (n=2), and DHAP (n=1). All 71 randomized patients were included in the intent-totreat (ITT) and safety populations. Eighteen of the 24 haematologica | 2019; 104(5)

patients (75%) from the investigator’s choice arm crossed over to receive mogamulizumab as administered in the randomized study.

Characteristics of the randomized patients are shown in Table 1. Despite randomization, there were imbalances between the treatment arms in known prognostic factors.15 The mogamulizumab arm had a higher median age (55.0 vs. 50.5 years), with a consequently higher proportion of patients aged >65 years (23% vs. 4%) and fewer aged <40 years (13% vs. 29%) compared to the investigator’s choice arm. More patients had an ECOG performance status of 2 in the mogamulizumab arm (40% vs.

29%). In addition, patients randomized to investigator’s choice of chemotherapy were more likely to have been responsive to their most immediate prior therapy versus those randomized to mogamulizumab (46% vs. 26%, respectively) and were less likely to have bone marrow involvement (33% vs. 57%). The treatment arms were well balanced with respect to other characteristics including gender, geographical region, ATL subtype, and prior ATL regimens. Tumor CCR4-positivity was 96% in each arm.

Despite amending the protocol to enroll patients less heavily pre-treated and more likely to benefit, the number of patients with ECOG performance status of 2 and those responding to immediate prior therapy was virtually the same pre- and post-amendment (Online Supplementary Table S1). Furthermore, the percentage of patients who completed ≤1 cycle was the same pre- and post-amendment (65% for both), indicative of the aggressive nature of their disease. These patients were considered non-responders in the ITT analysis.

Efficacy Confirmed ORR by Independent Review for mogamulizumab was 11% [1 complete response (CR), 4 partial response (PR); 95%CI: 4-23%] compared to 0% (95%CI: 0-14%) for the investigator’s choice arm. Best response was 28% (95%CI: 16-43%) versus 8% (95%CI: 1-27%) for mogamulizumab and investigator’s choice, respectively. A secondary analysis comparing cORR by treatment as assessed by Independent Review did not detect a significant difference (risk difference 10.6%; 95%CI: –14%34%).

By Investigator Assessment, cORR was 15% (95%CI: 6-28%) for mogamulizumab compared to 0% (95%CI: 0-14%) for the investigator’s choice arm. Best response was 34% (95%CI: 21-49%) versus 0% (95%CI: 0-14%) (Table 2), respectively.

Independent and investigator review identified a largely concordant group of responding patients, suggesting response assessment was not influenced by investigator bias. Because Investigator Assessments were considered to provide a more comprehensive evaluation of the patients' disease status and potential clinical improvement (investigators had access to all local labs, physical exam findings, and skin rash/infusion reaction, which was withheld from independent review to preserve blind conditions), the following, secondary efficacy results are described based upon Investigator Assessment only.

By compartment responses to mogamulizumab (Table 2 and Online Supplementary Table S3) were highest in blood (21/39; 54%, all CR) and skin (8/18; 44%). Responses by compartment to investigator’s choice of chemotherapy were only seen in skin (5/9, 56%) and blood (1/18, 6%). In 995

Adrienne A. Phillips et al.

Figure 1. CONSORT diagram. ITT: intent-to-treat.

the 18 patients crossed over to mogamulizumab, three (17%) demonstrated a response.

Responses to mogamulizumab were seen in all enrolled subtypes. Best and confirmed responses by ATL subtype to mogamulizumab were chronic 71% (5/7) and 43% (3/7); lymphoma 32% (6/19) and 5% /1/19); and acute 24% (5/21) and 5% (1/21), respectively. In the mogamulizumab arm, four out of the seven chronic patients had unfavorable characteristics; of those three had a response. Of the three patients with favorable characteristics, two had a response. Three chronic patients initially received Investigator’s Choice regimen. All three crossed over to treatment with mogamulizumab; there were no responses to either treatment in these patients.

In the mogamulizumab arm, best response was 46% (13/28) for patients with ECOG 0/1 and 16% (3/19) for ECOG 2. In the Investigator’s Choice group, no responses were observed.

Median time to response in the mogamulizumab arm was 1.13 (95%CI: 0.87-3.40) months, with most (75%) responses occurring by the first assessment at 4 weeks.

Median DoR in the mogamulizumab arm was 5.65 (95%CI: 3.63-not reached) months.

Median PFS was 0.93 (95%CI: 0.87-1.13) and 0.88 (95%CI: 0.50-0.93) months in the mogamulizumab and investigator’s choice arms, respectively. The observed hazard ratio (HR) for PFS was 0.71 (95%CI: 0.41-1.21) (Figure 2). As PFS may have been affected by imbalances in baseline prognostic characteristics, post hoc sensitivity analyses adjusting for these imbalances were performed (Figure 3). Adjusting for the imbalances in ECOG performance status and for response to last prior ATL therapy yielded an HR for PFS of 0.57 (95%CI: 0.327-0.983) and 996

0.58 (95%CI: 0.330-1.006), respectively. Survival analysis was confounded by the one-way crossover design; however, there was no apparent overall survival advantage or disadvantage associated with mogamulizumab use (Online Supplementary Figure S1).

Five patients (1 acute, 2 lymphoma, and 2 chronic) progressed per protocol in a single compartment but derived clinical benefit according to Investigator Assessment.

These patients were allowed to continue treatment after discussion with the study sponsor (Figure 4 and Online Supplementary Figure S2). These patients remained on mogamulizumab with clinical improvement and/or disease control for a median of 230 (range, 182-463) days.

Four of the five patients had blood disease, and response continued through to the end of data collection for this group. Of these four patients, one is alive 56 months post initial treatment with subsequent spot radiation to 3 skin lesions. Another subject is alive 41 months post initial treatment and progressed in lymph nodes per size criteria; however, the investigator felt these were more likely to be reactive nodes. Two patients progressed in skin and an additional patient in skin and nodes. No subjects directly bridged to transplant without subsequent therapy in either arm of the study (Figure 4).

Safety Mean (±Standard Deviation) duration of randomized treatment (78.0±141.51 vs. 26.5±33.61 days) and the number of treatment cycles initiated (3.1±4.60 vs. 1.5±0.98) were higher in the mogamulizumab arm than the investigator’s choice arm.

The overall incidence of treatment-related any-grade (83% vs. 88%), grade ≥3 (32% vs. 29%), or serious (23% haematologica | 2019; 104(5)

Mogamulizumab vs. investigator's choice in ATL Table 1. Patients’ demographic and clinical characteristics.

Characteristic Age (y) Median (range) >65 years <40 years Gender Male Female Race Black White Asian Other Unknown* Geographical region North America Europe South America and Caribbean ECOG performance status 0 1 2 ATL subtype at study entry Acute Lymphoma Chronic Disease site Lymph nodes Peripheral blood Bone marrow Skin Extranodal masses Spleen Liver Other None reported Median time from initial ATL diagnosis, months (range) CCR4 expression status Positive Negative Not done Number of prior ATL regimens, median (range) Prior ATL regimens AZT CHOP Interferon EPOCH Hyper-CVAD ICE Pralatrexate Autologous SCT Other Best response to immediate prior ATL therapy CR PR SD PD Unknown

Mogamulizumab (n = 47)

Investigator’s choice (n = 24)

55.0 (22-82) 11 (23) 6 (13)

50.5 (24-80) 1 (4) 7 (29)

24 (51) 23 (49)

10 (42) 14 (58)

32 (68) 6 (13) 2 (4) 1 (2) 6 (13)

15 (63) 5 (21) 1 (4) 0 3 (13)

25 (53) 14 (30) 8 (17)

14 (58) 7 (29) 3 (13)

12 (26) 16 (34) 19 (40)

11 (46) 6 (25) 7 (29)

21 (45) 19 (40) 7 (15)

12 (50) 9 (38) 3 (13)

41 (87) 37 (79) 27 (57) 13 (28) 12 (26) 10 (21) 2 (4) 1 (2) 0 9.1 (1.3-116.7)

20 (83) 17 (71) 8 (33) 9 (38) 8 (33) 4 (17) 3 (13) 0 1 (4)† 6.6 (1.3-150.6)

43 (96) 2 (4) 2 2.0 (1-6)

22 (96) 1 (4) 1 1.5 (1-5)

19 (40) 21 (45) 15 (32) 9 (19) 5 (11) 3 (6) 4 (9) 1 (2) 34 (72)

9 (38) 5 (21) 9 (38) 6 (25) 1 (4) 3 (13) 0 1 (4) 16 (67)

3 (6) 9 (19) 12 (26) 19 (40) 4 (9)

5 (21) 6 (25) 3 (13) 9 (38) 1 (4)

Data are given as n (%) unless otherwise stated. ATL: adult T-cell leukemia/lymphoma; AZT: zidovudine; CCR4: C-C chemokine receptor 4; CHOP: cyclophosphamide, doxorubicin, vincristine, and prednisone; CR: complete response; CVAD: cyclophosphamide, vincristine, dexamethasone, and doxorubicin; ECOG: Eastern Cooperative Oncology Group; EPOCH: etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin; ICE: ifosfamide, carboplatin, and etoposide; PD: progressive disease; PR: partial response; SCT: stem cell transplantation; SD: stable disease. *Not reported for those countries that do not allow race/ethnicity data to be collected. †This patient met eligibility criteria with disease in blood and not in lymph nodes according to the investigator but showed lymph node and no blood involvement on independent review.

haematologica | 2019; 104(5)

997

Adrienne A. Phillips et al.

vs. 17%) AEs were similar between mogamulizumab and investigator’s choice arms, respectively, while the overall incidence of treatment-related AEs leading to discontinuation (19% vs. 0%) was higher in the mogamulizumab arm and were most frequently due to infusion reactions and drug eruptions [2 patients (4.3%) each]. There were no treatment-related deaths during randomization or after crossover to mogamulizumab.

The most common treatment-related AEs during randomization and after crossover to mogamulizumab are summarized in Table 3. The most common treatment-related AEs (any grade) in the mogamulizumab arm were infusion-related reaction (47%), drug eruption (19%), thrombocytopenia (13%), and anemia (11%). The most common treatment-related AEs in the investigator’s choice arm were

neutropenia (25%), thrombocytopenia (21%), nausea (17%), diarrhea (17%), pyrexia (13%), headache (13%), constipation (13%), and vomiting (13%). The most common treatment-related AEs grade ≥3 in the randomized mogamulizumab arm were infusion-related reaction (9%) and thrombocytopenia (9%). The most common treatment-related AE grade ≥3 in the investigator’s choice arm was thrombocytopenia (17%). Treatment-related AEs (any grade or grade ≥3) after crossover were generally similar to those seen in randomized patients. The only treatmentrelated serious AE occurring in more than one patient in the mogamulizumab arm was pneumonia (n=3).

None of the patients developed detectable anti-mogamulizumab or neutralizing anti-mogamulizumab antibody following treatment.

Table 2. Best overall response and by disease compartment response according to investigator assessment during randomization and after crossover to mogamulizumab (ITT population).

Best response overall and by disease compartment Overall CR CRu PR SD PD Not evaluable† Blood CR CRu PR SD PD Not assessable* Lymph nodes CR CRu PR SD PD Not assessable* Skin CR CRu PR SD PD Not assessable*

Randomized

After crossover Mogamulizumab

Mogamulizumab

Investigator’s choice

n = 47 1 (2) 2 (4) 13 (28) 2 (4) 12 (26) 17 (36) n = 39 21 (54) 0 0 3 (8) 0 15 (39) n = 44 0 1 (2) 3 (7) 13 (30) 11 (25) 16 (36) n = 18 3 (17) 0 5 (28) 2 (11) 5 (28) 3 (17)

n = 24 0 0 0 6 (25) 11 (46) 7 (29) n = 18 1 (6) 0 0 10 (56) 4 (22) 3 (17) n = 22 0 0 0 10 (46) 8 (36) 4 (18) n=9 0 0 5 (56) 3 (33) 1 (11) 0

n = 18 0 1 ( 6) 2 (11) 1 (6) 6 (33) 8 (44) n = 14 7 (50) 0 0 1 (7) 0 6 (43) n = 17 0 0 0 5 (29) 4 (24) 7 (41) n=9 2 (22) 0 1 (11) 4 (44) 0 2 (22)

Data are given as n (%) unless otherwise stated. CR: complete response; CRu: uncertified CR; ITT: intent-to-treat; PD: progressive disease; PR: partial response; SD: stable disease. †All but one patient considered not evaluable for overall response received ≤1 cycle of treatment and did not have assessments for response. Of these, on the mogamulizumab arm, reasons for treament discontinuation from mogamulizumab were; adverse event (7), PD (6), death (2), withdrawal of consent (1), other (1); On the IC arm, PD (4), adverse event (2) withdrawal of consent (1). All were counted as non-responders for ORR in the ITT analysis.The patient on the IC arm who completed >1 treatment cycle, met eligibility criteria with disease in blood on local flow and not in lymph nodes according to the investigator but showed lymph node and no blood involvement on Independent Review and so was considered not evaluable for response by investigator assessment. One subject in crossover received 7 infusions of mogamulizumab and was discontinued from treatment due to an adverse event. Although this patient had a CR in blood and CR in skin, CT scan was not performed and so was not evaluable for overall response (See patient 19 in Figure 4). *If there was no post-baseline tumor assessment for response assessment, or there was no disease in that compartment, the response was designated not assessable.

998

haematologica | 2019; 104(5)

Mogamulizumab vs. investigator's choice in ATL

Discussion In this randomized phase II trial, the first of ATL outside Japan, mogamulizumab monotherapy demonstrated responses and predictable safety in patients with relapsed/refractory ATL, whereas the comparator arm (investigator’s choice of chemotherapy) showed almost no activity. cORR was higher in those randomized to mogamulizumab versus the investigator’s choice arm by blinded Independent Review for the ITT population: 11% vs. 0%.

This rate of response was less than that seen in the previous phase II study of mogamulizumab monotherapy in 26 evaluable (not in the ITT population) Japanese patients with relapsed CCR4+ ATL, which showed a 50% ORR.22 Several key differences may account for the discrepancy in activity. The Japanese study only included relapsed, not refractory patients, and confirmation of response (although not required) was evaluated after 4 weeks (compared to 8 weeks in our study). In addition, randomized patients in this study had a higher incidence of poor prog-

Figure 2. Kaplan-Meier analysis of progression-free survival during the randomized period.

Figure 3. Forest plot of progression-free survival during randomization adjusted for baseline characteristics. Age group = (i) < versus ≥40 years; age group (ii) = ≤65 versus ≥65 years; baseline Eastern Cooperative Oncology Group (ECOG): 0/1 versus 2; bone marrow in current sites: yes versus no; ATL subtype at consent: acute versus chronic versus lymphoma; best response to last ATL therapy: CR+PR versus SD+PD+unknown. ATL: adult T-cell leukemia/lymphoma; CI: confidence interval; CR: complete response; HR: hazard ratio; IC: investigator choice; PFS: progression-free survival; PD: progressive disease; PR: partial response; SD: stable disease.

haematologica | 2019; 104(5)

999

Adrienne A. Phillips et al.

Table 3. Most common* treatment-related adverse events.

Adverse event Non-hematologic Infusion-related reaction Drug eruption Pyrexia Nausea Headache ALT increased Diarrhea Fatigue Constipation AST increased Vomiting Weight decreased Decreased appetite Mucosal inflammation Tachycardia Asthenia Dyspnea Infections† Hematologic Neutropenia Thrombocytopenia Anemia Leukopenia

During randomization Mogamulizumab Investigator’s choice (n = 47) (n = 24) All grades Grade ≥3 All grades Grade ≥3

After crossover Mogamulizumab (n = 18) All grades Grade ≥3

22 (47) 9 (19) 3 (6) 2 (4) 2 (4) 2 (4) 0 2 (4) 0 2 (4) 0 1 (2) 2 (4) 0 0 0 0 7 (15)

4 (9) 0 0 0 0 1 (2) 0 0 0 1 (2) 0 0 0 0 0 0 0 5 (11)

0 0 3 (13) 4 (17) 3 (13) 2 (8) 4 (17) 2 (8) 3 (13) 2 (8) 3 (13) 2 (8) 1 (4) 2 (8) 0 0 0 3 (13)

0 0 1 (4) 1 (4) 1 (4) 1 (4) 0 0 0 1 (4) 0 0 0 0 0 0 0 0

8 (44) 4 (22) 1 (6) 0 0 0 2 (11) 2 (11) 1 (6) 0 0 0 2 (11) 0 2 (11) 2 (11) 2 (11) 2 (11)

1 (6) 1 (6) 0 0 0 0 0 1 (6) 0 0 0 0 0 0 0 0 1 (6) 0

2 (4) 6 (13) 5 (11) 3 (6)

1 (2) 4 (9) 1 (2) 0

6 (25) 5 (21) 1 (4) 0

0 4 (17) 0 0

3 (17) 1(6) 1 (6) 0

2 (11) 1 (6) 1 (6) 0

Data are given as n (%) unless otherwise stated. ALT: alanine aminotransferase; AST: aspartate aminotransferase. *Most common all grade adverse events that occurred in ≥5% of patients in either randomized group or ≥2 patients during crossover. †Incidence reported is for infections overall. Specific infections reported by ≤2 patients were: lower respiratory infection, oral candidiasis, cellulitis and neutropenic sepsis for investigator’s choice regimens; lower respiratory infection, oral candidiasis, pneumonia, breast abscess, candidiasis, viral conjunctivitis, Escherichia urinary tract infection, oropharyngeal candidiasis, pneumocystis jiroveci pneumonia and urosepsis for mogamulizumab in randomized period; lower respiratory infection and upper respiratory infection for mogamulizumab in crossover period.

nostic factors at baseline, including older age, higher ECOG performance status, and greater bone marrow involvement than in the Japanese study. The aggressiveness of the disease in the patients on this study was reflected in the high number of subjects (65%) that completed ≤1 treatment cycle. Lastly, our study enrolled a more ethnically diverse patient population, and differences in disease biology, clinical presentation, and response to treatment have been suggested in Japanese patients compared to those in other regions,15,26 although this has not been studied prospectively.

The Shimoyama classification of ATL12 and recommended response criteria for ATL25 have been useful for the standardization and comparison of outcomes of Japanese patients with those in the other countries.

However, a number of pitfalls in these schema have been reported27 and these were observed in this trial. Complex presentations with leukemic, lymphomatous, and skin compartments may complicate assessment, as disease control in one or more compartments, even alongside an increase in another compartment, may result in significant clinical improvement in a patient, although the patient technically meets progression criteria as the overall response.

1000

Protocol-defined progression was based on Tsukasaki criteria for composite scoring using data from all disease compartments (blood, skin, lymph nodes, extranodal masses, liver/spleen, and bone marrow), which are often involved to various degrees in a single patient and may have led to the clinical benefit being underestimated. In an aggressive, rapidly progressive disease such as ATL, clinical benefit may be apparent to the treating physician, and remains important even if a later blinded composite response is PD. Notable responses were observed in this study in peripheral blood (54%, all with CR) and skin (44%), and several subjects described were considered to be benefiting from mogamulizumab besides those represented in the cORR data.

There is no approved ATL treatment outside Japan.

Investigator’s choice of chemotherapy regimen in the trial was between GemOx, pralatrexate, and DHAP, with almost all (87%) allocated to GemOx. These regimens were most commonly used for the treatment of relapsed/refractory ATL in the countries where this study was conducted, although there is virtually no published evidence of clinical efficacy. Other studies or series have indicated little evidence of clinical efficacy in relapsed/refractory ATL with regimens such as cladribhaematologica | 2019; 104(5)

Mogamulizumab vs. investigator's choice in ATL

ine,28 irinotecan,29 or bortezomib.30 A study of 26 patients in Japan published subsequent to enrollment of our trial reported an ORR of 42% with lenalidomide,31 and a report on the use of alemtuzumab in relapsed/ refractory patients demonstrated an ORR of 50% in a lower-risk population.32 A recent phase I study examining the combination of romidepsin with pralatrexate included six relapsed/refractory ATL patients and reported a preliminary response rate of 50%.33 Landmark therapeutic trials leading to US Food and Drug Administration approval in the US of belinostat, pralatrexate, romidepsin, and brentuximab vedotin

in relapsed/refractory PTCL included solitary or no patients with ATL,34-37 precluding extrapolation of results to ATL. The safety profile for mogamulizumab was manageable and consistent with previous reports, with infusion-related reactions and drug eruptions as the most common AEs.22,38,39 The rate of discontinuation for drug eruption was similar to the recently reported phase III study of mogamulizumab in CTCL.39 As in that study, use of systemic steroids was not permitted by protocol and most rashes were successfully managed with topical steroids.

Figure 4. Duration on study for patients receiving â&#x2030;Ľ2 cycles of mogamulizumab. (Top) Initially randomized to mogamulizumab. (Bottom) Initially randomized to investigatorâ&#x20AC;&#x2122;s choice of chemotherapy and then crossed over to mogamulizumab. Response as assessed by investigator. *Indicates confirmed response. Patient 16 had salvage chemotherapy prior to transplant but no date was provided.

haematologica | 2019; 104(5)

1001

Adrienne A. Phillips et al.

Treatment-related AEs grade ≥3 were infrequent.

Comparison of mogamulizumab and investigator’s choice arms revealed little difference in the overall incidence of treatment-related AEs of any grade or grade ≥3 despite the fact that the duration of treatment exposure was approximately 3-fold longer in the mogamulizumab arm.

In summary, we have conducted the first, prospective, randomized therapeutic trial of ATL outside Japan.

Because of the rarity of the disease, the study required a major collaborative effort across multiple international centers to achieve the target accrual within 3 years.

Despite small numbers and unbalanced randomization, the trial demonstrated the efficacy of mogamulizumab (e.g. PFS, ORR, responses observed after crossover, durability of responses) in comparison to other frequently used agents. The safety profile in this ethnically diverse patient population with a high unmet medical need was manageable, while minimal benefit was demonstrated with commonly used chemotherapy agents. Given the rates of best response but overall short duration and PFS for many patients with this aggressive disease, future studies should explore combinations with other agents. For example, lenalidomide has demonstrated single agent activity and

References 1. Uchiyama T, Yodoi J, Sagawa K, et al. Adult T-cell leukemia: clinical and hematologic features of 16 cases. Blood. 1977; 50(30):481-492.

2. Poiesz BJ, Ruscetti FW, Gazdar AF, et al.

Detection and isolation of type C retrovirus particles from fresh and cultured lymphocytes of a patient with cutaneous T-cell lymphoma. Proc Natl Acad Sci U S A. 1980; 77(12):7415-7419.

3. Hinuma Y, Nagata K, Hanaoka M, et al.

Adult T-cell leukemia: Antigen in an ATL cell line and detection of antibodies to the antigen in human sera. Proc Natl Acad Sci U S A. 1981;78(10):6476-6480.

4. Yoshida M, Miyoshi I, Hinuma Y. Isolation and characterization of retrovirus from cell lines of human adult T-cell leukemia and its implication in the disease. Proc Natl Acad Sci U S A. 1982;79(6):2031-2035.

5. Vose J, Armitage J, Weisenburger D; International T-Cell Lymphoma Project: International peripheral T-cell and natural killer/T-cell lymphoma study. Pathology findings and clinical outcomes. J Clin Oncol. 2008;26(25):4124-4130.

6. Iwanaga M, Watanabe T, Yamaguchi K. Adult T-cell leukemia: a review of epidemiological evidence. Front Microbiol.

2012;3:322.

7. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Human immunodeficiciency viruses and human Tcell lymphotropic viruses. IARC Monogr Eval Carcinog Risks Hum. 1996;67:1-424.

8. Gessain A, Cassar O. Epidemiological aspects and world distribution on HTLV-1 infection. Front Microbiol. 2012;3:388.

9. Proietti FA, Carneiro-Proietti ABF, CatalanSoares BC, et al. Global epidemiology of HTLV-I infection and associated diseases.

Oncogene. 2005;24(39):6058-6068.

10. Yoshida N, Chihara D. Incidence of adult T-cell leukemia/lymphoma in nonendemic

1002

11.

12.

13.

14.

15.

16.

17.

18.

may potentiate ADCC in other non-Hodgkin lymphoma subtypes.31,40 Earlier lines of therapy, prior to the relapsed/refractory setting, where there is greater possibility of impacting the disease course should also be investigated.

Funding This study was supported by Kyowa Kirin (Princeton, NJ, USA). We thank the investigators and co-ordinators at each of the participating study centers, the patients who volunteered to participate in this study and their families, and the sponsor staff involved in data collection and analyses. Peter Todd (Tajut Ltd., Kaiapoi, New Zealand) provided editorial assistance in the development of the manuscript, for which he received financial compensation from Kyowa Kirin (Princeton, NJ, USA). The authors had complete access to all data and maintained control over the manuscript, including final wording and conclusions. AP was, in part, supported by The Harold Amos Medical Faculty Development Program (Robert Wood Johnson Foundation) for this research and career development. Researchers at the Memorial-Sloan Kettering Cancer Center were funded, in part, through the NIH/NCI Cancer Center Support Grant P30 CA008748.

areas. Curr Treat Options Oncol. 2015; 16(2):7.

Chihara D, Ito H, Matsuda T, et al.

Differences in incidence and trends of haematological malignancies in Japan and the United States. Br J Haematol. 2014; 164(4):536-545.

Shimoyama M. Diagnostic criteria and classification of clinical subtypes of adult T-cell leukaemia-lymphoma. A report from the Lymphoma Study Group (1984-87). Br J Haematol. 1991;79(3):428-437.

Tsukasaki K, Utsunomiya A, Fukuda H, et al. VCAP-AMP-VECP compared with biweekly CHOP for adult T-cell leukemialymphoma: Japan Clinical Oncology Group Study JCOG9801. J Clin Oncol. 2007; 25(34):5458-5464.

Takasaki Y, Iwanaga M, Imaizumi Y, et al.

Long-term study of indolent adult T-cell leukemia-lymphoma.

Blood.

2010; 115(22):4337-4343 Phillips AA, Shapira I, Willims RD, et al. A critical analysis of prognostic factors in North American patients with human Tcell lymphotropic virus type-1-associated adult T-cell leukemia/lymphoma: a multicenter clinicopathologic experience and new prognostic score. Cancer. 2010; 116(14):3438-3446.

Hishizawa M, Kanda J, Utsunomiya A, et al.

Transplantation of allogeneic hematopoietic stem cells for adult T-cell leukemia: a nationwide retrospective study.

Blood. 2010;116(8):1369-1376.

Ishida T, Hishizawa M, Kato K, et al.

Allogeneic hematopoietic stem cell transplantation for adult T-cell leukemia-lymphoma with special emphasis on preconditioning regimen: a nationwide retrospective study. Blood. 2012;120(8):1734-1741.

Fujiwara H, Fuji S, Wake A, et al; ATL Working Group of the Japan Society for Hematopoietic Cell Transplantation.

Dismal outcome of allogeneic hematopoietic stem cell transplantation for relapsed

19.

20.

21.

22.

23.

24.

25.

adult T-cell leukemia/lymphoma, a Japanese nation-wide study. Bone Marrow Transplant. 2017;52(3):484-488.

Yoshie O, Fujisawa R, Nakayama T, et al.

Frequent expression of CCR4 in adult T-cell leukemia and human T-cell leukemia virus type-1 transformed T-cells. Blood. 2002; 99(5):1505-1511.

Ishida T, Utsunomiya A, Iida S, et al.

Clinical significance of CCR4 expression in adult T-cell leukemia/lymphoma: Its close association with skin involvement and unfavorable outcome. Clin Cancer Res. 2003;9(10):3625-3634.

Ishii T, Ishida T, Utsunomiya A, et al.

Defucosylated humanized anti-CCR4 monoclonal antibody KW-0761 as a novel immunotherapeutic agent for adult T-cell leukemia/lymphoma. Clin Cancer Res. 2010;16(5):1520-1531.

Ishida T, Joh T, Uike N, et al. Defucosylated anti-CCR4 monoclonal antibody (KW0761) for relapsed adult T-cell leukemialymphoma: a multicenter phase II study. J Clin Oncol. 2012;30(8):837-842.

Ishida T, Jo T, Takemoto S, et al. Doseintensified chemotherapy alone or in combination with mogamulizumab in newly diagnosed aggressive adult T-cell leukaemia-lymphoma: a randomized phase II study. Br J Haematol. 2015;169(5):672682.

O’Connor OA, Horwitz S, Hamlin P, et al.

Phase II-I-II Study of two different doses and schedules of pralatrexate, a high-affinity substrate for the reduced folate carrier, in patients with relapsed or refractory lymphoma reveals marked activity in T-cell malignancies. J Clin Oncol. 2009; 27(26):4357-4364.

Tsukasaki K, Hermine O, Bazarbachi A, et al. Definition, prognostic factors, treatment, and response criteria of adult T-cell leukemia-lymphoma: a proposal from an international consensus meeting. J Clin Oncol. 2009;27(3):453-459.

haematologica | 2019; 104(5)

Mogamulizumab vs. investigator's choice in ATL

26. Hisada M, Stuver SO, Okayama A, Li HC, et al. Persistent paradox of natural history of human T lymphotropic virus type I: parallel analyses of Japanese and Jamaican carriers. J Infect Dis. 2004;190(9):1605-1609.

27. Bazarbachi A, Suarez F, Fields P, et al. How I treat adult T-cell leukemia/lymphoma.

Blood. 2011;118(7):1736-1745.

28. Tobinai K, Uike N, Saburi Y, et al; Cladribine/ATL Study Group, Japan. Phase II study of cladribine (2-chlorodeoxyadenosine) in relapsed or refractory adult T-cell leukemia-lymphoma. Int J Hematol.

2003;77(5):512-517.

29. Tsuda H, Takatsuki K, Ohno R, et al.

Treatment of adult T-cell leukaemia-lymphoma with irinotecan hydrochloride (CPT-11). CPT-11 Study Group on Hematological Malignancy. Br J Cancer.

1994;70(4):771-774.

30. Ishitsuka K, Utsunomiya A, Katsuya H, et al. A phase II study of bortezomib in patients with relapsed or refractory aggressive adult T-cell leukemia/lymphoma.

Cancer Sci. 2015;106(9):1219-1223.

31. Ishida T, Fujiwara H, Nosako K, et al.

haematologica | 2019; 104(5)

32.

33.

34.

35.

36.

Multicenter phase II study of lenalidomide in relapsed or recurrent adult T-cell leukemia/lymphoma: ATLL-002. J Clin Oncol. 2016;34(34):4086-4093.

Sharma K, Janik JE, Oâ&#x20AC;&#x2122;Mahoney D, et al.

Phase II study of alemtuzumab (CAMPATH-1) in patients with HTLV-1-associated adult T-cell leukemia/lymphoma. Clin Cancer Res. 2017;23(1):35-42.

Amengual JE, Lichtenstein R, Lue J, et al. A phase 1 study of romidepsin and pralatrexate reveals marked activity in relapsed and refractory T-cell lymphoma. Blood. 2018; 131(4):397-407.

O'Connor OA, Pro B, Pinter-Brown L, et al.

Pralatrexate in patients with relapsed or refractory peripheral T-cell lymphoma: results from the pivotal PROPEL study. J Clin Oncol. 2011;29(9):1182-1189.

Pro B, Advani R, Brice P, et al. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large-cell lymphoma: results of a phase II study. J Clin Oncol. 2012;30(18):2190-2196, O'Connor OA, Horwitz S, Masszi T, et al.

Belinostat in patients with relapsed or

37.

38.

39.

40.

refractory peripheral T-cell lymphoma: results of the pivotal phase II BELIEF (CLN19) Study. J Clin Oncol. 2015;33(23):24922499.

Foss F, Horwitz S, Pro B, et al. Romidepsin for the treatment of relapsed/refractory peripheral T cell lymphoma: prolonged stable disease provides clinical benefits for patients in the pivotal trial. J Hematol Oncol. 2017;10(1):154.

Sekine M, Kubuki Y, Kameda T, et al.

Effects of mogamulizumab in adult T-cell leukemia/lymphoma in clinical practice.

Eur J Haematol. 2017;98(5):501-507.

Kim YH, Bagot M, Pinter-Brown L, et al.

Mogamulizumab versus vorinostat in previously treated cutaneous T-cell lymphoma (MAVORIC): an international, open-label, randomised, controlled phase 3 trial. Lancet Oncol. 2018;19(9):1192-1204.

Wang M, Fowler N, Wagner-Bartak N, et al.

Oral lenalidomide with rituximab in relapsed or refractory diffuse large cell, follicular and transformed lymphoma: a phase II clinical trial. Leukemia. 2013;27(9);19021909.

1003

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):1004-1015

Chronic Lymphocytic Leukemia

The involvement of microRNA in the pathogenesis of Richter syndrome

Katrien Van Roosbroeck,1,2* Recep Bayraktar,1* Steliana Calin,3 Johannes Bloehdorn,4 Mihnea Paul Dragomir,1 Keishi Okubo,1 Maria Teresa Sabrina Bertilaccio,1 Simonetta Zupo,5 M. James You,3 Gianluca Gaidano,6 Davide Rossi,7 Shih-Shih Chen,8 Nicholas Chiorazzi,8 Philip A. Thompson,9 Alessandra Ferrajoli,9 Francesco Bertoni,7 Stephan Stilgenbauer,4 Michael J. Keating9 and George A. Calin1,9,10

Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Present address – Department of Investigational Cancer Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 3 Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 4Department of Internal Medicine III, University Hospital Ulm, Germany; 5 Molecular Diagnostic Laboratory, Pathology Department, IRCCS, Ospedale Policlinico San Martino, Genoa, Italy; 6Division of Hematology, Department of Translational Medicine, University of Eastern Piedmont, Novara, Italy; 7Università della Svizzera italiana, Institute of Oncology Research, Bellinzona, Switzerland; 8The Feinstein Institute for Medical Research, Northwell Health, Manhasset, NY, USA; 9Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA and 10Center for RNA Interference and Non-Coding RNAs, The University of Texas MD Anderson Cancer Center, Houston, TX, USA 1

*KVR and RB contibuted equally to this work.

ABSTRACT

R Correspondence: GEORGE A. CALIN gcalin@mdanderson.org Received: August 8, 2018.

Accepted: November 8, 2018.

Pre-published: November 8, 2018. doi:10.3324/haematol.2018.203828 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1004 ©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.

1004

ichter syndrome is the name given to the transformation of the most frequent type of leukemia, chronic lymphocytic leukemia, into an aggressive lymphoma. Patients with Richter syndrome have limited response to therapies and dismal survival. The underlying mechanisms of transformation are insufficiently understood and there is a major lack of knowledge regarding the roles of microRNA that have already proven to be causative for most cases of chronic lymphocytic leukemia. Here, by using four types of genomic platforms and independent sets of patients from three institutions, we identified microRNA involved in the transformation of chronic lymphocytic leukemia to Richter syndrome. The expression signature is composed of miR-21, miR-150, miR-146b and miR-181b, with confirmed targets significantly enriched in pathways involved in cancer, immunity and inflammation. In addition, we demonstrated that genomic alterations may account for microRNA deregulation in a subset of cases of Richter syndrome.

Furthermore, network analysis showed that Richter transformation leads to a complete rearrangement, resulting in a highly connected microRNA network. Functionally, ectopic overexpression of miR-21 increased proliferation of malignant B cells in multiple assays, while miR-150 and miR26a were downregulated in a chronic lymphocytic leukemia xenogeneic mouse transplantation model. Together, our results suggest that Richter transformation is associated with significant expression and genomic loci alterations of microRNA involved in both malignancy and immunity.

Introduction The most frequent type of adult leukemia, chronic lymphocytic leukemia (CLL), is a disease in which alterations of small non-coding RNA named microRNA (miRNA, miR) play a fundamental role: the miR-15a/16-1 cluster at the 13q deletion hotspot, which targets the oncogenic anti-apoptotic proteins BCL2 and MCL1, is deleted or downregulated in most and germline-mutated in some patients.1-3 Although these discoveries were made more than a decade ago haematologica | 2019; 104(5)

miRNA in Richter syndrome

as a first link between non-coding RNA alterations and human diseases,4,5 the mechanistic involvement of miRNA in the CLL patients with the worst prognosis, those whose disease transforms to Richter syndrome (RS), has not been reported to date. RS occurs in up to 8% of untreated CLL patients6 and in 5-16% of patients treated with targeted therapies, such as ibrutinib or venetoclax for relapsed CLL.7,8 Abnormalities of regulators of tumor suppression (TP53), cell proliferation (NOTCH1, MYC) and cell cycle (CDKN2A), have been reported in RS,9 but biomarkers to predict the occurrence of RS are lacking at present. RS is characterized by rapid progression and outcomes of patients treated with a variety of moderate or high-intensity chemoimmunotherapy regimens are uniformly dismal, with a median survival of less than 1 year,10-13 particularly for patients with clonally-related or TP53-mutated disease.14 Novel, molecularly targeted approaches are urgently required, but this is hampered by the limited understanding of the molecular pathogenesis of RS. The paucity of molecular studies is mainly due to the scarceness of biopsy materials. Furthermore, the availability of non-invasive methods of diagnosis (such as the use of 2deoxy-2-[(18)F] fluoroglucose/positron emission tomography,15 reduces the need for follow-up biopsies, which further limits the availability of material for research.

Therefore, there is a strong need to develop RS biomarkers and molecularly targeted therapies that could facilitate early and accurate diagnosis, as well as assist current treatment strategies. In the present study, we investigated the expression and potential roles of miRNA in the transformation from CLL to RS, as these miRNA could be therapeutically targeted.

Ulm University cohort We used peripheral blood from 58 fludarabine-resistant patients. Samples were taken at enrollment before treatment.

Eight of these 58 patients subsequently developed RS. These patients were described previously.16

The Bellinzona Institute of Oncology Research cohort This cohort of patients comprised 737 cases of mature lymphoid tumors including CLL, and were described previously.17 The study was approved by the institutional review boards and ethical committees of all the participating institutions.

Xenogeneic mouse transplantation We used a previously described mouse model of CLL for in vivo experiments.18-20

Firefly microRNA profiling assay We performed expression analysis of 40 human and viral miRNA known to be involved in the progression of CLL, associated with poor prognosis CLL, highly expressed in CLL as determined by a previously performed RNA-sequencing study,21 located in genomic regions reported to be deregulated in RS,17 or frequently reported in literature to be associated with CLL (Online Supplementary Table S3). We used a Firefly custom multiplex circulating miRNA assay (Abcam, Cambridge, MA, USA) for the extended cohort of CLL/RS samples collected at the UTMDACC, as described in the Online Supplementary Methods.

Polymerase chain reaction analysis and microRNA gene expression profiling Quantitative reverse transcription polymerase chain reaction (qRT-PCR) and miRNA gene expression profiling for miRNA profiling confirmation are described in detail in the Online Supplementary Methods.

Methods Genome-wide DNA profiles Patients’ samples The University of Texas MD Anderson Cancer Center (UTMDACC) cohort The “paired” set: 14 bone marrow samples from seven patients with RS were collected at the UTMDACC. For each patient, samples from the time of CLL diagnosis (group 1a) and Richter transformation (group 1b) were available. In addition, we collected 14 bone marrow samples from seven age-, sexand sample time-matched CLL control patients who did not develop RS over the course of follow-up at the UTMDACC. For each patient, a sample at the time of CLL diagnosis (group 2a) and at a time corresponding to the time of RS diagnosis of group 1 (group 2b) were available. Online Supplementary Table S1 shows that age at diagnosis, gender and time to transformation were not significantly different between patients of this paired RS/CLL cohort.

The “extended” set : we also extended our initial paired RS/CLL cohort to include samples from 27 patients with RS [25 samples at CLL diagnosis (group 1a) and 9 samples at the time of Richter transformation (group 1b)] and 23 control CLL patients [17 samples at CLL diagnosis (group 2a) and 14 samples at a time corresponding to the time of Richter transformation in the RS group (group 2b)]. All samples used were formalin-fixed paraffin-embedded (FFPE) bone marrow cores, except for one lymph node sample in group 1b. A schematic representation of the extended cohort is shown in Figure 1A,B, while the patients’ characteristics are presented in Table 1 and detailed in Online Supplementary Table S2. haematologica | 2019; 104(5)

Genome-wide DNA profiles were obtained from high-molecular-weight genomic DNA using the Genome-Wide Human SNP Array 6.0 (Affymetrix, Santa Clara, CA, USA), as previously reported (GSE50252).17

Network analyses The miRNA networks were generated as previously described;22 the method is detailed in the Online Supplementary Methods.

Results A microRNA signature is involved in the process of Richter transformation As illustrated in the workflow in Figure 1A, we first designed the UTMDACC cohort consisting of patients with RS and age-, sex- and sample time-matched CLL “controls”. For all of these cases FFPE bone marrow samples, taken at the time of CLL diagnosis and at the time of RS diagnosis, or a time corresponding to the RS diagnosis in the case of the matched CLL controls, were available and analyzed (Figure 1B and Online Supplementary Table S1). We decided to use the Firefly custom multiplex miRNA assay due to the best data generation/cost ratio, and well annotated and selected miRNA (see Methods). We identified nine miRNA potentially involved in the RS transformation process, 1005

K. Van Roosbroeck et al.

i.e. significantly differentially expressed (SDE, at P values <0.05) in bone marrow samples of patients with CLL at the time of Richter transformation when compared to their expression in samples obtained at the time of CLL diagnosis: SDE in group 1b versus group 1a, but not in group 2b versus group 2a (to exclude the time of evolution as a variable) (Figures 1C and 2A,B, Online Supplementary Figure S1 and Online Supplementary Table S2). We excluded miR-34a and miR-17 from the list of Richter-specific SDE miRNA, as these were also significantly upregulated in the control group of patients with CLL who did not develop RS over the course of followup, which was similar to the time to Richter transformation in the RS group. We then extended our cohort to include non-matched samples (Table 1 and Online Supplementary Table S2) and analyzed these with the Firefly custom multiplex miRNA assay as well. We considered the following groups (Figure 1B): group 1a - samples at the time of CLL diagnosis from patients with

CLL who later developed RS (n=25); group 1b - samples from patients with CLL at the time of Richter transformation/RS diagnosis (n=9); group 2a - control samples from patients with CLL, at the time of diagnosis, who did not develop RS over a follow-up period equal to or longer than the time between CLL and RS diagnosis of groups 1a and 1b, (n=16); and group 2b - at a time corresponding to RS transformation in group 1b (n=13).

This analysis revealed 15 miRNA that showed significantly different expression between samples at the time of CLL diagnosis (group 1a) and at the time of RS diagnosis (group 1b), of which seven are common with the paired analysis (Figure 2A-C and Online Supplementary Figure S2).

Next, we validated these results with a different assay and performed qRT-PCR expression analysis on 25 miRNA in the same extended CLL/RS dataset. This analysis confirmed three differentially expressed miRNA: miR-21 and miR-146b were upregulated and

A

B

C

Figure 1. Workflow of the microRNA screening process and composition of the University of Texas MD Anderson Cancer Center cohort of patients with Richter syndrome/chronic lymphocytic leukemia. (A) Three-step workflow of the miRNA screening process, consisting of miRNA profiling, array comparative genomic hybridization (aCGH) analysis and functional analysis. (B) Schematic representation of the extended University of Texas MD Anderson Cancer Center (MDACC) cohort of patients. This cohort consists of patients with Richter syndrome (RS) (group 1), as well as age-, sex- and sample time-matched controls with chronic lymphocytic leukemia (CLL) (group 2). For both groups, samples at CLL diagnosis (group 1a) and at the time of RS diagnosis (group 1b) or a time corresponding to RS diagnosis time in the case of the matched CLL controls (group 2b) are available. Time “t1” between CLL diagnosis and RS diagnosis time is similar for both groups. (C) miR-21, miR-146a, miR-150 and miR-181b are members of the four-miRNA “restricted signature”, while the eight-miRNA “enlarged signature” contains these four miRNA and additionally miR-24, miR-26a, miR-181a and miR-146a. miRNA highlighted in red are upregulated in RS, while miRNA highlighted in green are downregulated. For additional Information, see Online Supplementary Figure S1. IOR: Institute of Oncology Research; DLBCL: diffuse large B-cell lymphoma.

1006

haematologica | 2019; 104(5)

miRNA in Richter syndrome

miR-150 was downregulated at the time of Richter transformation (Figures 2A and 3A). Finally, to further confirm our findings, we analyzed the levels of the differentially expressed human miRNA in an independent set of 58 patients with CLL/RS from Ulm University16 for whom microarray data were available (Online Supplementary Table S4). We confirmed that three miRNA, miR-21, miR-146b and miR-181b, were all upregulated in RS when compared to CLL (Figures 2A and 3B,C). Taken together, miR-21, a well-known oncogene,23 was found by all four analyses to be significantly more highly expressed at the time of Richter transformation than at the time of CLL diagnosis, strongly suggesting a role of this miRNA in the process of transformation to RS. In addition, three miRNA were found to be differentially expressed by three analyses: miR-146b and miR-181b were upregulated and miR-150 was downregulated at the time of Richter transformation. We will refer to this signature of four miRNA as the “restricted signature”. In addition, four miRNA were significantly different in two

analyses: miR-26a was significantly downregulated, while miR-24, miR-146a and miR-181a were significantly upregulated at the time of RS diagnosis. This signature of eight miRNA was designated as the “enlarged signature” (Figures 1C and 2A). Of note, miRNA from the same families – miR-181a and miR-181b, and miR-146a and miR-146b - were present in both signatures. Three viral miRNA, the Epstein-Barr virus miRNA BART4 and BART16, and the Kaposi sarcoma herpes virus miRNA kshv-miR-K12-4 were differentially expressed in the Firefly analysis of all samples, but could not be evaluated in a cohort from an independent institution, as the array used to investigate the Ulm University cohort did not contain probes for non-human miRNA.

A

C

Genomic alterations may account for microRNA deregulation in a subset of cases of Richter syndrome miRNA are often located at fragile sites in human and mouse genomes.24,25 To investigate whether the differen-

B

Figure 2. microRNA profiling with the Firefly custom multiplex microRNA assay. (A) Summary of the significantly expressed miRNA in the profiling step of the project.

(B) miR-21 and miR-181b are significantly upregulated, and miR-150 is significantly downregulated at the time of Richter transformation (RT) when compared to at chronic lymphocytic leukemia (CLL) diagnosis (Dx) for the paired Richter syndrome (RS) samples, but not for the paired control CLL samples. (C) miR-146b, miR-21 and miR-181b are significantly upregulated, and miR-150 is significantly downregulated at RT time when compared to at CLL diagnosis in the extended set of RS samples, but not in the extended set of control CLL samples. *P<0.05; **P<0.01; ***P<0.001. For additional Information, see Online Supplementary Figure S2.

haematologica | 2019; 104(5)

1007

K. Van Roosbroeck et al.

tially expressed human miRNA are located in regions genetically altered in RS, we first analyzed the eight genomic loci of the enlarged signature-related miRNA in 15 paired CLL/RS cases for which samples at CLL diagnosis and at RS diagnosis were available.17 We found that, in RS samples, three cases (20%) had a gain of the miR181a/b clusters on either chromosome 1q32 or 9q33, three cases (20%) had a loss of miR-26a-1 and a-2 loci on either chromosome 3p22 or 12q14, two cases (13.5%) had a gain of miR-21 locus at 17q23, and one case (6.67%) had a loss of miR-150 locus at 19q13 (Figure 4A).

We then considered a larger series of cases of RS (n=59) and CLL phases (n=28). In accordance with the abovepresented expression data, we observed a similar pattern of miRNA-specific genetic aberrations: RS patients more frequently presented gains affecting miR-21, miR-146a and miR-181a/b genomic loci (Online Supplementary Figure S4A). Finally, we took advantage of a series of 737 genomic profiles obtained in mature lymphoid tumors to compare the frequency of DNA copy number aberrations with that observed in RS.17,26-29 The same miRNA loci

were sites of gains (miR-181a/b, 6% and 8% at 1q or 9q, respectively; miR-21, 4%; miR-146a, 3%; miR-146b, 0.2%) or losses (miR-150, 6%; miR-26a, 3% and 0.1% at 3p or 12q, respectively) in mature lymphoid tumors (Figure 4B and Online Supplementary Figure S4B), but at a significantly lower frequency than that observed in RS samples (P=0.017 and c2=5.669, see Online Supplementary Figure S4C). Altogether, these data suggest that, at least in some cases, Richter-related miRNA are deregulated due to DNA copy number changes, which occur during the transformation process, and are enriched with respect to non-Richter B-cell malignancies. The mechanism(s) for the miRNA expression deregulation for the majority of Richter cases has still to be identified.

The microRNA network is reprogrammed during Richter transformation Although it is known that changes in miRNA expression are involved in the initiation and development of CLL,30 we further investigated whether analysis of the miRNA interactor network could add more information

Table 1. Characteristics of patients with Richter syndrome and chronic lymphocytic leukemia in the cohort from the University of Texas MD Anderson Cancer Center.

Characteristic Age at diagnosis, years Median Range Sex Male Female Rai stage 0 1 2 3 4 NA FISH del13q NL cyto/FISH trisomy 12 del11q del17p NA ZAP70 Positive Negative NA CD38 Positive (>=20%) Negative (<20%) NA B2 microglobulin Positive (â&#x2030;Ľ2 mg/mL) Negative (<2 mg/mL) NA B2 microglobulin level Median Range

Richter (n=27) N. % 54 31-78

CLL (n=23) N. % 59 43-70

Characteristic 0.1973

17 10

63.0 37.0

14 9

60.9 39.1

>0.9999

4 11 5 4 1 2

14.8 40.7 18.5 14.8 3.7 7.4

7 13 1 0 2 0

30.5 56.5 4.3 0.0 8.7 0.0

0.0947

8 4 8 6 5 4

29.6 14.8 29.6 22.2 18.5 14.8

16 0 5 3 1 2

69.6 0.0 21.8 13.1 4.35 8.7

0.0328

6 0 21

22.2 0.0 77.8

8 10 5

34.8 43.5 21.8

0.0239

17 3 7

63.0 11.1 25.9

4 15 4

17.4 65.3 17.4

<0.0001

25 1 1

92.6 3.7 3.7

18 2 3

78.3 8.7 13

0.5718

3.6 1.6-10

2.4 1-4.1

continued from the previous coloum

P-value

0.0051

Richter (n=27) N. %

IGHV Unmutated 18 Mutated 5 NA 4 Time to transformation (months) Median 69 Range 8-213 Survival Dead 22 Alive at last follow-up 5 Survival time (months) Median 108 Range 20-245 LDH at diagnosis Normal (â&#x2030;¤18 UI/L) 12 Elevated (>618 UI/L) 15 NA 0 Median 647 Range 303-1587 Time to first treatment (months) Median 10 Range 0-156 Number of treatments (including SCT) 0 1 1 6 2 5 3 2 4 6 5 4 6 3 NA 0

66.7 18.5 14.8

CLL (n=23) N. % 7 7 9

P-value

30.45 30.45 39.1

0.1459

4.35 95.7

<0.0001

NA NA 81.5 18.5

1 22 96 15-234

44.4 55.6 0.0

21 2 0 513 357-1175

0.3821

91.3 8.7 0.0

0.0079

39 0-156 3.7 22.2 18.5 7.4 22.2 14.8 11.1 0

4 13 1 1 1 0 0 3

0.0007

0.0083

17.4 56.55 4.35 4.35 4.35 0.0 0.0 13.1

0.0083

NA: not available; FISH: fluorescence in situ hybridization; LDH: lactate dehydrogenase; SCT: stem cell transplantation.

continued in the next coloum

1008

haematologica | 2019; 104(5)

miRNA in Richter syndrome

to our scarce understanding of the transformation of CLL to RS. We investigated the miRNA regulatory networks composed of nodes and edges, where the nodes are miRNA genes, and the edges (links) are the molecular interactions (a high statistical correlation between two miRNA nodes in a given set of patients). It is important to note that this method also recognizes miRNA that are not SDE as key elements. The network analysis provides a different perspective on the role of a miRNA than the commonly used expression profiling, and the results do not always overlap. By using the qRT-PCR expression data, we generated a 25-miRNA expression network for each of the four groups of patients of the extended UTMDACC cohort (Figure 5A). We observed that the number of edges increased significantly from 29 in the group 1a network to 55 in the group 1b network compared to the group 2a and group 2b networks, in which only one new extra-edge appeared, with the number of edges increasing from 53 to 54 (P<0.05, c2=3.913) (Figure 5B, left panel). When comparing the behavior of the miRNA nodes in patients who underwent Richter transformation, i.e., groups 1a and 1b, we observed an increase in the connectivity of nodes (P=0.0007), indicating a complete reprogramming of the

miRNA network during the transformation (Figure 5B, middle panel). When we performed the same comparison for the miRNA networks of the “control” CLL patients who did not develop RS, i.e., groups 2a and 2b, we observed no significant change in the connectivity of the nodes (Figure 5B, right panel). The miRNA network of group 1a is a disjointed graph with many isolated nodes which, after Richter transformation (group 1b), becomes a highly-connected graph, with only two isolated miRNA (miR-23b and miR-155) (Figure 5A). The hubs (defined as the nodes with the highest connectivity, i.e., the miRNA best connected in the network) specific for Richter transformation (hubs in group 1b versus group 1a, but not in group 2a versus group 2b) were miR-191, miR-17 and miR-29c, which we named hubspecific (HUS) miRNA. Additionally, these data were confirmed when we generated an independent 40-miRNA network for the four groups of patients by using the Firefly assay expression data (Online Supplementary Figure S5). Therefore, both qRTPCR-based and Firefly-based networks confirmed that Richter transformation leads to a complete rearrangement of the miRNA network, with a significant increase in the number of edges and the addition of new miRNA hubs.

B

A

C

Figure 3. microRNA expression validation by quantitative reverse transcription polymerase chain reaction and gene expression profiling. (A) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) expression analysis shows that miR-21 and miR-146b are significantly upregulated, and miR-150 is significantly downregulated at the time of Richter transformation (RT) when compared to their expression at the time of chronic lymphocytic leukemia (CLL) diagnosis (Dx) in the extended set of Richter syndrome (RS) samples, but not in the extended set of control CLL samples. (B) Gene expression profiling analysis in an independent set of RS/CLL samples from Ulm University shows significant upregulation of miR-21, miR-146b and miR-181b in RS when compared to CLL. (C) Heatmap showing the miRNA signature for CLL with subsequent transformation (“Richter”) and CLL without transformation (“CLL”) in the validation cohort (“Ulm University”) after hierarchical clustering on genes (Pearson correlation, average linkage). *P<0.05; **P<0.01. For additional Information, see Online Supplementary Figure S3.

haematologica | 2019; 104(5)

1009

K. Van Roosbroeck et al.

The significantly differently expressed and hub-specific microRNA signature targets are enriched in pathways involved in cancer, immunity and inflammation To understand the biological significance of the differentially expressed miRNA as well as of the RS-specific miRNA hubs from the above network analyses, we performed ingenuity pathway analysis for the validated targets (Online Supplementary Table S5A,B). Ingenuity pathway analysis of all targets of the four SDE miRNA of the restricted signature revealed the involvement of signaling pathways with a strong enrichment for pathways involved in cancer and also in immune responses (Figure 5C). We performed the same ingenuity pathway analysis on the three HUS Richter-specific miRNA (Figure 5D), and again found a strong enrichment towards cancer and autoimmune disorders/immunity-specific pathways.

Interestingly, seven of the top 20 canonical pathways were common between the SDE miRNA and HUS

A

miRNA (Figure 5C,D). The connection with the p53 signaling pathway is compelling and can be considered as a â&#x20AC;&#x153;positiveâ&#x20AC;? control since this protein has a well-known role in the pathogenesis of RS, as well as aggressive CLL.14

Functional characterization of the miRNA involved in Richter transformation In order to start the functional characterization of the miRNA, we took advantage of a well characterized CLL xenograft model that employs primary CLL cells18-20 (Figure 6A). We checked the expression levels of the eight miRNA in the Richter transformation restricted and enlarged signatures. Since human CLL cells proliferate faster after transplantation into mice18-20,31 similar to what is observed after Richter transformation in humans, we hypothesized that this model partially resembles the functional phenotype observed in Richter transformation, characterized by higher proliferation rates compared to

Figure 4. Genomic microRNA profiling in lymphoid tumors. (A) Analysis of miR-150, miR-181a/b, miR-21 and miR-26a genomic loci in a subset of 15 paired chronic lymphocytic leukemia (CLL) phase/Richter syndrome (RS) phase cases. (B) The genomic regions containing miR-21, miR-146a and miR-181a/b-1 are more often gained than lost in a set of 737 cases of mature lymphoid tumors.17,26-29 For additional Information, see Online Supplementary Figure S4.

B

1010

haematologica | 2019; 104(5)

miRNA in Richter syndrome

A

B

C

D

Figure 5. microRNA network reprogramming during Richter transformation. (A) The 25-miRNA quantitative reverse transcription polymerase chain reaction (qRTPCR) expression network in the four groups analyzed. (B) Analysis of the number of edges (left panel) and connectivity (middle and right panels). (C) Ingenuity canonical pathway analysis â&#x20AC;&#x201C; top 20 signaling pathways for validated targets of miR-21, miR-146b, miR-150 and miR-181b and (D) for the miRNA hubs specific for Richter transformation (miR-191, miR-17 and miR-29c). The heatmap shows the most significantly involved pathways, considering validated targets of all miRNA (last column). White to purple boxes represent log(P-value) with the darkest boxes being the statistically most significant. P-values for targets of each miRNA separately are given in the first four columns. Due to the large number of targets, to increase the confidence of the data, we considered a P<0.01 as an inclusion criterion. For additional Information, see Online Supplementary Figure S5.

haematologica | 2019; 104(5)

1011

K. Van Roosbroeck et al.

A

B

C

Figure 6. microRNA analysis after transfer of chronic lymphocytic leukemia B cells into NSG mice and in vitro functional studies. (A) Schematic representation of the chronic lymphocytic leukemia (CLL) mouse model. (B) miR-150 and miR-26a are significantly downregulated in post-transfer peripheral blood mononuclear cells in mice compared with pre-transfer CLL samples. (C) Two different CLL-specific cell lines – MEC1 and HG3, and one diffuse large B-cell lymphoma cell line – HB – were used for functional assays (upper panel). Proliferation of MEC1, HG3 and HB cells was assessed by a CellTiter 96 AQueous One Solution assay. The cells were transfected with scrambled mimic or miR-21 mimic for 24, 48 and 72 h (middle panel). Representative images and graphs representing at least three independent experiments of the soft agar colony-formation assay for miR-21 mimic and control mimic transfected MEC1, HG3 and HB cells (lower panel). *P<0.05; **P<0.01; ***P<0.001. For additional Information, see Online Supplementary Figure S6.

1012

haematologica | 2019; 104(5)

miRNA in Richter syndrome

CLL.9 The levels of expression of miR-150 and miR-26a were found to be significantly downregulated in posttransfer samples compared with pre-transfer samples, independently of the IGHV mutated/unmutated status of the pre-transfer CLL samples (Figure 6B). This reduction was concordant with what we observed in the analyzed human cohorts (Figure 2A). Among the other six miRNA, we identified the same variation trends in miR-21 (2/5 cases) and miR-181b (3/5 cases) although statistical significance was not reached (Online Supplementary Figure 6A).

Among canonical pathways common to both the miR150 and miR-26a validated targets, we identified mostly cancer-related pathways, including the TP53 pathway (Online Supplementary Figure 6B).

Since no RS cell lines have been reported before, while performing these studies, we used two different CLL cell lines (MEC1 and HG3) and one diffuse large B-cell lymphoma (DLBCL) cell line (HB) (Figure 6C) to investigate the role of miRNA from the “restricted signature” in cell proliferation and colony formation: only the ectopic overexpression of miR-21 through a miR-mimic significantly increased cell proliferation in all three cell lines, when compared with scrambled controls (Figure 6C). We further examined the long-term effects (up to 21 days) of miR-21 on CLL and DLBCL cell proliferation using a soft agar colony-formation assay. Overexpression of miR-21 significantly increased colony formation in all cell models when compared with scrambled controls (Figure 6C).

There have been reports that aberrant expression of miR21 is associated with poor clinical outcome in patients with CLL.32 These data support a functional effect of miR21 in both CLL and DLBCL cells that needs to be tested in future RS cell models.

Discussion By using a stringent training/validation workflow on multiple, independent CLL/RS cohorts from several institutions, analyzed by three different expression profiling methods, we identified a miRNA signature that is involved in the process of transformation of CLL to RS. This signature is composed of the overexpressed miR-21, miR-146b and miR-181b, and by the downregulated miR150. These miRNA are SDE at the time of Richter transformation, but not after the same time of evolution of CLL in patients who never underwent Richter transformation (“control” group). miR-21 is a well-known oncogene, overexpressed and associated with poor prognosis in CLL,32,33 and highly expressed in patients who do not respond to fludarabine treatment.34 In addition, miR-21 transgenic mice spontaneously develop a pre-B-cell lymphoblastic lymphoma/leukemia.23 miR-150, on the other hand, is lowly expressed in CLL proliferation centers35,36 and shows reduced expression in poor prognosis CLL.37 Recently, Balatti et al. reported that a signature of 23 miRNA is differentially expressed in CLL samples which developed RS compared with CLL samples which did not develop RS. In their study, it was demonstrated that miR125a-5p is highly expressed, while miR-34a-5p is downregulated in pre-RS samples compared with samples from a control group and that deregulation of miR-34a and miR-125a-5p can predict ~50% of RS with a false positive rate of ~9%.38 The different results we obtained can be explained by the fact that we focused on miRNA which haematologica | 2019; 104(5)

are deregulated in the stage of RS versus pre-RS (group 1b versus group 1a) and not on miRNA that can predict Richter transformation (group 1a versus group 2a).

Of interest, the same four miRNA from this “restricted” signature are reported in literature as being differentially expressed in another fatal condition, sepsis (Online Supplementary Table S6), in the same way as in RS, making this signature similar to that of an acute infectious disease characterized by multiple organ failure. For example, tumor-secreted miR-21 can bind to toll-like receptors (TLR) on immune cells, triggering an inflammatory response that may contribute to tumor growth and metastasis.39,40 In addition, miR-21 suppresses T-cell apoptosis,41 and promotes Th17 cell differentiation, which is important for the development of multiple autoimmune diseases.42,43 miR-146b modulates the TLR4 signaling pathway through targeting of TLR4, MyD88, IRAK-1 and TRAF6, and has anti-inflammatory activity by reducing several pro-inflammatory cytokines, such as interleukin6, tumor necrosis factor-a, interleukin-8 and CCL2/3/7.44,45 Furthermore, miR-150 is important for the development of mature natural killer cells, which are of primordial importance for the induction of adaptive immune responses, and defense against pathogens.46 miR150 can also decrease the production of inflammatory cytokines, such as interleukin-2 and tumor necrosis factor-a, through disruption of CD28/B7 co-stimulatory signal transduction, resulting in immune tolerance.47 In the context of acute lung injury, miR-181b may stimulate inflammation through activation of nuclear factor-κB,48 a pathway commonly deregulated in CLL, which is currently being investigated as a therapeutic target for the treatment of CLL and RS.49 To address our miRNA signature in the setting of markedly enhanced CLL B-cell proliferation, as occurs in patients with Richter transformation, we used a CLL primary xenograft mouse model and found that the levels of expression of miR-26a and miR-150 were significantly reduced in proliferating CLL cells purified from the spleen of xenografted mice compared to the levels in pre-transfer cells. miR-26a reduces the expansion/accumulation of leukemic cells of Em-TCL1 mice and in vivo administration of miR-26a promotes apoptosis in mice.50 Further support for this new view of the pathogenesis of RS, as a disease related to both malignant and immunerelated processes, comes from the network analysis we performed. miRNA networks in cancer have been described to be disjointed and composed of multiple subnetworks.51 In contrast, the miRNA network after Richter transformation (group 1b) contains a significantly higher number of edges, with almost all nodes connected to the main network.

A noteworthy observation is that the expression of viral miRNA, detected by the Firefly assay, shows a very high level of correlation, leading to the construction of a highly connected viral miRNA network. In all four networks (groups 1a, 1b, 2a, and 2b, see Online Supplementary Figure S5) the viral miRNA are all part of the main graph and miR-4286 and miR-1260a (a tRNA-derived miRNA)52 are always joining this main network. This observation leads us to the supposition that there is an unknown mechanism that controls the expression of the different viral miRNA simultaneously. Moreover, it has recently been shown in an in vivo model that immunosuppressive therapy leads to the reactivation of Epstein-Barr virus in CLL 1013

K. Van Roosbroeck et al.

xenografts, increasing the risk of DLBCL (Richter transformation).53 The viral miRNA that correlate in our networks could be a marker of viral reactivation.

Several questions remain to be answered. First, what are the mechanisms that lead to the high level of correlation between miRNA during Richter transformation?

Further investigation of the exact role of each of the signature miRNA and their mechanisms of action in the Richter transformation process is warranted. For this, the development of in vitro and in vivo models of CLL to RS transformation is imperative. Second, it was surprising that, out of 15 deregulated miRNA identified by the Firefly assay in the extended cohort, only three could be confirmed by qRT-PCR in the same dataset. This is most likely due to a combination of factors, such as the poor quality of RNA from FFPE samples, sensitivity of the assays, and the difficulty in identifying good normalizers.

In the Firefly assay, a combination of miR-15a, miR-191 and miR-26a were found to be the best for normalization, while for the qRT-PCR experiments, the geometric mean of U6 and RNU48 was found to have the least variability.

In addition, the Firefly platform is not based on PCR amplification, while qRT-PCR is, which may also partially explain the differences in results.

Some limitations need to be acknowledged with regard to this study. The number of patients included was small, so the statistical power of the analysis is limited. RS is a relatively rare disease and it is very difficult to obtain a large sample set. Furthermore, we analyzed only 40 human and viral miRNA using the Firefly assay and 25 miRNA using qRT-PCR, thus offering a limited view of the miRNA expression deregulation and miRNA network in RS. It is likely that other miRNA, which we overlooked by analyzing a restricted panel of miRNA, could be involved in the pathogenesis of RS. We tried to overcome this limitation by choosing the most suitable miRNA candidates and using previous RNA-sequencing data,20 and

References 1. Ciccone M, Ferrajoli A, Keating MJ, Calin GA. SnapShot: chronic lymphocytic leukemia. Cancer Cell. 2014;26(5):770770.e1.

2. Kipps TJ, Stevenson FK, Wu CJ, et al.

Chronic lymphocytic leukaemia. Nat Rev Dis Primers. 2017;3:16096.

3. Bullrich F, Fujii H, Calin G, et al.

Characterization of the 13q14 tumor suppressor locus in CLL: identification of ALT1, an alternative splice variant of the LEU2 gene. Cancer Res. 2001;61(18):6640-6648.

4. Calin GA, Croce CM. Genomics of chronic lymphocytic leukemia microRNAs as new players with clinical significance. Semin Oncol. 2006;33(2):167-173.

5. Calin GA, Dumitru CD, Shimizu M, et al.

Frequent deletions and down-regulation of micro- RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia.

Proc Natl Acad Sci U S A. 2002;99(24): 15524-15529.

6. Aydin S, Rossi D, Bergui L, et al. CD38 gene polymorphism and chronic lymphocytic leukemia: a role in transformation to

1014

7.

8.

9.

10.

11.

12.

literature research of miRNA implicated in CLL. Finally, the identification of SDE miR-21 (overexpressed) and miR-150 (downregulated) and of HUS miR17, -29c and -191 demonstrates that RS is a genetically heterogeneous disease and offers new targets for the therapeutic development of anti-miRNA54 in this disease which currently does not have a confirmed treatment to increase the poor life expectancy.

Acknowledgments We thank Mike Tackett from FirePlex Service at Abcam (Cambridge, MA, USA) for the technical support.

Funding Dr. Calin is the Felix L. Haas Endowed Professor in Basic Science. Work in his laboratory is supported by National Institutes of Health (NIH/NCATS) grant UH3TR00943-01 through the NIH Common Fund, Office of Strategic Coordination (OSC), National Cancer Institute (NCI) grants 1R01 CA182905-01 and 1R01CA222007-01A1, an NIGMS 1R01GM122775-01 grant, a U54 grant â&#x20AC;&#x201C; UPR/MDACC Partnership for Excellence in Cancer Research 2016 Pilot Project, a Team DOD (CA160445P1) grant, a Ladies Leukemia League grant, a CLL Moonshot Flagship project, a SINF 2017 grant, and the Estate of C. G. Johnson, Jr. Work by Dr. Van Roosbroeck was supported in part by the Lauri Strauss Leukemia Foundation. Dr. Dragomir was supported by POC grant n.35/01.09.2016, ID 37_796. Dr. Bertoni was supported by grants from the Helmut Horten Foundation, the San Salvatore Foundation, and by Oncosuisse (OCS - 02296-082008). Dr. Stilgenbauer and Dr. Bloehdorn were supported by the DFG through SFB1074 subprojects B1 and B2. Dr. You was supported in part by NIH/NCI R01 CA164346 and IRG of the University of Texas MD Anderson Cancer Center. Dr. Gaidano was supported by AIRC 5 x 1000 project 21198, AIRC, Milan, Italy. Dr. Zupo was supported by Italian Ministry of Health 5 Ă&#x2014; 1000 funds 2014 and 2015.

Richter syndrome? Blood. 2008;111(12): 5646-5653.

Maddocks KJ, Ruppert AS, Lozanski G, et al. Etiology of ibrutinib therapy discontinuation and outcomes in patients with chronic lymphocytic leukemia. JAMA Oncol.

2015;1(1):80-87.

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.

Rossi D, Gaidano G. Richter syndrome: pathogenesis and management. Semin Oncol. 2016;43(2):311-319.

Coiffier B, Lepage E, Briere J, et al. CHOP chemotherapy plus rituximab compared with CHOP alone in elderly patients with diffuse large-B-cell lymphoma. N Engl J Med. 2002;346(4):235-242.

Dabaja BS, O'Brien SM, Kantarjian HM, et al. Fractionated cyclophosphamide, vincristine, liposomal daunorubicin (daunoXome), and dexamethasone (hyperCVXD) regimen in Richter's syndrome. Leuk Lymphoma. 2001;42(3):329337.

Tsimberidou AM, O'Brien SM, Cortes JE, et

13.

14.

15.

16.

al. Phase II study of fludarabine, cytarabine (Ara-C), cyclophosphamide, cisplatin and GM-CSF (FACPGM) in patients with Richter's syndrome or refractory lymphoproliferative disorders. Leuk Lymphoma.

2002;43(4):767-772.

Tsimberidou AM, Wierda WG, Wen S, et al.

Phase I-II clinical trial of oxaliplatin, fludarabine, cytarabine, and rituximab therapy in aggressive relapsed/refractory chronic lymphocytic leukemia or Richter syndrome.

Clin Lymphoma Myeloma Leuk. 2013;13 (5):568-574.

Rossi D, Spina V, Deambrogi C, et al. The genetics of Richter syndrome reveals disease heterogeneity and predicts survival after transformation. Blood. 2011;117(12): 3391-3401.

Falchi L, Keating MJ, Marom EM, et al.

Correlation between FDG/PET, histology, characteristics, and survival in 332 patients with chronic lymphoid leukemia. Blood.

2014;123(18):2783-2790.

Stilgenbauer S, Zenz T, Winkler D, et al.

Subcutaneous alemtuzumab in fludarabinerefractory chronic lymphocytic leukemia: clinical results and prognostic marker analy-

haematologica | 2019; 104(5)

miRNA in Richter syndrome

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

ses from the CLL2H study of the German Chronic Lymphocytic Leukemia Study Group. J Clin Oncol. 2009;27(24):39944001.

Chigrinova E, Rinaldi A, Kwee I, et al. Two main genetic pathways lead to the transformation of chronic lymphocytic leukemia to Richter syndrome. Blood. 2013;122(15): 2673-2682.

Bagnara D, Kaufman MS, Calissano C, et al.

A novel adoptive transfer model of chronic lymphocytic leukemia suggests a key role for T lymphocytes in the disease. Blood.

2011;117(20):5463-5472.

Chen SS, Chiorazzi N. Murine genetically engineered and human xenograft models of chronic lymphocytic leukemia. Semin Hematol. 2014;51(3):188-205.

Patten PE, Ferrer G, Chen SS, et al. Chronic lymphocytic leukemia cells diversify and differentiate in vivo via a nonclassical Th1dependent, Bcl-6-deficient process. JCI Insight. 2016;1(4).

Ferrajoli A, Ivan C, Ciccone M, et al.

Epstein-Barr virus microRNAs are expressed in patients with chronic lymphocytic leukemia and correlate with overall survival. EBioMedicine. 2015;2(6):572-582.

Vasilescu C, Dragomir M, Tanase M, et al.

Circulating miRNAs in sepsis. A network under attack: an in-silico prediction of the potential existence of miRNA sponges in sepsis. PLoS One. 2017;12(8):e0183334.

Medina PP, Nolde M, Slack FJ. OncomiR addiction in an in vivo model of microRNA21-induced pre-B-cell lymphoma. Nature.

2010;467(7311):86-90.

Calin GA, Sevignani C, Dumitru CD, et al.

Human microRNA genes are frequently located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci U S A. 2004;101(9):2999-3004.

Sevignani C, Calin GA, Nnadi SC, et al.

MicroRNA genes are frequently located near mouse cancer susceptibility loci. Proc Natl Acad Sci U S A. 2007;104(19):80178022.

Boi M, Rinaldi A, Kwee I, et al.

PRDM1/BLIMP1 is commonly inactivated in anaplastic large T-cell lymphoma. Blood.

2013;122(15):2683-2693.

Martinez N, Almaraz C, Vaque JP, et al.

Whole-exome sequencing in splenic marginal zone lymphoma reveals mutations in genes involved in marginal zone differentiation. Leukemia. 2014;28(6):1334-1340.

Rinaldi A, Kwee I, Young KH, et al.

Genome-wide high resolution DNA profiling of hairy cell leukaemia. Br J Haematol.

2013;162(4):566-569.

haematologica | 2019; 104(5)

29. Rossi D, Trifonov V, Fangazio M, et al. The coding genome of splenic marginal zone lymphoma: activation of NOTCH2 and other pathways regulating marginal zone development. J Exp Med. 2012;209(9):15371551.

30. Van Roosbroeck K, Calin GA. MicroRNAs in chronic lymphocytic leukemia: miRacle or miRage for prognosis and targeted therapies? Semin Oncol. 2016;43(2):209-214.

31. Chiorazzi N. Cell proliferation and death: forgotten features of chronic lymphocytic leukemia B cells. Best Pract Res Clin Haematol. 2007;20(3):399-413.

32. Rossi S, Shimizu M, Barbarotto E, et al. microRNA fingerprinting of CLL patients with chromosome 17p deletion identify a miR-21 score that stratifies early survival.

Blood. 2010;116(6):945-952.

33. Fulci V, Chiaretti S, Goldoni M, et al.

Quantitative technologies establish a novel microRNA profile of chronic lymphocytic leukemia. Blood. 2007;109(11):4944-4951.

34. Ferracin M, Zagatti B, Rizzotto L, et al.

MicroRNAs involvement in fludarabine refractory chronic lymphocytic leukemia.

Mol Cancer. 2010;9:123.

35. Szurian K, Csala I, Piurko V, et al.

Quantitative miR analysis in chronic lymphocytic leukaemia/small lymphocytic lymphoma - proliferation centres are characterized by high miR-92a and miR-155 and low miR-150 expression. Leuk Res. 2017;58:39-42.

36. Wang M, Tan LP, Dijkstra MK, et al. miRNA analysis in B-cell chronic lymphocytic leukaemia: proliferation centres characterized by low miR-150 and high BIC/miR-155 expression. J Pathol. 2008;215(1):13-20.

37. Mraz M, Chen L, Rassenti LZ, et al. miR150 influences B-cell receptor signaling in chronic lymphocytic leukemia by regulating expression of GAB1 and FOXP1. Blood.

2014;124(1):84-95.

38. Balatti V, Tomasello L, Rassenti LZ, et al.

MiR-125a and MiR-34a expression predicts Richter syndrome in chronic lymphocytic leukemia patients. Blood. 2018;132(20): 2179-2182.

39. Bayraktar R, Van Roosbroeck K, Calin GA. Cell-to-cell communication: microRNAs as hormones. Mol Oncol. 2017;11(12):16731686.

40. Fabbri M, Paone A, Calore F, et al.

MicroRNAs bind to Toll-like receptors to induce prometastatic inflammatory response. Proc Natl Acad Sci U S A. 2012; 109(31):E2110-2116.

41. Meisgen F, Xu N, Wei T, et al. MiR-21 is upregulated in psoriasis and suppresses T cell

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

apoptosis. Exp Dermatol. 2012;21(4):312314.

Wang S, Wan X, Ruan Q. The microRNA21 in autoimmune diseases. Int J Mol Sci. 2016;17(6).

Xu WD, Pan F, Li JH & Ye DQ. MicroRNA21 with therapeutic potential in autoimmune diseases. Expert Opin Ther Targets.

2013;17(6):659-665.

Curtale G, Mirolo M, Renzi TA, et al.

Negative regulation of Toll-like receptor 4 signaling by IL-10-dependent microRNA146b. Proc Natl Acad Sci U S A. 2013;110(28):11499-11504.

Taganov KD, Boldin MP, Chang KJ, Baltimore D. NF-kappaB-dependent induction of microRNA miR-146, an inhibitor targeted to signaling proteins of innate immune responses. Proc Natl Acad Sci U S A. 2006;103(33):12481-12486.

Bezman NA, Chakraborty T, Bender T, Lanier L L. miR-150 regulates the development of NK and iNKT cells. J Exp Med. 2011;208(13):2717-2731.

Sang W, Wang Y, Zhang C, et al. MiR-150 impairs inflammatory cytokine production by targeting ARRB-2 after blocking CD28/B7 costimulatory pathway. Immunol Lett. 2016;172:1-10.

Wang Y, Mao G, Lv Y, Huang Q, Wang G. MicroRNA-181b stimulates inflammation via the nuclear factor-kappaB signaling pathway in vitro. Exp Ther Med. 2015;10(4): 1584-1590.

Vaisitti T, Gaudino F, Ouk S, et al. Targeting metabolism and survival in chronic lymphocytic leukemia and Richter syndrome cells by a novel NF-kappaB inhibitor.

Haematologica. 2017;102(11):1878-1889.

D'Abundo L, Callegari E, Bresin A, et al.

Anti-leukemic activity of microRNA-26a in a chronic lymphocytic leukemia mouse model. Oncogene. 2017;36(47):6617-6626.

Volinia S, Galasso M, Costinean S, et al.

Reprogramming of miRNA networks in cancer and leukemia. Genome Res. 2010;20(5): 589-599.

Venkatesh T, Suresh PS, Tsutsumi R. tRFs: miRNAs in disguise. Gene. 2016;579(2):133138.

Garcia-Barchino MJ, Sarasquete ME, Panizo C, et al. Richter transformation driven by Epstein-Barr virus reactivation during therapy-related immunosuppression in chronic lymphocytic leukaemia. J Pathol. 2018;245 (1):61-73.

Shah MY, Ferrajoli A, Sood AK, LopezBerestein G, Calin GA. microRNA Therapeutics in cancer - an emerging concept. EBioMedicine. 2016;12:34-42.

1015

ARTICLE Ferrata Storti Foundation

Chronic Lymphocytic Leukemia

Targeting intermediary metabolism enhances the efficacy of BH3 mimetic therapy in hematologic malignancies

Aoula Al-Zebeeby,1 Meike Vogler,2 Mateus Milani,1 Caitlin Richards,1 Ahoud Alotibi,1 Georgia Greaves,1 Martin J.S. Dyer,3 Gerald M. Cohen1,4 and Shankar Varadarajan1,4*

Department of Molecular and Clinical Cancer Medicine, Institute of Translational Medicine, University of Liverpool, UK; 2Institute for Experimental Cancer Research in Pediatrics, Goethe-University, Frankfurt, Germany; 3Ernest and Helen Scott Haematological Research Institute, Leicester Cancer Research Centre, University of Leicester, Leicester Royal Infirmary, UK and 4Department of Molecular and Clinical Cancer Pharmacology, Institute of Translational Medicine, University of Liverpool, UK 1

Haematologica 2019 Volume 104(5):1016-1025

ABSTRACT

B

Correspondence: SHANKAR VARADARAJAN svar@liverpool.ac.uk Received: August 16, 2018.

Accepted: November 20, 2018.

Pre-published: November 22, 2018.

H3 mimetics are novel targeted drugs with remarkable specificity, potency and enormous potential to improve cancer therapy.

However, acquired resistance is an emerging problem. We report the rapid development of resistance in chronic lymphocytic leukemia cells isolated from patients exposed to increasing doses of navitoclax (ABT-263), a BH3 mimetic. To mimic such rapid development of chemoresistance, we developed simple resistance models to three different BH3 mimetics, targeting BCL-2 (ABT-199), BCL-XL (A-1331852) or MCL-1 (A-1210477), in relevant hematologic cancer cell lines. In these models, resistance could not be attributed to either consistent changes in expression levels of the anti-apoptotic proteins or interactions among different pro- and anti-apoptotic BCL-2 family members. Using genetic silencing, pharmacological inhibition and metabolic supplementation, we found that targeting glutamine uptake and its downstream signaling pathways, namely glutaminolysis, reductive carboxylation, lipogenesis, cholesterogenesis and mammalian target of rapamycin signaling resulted in marked sensitization of the chemoresistant cells to BH3 mimeticmediated apoptosis. Furthermore, our findings highlight the possibility of repurposing widely used drugs, such as statins, to target intermediary metabolism and improve the efficacy of BH3 mimetic therapy.

doi:10.3324/haematol.2018.204701 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1016 Š2019 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.

1016

Introduction Failure to undergo apoptosis is a cardinal feature of cancer and several targeted therapies, such as the small molecule inhibitors targeting specific members of the anti-apoptotic BCL-2 family - navitoclax/ABT-263 (targeting BCL-2, BCL-XL and BCL-w) and venetoclax/ABT-199 (BCL-2 specific) - are aimed at facilitating cancer cell clearance by enhanced apoptosis.1-4 Recently, selective inhibitors of BCL-XL (A1331852) and MCL-1 (A-1210477 and S63845) have also been synthesized.5-7 Despite their selectivity in targeting distinct anti-apoptotic BCL-2 family members, and remarkable potency in inducing rapid and extensive apoptosis in a wide variety of malignancies, resistance to BH3 mimetics, in particular venetoclax, is starting to be reported in the clinic. Elevated levels of multiple members of the anti-apoptotic BCL-2 family proteins, including BCL-XL and MCL-1, are often implicated in such chemoresistance.8-13 Although it may be possible to target these proteins with a combination of selective BH3 mimetics, the potential toxicities associated with such combination therapy may be problematic.

Altered metabolism is a promising approach to enhance the efficacy of chemotherapeutic agents, as a requirement for intermediary metabolites, such as glucose and glutamine, for the survival and proliferation of cancer cells is well dochaematologica | 2019; 104(5)

Modulation of intermediary metabolism in cancer therapy

umented.1,14-19 This is a promising approach, as drugs targeting different stages of intermediary metabolism are already approved or in trials for treating different malignancies.20,21 In this study, we found a low level of resistance that developed in cells from patients with chronic lymphocytic leukemia (CLL) exposed to navitoclax. To mimic this modest resistance, we developed simple models of resistance to different BH3 mimetics and demonstrated that downregulating glutamine uptake or metabolism as well as its downstream signaling cascades, such as reductive carboxylation, lipogenesis and cholesterogenesis, result in enhanced apoptosis of cancer cells resistant to different BH3 mimetics, thus highlighting the possibility that inhibition of key regulatory enzymes of these metabolic pathways may enhance sensitivity to BH3 mimetic therapy.

Methods Reagents and antibodies ABT-263, A-1331852 and A-1210477 were from AbbVie (North Chicago, IL, USA), ABT-199, epigallocatechin gallate (EGCG), CB-839, simvastatin, rapamycin and torin-1 from Selleck Chemicals (Houston, TX, USA), gamma-L-glutamyl-pnitroanilide (GPNA) from Insight Biotechnology (Wembley, Middlesex, UK), azaserine from Cambridge Bioscience (Cambridge, UK), aminooxyacetate (AOA), sodium palmitate, dimethyl a-ketoglutarate, oxaloacetate and citrate from SigmaAldrich (Gillingham, UK), L-glutamine from Life Technologies (Paisley, UK) and GSK2194069, SB204990, atorvastatin, pitavastatin and bafilomycin A1 from Tocris (Abingdon, UK).

Antibodies against PARP, BCL-2, MCL-1, BAX, BAK and GAPDH were from Santa Cruz Biotechnology (Santa Cruz, CA, USA), caspase-3, caspase-9, BCL-XL, BCL-w, BIM, PUMA, BAD, IDH2, ACL, ACO2, ATG5 and ATG7 from Cell Signaling Technology (MA, USA), BID from Prof. J. Borst (The Netherlands Cancer Institute, Amsterdam, the Netherlands), NOXA from Millipore (Watford, UK) and SLC1A5, GLS, GFAT, GLUD1, IDH3, FASN and HMGR from Abcam (Cambridge, UK).

Primary chronic lymphocytic leukemia cells and cell lines Peripheral blood samples from CLL patients were obtained with the patients’ consent and ethics committee approval (06_Q2501_122) from Leicester Haematological Tissue Bank and cultured as described elsewhere.38 CLL samples were obtained from patients enrolled in a phase I/IIa study of ABT-263 (navitoclax) in patients with relapsed or refractory CLL (NCT00481091). Lymphocytes were purified and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (Life Technologies Inc.). Alternatively, blood from patients was incubated at 37°C in 48-well plates and apoptosis assessed as described previously.39,40 Blood samples were collected prior to the first in vivo dose of navitoclax or 4 h after dosing during the lead-in period (day 1 of the lead-in period; L1D1), or day 1 of cycle 1 (C1D1), cycle 3 (C3D1) or cycle 5 (C5D1). Samples were collected 4 h after dosing as blood concentrations of ABT-263 were maximal at this time.41 For culture of CLL cells, mouse fibroblast L cells were irradiated with 75 Gy and seeded in 24well plates (3 x105 cells/well). CLL cells were cultured at 1.5 x 106 cells/well on the L cells and removed when required by gentle washing with RPMI before treatment. Mantle cell lymphoma (MAVER-1), chronic myeloid leukemia (K562) and multiple myeloma (NCI-H929) cell lines were cultured in RPMI 1640 haematologica | 2019; 104(5)

medium but the medium was supplemented with 0.02% 2-mercaptoethanol for culturing H929 cells. Cell lines were from either the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DMSZ; Braunshweig, Germany) or the American Type Culture Collection (ATCC; Middlesex, UK) and subjected to short tandem repeat profiling to confirm their identity.

Resistance models The different resistance models to relevant BH3 mimetics were developed by treating control cells (represented as A in all schemes) of MAVER-1, K562 and H929 to the relevant BH3 mimetics, ABT-199 (10 nM), A-1331852 (10 nM) or A-1210477 (5 µM), respectively. In the first resistance model, cells were exposed to their appropriate BH3 mimetic for 24 h followed by 2 weeks without drug resulting in the cells depicted as B. These cells were then exposed to their appropriate BH3 mimetic for a further 24 h followed again by 2 weeks without drug resulting in C. This procedure was repeated twice more, resulting in E. In the second resistance model, cells were exposed to their appropriate BH3 mimetic for 24 h followed by 8 weeks without drug, resulting in the cells depicted as A4. The cells were collected every 2 weeks and labeled as A1, A2 and A3, respectively. In the third resistance model, cells were exposed to increasing concentrations of the appropriate BH3 mimetic every 5 days, resulting in cells depicted as A-a, A-b, A-c and A-d. The fourth model of resistance was made in a similar manner, but the 5-day treatment period was split into 2 days of treatment, followed by 3 days without drug, resulting in cells depicted as A-i, A-ii, A-iii and A-iv.

Metabolic deprivation, supplementation and apoptosis measurements For glutamine deprivation experiments, cells were washed with phosphate-buffered saline and re-suspended in SILAC RPMI 1640 Flex Media (Life Technologies Inc.), supplemented with glucose (2 mg/mL) and 10% fetal bovine serum, for 16 h.

For supplementation studies, the indicated concentrations of metabolites were added to the glutamine-free media, immediately before glutamine deprivation. For lipid supplementation studies, sodium palmitate [dissolved in water at 70°C to form a stock concentration of 100 mM and added dropwise into fatty acid-free bovine serum albumin (10%) to produce a final concentration of 10 mM] was supplemented in the culture media.

The extent of apoptosis in cells following different treatments was quantified by fluorescence activated cell sorting (FACS) after having stained the cells with annexin V-fluorescein isothiocyanate and propidium iodide to measure phosphatidylserine externalization, as previously described.42

Short interfering RNA knockdown, immunoprecipitation and western blotting Cells were transfected with 10 nM of short interfering RNA (siRNA) against SLC1A5 (SI00079730), GLS (SI03155019), GFAT (SI03246355), GLUD1 (SI02654743), IDH2 (SI02654820), IDH3 (SI00300524), ACO2 (SI03019037), ACLY (SI02663332), FASN (SI00059752), HMGR (SI00017136), ATG5 (SI02633946) and ATG7 (SI04344830) from Qiagen Ltd. (Manchester, UK) using interferin (Polyplus Transfection Inc, NY, USA), according to the manufacturer's protocol and processed 72 h after transfection.

Immunoprecipitation and western blotting were carried out according to standard protocols.43

Statistical analysis One-way analysis of variance (ANOVA) multiple comparisons and the Fisher least significant difference test (P≤0.01) were per1017

A. Al-Zebeeby et al.

formed to compare sensitive and resistant cells. For samples from CLL patients, one-way repeated measures ANOVA with the Fisher least significant difference test (P≤0.01) was used and statistics analyzed using GraphPad Prism 6 software (La Jolla, CA, USA).

Results Hematologic malignancies rapidly acquire resistance to BH3 mimetics The potential of BH3 mimetic therapy in cancer was first demonstrated in the treatment of BCL-2-dependent CLL using navitoclax/ABT-263. In a phase I/II clinical trial of navitoclax, CLL patients were treated for an initial lead-in period of 7 days with a low dose of navitoclax (100 mg daily) followed by five to seven cycles of treatment, with each cycle lasting 21 days during which the patients received 250 mg navitoclax daily. Analysis of blood samples collected from these patients, either prior to the first in vivo dose of navitoclax or 4 h after dosing, during the different cycles of therapy revealed marked

A

D

B

changes in the ability of navitoclax to induce apoptosis in the CLL cells (Figure 1A). The first in vivo dose of navitoclax on day 1 of the lead-in period (L1D1) resulted in a time-dependent induction of apoptosis, as assessed by phosphatidylserine externalization and ultrastructural changes, in comparison to that of CLL cells from the same patients prior to treatment (Figure 1A and Online Supplementary Figure S1). A progressive increase in resistance to navitoclax was observed in CLL cells in vivo during the different cycles of treatment (Figure 1A). Since these studies were carried out in whole blood, we wanted to ascertain whether the decrease in ABT-263-induced apoptosis in CLL cells could be attributed to chemoresistance. To test this, CLL cells were isolated from these patients at the beginning of each treatment cycle and exposed to increasing concentrations of ABT-263. A significant decrease [3-fold difference in the half maximal inhibitory concentration (IC50) values between the lead-in period and cycle 5] was observed in their ability to undergo ABT-263-induced apoptosis (Figure 1B), demonstrating that continued dosing of patients with ABT-263 resulted in a modest, yet significant increase in chemoresistance.

C

E

F

Figure 1. Hematologic malignancies rapidly acquire resistance to BH3 mimetics. (A) Blood samples collected from patients with chronic lymphocytic leukemia (CLL) (n=5), either prior to the first in vivo dose of navitoclax or 4 h after dosing during different stages of treatment - day 1 of the initial lead-in-period (L1D1), day 1 of cycle 1 (C1D1), day 1 of cycle 3 (C3D1) or day 1 of cycle 5 (C5D1) - were incubated ex vivo and the extent of apoptosis in the CD19+ CLL cells was assessed at the indicated time points by measuring phosphatidylserine (PS) externalization. (B) CLL cells isolated from these patients at the beginning of each treatment cycle, as indicated in the figure, were exposed in vitro to increasing concentrations of ABT-263 and the extent of apoptosis was assessed: half maximal inhibitory concentration (IC50) values are shown. (C) Scheme for establishing resistance to specific BH3 mimetics in relevant hematologic cell lines, as explained in the Methods section.

Sensitive [A] and resistant [E] cells of MAVER-1, K562 and H929 cell lines were exposed for 4 h to ABT-199 (10 nM), A-1331852 (10 nM) and A-1210477 (5 mM), respectively, and apoptosis was assessed. (D-F) Combinations with some but not all BH3 mimetics restored apoptotic sensitivity of resistant [E] MAVER-1, K562 and H929 cells exposed for 4 h to ABT-199 (10 nM), A-1331852 (10 nM) or A-1210477 (5 mM), respectively. ***P⩽0.001, **P⩽0.01. Error bars = mean ± standard error of mean (n=3).

1018

haematologica | 2019; 104(5)

Modulation of intermediary metabolism in cancer therapy

Since chemoresistance is an emerging problem in BH3 mimetic therapy, we extended these studies to more selective BH3 mimetics, such as ABT-199, which has replaced navitoclax owing to the dose-limiting thrombocytopenia associated with BCL-XL inhibition.3 Moreover, other selective BH3 mimetics that target BCL-XL (A1331852) and MCL-1 (A-1210477 and S63845) have been introduced for use in several other malignancies.5-7 Using these BH3 mimetics and relevant cancer cell lines, we tried to mimic the rapid resistance observed in CLL patients following navitoclax treatment (Figure 1A,B), in order to identify ways to tackle chemoresistance, as it emerges. For this, we chose the BCL-2-dependent MAVER-1, BCL-XL-dependent K562 and MCL-1-dependent H929 cell lines and exposed them to ABT-199, A1331852 and A-1210477, respectively, to generate different models of resistance (Figure 1C and Online Supplementary Figure S2). Initial exposure of the relevant cell lines to the corresponding BH3 mimetic resulted in a rapid, time-dependent induction of apoptosis as assessed by the activation of caspase-9 and caspase-3 as well as cleavage of the canonical caspase substrate, PARP (Online Supplementary Figure S2A). Resistance to BH3 mimetics in these cells was generated by following the scheme presented in Figure 1C, when the initially sensitive cells [A] became relatively resistant [E], after four exposures (within 8 weeks) to their respective BH3 mimetic (Figure 1C).

Similarly, a rapid resistance to the different BH3 mimetics was also observed using the other three resistance models (Online Supplementary Figure S2B-D). The rapid and modest resistance to the different BH3 mimetics in these cell lines was comparable to the extent of resistance observed in CLL cells during navitoclax therapy (Figure 1B).

Resistance to BH3 mimetics can be overcome by inhibiting multiple BCL-2 family members Since resistance to BH3 mimetics has often been attributed to elevated expression levels of one or more antiapoptotic BCL-2 family members, we wanted to identify whether such changes could be responsible for the observed resistance. Comparison of the sensitive [A], intermediate [C] and resistant [E] cells from the different cell lines did not reveal any consistent differences in BCL2 family expression to explain the resistance (Online Supplementary Figure S3). We, therefore, sought to identify whether changes in protein-protein interactions among different pro-apoptotic BH3-only members and their antiapoptotic counterparts could explain the resistance to BH3 mimetics. To do this, we performed immunoprecipitation studies to isolate the anti-apoptotic proteins bound to BIM and PUMA, which were abundantly expressed in the three different cell types. However, in the sensitive [A] and resistant [E] MAVER-1 cells, immunoprecipitation of BIM and PUMA revealed similar binding of BCL-2 and BCL-XL and little or no binding to MCL-1 (Online Supplementary Figure S4). Likewise, no differences were observed in the binding of BIM and PUMA to BCL-XL and MCL-1 in sensitive and resistant K562 or H929 cells (Online Supplementary Figure S4).

Although the protein expression levels and immunoprecipitation studies did not support an involvement of other BCL-2 family proteins in the observed resistance, the resistance to ABT-199 observed in MAVER-1 cells was completely overcome by a combination of ABT-199 with either A-1331852 or A-1210477, but not by either Ahaematologica | 2019; 104(5)

1331852 or A-1210477 alone, suggesting that the resistant cells depend not only on BCL-2 but also on BCL-XL and/or MCL-1 for survival (Figure 1D). Furthermore, a combination of all three BH3 mimetics induced apoptosis in all the resistant cells, emphasizing the importance of all three anti-apoptotic BCL-2 family members in chemoresistance in these cells (Figure 1D). In K562 and H929 cells, the resistance was overcome by the combination of A1331852 and A-1210477, but not ABT-199, thus implicating primary roles for BCL-XL and MCL-1 in chemoresistance (Figure 1E,F). Similar to the MAVER-1 cells, the chemoresistant K562 cells also exhibited enhanced apoptosis following treatment with a combination of all three BH3 mimetics (Figure 1E), suggesting that some contribution of BCL-2 could not be totally excluded in these cells.

These observations were almost entirely reproducible in the other three models of resistance (Online Supplementary Figure S5), supporting the notion that BCL-XL and/or MCL-1 contributed significantly to the observed chemoresistance in the different models.

Modulation of glutamine uptake and/or metabolism enhances sensitivity to BH3 mimetics Although the above results demonstrate that a combination of BH3 mimetics can overcome resistance, such an approach targeting multiple members of the BCL-2 family requires careful evaluation of the therapeutic index, as these proteins perform redundant functions in the maintenance of normal cellular homeostasis. An alternative strategy to overcome chemoresistance to BH3 mimetics could be achieved by altered metabolism, as depriving cells of glutamine has recently been shown to overcome MCL-1-mediated chemoresistance in multiple myeloma.19 In our experiments, glutamine deprivation for 16 h alone did not exhibit any effect on overall cell survival and yet sensitized both the sensitive [A] and resistant [E] cells to BH3 mimetic-mediated apoptosis (Figure 2A). The increase in apoptosis observed in both sensitive and resistant cells indicates that glutamine deprivation most likely provides an additional cytotoxic cue that induces apoptosis in the sensitive and resistant cells, but could also bypass the resistance mechanism in the resistant cells. Nevertheless, our results suggest that targeting the glutamine metabolic pathway could enhance apoptosis and circumvent chemoresistance to BH3 mimetics in all our resistance models (Figure 2A and Online Supplementary Figure S6). To investigate the therapeutic potential of this approach, we wished to further understand how changes in glutamine metabolism might alter BH3 mimetic-mediated apoptosis.

Glutamine is transported into cells primarily via the SLC1A5 transporter and metabolized to glutamate, primarily via glutaminase (GLS)-mediated glutaminolysis.22 Alternatively, glutamate can be generated from glutamine as a by-product of the hexosamine biosynthetic pathway, during the conversion of fructose-6-phosphate to glucosamine-6-phosphate, catalyzed by the enzyme, glutamine:fructose-6-phosphate-amidotransferase (GFAT) (Figure 2B).22 Glutamate can then generate a-ketoglutarate (a-KG) either via glutamate dehydrogenase (GLUD)-mediated oxidative deamination or a series of aminotransferase reactions (Figure 2B).14,22 Downregulation by RNA interference or pharmacological inhibition of key players involved in both glutamine uptake and its subsequent metabolism restored sensitivi1019

A. Al-Zebeeby et al.

ty of chemoresistant K562 cells to A-1331852-mediated apoptosis, albeit to varying degrees (Figure 2C,D). While downregulation of SLC1A5 and GLS resulted in enhanced sensitivity to A-1331852-mediated apoptosis in the different cell lines tested, inhibition of other enzymes in the glutamine metabolic pathway produced more modest effects (Figure 2C,D and Online Supplementary Figure S7).

Targeting reductive carboxylation enhances sensitivity to BH3 mimetics Metabolic supplementation of the glutamine-deprived cells with either glutamine or a-KG restored the resistance of K562 cells to A1331852-induced apoptosis (Figure 3A). Since glutamine-derived a-KG feeds into the tricarboxylic acid cycle, we explored the functions of this cycle and its intermediates in chemoresistance to BH3 mimetics. For this, we supplemented glutamine-deprived cells with tricarboxylic acid intermediates, such as oxaloacetate and citrate. Strikingly, supplementation with citrate, but not oxaloacetate, restored the resistance of K562 cells to A-1331852-induced apoptosis (Figure 3A). These results suggest that conversion of a-KG to citrate via reductive carboxylation may play a role in regulating sensitivity to BH3 mimetics.

Reductive carboxylation involves the conversion of aKG to isocitrate (catalyzed by isocitrate dehydrogenases

1 and 2; IDH1/2), which then generates citrate (catalyzed by aconitase) (Figure 3B).23 While IDH1/2 catalyze reductive carboxylation of a-KG, another isoform of isocitrate dehydrogenase, IDH3, catalyzes the reverse conversion of isocitrate to a-KG.23 Silencing the expression of IDH2 and aconitase, but not IDH3, restored the sensitivity of chemoresistant K562 cells to A-1331852-mediated apoptosis (Figure 3C,D), suggesting that the availability of citrate could be associated with the chemoresistance phenotype. To test this, IDH2-downregulated K562 cells were supplemented with citrate to identify whether addition of citrate could overcome the inhibition of reductive carboxylation and revert the associated increase in A1331852-induced apoptosis. Supplementation with citrate, but not glutamine or a-KG, did indeed restore the chemoresistance of IDH2-downregulated cells (Figure 3E), thus confirming the involvement of reductive carboxylation and the availability of citrate as crucial players in the observed chemoresistance.

Downregulation of lipogenesis and cholesterogenesis enhances sensitivity to BH3 mimetics Since citrate generated as a consequence of reductive carboxylation of a-KG is a major source of carbon for lipid synthesis, we investigated whether inhibition of lipogenesis could enhance sensitivity to BH3 mimetics

A

B

C

D

Figure 2. Inhibition of glutamine uptake and metabolism enhances sensitivity to BH3 mimetics. (A) Deprivation of glutamine (Gln) for 16 h restores the apoptotic sensitivity of resistant [E] MAVER-1, K562 and H929 cells to the indicated BH3 mimetic for 4 h. (B) Scheme representing the pathway of glutamine uptake and metabolism. (C) Apoptotic sensitivity of K562 resistant [E] cells exposed to A-1331852 (10 nM) for 4 h was restored following genetic knockdown for 72 h with the indicated short interfering (si) RNA. (D) Apoptotic sensitivity of K562 resistant [E] cells exposed to A-1331852 (10 nM) for 4 h was restored following pharmacological inhibition of glutamine uptake or metabolism with GPNA (5 mM) for 48 h, CB-839 (10 mM) for 72 h, azaserine (25 mM) for 16 h and AOA (500 mM) for 24 h but not with EGCG (50 mM) for 24 h. Western blots confirmed the knockdown efficiency of the different siRNA. ***P⊽0.001, **P⊽0.01. Error bars = mean ¹ standard error of mean (n=3). PS: phosphatidylserine; DMSO: dimethylsulfoxide.

1020

haematologica | 2019; 104(5)

Modulation of intermediary metabolism in cancer therapy

(Figure 4A). Using a complementary approach of genetic and pharmacological inhibition of ATP-citrate lyase (ACLY), which catalyzes the conversion of citrate to acetyl-CoA,24,25 as well as fatty acid synthase (FASN), which synthesizes long chain fatty acids following the condensation of acetyl-CoA and malonyl-CoA,26,27 we identified that modulation of the lipogenesis pathway, using either genetic silencing or pharmacological inhibition of ACLY (with SB204990) or FASN (with GSK2194069) could enhance sensitivity of cells to BH3 mimetics (Figure 4B,C and Online Supplementary Figure S7C,D). Furthermore, metabolic supplementation with palmitate (the product of FASN (Figure 4A) in cells treated with GSK2194069 reverted the sensitized cells to their original chemoresistant phenotype (Figure 4D), thus obviating a requirement for FASN. These findings conclusively demonstrated that enhanced lipogenesis was associated with chemoresistance to BH3 mimetics and targeting lipogenesis could circumvent such resistance by enhancing BH3 mimetic-mediated apoptosis.

Acetyl-CoA generated from citrate can also feed into the cholesterol biosynthetic pathway, thus resulting in enhanced cholesterol production in cells. Targeting the rate-limiting step of cholesterol biosynthesis (catalyzed

B

A

D

by HMG-CoA reductase; HMGR), either by genetic knockdown (Figure 4E) or pharmacological inhibition, using three widely used statins, simvastatin, atorvastatin and pitavastatin (Figure 4F), reversed resistance and restored the sensitivity of cells to BH3 mimetics (Figure 4E,F and Online Supplementary Figure S7E). Taken together, these data demonstrate that inhibition of several key players in lipid synthesis, including ACLY, FASN and HMGR, enhances the sensitivity to BH3 mimetics.

Targeting the mammalian target of rapamycin signaling cascade enhances sensitivity to BH3 mimetics Since glutamine metabolism has been extensively implicated in mammalian target of rapamycin (mTOR) signaling,22,28 we speculated whether targeting mTOR kinases could enhance sensitivity to BH3 mimetics.

Inhibition of mTOR kinases with rapamycin and torin-1 resulted in significant sensitization of cells to BH3-mediated apoptosis (Figure 5A and Online Supplementary Figure S7F). To identify whether torin-1-mediated sensitization of cells to apoptosis was due to autophagy, we exposed the sensitive and resistant cells to bafilomycin A1 (Baf A1), which blocks autophagic flux by preventing lysosomal fusion of the autophagosomes. Exposure to

C

E

Figure 3. Modulation of reductive carboxylation enhances sensitivity to BH3 mimetics. (A) K562 sensitive [A] and resistant [E] cells were cultured in normal RPMI medium or glutamine-free medium with and without the supplementation of glutamine (2 mM), exposed to A-1331852 (10 nM) for 4 h and the extent of apoptosis assessed. Addition of citrate (4 mM) and ι-ketoglutarate (a-KG) (4 mM) but not oxaloacetate (4 mM) for 16 h reversed the sensitivity of the resistant [E] cells in glutamine-deprived media. (B) Scheme representing the link between the tricarboxylic acid (TCA) cycle and reductive carboxylation. (C) K562 sensitive [A] and resistant [E] cells were transfected with short interfering (si) RNA against IDH2, IDH3 and aconitase for 72 h, followed by exposure for 4 h to A-1331852 and then apoptosis was assessed. (D) Western blots confirmed the knockdown efficiency of the different siRNA. (E) K562 [A] and [E] cells, transfected with a siRNA against IDH2 for 72 h, were glutamine-deprived and then given or not supplementation with glutamine (2 mM), a-ketoglutarate or citrate (both at 4 mM) for 16 h and the extent of apoptosis following exposure to A-1331852 (10 nM) for 4 h was assessed. ***P⊽0.001. Error bars = mean ¹ standard error of mean (n=3). PS: phosphatidylserine; DMSO: dimethylsulfoxide.

haematologica | 2019; 104(5)

1021

A. Al-Zebeeby et al.

bafilomycin A1 failed to revert torin-1-mediated chemosensitization, suggesting that this effect could be independent of autophagy (Figure 5B). Furthermore, genetic silencing of autophagy proteins, ATG5 and ATG7, which are critical for the induction of autophagy, also failed to revert torin-1-mediated sensitization (Figure 5C), confirming our finding that mTOR inhibition circumvented resistance and enhanced sensitivity to BH3 mimetics independently of autophagy. In summary, our findings demonstrate that modulation of glutamine metabolism and its downstream signaling pathways, namely reductive carboxylation, lipogenesis and cholesterogenesis, as well as inhibition of mTOR signaling could enhance the therapeutic efficacy of BH3 mimetic therapy thereby circumventing chemoresistance to BH3 mimetics (Figure 5D).

tions in cell lines to primary samples from patients, we used CLL cells isolated from patients during the lead-in period (L1D1) as well as cells from the same patients after five cycles of navitoclax therapy (C5D1), as previously detailed in Figure 1. Using these samples, we wanted to determine whether modulating glutamine metabolism would enhance apoptosis mediated by navitoclax. For this, we exposed CLL cells to CB-839 and simvastatin for 24 h followed by navitoclax for 4 h and assessed the extent of apoptosis. In agreement with our cell line data, both CB-839 and statins overcame the resistance to navitoclax-mediated apoptosis in primary CLL cells (Figure 6), supporting the therapeutic translatability of our data from cell lines to patients.

Discussion Targeting intermediary metabolism enhances sensitivity to navitoclax in primary samples from patients with chronic lymphocytic leukemia Our results indicate that targeting various facets of intermediary metabolism enhanced sensitivity to different BH3 mimetics in cell lines derived from relevant hematologic malignancies. To further extend our observa-

C

B

A

D

Anti-apoptotic BCL-2 family members are attractive drug targets both because of their high expression levels in several cancers and because of their well-characterized pro-survival roles. Even with extensive supportive in vitro data, the use of BH3 mimetics in treating cancer patients is still in its infancy, with venetoclax, a BCL-2-specific

E

F

Figure 4. Inhibition of lipogenesis and cholesterogenesis enhances sensitivity to BH3 mimetics. (A) Scheme representing reductive carboxylation, lipogenesis and cholesterogenesis. (B) Apoptotic sensitivity of K562 resistant [E] cells exposed to A-1331852 (10 nM) for 4 h was restored following genetic knockdown for 72 h of key enzymes in fatty acid synthesis. Western blots confirmed the knockdown efficiency of the different short interfering (si) RNA. (C) Apoptotic sensitivity of K562 resistant [E] cells exposed to A-1331852 (10 nM) for 4 h was restored following pharmacological inhibition of key enzymes in fatty acid synthesis using SB204990 (1 mM) for 72 h or GSK2194069 (100 nM) for 48 h. (D) Metabolic supplementation of K562 sensitive [A] and resistant [E] cells with palmitate (50 mM) for 48 h prior to the exposure of cells to GSK2194069 (100 nM) overcame the sensitizing effect of GSK2194069 on A-1331852-mediated apoptosis. (E) Genetic knockdown for 72 h of HMGR or (F) pharmacological inhibition of HMGR by simvastatin (250 nM) for 72 h, atorvastatin (10 mM) for 48 h or pitavastatin (1 mM) for 72 h. ***P⊽0.001; **P⊽0.01. Error bars = mean ¹ standard error of mean (n=3). PS: phosphatidylserine; DMSO: dimethylsulfoxide.

1022

haematologica | 2019; 104(5)

Modulation of intermediary metabolism in cancer therapy

A

B

C

D

Figure 5. Modulation of mammalian target of rapamycin signaling enhances sensitivity to BH3 mimetics independently of autophagy. (A) Apoptotic sensitivity of K562 resistant [E] cells exposed to A-1331852 (10 nM) for 4 h was restored following pharmacological inhibition of mTOR signaling using rapamycin (100 nM) or torin-1 (10 nM) for 16 h. (B) Inhibition of mTOR-regulated autophagy using 3-MA (10 mM) or bafilomycin A1 (100 nM) for 1 h, followed by torin-1 (10 nM) for a further 16 h, resulted in varying effects on A-1331852-mediated apoptosis. (C) Genetic knockdown of ATG5 and ATG7 for 72 h failed to revert torin-1 (10 nM)-mediated sensitization of apoptosis in K562 resistant [E] cells, following A-1331852 (10 nM) for 4 h. Western blots confirmed the knockdown efficiency of ATG5 and ATG7 short interfering (si) RNA. ***P⩽0.001. Error bars = mean ± standard error of mean (n=3). (D) Scheme representing glutamine uptake by SLC1A5 (inhibited by GPNA), glutaminolysis (inhibited by CB-839) to generate α-ketoglutarate, reductive carboxylation of a-ketoglutarate to generate citrate, which produces acetyl-CoA by a reaction catalyzed by ACLY (inhibited by SB204990), which eventually results in lipogenesis (inhibited by GSK2194069) and cholesterogenesis (inhibited by statins).

Glutamine uptake, metabolism and its downstream signaling cascade can feed into mTOR signaling (inhibited by torin-1), all of which promote cell growth. In this study, we demonstrate that modulation of these distinct intermediary metabolic pathways could successfully sensitize cancer cells to BH3 mimetic-mediated apoptosis. PS: phosphatidylserine; DMSO: dimethylsulfoxide; TCA: tricarboxylic acid.

A

haematologica | 2019; 104(5)

B

Figure 6. Inhibition of glutaminase and HMG-CoA reductase circumvents resistance to navitoclax-mediated apoptosis in primary chronic lymphocytic leukemia cells. Chronic lymphocytic leukemia cells isolated from five patients during the initial lead-in-period (L1D1) or day 1 of cycle 5 (C5D1) were cultured ex vivo on a feeder layer for 24 h and then exposed for a further 24 h to (A) CB-839 (50 nM) or (B) simvastatin (10 nM), and removed from the feeder layer for further exposure to navitoclax (50 nM) for 4 h. The extent of apoptosis was assessed as before. *P⩽0.05.

Error bars = mean ± standard error of mean (n=5). PS: phosphatidylserine; DMSO: dimethylsulfoxide.

1023

A. Al-Zebeeby et al.

inhibitor, only recently having received approval for treatment of refractory CLL.4 The development of BH3 mimetics to target BCL-XL and MCL-1 in patients will be extremely valuable in the treatment of several types of cancer. However potential mechanisms of resistance to BH3 mimetics need to be recognized as they emerge and ways to circumvent resistance identified. Several resistance mechanisms, including mutations of the target site,29 post-translational modifications,30,31 and elevated levels of anti-apoptotic BCL-2 family members,8,11,32,33 have already been identified. While some of these resistance mechanisms could be overcome by co-administration of other specific BH3 mimetics that target BCL-XL and/or MCL-1,57 such inhibitors are not yet clinically available and the potential toxicities associated with the simultaneous inhibition of multiple BCL-2 family members are not known.

Attempts to identify measures that could overcome chemoresistance have led to exploration of the therapeutic potential of modulating intermediary metabolism in BH3 mimetic-mediated apoptosis.19,20,34 Although the mechanisms by which glutamine could regulate cancer cell proliferation have been extensively studied, the interrelationship between glutamine metabolism and apoptosis requires further study. It has been previously reported that glutamine-mediated apoptosis is dependent on Myc14 and that c-Myc activates glutaminolysis by upregulating both the glutamine transporter, SLC1A5, and glutaminase, GLS-1.35,36 However, we were unable to detect an increase in expression levels of Myc, SLC1A5 or GLS-1 in our resistance models (Figure 3 and data not shown). The ability of glutamine to regulate apoptosis and/or chemoresistance could also be due to its regulatory effect on mitochondrial oxidative phosphorylation.20

References 1 Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144 (5):646–674.

2 Tse C, Shoemaker AR, Adickes J, et al. ABT263: a potent and orally bioavailable Bcl-2 family inhibitor. Cancer Res. 2008;68(9): 3421–3428.

3 Souers AJ, Leverson JD, Boghaert ER, et al.

ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing platelets. Nat Med. 2013;19(2):202– 208.

4 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.

5 Leverson JD, Phillips DC, Mitten MJ, et al.

Exploiting selective BCL-2 family inhibitors to dissect cell survival dependencies and define improved strategies for cancer therapy. Sci Transl Med. 2015;7(279):279ra40.

6 Leverson JD, Zhang H, Chen J, et al. Potent and selective small-molecule MCL-1 inhibitors demonstrate on-target cancer cell killing activity as single agents and in combination with ABT-263 (navitoclax). Cell Death Dis. 2015;6:e1590.

7 Kotschy A, Szlavik Z, Murray J, et al. The MCL1 inhibitor S63845 is tolerable and effective in diverse cancer models. Nature.

2016;538(7626):477–482.

1024

Although we do not entirely understand how glutamine metabolism impinges on apoptosis at this point, our data strongly support the notion that modulating glutamine metabolism and its related signaling pathways, such as reductive carboxylation, lipogenesis, cholesterogenesis and mTOR signaling, could enhance BH3 mimetic-mediated apoptosis in several hematologic malignancies (Figures 3-6). This is particularly promising, as glutaminase inhibitors, such as CB-839 and related drugs are already in clinical trials for the treatment of several malignancies20,37 and other drugs targeting cholesterogenesis, such as statins are the most commonly prescribed drugs to millions of people worldwide. While this manuscript was in preparation, an independent study comparing a large cohort of CLL patients, many of whom were statin users, found that response to venetoclax/ ABT-199 was enhanced among statin users in three different clinical trials.44 These findings highlight the possibility of repurposing several drugs targeting the intermediary metabolic pathways in conjunction with BH3 mimetic therapy to enhance therapeutic effectiveness and overcome the emerging chemoresistance in several cancers.

Acknowledgments We thank AbbVie for inhibitors and Prof. J. Borst for antibodies. This work was supported by a NorthWest Cancer Research grant CR1040 (to SV and GMC), a studentship from the Ministry of Higher Education and Scientific Research and the University of Al-Qadisiyah, Iraq (fpr AA-Z), a Science Without Borders studentship, CNPq 233624/2014-7, from the Ministry of Education, Brazil (for MM) and a studentship from the Prince Sattam Bin Abdulaziz University, Saudi Arabia (for AA).

8 van Delft MF, Wei AH, Mason KD, et al. The BH3 mimetic ABT-737 targets selective Bcl-2 proteins and efficiently induces apoptosis via Bak/Bax if Mcl-1 is neutralized. Cancer Cell. 2006;10(5):389–399.

9 Zhang H, Guttikonda S, Roberts L, et al.

Mcl-1 is critical for survival in a subgroup of non-small-cell lung cancer cell lines.

Oncogene. 2010;30(16):1963–1968.

10 Gores GJ, Kaufmann SH. Selectively targeting Mcl-1 for the treatment of acute myelogenous leukemia and solid tumors. Genes Dev. 2012;26(4):305–311.

11 Vogler M, Butterworth M, Majid A, et al.

Concurrent up-regulation of BCL-XL and BCL2A1 induces approximately 1000-fold resistance to ABT-737 in chronic lymphocytic leukemia. Blood. 2009;113(18):4403–4413.

12 Tahir SK, Smith ML, Hessler P, et al.

Potential mechanisms of resistance to venetoclax and strategies to circumvent it. BMC Cancer. 2017;17(1):399.

13 Bose P, Grant S. Mcl-1 as a therapeutic target in acute myelogenous leukemia (AML). Leuk Res Rep. 2013;2(1):12-14.

14 Yuneva M, Zamboni N, Oefner P, Sachidanandam R, Lazebnik Y. Deficiency in glutamine but not glucose induces MYCdependent apoptosis in human cells. J Cell Biol. 2007;178(1):93–105.

15 Wise DR, Thompson CB. Glutamine addiction: a new therapeutic target in cancer.

Trends Biochem Sci. 2010;35(8):427–433.

16 Graham NA, Tahmasian M, Kohli B, et al.

17

18 19

20

21

22

23 24

Glucose deprivation activates a metabolic and signaling amplification loop leading to cell death. Mol Syst Biol. 2012;8:589.

Son J, Lyssiotis CA, Ying H, et al. Glutamine supports pancreatic cancer growth through a KRAS-regulated metabolic pathway. Nature.

2013;496(7443):101–105.

Still ER, Yuneva MO. Hopefully devoted to Q: targeting glutamine addiction in cancer.

Br J Cancer. 2017;116(11):1375–1381.

Bajpai R, Matulis SM, Wei C, et al. Targeting glutamine metabolism in multiple myeloma enhances BIM binding to BCL-2 eliciting synthetic lethality to venetoclax. Oncogene.

2016;35(30):3955–3964.

Jacque N, Ronchetti AM, Larrue C, et al.

Targeting glutaminolysis has antileukemic activity in acute myeloid leukemia and synergizes with BCL-2 inhibition. Blood.

2015;126(11):1346–1356.

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):890–901.

Altman BJ, Stine ZE, Dang CV. From Krebs to clinic: glutamine metabolism to cancer therapy. Nat Rev Cancer. 2016;16(11):619– 634.

Al-Khallaf H. Isocitrate dehydrogenases in physiology and cancer: biochemical and molecular insight. Cell Biosci. 2017;7:37.

Zaidi N, Swinnen JV, Smans K. ATP-citrate lyase: a key player in cancer metabolism.

Cancer Res. 2012;72(15):3709–3714.

haematologica | 2019; 104(5)

Modulation of intermediary metabolism in cancer therapy

25 Khwairakpam AD, Shyamananda MS, Sailo BL, et al. ATP citrate lyase (ACLY): a promising target for cancer prevention and treatment. Curr Drug Targets. 2015;16(2):156– 163.

26 Menendez JA, Lupu R. Fatty acid synthase and the lipogenic phenotype in cancer pathogenesis. Nat Rev Cancer. 2007;7(10): 763–777.

27 Mashima T, Seimiya H, Tsuruo T. De novo fatty-acid synthesis and related pathways as molecular targets for cancer therapy. Br J Cancer. 2009;100(9):1369–1372.

28 Saxton RA, Sabatini DM. mTOR signaling in growth, metabolism, and disease. Cell.

2017; 168(6):960–976.

29 Fresquet V, Rieger M, Carolis C, GarcíaBarchino MJ, Martinez-Climent JA. Acquired mutations in BCL2 family proteins conferring resistance to the BH3 mimetic ABT-199 in lymphoma.

Blood.

2014;123(26): 4111–4119.

30 Konopleva M, Contractor R, Tsao T, et al.

Mechanisms of apoptosis sensitivity and resistance to the BH3 mimetic ABT-737 in acute myeloid leukemia. Cancer Cell.

2006;10(5):375–388.

31 Mazumder S, Choudhary GS, Al-Harbi S, Almasan A. Mcl-1 phosphorylation defines ABT-737 resistance that can be overcome by increased NOXA expression in leukemic B cells. Cancer Res. 2012;72(12):3069–3079.

haematologica | 2019; 104(5)

32 Chen S, Dai Y, Harada H, Dent P, Grant S. Mcl-1 down-regulation potentiates ABT-737 lethality by cooperatively inducing Bak activation and Bax translocation. Cancer Res. 2007;67(2):782–791.

33 Lin KH, Winter PS, Xie A, et al. Targeting MCL-1/BCL-XL forestalls the acquisition of resistance to ABT-199 in acute myeloid leukemia. Sci Rep. 2016;6:27696.

34 Chan SM, Thomas D, Corces-Zimmerman MR, et al. Isocitrate dehydrogenase 1 and 2 mutations induce BCL-2 dependence in acute myeloid leukemia. Nat Med. 2015;21(2):178–184.

35 Wise DR, DeBerardinis RJ, Mancuso A, et al.

Myc regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction. Proc Natl Acad Sci U S A. 2008;105(48):18782–18787.

36 Liu W, Le A, Hancock C, et al.

Reprogramming of proline and glutamine metabolism contributes to the proliferative and metabolic responses regulated by oncogenic transcription factor c-MYC. Proc Natl Acad Sci U S A. 2012;109(23):8983–8988.

37 Vander Heiden MG, DeBerardinis RJ. Understanding the intersections between metabolism and cancer biology. Cell.

2017;168(4):657–669.

38 Vogler M, Weber K, Dinsdale D, et al.

Different forms of cell death induced by putative BCL2 inhibitors. Cell Death Differ.

2009;16(7):1030–1039.

39 Vogler M, Dinsdale D, Dyer MJS, Cohen GM. ABT-199 selectively inhibits BCL2 but not BCL2L1 and efficiently induces apoptosis of chronic lymphocytic leukaemic cells but not platelets. Br J Haematol. 2013; 163(7):139–142.

40 Vogler M, Furdas SD, Jung M, Kuwana T, Dyer MJS, Cohen GM. Diminished sensitivity of chronic lymphocytic leukemia cells to ABT-737 and ABT-263 due to albumin binding in blood. Clin Cancer Res. 2010;16(16): 4217–4225.

41 Roberts AW, Seymour JF, Brown JR, et al.

Substantial susceptibility of chronic lymphocytic leukemia to BCL2 inhibition: results of a phase I study of navitoclax in patients with relapsed or refractory disease.

J Clin Oncol. 2012;30(5):488–496.

42 Varadarajan S, Poornima P, Milani M, et al.

Maritoclax and dinaciclib inhibit MCL-1 activity and induce apoptosis in both a MCL-1-dependent and -independent manner. Oncotarget. 2015;6(14):12668–12681.

43 Lucas CM, Milani M, Butterworth M, et al.

High CIP2A levels correlate with an antiapoptotic phenotype that can be overcome by targeting BCL-XL in chronic myeloid leukemia. Leukemia. 2016;30(6):1273–1281.

44. Lee JS, Roberts A, Juarez D, et al. Statins enhance efficacy of venetoclax in blood cancers. Sci Transl Med. 2018;10(445).

1025

ARTICLE Ferrata Storti Foundation

Plasma Cell Disorders

Efficacy of first-line treatments for multiple myeloma patients not eligible for stem cell transplantation: a network meta-analysis Hedwig M. Blommestein,1,2* Chrissy H.Y. van Beurden-Tan,3* Margreet G. Franken,1 Carin A. Uyl-de Groot,1,2 Pieter Sonneveld3 and Sonja Zweegman4 *HB and CvBT contributed equally to this work

Haematologica 2018 Volume 104(5):1026-1035

Erasmus School of Health Policy & Management, Institute for Medical Technology Assessment, Erasmus University Rotterdam; 2Comprehensive Cancer Organisation, Utrecht; 3Erasmus MC Cancer Institute, Rotterdam and 4Department of Hematology, Amsterdam UMC, the Netherlands 1

ABSTRACT

D

Correspondence: HEDWIG M. BLOMMESTEIN blommestein@eshpm.eur.nl Received: September 14, 2018.

Accepted: January 2, 2019.

Pre-published: January 3, 2019. doi:10.3324/haematol.2018.206912 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1026

ecision making for patients with multiple myeloma (MM) not transplant eligible (NTE) is complicated by a lack of head-tohead comparisons of standards of care, the increase in the choice of treatment modalities, and the promising results that are rapidly evolving from studies with novel regimens. To support evidence-based decision making, we performed a network meta-analysis for NTE MM patients that synthesizes direct and indirect evidence and enables a comparison of all treatments. Relevant randomized clinical trials were identified by a systematic literature review in EMBASE®, MEDLINE®, MEDLINE®-in-Process and the Cochrane Central Register of Controlled Trials for January 1999 to March 2016. Efficacy outcomes [i.e. the hazard ratio (HR) and 95% confidence interval (95%CI) for progression-free survival] were extracted and synthesized in a random effects network-meta analysis. In total, 24 studies were identified including 21 treatments.

According to the network-meta analysis, the HR for progression-free survival was favorable for all NTE MM treatments compared to dexamethasone (HR: 0.19-0.90). Daratumumab-bortezomib-melphalan-prednisone and bortezomib-melphalan-prednisone-thalidomide with bortezomib-thalidomide maintenance were identified as the most effective treatments (HR: 0.19, 95%CI: 0.08-0.45 and HR: 0.22, 95%CI: 0.10-0.51, respectively). HR and 95%CI for currently recommended treatments, bortezomib-lenalidomide-dexamethasone, bortezomib-melphalanprednisone, and lenalidomide-dexamethasone compared to dexamethasone, were 0.31 (0.16-0.59), 0.39 (0.20-0.75), and 0.44 (0.29-0.65), respectively. In addition to identifying the most effective treatment options, we illustrate the additional value and evidence of network meta-analysis in clinical practice. In the current treatment landscape, the results of network meta-analysis may support evidence-based decisions and ultimately help to optimize treatment and outcomes of NTE MM patients.

©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.

1026

Introduction Multiple myeloma (MM) is a hematologic disease characterized by the proliferation of malignant plasma cells, causing disease-related symptoms such as anemia, hypercalcemia, renal and bone disease. The age-standardized incidence rate is 4.5 per 100,000.1 Incidence increases with age and two-thirds of the patients diagnosed with MM are over 65 years of age.2 The treatment armamentarium has been greatly increased in the last decade, with novel proteasome inhibitors (PIs), immunomodulatory drugs (IMiDs), and monoclonal antibodies now being incorporated in first-line treatment regimens, which have considerably improved progression-free survival (PFS) and overall survival (OS) of MM. Given the median age haematologica | 2019; 104(5)

Efficacy of MM treatments

of 70 years at diagnosis, the majority of newly diagnosed (ND) MM patients are not transplant eligible (NTE) for stem cell transplant (SCT). Current standards of care for NTE NDMM patients are bortezomib-melphalan-prednisone (VMP), lenalidomide-dexamethasone (Rd), and in the USA bortezomib-Rd (Vrd),3 supported by randomized phase III trials.4-6 Recently, better PFS was demonstrated for daratumumab-VMP (DaraVMP) compared to VMP.7 Although randomized clinical trials (RCTs) remain the gold standard to define standards of care, we predict that in the current treatment landscape the role of network meta-analysis (NMA) will become increasingly important.

Firstly, today there is more than one standard of care, but a randomized study between two registered standards of care is highly unlikely to be performed because of the reluctance of pharmaceutical industries to support such studies.8 Therefore, head-to-head comparisons of VMP versus Rd or VRd versus VMP are not likely to be initiated.9 NMA can help to discriminate between efficacy of nonhead-to-head compared regimens. Secondly, with the increase in the number of treatment modalities, the number of smaller randomized Phase II studies is expected to increase at the cost of Phase III RCTs. NMA provides more solid estimates of treatment effects by combining RCTs that provide direct and indirect evidence for effectiveness and allows competing treatments to be ranked.10 Thirdly, with the high number of studies currently enrolling patients, standard of care arms are expected to change within short time frames.8 This hampers the development of classical phase III trials, as at the end of the study it might appear that the standard arm of the study no longer reflects the reality in the clinic. Finally, the heterogeneous biological characteristics of MM and the clonal evolution of the disease will lead to studies with a smaller sample size that will not allow for randomization, increasing the need for indirect comparisons.

There are currently two systematic literature reviews (SLRs) and NMAs available for first-line NTE NDMM treatments.11,12 Due to the timing of their searches and selection criteria, these reviews did not, however, include all currently available treatments (e.g. VRd, VMPT-VT, DaraVMP) and RCT evidence (e.g. HOVON87 comparing MPT-T and MPR-R13). To support evidence-based decision making in clinical practice, we performed an SLR and an NMA synthesizing all direct and indirect evidence from phase III RCTs that is currently available and compared the outcome of all treatment options for NTE NDMM patients.

Methods Systematic literature review An SLR was conducted in the databases EMBASE®, MEDLINE®, MEDLINE®-in-Process and the Cochrane Central Register of Controlled Trials for the period January 1, 1999 to March 1, 2016 to identify relevant studies (Online Supplementary Appendix 1). Studies were included if they described a Phase III RCT among newly diagnosed adult patients with MM. Furthermore, one of the pre-specified treatments (Online Supplementary Appendix 2) had to be part of the regimens of the RCT. After removing duplicates, citations were first screened on the basis of title and abstract and then screened on the contents of their full text. Citations were excluded due to the following reasons: not in English, review, study phase, intervention, dishaematologica | 2019; 104(5)

ease, study design, meta-analysis, patient population, economic outcomes, meta-analysis, and other. (For a detailed description of the exclusion categories see Online Supplementary Appendix 2).

To incorporate the latest clinical developments, the publication of the pre-specified interim analysis of the phase III ALCYONE RCT comparing DaraVMP to VMP7 was added as additional record.

Data extraction Data were extracted on trial details (i.e. publication source, trial ID, trial number, research, and comparator treatment(s), number of patients, median age, and primary outcome, and follow up) and efficacy outcomes. Efficacy outcomes included PFS and OS. For OS we obtained median survival. For PFS we obtained the median survival, 95% confidence interval (CI) and hazard ratio (HR) and 95%CI of the HR. In cases in which HRs and/or 95%CI for PFS were not reported, we estimated the missing data with the available Kaplan-Meier curves using the methods described by Tierney et al.14 In cases in which multiple sources reported on the same trial, the most recently published PFS data were extracted. Risk of bias in randomized trials was assessed using the Cochrane Collaboration tool15 (Online Supplementary Appendix 3).

Network meta-analysis A network was made from the identified treatment options in the SLR. It includes the HRs for PFS from the trials for treatments that were compared head-to-head. A comparison between all treatments can be made based on a common comparator (i.e. reference treatment). The choice of the reference treatment does not influence the outcomes of the study and final results can be presented relative to all included treatments. The oldest treatment (i.e. dexamethasone) was selected as a reference treatment from which the relative effectiveness of all treatments was estimated.

We performed a similar analysis with MPT as reference treatment, given that this regimen was used as (comparator) treatment in several RCTs. Treatments were sorted based on their P-score.

This P-score measures the average proportion of treatments worse than the respective treatment where 1 means theoretically the best and 0 means the worst.16 To conduct an NMA for 2- and multi-arm studies, we used the netmeta package v.0.9-7 in R version 3.3.1 (Online Supplementary Appendix 4). We ran a random effects model assuming that the included studies represent a random sample of effect sizes that could have been observed and that the effect can best be estimated by the mean of all available studies. A random effects model was deemed appropriate since there were multiple trials available for some comparisons (e.g. MPT with MP) and sampling error was not considered to be the most plausible explanation for the observed variation. With a random effects model we allow for differences in the patient population and implementations of interventions.17 The netmeta package uses a frequentist approach based on the graph-theoretical methods routinely applied in electrical networks.18,19 In contrast to the Bayesian approach that produces credible intervals, analysis based on the frequentist approach produces 95%CIs and, as all CIs, these should be interpreted as follows: 95% of the produced CIs would contain the true value if the analysis were repeated many times.20 Face-validity of the NMA results was checked by comparing the computed HRs by the NMA with the HRs reported in the publications of the trials. To validate our outcomes to a previously reported NMA,12 we performed a scenario analysis with different treatment groups (separating MPT and MPT-T) and a scenario with a limited number of studies. In the third scenario analysis, we used a fixed effect model instead of a random 1027

H.M. Blommestein et al.

Table 1. Data extraction of the included trials.

Trial reference Trial ID NCT number

Primary Randomized / Treatment Median outcome enrolled age research patients treatment years (range)

Facon 2006

OS

500

IFM 95/01 n/r

447

IFM 99–06 NCT00367185

Morgan 2013

PFS, OS

MRC M IX ISRCTN68454111 Rajkumar 2008

856

470

MM-003 NCT00057564 Ludwig 2009

12.2 (10.2-14.2)

MP

70 (68-72) 69 (68-72) 69 (67-72) n/r (65-75¹)

122

21.1 (17.8-24.4) 22.9 (19.0-26.8) 15.2 (9.9-20.5) 27.5 (23.4-31.6)

0.75 (0.62-0.91) 1.15 (0.93-1.42) {MP vs. D} {MP vs. MD} 0.66 (0.53-0.81) 1.45 (1.17-1.79) {MD vs. D} {DI vs MD} 0.92 (0.76-1.11) 1.26 (1.04-1.53) {DI vs. D} {DI vs. MP} 0.59 (0.44-0.78) {MPT vs. M100}

MPT

289

NCT00205751 Palumbo 2008

331

GIMEMA NCT00232934 Hulin 2009 IFM 01/01 Trial n/r Waage 2010 NMSG NCT00218855 Beksac 2010 TMSG NCT00934154 Wijermans 2010 HOVON-49 ISRCTN90692740 Sacchi 2011

OS

232

363

NCT01532856

32

82.8

51.6

51.5

M100

n/r (65-75³) 73 (57-89) 73 (58-87) 64 (39-86) 64 (31-84)

126

19.4 0.87 (0.68-1.1) (17.4-21.4) {M100 vs. MP} 12 0.81 (0.69-0.94) n/r {CTD(a) vs. MP} 13 n/r 14.9 0.5 (0.38-0.64) n/r {TD vs. D} 6.5 n/r

38.3

51.5

32

70.8

34

70.8

NR

17

30

18

145

28.1

49.4

28.1

167

16.7 n/r 20.7 n/r 21.8 (19.6-26.1)

41.5

MPT-T

72 (54-86) 72 (55-86) 72

45

38.4

MP

72

164

47.6

37.7

MPT

79 (75-89)

115

44

47.5

29.1

47.5

29

42

32

42

26

35

28 40

23 39

31

39

52

30

32

30

32.4

37.5

54.6

37.5

42

37.5

MP

TD

TD

423 426 235 235

144

117

MPT-T

75

184

MP

74

179

69

60

EFS

344

MP MPT-T

72 72 (65-87) 73 (65-84) 76 (66–89) MP

62 171

MP

82

82.8

51.5

MPT

ORR

39.6

33.2

122

135

82.8

0.51 (0.39-0.66) {MPT vs. MP}

Treatment response, toxicities

n/r

34

17.8 (15.1-20.5)

MPT

n/r Hungria 2016

125

82.8

196

MP OS

121

33.4

n/r (65-75²)

MP RR, PFS

118

Median follow up

MP

D

PFS, tolerance

Median OS

127

CTDa TTP

HRs (95% CI) {research vs. comparator treatment}

70 (67-73)

DI OS

Median PFS 95% CI

D

MD

Facon 2007

N itt

173 70

CTD

70

79 (68–88) 32

TD

72

18

MPT

72

32

14.5 (12.2-17) 24.1 (19.4-29) 18.5 (14.6-21.3) 15 (12-19) 14 (11-18) n/r

n/r 15 n/r 11 n/r 33 n/r 65 25.9 n/r 21.5 n/r 38.5

1.3 (0.95-1.78) {TD vs. MP}

0.63 (0.48-0.81) {MPT vs. MP}

0.61 (0.46-0.82) {MPT vs. MP}

0.89 (0.7-1.13) {MPT vs. MP}

0.7 (0.42-1.17) {MPT vs. MP}

0.79 (0.62-1) {MPT vs. MP}

0.67 (0.38-1.18) {MPT vs. MP} 22 n/r 0.89 (0.48-1.64) {MPT vs. CTD} 1.1 (0.53-2.31) {TD vs. CTD} 0.73 (0.34-1.59)

continued on the next page

1028

haematologica | 2019; 104(5)

Efficacy of MM treatments continued from the previous page

San Miguel 2008 VISTA NCT00111319 Mateos 2014 GEM2005 NCT00443235 Niesvizky 2015

TTP

682

VMP

RD

71 (57–90) 71 (48–91) 73 (69–76) 73 (68–77) 75 (67-79) 73 (66-77) 72 (68-77) 71 (68-75) 71 (68-75) n/r⁴

D

n/r⁵

99

1623

Rd

535

Rd18

73 (40–89) 73 (51–92) MPT-T

73 (44–91) 541

MP n/r

260

VTP VMP

PFS

502

UPFRONT NCT00507416

VD VTD VMP

Palumbo 2014 GIMEMA0305 NCT01063179 Zonder 2011

PFS

511

VMP PFS

S0232 NCT00064038 Benboubkher 2014 PFS FIRST/MM-020 NCT00689936

198

MPT Zweegman 2016

PFS

HOVON-87 EUDRACT 2007-004007-34 Stewart 2015 PFS

568

MPR-R 306

ECOG E1A06 NCT00602641 Magarotto 2016

VMPT-VT

MPT-T

MPR-R PFS

654

EMN01 NCT01093196

MPR-R

CPR Rd

Palumbo 2012

PFS

459

MM-015 NCT00405756

MPR-R MP MPR

Durie 2017ǁ SWOG S0777 NCT00644228 Mateos 2018

PFS

525

VRd Rd

PFS

706

ALCYONE NCT02195479

DaraVMP

VMP

344 338 130 130 168 167 167 254 257 99

547 72 (60-91) 73 (60-87) 76 (54-92)

280

77 (63-92) 74 (63-91)

152

73 (63-87) 73 (50-89) 71 (65–87) 72 (65–91) 71 (65–86) n/r (≥18⁶) n/r (≥18⁷) 71 (40-93) 71 (50-91)

280 154

218

222 222 152 154 153 264 261 350

356 (16.5-19.9)

n/r {MPT vs. TD} 21.7 0.56 (0.4-0.79) n/r {VMP vs. MP} 15.2 n/r 23 0.8 (0.61-1.04) n/r {VMP vs. VTP} 32 n/r 14.7 1.12 (0.83-1.51) (12-18.6) {VD vs. VTD} 15.4 0.89 (0.66-1.21) (12.6-24.2) {VTD vs. VMP} 17.3 1.11 (0.84-1.48) (14.8-20.3) {VD vs. VMP} 35.3 0.58 (0.47-0.71) n/r {VMPT-VT vs. VMP} 24.8 n/r 39 0.56 (0.39-0.79) (26-53) {RD vs. D} 15 (8-23) 25.5 0.97 (0.83-1.12) n/r {MPT vs. RD18} 20.7 1.43 (1.22-1.67) n/r {RD18 vs. RD} 21.2 1.39 (1.18-1.64) n/r {MPT vs. RD} 20 0.87 (0.72-1.04) (18-23) {MPR-R vs. MPT-T} 22 (19-27) 21 0.84 (0.64-1.09) (18-27) {MPT-T vs. MPR-R} 18.7 (16-22) 24 n/r 20 n/r 21 n/r 31 n/r 13 n/r 14 n/r 43 (39-52) 30 (25-39) NR

0.81 (0.63-1.03) {MPR-R vs. RD} 1.01 (0.9-1.13) {CPR vs. RD} 0.8 (0.63-1.02) {MPR-R vs. CPR} 0.49 (0.35-0.69) {MPR-R vs. MPR} 1.19 (0.94-1.5) {MP vs. MPR} 0.4 (0.29-0.54) {MPR-R vs. MP} 0.71 (0.56-0.91) {VRd vs. Rd}

0.50 (0.38-0.65) {DaraVMP vs. VMP}

56.4

60.1

43.1

60.1

43

72

63

72

49.8

44.3

51.5

41.3

53.1

43.4

NR

54

60.6

54

NR

45.4

NR

45.4

58.9

45.5

56.7

45.5

48.5

45.5

49

32.6

50

32.6

52.6

40.7₸

47.7 NR

39

NR

39

NR

39

56

53

52

53

54

53

52

54

38

56

NR

16.5

18.1 NR

CI: confidence interval; N: number; itt: intention to treat; n/r: not reported; NR: not reached; PFS: progression-free survival; OS: overall survival; TTP: time to progression; EFS: event-free survival; ORR: overall response rate; RR: response rate, D: Dexamethasone; DI: Dexamethasone-Interferon alpha; M100: Melphalan 100; MD: Melphalan-Dexamethasone; MP: MelphalanPrednisone; TD: Thalidomide-Dexamethasone; CTD: Cyclophosphamide-Thalidomide-Dexamethasone; CTD(a): Cyclophosphamide-Thalidomide-Dexamethasone (attenuated); Melphalan-Prednisone-Thalidomide/Melphalan-Prednisone-Thalidomide and Thalidomide maintenance (MPT/MPT-T); VD: Bortezomib-Dexamethasone; VTD: BortezomibThalidomide-Dexamethasone; VMP: Bortezomib-Melphalan-Prednisone; VTP: Bortezomib-Thalidomide-Prednisone; VMPT-VT: Bortezomib-Melphalan-Prednisone-Thalidomide and Bortezomib-Thalidomide; CPR: Cyclophosphamide-Prednisone-Lenalidomide; Rd: Lenalidomide-Dexamethasone; Rd18: 18 cycles Lenalidomide-Dexamethasone; MPR: MelphalanPrednisone-Lenalidomide; MPR-R: Melphalan-Prednisone-Lenalidomide and Lenalidomide maintenance; VRd: Bortezomib-Lenalidomide-Dexamethasone; DaraVMP: DaratumumabBortezomib-Melphalan-Prednisone; ¹40% ≥70 years ²43% ≥70 years ³39% ≥70 years.⁴ 49% ≥65 years.⁵ 47% ≥65 years.⁶ 38% ≥65 years.⁷ 48% ≥65 years Source HR: from published trial (MM-003, Ludwig 2009, GIMEMA, MRC-MIX, GIMEMA0305, HOVON87, S0777, E1A06, ALCYONE, IFM-99/06, EMN01, FIRST), obtained from a previous patient-level meta-analysis5 (IFM01/01, NMSG,TMSG, HOVON49), from a previous NMA15 (Sacchi 2011) and data on file from investigators (Hungria 2016). Calculations were made using the published HR and P value (VISTA), Kaplan-Meier curves (IFM95/01 and the MM-15) and P-value and number of events (GEM2005, Upfront, s0232). Table 1 presents the extracted and calculated data.

haematologica | 2019; 104(5)

1029

H.M. Blommestein et al.

effects model. Heterogeneity and inconsistency were assessed by decomposing the Q statistic21,22 and quantified by the I2-statistic,23 which presents the percentage of the variability in effects due to heterogeneity rather than chance.24

Results Systematic literature review Figure 1 presents the PRISMA flow diagram; the PRISMA checklist is presented in Online Supplementary Appendix 3. The SLR identified a total of 19,773 citations from the databases. One additional recent record was included (i.e. the ALCYONE trial7). After removing duplicates, 18,752 citations remained. Based on title and abstract, 17,741 citations were excluded for further analysis. The full texts of 1011 citations were reviewed and, based on this assessment, 944 citations were excluded. In the second full text

review of the remaining 67 citations, 43 citations were excluded because these did not report the most recent results (e.g. extended follow-up results were available). After the entire assessment, 24 RCTs remained and were included for data extraction and the NMA. See Figure 1 for further details of the reasons for exclusion.

These 24 RCTs included 21 treatment options: 1) Dexamethasone (D), 2) Dexamethasone-Interferon alpha (DI), 3) Melphalan 100 (M100), 4) Melphalan-Dexamethasone (MD), 5) Melphalan-Prednisone (MP), 6) Thalidomide-Dexamethasone (TD), 7) Cyclophosphamide-Thalidomide-Dexamethasone (CTD), 8) Cyclophosphamide-Thalidomide-Dexamethasone (attenuated) [CTD(a)], 9) Melphalan-Prednisone-Thalidomide/ Melphalan-Prednisone-Thalidomide and Thalidomide maintenance (MPT/MPT-T), 10) Bortezomib-Dexamethasone (VD), 11) Bortezomib-Thalidomide-Dexamethasone (VTD), 12) Bortezomib-Melphalan-Prednisone (VMP), 13) BortezomibThalidomide-Prednisone (VTP), 14) Bortezomib-Melphalan-

Figure 1. PRISMA 2009 flow diagram: transplant not eligible multiple myeloma (TNEMM) Phase III randomized controlled trials (RCTs). n: number. From Moher et al. 2009.52

1030

haematologica | 2019; 104(5)

Efficacy of MM treatments

Prednisone-Thalidomide and Bortezomib-Thalidomide (VMPTVT), 15) Cyclophosphamide-Prednisone-Lenalidomide (CPR), 16) Lenalidomide-Dexamethasone (Rd), 17) 18 cycles LenalidomideDexamethasone (Rd18), 18) Melphalan-Prednisone-Lenalidomide (MPR), 19) Melphalan-Prednisone-Lenalidomide and Lenalidomide maintenance (MPR-R), 20) BortezomibLenalidomide-Dexamethasone (VRd), 21) DaratumumabBortezomib-Melphalan-Prednisone (DaraVMP).

Data extraction Table 1 provides the details, and extracted and calculated data of the included trials. Most trials (21 out of 24) investigated iMIDbased regimens (thalidomide or lenalidomide). Since MP has been the standard treatment for decades,25 MP was the comparator in 12 trials. PFS was the primary end point for 13 trials. The median age of the patient population was reported by most trials and ranged from 64 to 79 years. While some trials included patients aged <65 years, either because of choosing broader age limits or because of including patients who were not eligible for SCT independent of age, most trials only included patients aged ≥65 years.

The IFM99-0626 and IFM01/0127 only focused on patients aged ≥70 and ≥75, respectively.

Network meta-analysis network All identified RCTs (n=24) and treatments (n=21) were incorporated within one network (Figure 2). We combined MPT and MPT-T. The duration of induction therapy with thalidomide var-

ied leading to a clear overlap in planned thalidomide use between protocols with and without maintenance, preventing a clear discrimination between MPT with and without thalidomide maintenance.

Figure 2 presents the obtained HR(s) from the trial(s) and the HR obtained from the NMA for each of the connections (i.e. treatment comparisons) in our network. In order to validate our data, we compared the HR from treatments for which only direct evidence from a single RCT was available. The HR obtained from the NMA should be equal to the HR obtained from the RCT. The HR from the NMA was indeed similar to the HR from the trials for six comparisons5-7,28-30 [i.e. CTD(a) vs. MP, VMP vs. MP, DaraVMP vs.

VMP, VRd vs. Rd, VMPT-VT vs. VMP and VMP vs. VTP] (Online Supplementary Appendix 5). In addition, our network includes several treatments for which both direct and indirect evidence were available. Online Supplementary Appendix 5 presents the HRs based on direct and indirect evidence and shows that none of the P-values for disagreement was lower than 0.05.

The percentage of the variability in effect estimates due to heterogeneity rather than sampling error (=I2) was 72% indicating substantial between-study heterogeneity (i.e. within the 50-90% range can be quantified as substantial heterogeneity24). We allowed for between-study heterogeneity by using the random effects model. Heterogeneity could be reduced by excluding some of the trials; however, because of a lack of valid reasons (e.g. patients' characteristics, treatment dosing or follow up) for excluding trials, we decided not to perform analyses of this kind.

Figure 2. Network of the studies included in the network meta-analysis (NMA). White boxes represent treatments and reference numbers using the following abbreviations. 1) Dexamethasone (D); 2) Dexamethasone-Interferon alpha (DI); 3) Melphalan 100 (M100); 4) Melphalan-Dexamethasone (MD); 5) Melphalan-Prednisone (MP); 6) Thalidomide-Dexamethasone (TD); 7) Cyclophosphamide-Thalidomide-Dexamethasone (CTD); 8) Cyclophosphamide-Thalidomide-Dexamethasone (attenuated) [CTD(a)]; 9) Melphalan-Prednisone-Thalidomide / Melphalan-Prednisone-Thalidomide and Thalidomide maintenance (MPT/MPT-T); 10) BortezomibDexamethasone (VD); 11) Bortezomib-Thalidomide-Dexamethasone (VTD); 12) Bortezomib-Melphalan-Prednisone (VMP); 13) VTP: Bortezomib-ThalidomidePrednisone (VTP); 14) Bortezomib-Melphalan-Prednisone-Thalidomide and Bortezomib-Thalidomide (VMPT-VT); 15) Cyclophosphamide-Prednisone-Lenalidomide (CPR); 16) Lenalidomide-Dexamethasone (Rd); 17) 18 cycles Lenalidomide-Dexamethasone (Rd18); 18) Melphalan-Prednisone-Lenalidomide (MPR); 19) MelphalanPrednisone-Lenalidomide and Lenalidomide maintenance (MPR-R); 20) Bortezomib-Lenalidomide-Dexamethasone (VRd); 21) Daratumumab-Bortezomib-MelphalanPrednisone (DaraVMP). Black box represents the reference treatment in the network meta-analysis. Gray boxes include the trial reference and hazard ratio (HR) for progression-free survival on the top row(s). Bottom row shows HR according to the NMA. *HR not statistically significant at 5%.

haematologica | 2019; 104(5)

1031

H.M. Blommestein et al.

Results versus dexamethasone Figure 3 presents the HRs with the corresponding 95%CI for PFS and the P-score of the NMA in which dexamethasone was used as comparator for the remaining 20 “other treatment” options. HRs above 1 indicate that the “other treatment” is less effective than the comparator treatment, dexamethasone; HRs below 1 indicate that the “other treatment” is more effective than dexamethasone. All first-line NTE NDMM treatment options were better compared to the reference treatment dexamethasone (i.e. reducing the risk of progression or death compared to dexamethasone). HRs ranged between 0.19-0.90; however, not all treatments were statistically significantly different from dexamethasone, because of wide 95%CIs. DaraVMP and VMPT-VT were identified as the most effective treatment options as they had the highest and almost similar P-scores (i.e. a 96% and 93% certainty that this treatment is better than another treatment, averaged over all competing treatments) and most favorable relative treatment effects compared to dexamethasone (i.e. HR: 0.19, 95%CI: 0.080.45 and HR 0.22, 95%CI: 0.10-0.51 for DaraVMP and VMPT-VT, respectively). The HRs and 95%CIs for currently recommended treatments, VRd, VMP and Rd compared to dexamethasone, were 0.31 (95%CI: 0.16-0.59), 0.39 (95%CI: 0.20-0.75), and 0.44 (95%CI: 0.29-0.65), respectively. Selecting MPT as a reference treatment does not change the hierarchy of the treatments as the

P-score values do not change if one considers a different reference treatment. Compared to MPT, only DaraVMP had a statistically lower HR for PFS (HR 0.41, 95%CI: 0.19-0.91; P<0.05) (Online Supplementary Appendix 6).

Scenario analysis network meta-analysis In order to rule out that grouping of MPT and MPT-T would affect the outcome of the analysis, we performed a scenario in which we grouped IFM 01/01, IFM 99/06 and Sacchi et al. 2011, as MPT and GIMEMA, HOVON49, TMSG and NMSG as MPT-T. The MPT-T group was connected in the network to the MPT-T arm from the HOVON87 trial and the ECOG E1A06 trial. Overall, the results were comparable to the base case (Online Supplementary Appendix 7). We found similar results for MPT (HR 0.46, 95%CI: 0.30-0.71) and MPT-T (HR 0.47, 95% CI 0.30-0.73) compared to D. The second scenario, based on the trials included by Weisel et al.,12 showed lower HRs for PFS for Rd compared to VMP, MPT and MP, but the 95%CI for VMP overlapped with Rd [Rd vs. VMP: HR 0.73, 95%CI: 0.48-1.11 (Online Supplementary Appendix 8)].

Results from the third scenario analysis (fixed effect model instead of random effects model) are presented in Online Supplementary Appendix 9. While the HRs from the fixed effect model are quite similar, the 95%CIs are smaller, as is typical for fixed effect models.

Figure 3. Results of the network meta-analysis in which dexamethasone was used as comparator. HR: hazard ratio; CI: confidence interval. For abbreviations for treatments, see Figure 2.

1032

haematologica | 2019; 104(5)

Efficacy of MM treatments

Discussion Current clinical decision making in MM is complicated by a lack of head-to-head comparisons of standards of care, an increasing number of treatment modalities, and rapidly evolving promising results of studies with novel regimens (among smaller subpopulations). In this treatment landscape, we believe the role of NMA will become increasingly important, although it cannot replace RCTs, and these will still represent the gold standard.

Firstly, NMAs are able to provide data where head-tohead comparisons are lacking.20,24 For NTE NDMM, there have been no head-to-head comparisons between the current three standard of care regimens (i.e. VRd, VMP and Rd). Only VRd has been compared head-to-head with Rd, and there are no studies comparing VMP with VRd or Rd. With our NMA, we show that the HR of VRd was lower than VMP and Rd, and VRd also had the highest P-score.

We present similar HRs and P-scores for VMP and Rd. However, we also show considerable overlap of the 95%CIs of VRd, Rd and VMP. Our NMA does not support the use of one over the other regimens, thus leaving three valuable options for clinical practice. The choice of therapy will be guided by the patient's characteristics. For example, a proteasome inhibitor (PI)-based regimen in high-risk cytogenetic disease, and a preference for lenalidomide without bortezomib in patients with neuropathy.31-34 According to the ranking based on their P-scores and comparative effectiveness estimates, DaraVMP and VMPT-VT were identified as the most effective treatments. Although there is one RCT already showing better PFS and OS28 for VMPT-VT when compared with VMP, we now add data showing comparable efficacy to DaraVMP, which is expected to become an important standard of care. This finding is important given the pronounced differences in global access to expensive treatment regimens. As all drugs in the VMPT-VT regimen will soon be available as generic compounds, this regimen is a valuable option in clinical practice as well. In addition, the pronounced efficacy of VMPT-VT highlights the use of maintenance therapy following PI-based induction regimens. In addition, in a non-head-to-head comparison with VMP, the study of the PETHEMA group showed that maintenance therapy did result in a substantially longer PFS.35 We now add further evidence for maintenance therapy with PIs by showing high efficacy of VMPT-VT as compared to VMP. This is important because the European Medicines Agency has still not approved maintenance therapy with bortezomib, given that no head-tohead comparisons of maintenance versus no maintenance therapy have been made.

Secondly, NMAs provide more solid and precise effectiveness estimates when head-to-head data from multiple RCTs are available.20,24 Our network included several trials investigating MPT/MPT-T versus MP. Some of these trials showed superiority of MPT/MPT-T over MP,26,27,36 while other trials found no difference.37-40 NMA enables this evidence to be synthesized and, according to our analysis, MPT/MPT-T was superior over MP (HR 0.67, 95%CI: 0.55-0.81).

Thirdly, NMA calculates effectiveness estimates including direct and indirect evidence from RCTs providing additional evidence when head-to-head data are only available from one single RCT. Due to the rapid evolution haematologica | 2019; 104(5)

of the treatment armamentarium, efficacy evidence is increasingly based on a single RCT, and this is often from only one institute or region in the world. There is increasing evidence of RCTs investigating a similar treatment comparison providing contradictory results41 and this may increase the interest in indirect evidence. Indirect evidence may confirm or alter the results from a single RCT, as we have shown for MPR-R compared to MPT. Although there was no statistically significant difference between MPR-R and MPT-T based on direct evidence from two RCTs, synthesizing direct and indirect evidence resulted in a statistically significant HR for MPR-R compared to MPT/MPT-T. Favorable indirect evidence for MPR-R compared to MPTT was obtained through the comparison with MP. MPR-R compared more favorably to MP (according to the MM-15 HR MPR-R vs. MP 0.4) than MPT (HR MPT vs. MP 0.67 according to multiple trials). However, it should be noted that the direct evidence for MPR-R compared to MP was based on a single RCT while MPT/MPT-T versus MP was studied in seven RCTs, and therefore the evidence for the latter comparison is believed to be more solid.24,41 Indirect evidence is not always available, for example, for the comparison between VRd and Rd there is only direct evidence from a single study.6 While a fixed effect NMA will produce similar results to the trial (HR 0.71, 95%CI: 0.57-0.9), a random effects NMA obtains larger 95%CIs (HR 0.71, 95%CI: 0.43-1.17), as it includes two levels of uncertainty: within and between study variances.17 Therefore, there is less likelihood of significant differences between treatments.

Two other NMAs are available for newly diagnosed NTE NDMM patients. Our results align with the results from Kuhr et al.11 in that VMP and MPT are more effective than MP. Our results also confirm the conclusion from Weisel et al.12 that Rd is more favorable than MP [HR 0.63, 95%CI: 0.44-0.89 (Online Supplementary Appendix 5)].

However, in contrast to their findings, we found that Rd and VMP have comparable effectiveness outcomes (i.e. small difference in HR for PFS compared to D but largely overlapping CIs). The primary analysis of Weisel et al. included a limited number of treatments (i.e. VMP, MP, MPT and Rd) and RCTs (i.e. VISTA, IFM01/01, IFM 99/06, Sacchi, FIRST) as Phase III trials not using dosing schemes in line with the summary of product characteristics (SmPC) were excluded. There are several arguments against this restriction. Firstly, although dosing schemes in line with the SmPC might be recommended in the selected trials by Weisel et al., it is debatable whether this ensures treatments are identical within a network, especially because of variation in clinical practice due to either physician's preference or patient-related factors such as age, co-morbidities and toxicities. For example, the trial of Sacchi et al. 2011 was grouped with MPT studies while maintenance was only provided in a limited number of centers. Furthermore, the administered and planned dose may differ, as, for example, illustrated by the HOVON87 in which relative dose intensity varied between 0.540.96.13 Since there is no evidence on the impact of dosing schemes, we believe that a more comprehensive network (e.g. our network, including 19 additional trials) provides more solid evidence. The reason Weisel et al.12 did not find an overlap between VMP and Rd in their sensitivity analyses including six and twelve additional studies, is most likely because they used a fixed effect model for their analysis. A random effects model, like that used in our 1033

H.M. Blommestein et al. analysis and by Kuhr et al.,11 is, however, more appropriate, as this model allows for the between study-heterogeneity in the additional studies.

One might argue that while our NMA provides additional evidence in different circumstances, we had to make assumptions in order to conduct the analysis, and that this introduces a level of uncertainty. Firstly, we grouped MPT and MPT-T studies together since we could not make an unambiguous distinction between them. For example, thalidomide was prescribed until disease progression in the HOVON49 and GIMEMA trial but prescribed “continuously” for up to a maximum of 12 months in the TMSG trial. In the NMSG trial, it was even recommended to continue thalidomide maintenance until second relapse. However, most investigators discontinued thalidomide at first relapse. Prescription of thalidomide was also not consistent within a trial.38 Sacchi et al.38 described that, although planned, maintenance was only provided to 18% of the patients and in a limited number of centers. Their results, however, showed that there was no difference in PFS between maintenance and no-maintenance approaches.42 Therefore, we believe that it is appropriate to combine these trials, as performed previously,11,43 and the results of our sensitivity analysis confirm this assumption (see Online Supplementary Appendix 7).

Secondly, the validity of the outcomes of NMA depends on the comparability between studies. Our analysis focused on treatments for NTE NDMM patients studied in phase III RCTs. Although it is possible to include non-randomized evidence in NMA,45 and this could have provided additional information regarding effectiveness in clinical practice46-48 or treatments not analyzed in a phase III RCT (e.g. bortezomib-cyclophosphamide-dexamethasone, VCD49), we believe that limiting our analysis to the relative effectiveness of RCT evidence reduces the risk of bias and systematic errors.44 Further research to improve methodologies for conducting, evaluating and interpreting non-randomized evidence is recommended.44 We focused on NTE NDMM treatment to increase homogeneity between the patient populations in the study. We observed between-study heterogeneity comparable to the proportions previously

References 1. Ferlay J, Steliarova-Foucher E, LortetTieulent J, et al. Cancer incidence and mortality patterns in europe: Estimates for 40 countries in 2012. Eur J Cancer.

2013;49(6):1374-1403.

2. Mateos MV, San Miguel JF. How should we treat newly diagnosed multiple myeloma patients? Hematology Am Soc Hematol Educ Program. 2013;2013:488-495.

3. Moreau P, San Miguel J, Sonneveld P, et al.

Multiple myeloma: ESMO clinical practice guidelines for diagnosis, treatment and follow-updagger.

Ann Oncol.

2017; 128(suppl_4):iv52-iv61.

4. Benboubker L, Dimopoulos MA, Dispenzieri A, et al. Lenalidomide and dexamethasone in transplant-ineligible patients with myeloma. N Engl J Med. 2014;371(10):906-917.

5. San Miguel JF, Schlag R, Khuageva NK, et al. Bortezomib plus melphalan and prednisone for initial treatment of multiple

1034

6.

7.

8.

9.

reported by Kuhr et al.11 We allow for this heterogeneity by using a random effects instead of a fixed effect model.

However, the consequence of this is larger 95%CIs. A potential limitation of our search strategy is that we only included English language publications. To the best of our knowledge, this does not, however, lead to the exclusion of relevant studies or treatments. Furthermore, our NMA was limited to the intermediate outcome PFS and did not include other outcomes of interest such as OS, adverse events, quality of life, cost, and cost-effectiveness.

While OS may even be the most important subject of investigation for patients and health care decision makers, we believe a comparison of OS for first-line therapies with the currently available data is prone to bias due to crossover, heterogenous and limited follow up (e.g. especially for DaraVMP: median OS was not reached at 16.5 months follow up), and different subsequent treatment lines.50,51 In the context of increasing health care expenditures, costeffectiveness is also another relevant and important outcome, and this remains a subject for further research.

Several treatment options showed comparable effectiveness outcomes, but costs could very well differ due to drug prices, treatment duration, and route of administration. Our study facilitates cost-effectiveness research on first-line NTE treatments.

The treatment armamentarium is rapidly increasing and evolving for NTE NDMM patients, and NMAs will, therefore, become increasingly important. We illustrate the additional value and evidence that can be provided.

NMAs support evidence-based decision making and may help optimize treatment and outcomes of NTE NDMM patients in clinical practice.

Funding This work was supported by a grant from ZonMw, the Netherlands Organisation for Health Research and Development, project number 152001020, project title “Treatment Sequencing in Multiple Myeloma: modeling the disease and evaluating cost-efficacy vs. cost-effectiveness”. The funding source had no role in writing the manuscript or decision to submit for publication.

myeloma. N Engl J Med. 2008;359(9):906917.

Durie BG, Hoering A, Abidi MH, et al.

Bortezomib with lenalidomide and dexamethasone versus lenalidomide and dexamethasone alone in patients with newly diagnosed myeloma without intent for immediate autologous stem-cell transplant (SWOG S0777): A randomised, open-label, phase 3 trial. Lancet. 2017;389(10068):519527.

Mateos MV, Dimopoulos MA, Cavo M, et al. Daratumumab plus bortezomib, melphalan, and prednisone for untreated myeloma. N Engl J Med. 2018;378(6):518528.

Bothwell LE, Greene JA, Podolsky SH, Jones DS. Assessing the gold standard-lessons from the history of RCTs. N Engl J Med. 2016;374(22):2175-2181.

Gentile M, Magarotto V, Offidani M, et al.

Lenalidomide and low-dose dexamethasone (rd) versus bortezomib, melphalan, prednisone (VMP) in elderly newly diag-

10.

11.

12.

13.

nosed multiple myeloma patients: A comparison of two prospective trials. Am J Hematol. 2017;92(3):244-250.

Neupane B, Richer D, Bonner AJ, Kibret T, Beyene J. Network meta-analysis using R: A review of currently available automated packages. PLoS One. 2014;9(12):e115065.

Kuhr K, Wirth D, Srivastava K, Lehmacher W, Hellmich M. First-line therapy for nontransplant eligible patients with multiple myeloma: Direct and adjusted indirect comparison of treatment regimens on the existing market in germany. Eur J Clin Pharmacol. 2016;72(3):257-265.

Weisel K, Doyen C, Dimopoulos M, et al.

A systematic literature review and network meta-analysis of treatments for patients with untreated multiple myeloma not eligible for stem cell transplantation. Leuk Lymphoma. 2017;58(1):153-161.

Zweegman S, van der Holt B, Mellqvist UH, et al. Melphalan, prednisone, and lenalidomide versus melphalan, prednisone, and thalidomide in untreated mul-

haematologica | 2019; 104(5)

Efficacy of MM treatments

tiple myeloma. Blood. 2016;127(9):11091116.

14. Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials. 2007;8:16.

15. Higgins JP, Altman DG, Gotzsche PC, et al.

The cochrane collaboration's tool for assessing risk of bias in randomised trials.

BMJ. 2011;343:d5928.

16. Rucker G, Schwarzer G. Ranking treatments in frequentist network meta-analysis works without resampling methods. BMC Med Res Methodol. 2015;15:58.

17. Borenstein M, Hedges LV, Higgins JP, Rothstein HR. A basic introduction to fixed-effect and random-effects models for meta-analysis. Res Synth Methods. 2010;1 (2):97-111.

18. Rucker G. Network meta-analysis, electrical networks and graph theory. Res Synth Methods. 2012;3(4):312-324.

19. Rucker G, Schwarzer G. Reduce dimension or reduce weights? comparing two approaches to multi-arm studies in network meta-analysis. Stat Med. 2014;33(25): 4353-4369.

20. Bhatnagar N, Lakshmi PV, Jeyashree K. Multiple treatment and indirect treatment comparisons: An overview of network meta-analysis.

Perspect Clin Res. 2014;5(4):154-158.

21. Higgins JP, Jackson D, Barrett JK, Lu G, Ades AE, White IR. Consistency and inconsistency in network meta-analysis: Concepts and models for multi-arm studies. Res Synth Methods. 2012;3(2):98-110.

22. Krahn U, Binder H, Konig J. A graphical tool for locating inconsistency in network meta-analyses. BMC Med Res Methodol.

2013;13:35.

23. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in metaanalyses. BMJ. 2003;327(7414):557-560.

24. Higgins JP, Green S, eds. Cochrane handbook for systematic reviews of interventions version 5.1.0. [updated March 2011] ed. The Cochrane Collaboration; 2011.

Available from www.handbook.cochrane.org. Available from www.handbook.cochrane.org.

25. Alexanian R, Haut A, Khan AU, et al.

Treatment for multiple myeloma. combination chemotherapy with different melphalan dose regimens. JAMA. 1969; 208(9):1680-1685.

26. Facon T, Mary JY, Hulin C, et al. Melphalan and prednisone plus thalidomide versus melphalan and prednisone alone or reduced-intensity autologous stem cell transplantation in elderly patients with multiple myeloma (IFM 99-06): A randomised trial. Lancet. 2007; 370(9594): 1209-1218.

27. Hulin C, Facon T, Rodon P, et al. Efficacy of melphalan and prednisone plus thalidomide in patients older than 75 years with newly diagnosed multiple myeloma: IFM 01/01 trial. J Clin Oncol. 2009;27(22):36643670.

28. Palumbo A, Bringhen S, Larocca A, et al.

Bortezomib-melphalan-prednisonethalidomide followed by maintenance with bortezomib-thalidomide compared with

haematologica | 2019; 104(5)

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

bortezomib-melphalan-prednisone for initial treatment of multiple myeloma: Updated follow-up and improved survival.

J Clin Oncol. 2014;32(7):634-640.

Morgan GJ, Davies FE, Gregory WM, et al.

Long-term follow-up of MRC myeloma IX trial: Survival outcomes with bisphosphonate and thalidomide treatment. Clin Cancer Res. 2013;19(21):6030-6038.

Mateos MV, Richardson PG, Schlag R, et al.

Bortezomib plus melphalan and prednisone compared with melphalan and prednisone in previously untreated multiple myeloma: Updated follow-up and impact of subsequent therapy in the phase III VISTA trial. J Clin Oncol. 2010;28(13):2259-2266.

Dimopoulos MA, Terpos E, Chanan-Khan A, et al. Renal impairment in patients with multiple myeloma: A consensus statement on behalf of the international myeloma working group. J Clin Oncol. 2010; 28(33):4976-4984.

Larocca A, Offidani M, Musto P, et al. 744 impact of bortezomib- or lenalidomidebased induction treatment on high risk cytogenetic transplant-ineligible patients with newly diagnosed multiple myeloma enrolled in the gimema-MM-03-05 and EMN01 trials. Blood. 2017;130(Suppl 1):744.

Sonneveld P, Avet-Loiseau H, Lonial S, et al.

Treatment of multiple myeloma with highrisk cytogenetics: A consensus of the international myeloma working group. Blood.

2016;127(24):2955-2962.

Terpos E, Kleber M, Engelhardt M, et al.

European myeloma network guidelines for the management of multiple myelomarelated complications. Haematologica.

2015;100(10):1254-1266.

Mateos MV, Oriol A, Martinez-Lopez J, et al. Outcomes with two different schedules of bortezomib, melphalan, and prednisone (VMP) for previously untreated multiple myeloma: Matched pair analysis using long-term follow-up data from the phase 3 VISTA and PETHEMA/GEM05 trials. Ann Hematol. 2016;95(12):2033-2041.

Palumbo A, Bringhen S, Liberati AM, et al.

Oral melphalan, prednisone, and thalidomide in elderly patients with multiple myeloma: Updated results of a randomized controlled trial. Blood. 2008;112(8):31073114.

Wijermans P, Schaafsma M, Termorshuizen F, et al. Phase III study of the value of thalidomide added to melphalan plus prednisone in elderly patients with newly diagnosed multiple myeloma: The HOVON 49 study. J Clin Oncol. 2010;28(19):3160-3166.

Sacchi S, Marcheselli R, Lazzaro A, et al. A randomized trial with melphalan and prednisone versus melphalan and prednisone plus thalidomide in newly diagnosed multiple myeloma patients not eligible for autologous stem cell transplant. Leuk Lymphoma. 2011;52(10):1942-1948.

Beksac M, Haznedar R, Firatli-Tuglular T, et al. Addition of thalidomide to oral melphalan/prednisone in patients with multiple myeloma not eligible for transplantation: Results of a randomized trial from the turkish myeloma study group. Eur J Haematol.

2011;86(1):16-22.

40. Waage A, Gimsing P, Fayers P, et al.

Melphalan and prednisone plus thalidomide or placebo in elderly patients with multiple myeloma. Blood. 2010; 116(9): 1405-1412.

41. Ioannidis JP. Why most published research findings are false. PLoS Med. 2005; 2(8):e124.

42. Mateos MV, Oriol A, Martinez-Lopez J, et al. Bortezomib, melphalan, and prednisone versus bortezomib, thalidomide, and prednisone as induction therapy followed by maintenance treatment with bortezomib and thalidomide versus bortezomib and prednisone in elderly patients with untreated multiple myeloma: A randomised trial.

Lancet Oncol. 2010;11(10):934-941.

43. Fayers PM, Palumbo A, Hulin C, et al.

Thalidomide for previously untreated elderly patients with multiple myeloma: Meta-analysis of 1685 individual patient data from 6 randomized clinical trials.

Blood. 2011;118(5):1239-1247.

44. Efthimiou O, Mavridis D, Debray TP, et al.

Combining randomized and non-randomized evidence in network meta-analysis.

Stat Med. 2017;36(8):1210-1226.

45. Mohty M, Terpos E, Mateos MV, et al.

Multiple myeloma treatment in real-world clinical practice: Results of a prospective, multinational, noninterventional study.

Clin Lymphoma Myeloma Leuk. 2018; 18(10):e401-e419.

46. Schmitz S, Maguire A, Morris J, et al. The use of single armed observational data to closing the gap in otherwise disconnected evidence networks: A network meta-analysis in multiple myeloma. BMC Med Res Methodol. 2018;18(1):66.

47. Verelst SGR, Blommestein HM, de Groot S, et al. Long-term outcomes in patients with multiple myeloma: A retrospective analysis of the dutch population-based HAematological regustry for observational studies (PHAROS). HemaSphere. 2018; 2(4):1.

48. Jimenez-Zepeda VH, Duggan P, Neri P, Tay J, Bahlis NJ. Bortezomib-containing regimens (BCR) for the treatment of non-transplant eligible multiple myeloma. Ann Hematol. 2017;96(3):431-439.

49. Arditi C, Burnand B, PeytremannBridevaux I. Adding non-randomised studies to a cochrane review brings complementary information for healthcare stakeholders: An augmented systematic review and meta-analysis. BMC Health Serv Res. 2016; 16(1):598.

50. Blommestein HM, Verelst SG, de Groot S, Huijgens PC, Sonneveld P, Uyl-de Groot CA. A cost-effectiveness analysis of realworld treatment for elderly patients with multiple myeloma using a full disease model. Eur J Haematol. 2015;96(2):198208.

51. Zheng Y, Pan F, Sorensen S. Modeling treatment sequences in pharmacoeconomic models.

Pharmacoeconomics.

2017; 35(1):15-24.

52. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med. 2009;6(7):e1000097.

1035

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):1036-1045

Platelet Biology & its Disorders

Sphingolipid dysregulation due to lack of functional KDSR impairs proplatelet formation causing thrombocytopenia

Tadbir K. Bariana,1.2.3,4 Veerle Labarque,5 Jessica Heremans, 5 Chantal Thys,4,5 Mara De Reys,5 Daniel Greene,3,4,6,7 Benjamin Jenkins,8 Luigi Grassi,3,4,6,7 Denis Seyres,3,4,6,7 Frances Burden,3,4,6 Deborah Whitehorn,3,4,6, Olga Shamardina,3,4,6 Sofia Papadia,3,4,6 Keith Gomez,1,2,4 NIHR BioResource,4 Chris Van Geet,4,5 Albert Koulman,8 Willem H. Ouwehand,3,4,6,9,10 Cedric Ghevaert,3,6,9 Mattia Frontini,3,4,6,9 Ernest Turro3,4,6,7 and Kathleen Freson4,5

Department of Haematology, University College London, UK; 2The Katharine Dormandy Haemophilia Centre and Thrombosis Unit, Royal Free London NHS Foundation Trust, UK; 3Department of Haematology, University of Cambridge, Cambridge Biomedical Campus, UK; 4NIHR BioResource, Cambridge University Hospitals, Cambridge Biomedical Campus, UK; 5Department of Cardiovascular Sciences, Center for Molecular and Vascular Biology, University of Leuven, Belgium; 6NHS Blood and Transplant, Cambridge Biomedical Campus, UK; 7Medical Research Council Biostatistics Unit, Cambridge Institute of Public Health, Cambridge Biomedical Campus, UK; 8NIHR Biomedical Research Centre Core Metabolomics and Lipidomics Laboratory, University of Cambridge, Cambridge Biomedical Campus, UK; 9British Heart Foundation Centre of Excellence, Division of Cardiovascular Medicine, Cambridge University Hospitals, Cambridge Biomedical Campus, UK and 10Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton, Cambridge, UK 1

ABSTRACT

Correspondence: KATHLEEN FREESON kathleen.freson@med.kuleuven.be Received: August 17, 2018.

Accepted: November 19, 2018.

Pre-published: November 22, 2018. doi:10.3324/haematol.2018.204784 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1036 Š2019 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.

1036

S

phingolipids are fundamental to membrane trafficking, apoptosis, and cell differentiation and proliferation. KDSR or 3-keto-dihydrosphingosine reductase is an essential enzyme for de novo sphingolipid synthesis, and pathogenic mutations in KDSR result in the severe skin disorder erythrokeratodermia variabilis et progressiva-4. Four of the eight reported cases also had thrombocytopenia but the underlying mechanism has remained unexplored. Here we expand upon the phenotypic spectrum of KDSR deficiency with studies in two siblings with novel compound heterozygous variants associated with thrombocytopenia, anemia, and minimal skin involvement. We report a novel phenotype of progressive juvenile myelofibrosis in the propositus, with spontaneous recovery of anemia and thrombocytopenia in the first decade of life. Examination of bone marrow biopsies showed megakaryocyte hyperproliferation and dysplasia. Megakaryocytes obtained by culture of CD34+ stem cells confirmed hyperproliferation and showed reduced proplatelet formation. The effect of KDSR insufficiency on the sphingolipid profile was unknown, and was explored in vivo and in vitro by a broad metabolomics screen that indicated activation of an in vivo compensatory pathway that leads to normalization of downstream metabolites such as ceramide. Differentiation of propositus-derived induced pluripotent stem cells to megakaryocytes followed by expression of functional KDSR showed correction of the aberrant cellular and biochemical phenotypes, corroborating the critical role of KDSR in proplatelet formation. Finally, Kdsr depletion in zebrafish recapitulated the thrombocytopenia and showed biochemical changes similar to those observed in the affected siblings. These studies support an important role for sphingolipids as regulators of cytoskeletal organization during megakaryopoiesis and proplatelet formation. haematologica | 2019; 104(5)

Mechanism of KDSR-associated thrombocytopenia

Introduction 3-keto-dihydrosphingosine reductase (KDSR) is an early, essential enzyme in the pathway of de novo sphingolipid synthesis that catalyzes the conversion of 3-ketodihydrosphingosine (KDS) to dihydrosphingosine (DHS) on the cytosolic leaflet of the endoplasmic reticulum.1 The canonical transcript for KDSR encodes a 332 amino acid protein. The gene is widely transcribed,1-3 consistent with the integral roles of the sphingolipid family in forming lipid rafts that facilitate membrane trafficking and in the regulation of fundamental cellular functions that include apoptosis, differentiation, and proliferation.4 The importance of sphingolipid synthesis for normal cellular functions is illustrated by the complex multisystem phenotypes of null mice for key enzymes or receptors in the pathway, including defective platelet activation and thrombus formation.5,6 A pathway for de novo synthesis of sphingolipids in a megakaryocytic cell line has been shown, but this plays a minimal role in mature platelets, which instead acquire essential sphingolipids by incorporating them from plasma or recycling plasma membrane sphingomyelins, both largely independently of KDSR.7 Consistent with these important roles of sphingolipids, compound heterozygous variants in KDSR (Figure 1) have recently been identified as causing the severe skin disorder erythrokeratodermia variabilis et progressiva 4 (EKVP4, OMIM617526), a condition characterized by neonatal onset of thick, scaly skin on the face and genitals, and milder erythematous palmo-plantar scaling.8 This observation established a role for KDSR in the homeostasis of keratinization; however, it was unclear whether these cases had hematologic pathologies. A more recent study described four probands with EKVP4 caused by KDSR variants accompanied by severe thrombocytopenia and platelet dysfunction in infancy.9 A reduction in plasma S1P and surface-exposed ceramide in human platelets, as well

as diminished ceramide levels in affected skin, were reported. Bone marrow (BM) morphology in one patient was normal and in a second patient demonstrated increased megakaryopoiesis. For this patient, a diagnosis of immune-mediated thrombocytopenia was made with no response to corticosteroid treatment and minimal response to splenectomy. No further exploration of the molecular mechanism underlying the thrombocytopenia was undertaken.9 Here we provide evidence that in this pedigree, KDSR plays a fundamental role in megakaryopoiesis, cytoplasmic organization, and proplatelet formation. We describe a pedigree in which compound heterozygous variants in KDSR segregate with severe thrombocytopenia and minimal or no skin involvement. We report novel phenotypes of progressive juvenile myelofibrosis in the propositus, who is older, and anemia in both siblings. Broad metabolic profiling complemented by targeted mass spectrometry assays confirm KDSR hypofunction and suggest activation of an alternative, compensatory pathway in vivo.

Depletion of kdsr in zebrafish and studies with CD34+ stem cell- and induced pluripotent stem cell (iPSC)-derived MK show cellular and biochemical signatures in common with those observed in our patients, showing the mechanism by which KDSR variants mediate thrombocytopenia.

Methods Recruitment and sequencing Following informed, written consent (ethical study approval ML3580), the propositus was recruited to the Bleeding, thrombotic and Platelet Disorders (BPD) domain of the NIHR BioResourceRare Diseases study (UK Research Ethics Committee 13/EE/0325, https://bioresource.nihr.ac.uk). Further details are provided in the Online Supplementary Methods.

Figure 1. Reported KDSR variant genotypes and phenotypes in the context of 3-keto-dihydrosphingosine reductase (KDSR) structure and function. Protein and cDNA schematic adapted from Gupta et al.20 demonstrating location of known pathogenic KDSR variants with documented phenotypes in skin only (black), skin and platelets (red), and the novel variants reported in this manuscript in bold and underlined. Variants are linked by brackets where present in compound heterozygosity in an individual. Key structural elements of KDSR are illustrated: transmembrane anchors (blue, purple), the Rossman folds (red), and a highly conserved domain containing three putative catalytic sites (yellow). The novel p.Arg154Trp variant is within the catalytic domain.

haematologica | 2019; 104(5)

1037

T.K. Bariana et al.

Platelet studies Aggregation and transmission electron microscopy (EM) studies were performed as described previously.10

Metabolic profiling Global metabolic profiling of plasma was performed by Metabolon Inc. (Durham, NC, USA) using the DiscoveryHD4 liq-

B

Hemoglobin

A

uid chromatography tandem mass spectrometry (LC-MS/MS) platform, as previously described.11 Results of study participants were compared with 496 subjects between the ages of 4 and 55 years without thrombocytopenia. A separate LC-MS platform method, previously described,12,13 was used for specific confirmation of the global sphingolipid profile. Further details are provided in the Online Supplementary Methods.

C

D

E

Figure 2. Clinical characteristics associated with loss-of-function KDSR variants. (A) Pedigree and variants identified in KDSR. â&#x20AC;&#x2DC;+â&#x20AC;&#x2122; denotes the major allele. The propositus and affected sibling, but not the healthy sibling, carry the missense variant 18:61018270 G>A (p.Arg154Trp) and the nonsense variant 18:61006104 G>A (p.Arg236*). Co-segregation analysis demonstrated that the father carries the former and the mother the latter variant. (B) Serial blood counts are shown for the two affected siblings and a single value for the healthy brother. Fluctuating anemia and thrombocytopenia was observed, without evidence of neutropenia. (C) Bone marrow biopsy. (Left) Numerous dysplastic megakaryocytes made visible with linker for activation of T cell (LAT) staining are present. (Right) Marrow fibrosis with strong stromal reticulin staining. Magnification x40. Further images can be found in the Online Supplementary Appendix. (D) The affected sibling was born during the course of this study and presented at birth with thrombocytopenia and mild ichthyosis in her left axilla. The skin symptoms improved spontaneously over the first month. (E) Platelets were examined by electron microscopy for an unrelated healthy control, the propositus, and the affected sibling. There were no marked morphological differences. Arrowed magnifications show delta granules. Magnification x 12,000.

1038

haematologica | 2019; 104(5)

Mechanism of KDSR-associated thrombocytopenia

Stem cell differentiation assays CD34+ hematopoietic stem cells (HSC) were isolated by magnetic cell sorting (Miltenyi Biotec, Bergisch Gladback, Germany) from BM aspirates from the propositus (at 5 years of age) and an unrelated control, and from peripheral blood (PB) from the propositus (at 8.5 years of age), his affected sister (at 5 months of age), and an unrelated control.

In addition, expanded BM- and PM-derived12,13 HSC at day 3 of differentiation were used for liquid MK cultures in two experiments. In the first experiment, HSC obtained from the BM of the propositus were differentiated in parallel to a control. For the second experiment, HSC obtained from the PB of the propositus and his affected sister were cultured in parallel with a different control.

Details of the differentiation protocols, colony assays and statistical analysis of megakaryocyte (MK) immunostaining are provided in the Online Supplementary Methods.

Zebrafish analysis Tg(cd41:EGF) embryos14 were injected at the one-cell stage with a kdsr ATG morpholino (MO) (5â&#x20AC;&#x2122; ctcagaggacatgggtcaacctgat, KdsrMO) purchased from Gene Tools LLC (Philomath, OR, USA) or with buffer (control). Zebrafish kdsr has ZFIN accession number

ZDB-GENE-040426-853. Thrombocyte formation was analyzed as described previously.15,16 Immunoblots were developed with goat anti-GFP (Rockland) and anti-FVT1/KDSR (Clone H-149; Santa Cruz, CA, USA). All animal protocols were approved by the Ethical Committee of KU Leuven, Belgium.

Lentiviral reference 3-keto-dihydrosphingosine reductase transcript expression in induced pluripotent stem cells Induced pluripotent stem cells (iPSC) were prepared by the Cambridge Biomedical Research Centre iPSC core laboratory as described in the Online Supplementary Methods. iPSC were transduced with the lentiviral vector to express the open reading frame (ORF) of KDSR (pLenti-EF1a-KDSR-myc-DDK-IRES-Puro, Origene) and the un-cloned destination vector PS10085 (Origene) to generate the reference transcript rescue line (Kresc) and empty vector control line (Kev), respectively. The ORF is identical to transcript ENST00000591902 (RefSeq accession n. NM_002035, Origene TrueClone cDNA cat. RC201153), which has the highest reported expression in MK.2 Details of lentiviral particle production, transduction and selection are given in the Online Supplementary Methods.

A

B

C

Figure 3. Metabolic profiling shows that the KDSR variants are associated with loss-of-function and downstream sphingolipid pathway compensation. (A) Simplified sphingolipid pathway highlighting the role of the 3-keto-dihydrosphingosine reductase (KDSR) enzyme in de novo synthesis (black arrows) and the generation of sphingolipid intermediates from the recycling of complex sphingolipids and sphingomyelins (green arrows). (B) Mass spectrometry using the Metabolon platform shows the major chromatographic peak of 3-keto-dihydrosphingosine (KDS) in the plasma of the propositus, but not of the unaffected pedigree members (shown) or the controls (data not shown). (C) KDSR hypofunction was confirmed in the propositus and affected sister using a second mass spectrometry platform for targeted sphingolipid profiling. The chromatogram shows that KDS was detected in the plasma from the propositus and his affected sister but not in the plasma from the healthy brother, parents (shown), and two controls (data not shown).

haematologica | 2019; 104(5)

1039

T.K. Bariana et al.

Results

Forward programming to MK Induced pluripotent stem cells were reprogrammed to MK (named iMK hereafter) using a protocol for generating MK described by Moreau et al.17 Kresc and Kev iMK were generated in three independent experiments. Details of iMK reprogramming, immunophenotyping, the proplatelet assay, and RNA sequencing are given in the Online Supplementary Methods.

A

Clinical characteristics The 8-year old male propositus was born to healthy, unrelated parents of European ancestry (Figure 2A). At 4 months of age he presented with a viral infection and was found to have a platelet count of 65x109/L and mild nor-

B

C

D

E

1040

Figure 4. Kdsr morpholino knockdown is associated with reduced thrombopoiesis in zebrafish.

(A) Tg(cd41:EGF) embryos were injected with a kdsr ATG-MO (1000 mM) or with buffer (control).

Embryos were lysed 72 hpf and used for immunoblotting. GFP and Kdsr proteins were reduced in the kdsr knockdown condition. Equal amounts (50 mg) were loaded (5 randomly selected embryos for each of the 4 conditions). Staining of Gapdh was used as loading control. (B) Quantification of immunoblot after normalization for Gapdh. Mean values are plotted and error bars show the standard deviation, analyzed by one-way ANOVA. (C) Quantification by flow cytometry of the number of GFP-labeled thrombocytes in Tg(cd41:EGFP) Danio rerio embryos at 72 hpf for kdsr-MO (800 or 1000 mM) or buffer (control) injected fish. Values are means and Standard Deviations as quantified for 10 randomly selected embryos for each condition, performed in triplicate. Results were analyzed by one-way ANOVA. (D) Grayscale stereo-microscope images (x20 original magnification) in the tail region at 72 hpf showed a reduced number of GFP-labeled thrombocytes (in white). (E) KDS levels in lysates from 72 hpf embryos (20/condition) for kdsrMO (800 mM) or buffer (control) injected fish. KDS was detected in the lysates from the MO-injected embryos but not in the control-injected embryos.

haematologica | 2019; 104(5)

Mechanism of KDSR-associated thrombocytopenia

mocytic, normochromic anemia with normal iron and hematinic levels (Figure 2B, Online Supplementary Table S1A and Online Supplementary Figure S1). Subsequent complete blood counts showed on several occasions platelet counts <100x109/L accompanied by rectal and gingival bleeding, excessive ecchymosis, and recurrent epistaxis when platelet counts were <10x109/L. Possible skin involvement was limited to a slow-to-heal perianal wound following rectal manometry. Serial BM examinations revealed increased numbers of dysplastic MK and progressive severe myelofibrosis (Figure 2C, Online Supplementary Table S1B and Online Supplementary Figure S2); despite this we observed significant fluctuation in the propositus’s thrombocytopenia and normalization of the hemoglobin level over the course of his first decade (Figure 2B and Online Supplementary Table S1A). The mechanism of the improvement is unclear, and occurred in the absence of identifiable environmental, therapeutic, or dietary interventions. Genetic analyses of BM DNA excluded known somatic mutations causal of myelodysplasia or primary myelofibrosis (Online Supplementary Table S1B). Light transmission platelet aggregation was normal with the exception of an attenuated response to stimulation with collagen at low dose (Online Supplementary Table S1C). The propositus’s older brother and his parents were unaffected (Figure 2A). His sister presented at birth with thrombocytopenia (Figure 2A and B, and Online Supplementary Table S1A) and mild ichthyosis in her left axilla (Figure 2D), but the skin symptoms resolved spontaneously over the first month. At 5 months of age she also developed persistent, normocytic, normochromic

anemia (Figure 2B and Online Supplementary Table S1A).

Transmission EM analysis showed platelets of normal size (Figure 2E). Delta (δ)-granules appeared diminished; however, it was not possible to count these accurately in the absence of whole-mount EM or a specific δ-granule marker (CD63 also stains lysosomal structures). There were no other marked ultrastructural abnormalities.

Pathogenic variants in KDSR The propositus and his affected sister carry a maternally inherited nonsense variant 18:61006104 G>A (p.Arg236*) and a paternally inherited missense variant 18:61018270 G>A (p.Arg154Trp) in KDSR (Figure 2A). The variants were confirmed by Sanger sequencing18,19 (Online Supplementary Figure S3) and have minor allele frequencies in Europeans of 4.82x10-5 and 2.32x10-4, respectively. The missense variant p.Arg154Trp is localized in the catalytic domain of KDSR (Figure 1)20 and both are found in the most abundant KDSR transcripts in MK (Online Supplementary Figure S4).2 The nonsense variant is absent from two out of three major platelet transcripts, which is in keeping with minimal de novo sphingolipid synthesis in mature platelets (Online Supplementary Figure S4).7 The results of co-segregation study were concordant with an autosomal recessive mode of inheritance (Figure 2A).

Sphingolipid profiles We reasoned that the variants would cause reduced enzymatic function, leading to a build-up of the substrate KDS (Figure 3A). Indeed, global metabolic profiling showed KDS to be detectable in plasma from the proposi-

B

A

C

Figure 5. KDSR variants are associated with reduced proplatelet formation by megakaryocytes (MK). (A) Quantification of proplatelet formation by MK at day 11 of differentiation. On the left are the results of differentiation of bone marrow (BM)-derived hematopoietic stem cells (HSC) from the propositus and control. (Right) Results of differentiation of HSC obtained from the blood of the propositus, his affected sibling, and a second control. All MK with proplatelets and membrane budding were counted as positive. Values plotted are means and Standard Deviations (SD) as quantified on 20 images. Results were analyzed by the unpaired, twotailed t-test. (B) MK at day 11 derived from BM HSC from the propositus and a control. MK are stained for the cytoskeletal marker F-actin (red) and lysosome and delta granule marker CD63 (green). MK from the propositus and affected sibling have irregular cytoskeletal structures with lamellipodia (arrows). Further images can be found in Online Supplementary Figure S10. (C) MK area was quantified by automated analysis. Modeling was performed using a linear mixed effects model and associated P-values were computed by a likelihood ratio test. MK from affected cases were smaller compared to unrelated controls (P=0.01473).

haematologica | 2019; 104(5)

1041

T.K. Bariana et al.

A

B

C

D

1042

Figure 6. KDSR reference allele expression rescues ineffective proplatelet formation.

Proplatelet formation by induced pluripotent stem cells reprogrammed to MK (iMK).

One hundred percent of live MK plated for the proplatelet assay were CD41 positive and 75% were dual positive for CD41 and CD42 by flow cytometry (Online Supplementary Figure S11A). Cytoskeletal marker a-tubulin was stained with antibodies in green and nuclei were stained with DAPI in blue. Proplatelet formation was counted manually. Values shown were analyzed using the paired, two-tailed Student t-test plotted as means and standard deviations. *P<0.05 was considered statistically significant. (A) The number of proplatelets formed at 4 hours (h) per proplatelet-forming MK (PPFMK) by the rescued and non-rescued iMK. The differences were significant at 4 h (P=0.047) but not at 24 h (P=0.20). (B) There was no significant difference in the number of PPFMK- at 4 h between the two groups (7â&#x2C6;&#x2122;1% vs.

42â&#x2C6;&#x2122;4%; P=0.10), but at 24 h the rescued iMK showed less PPFMK (P=0.03).

(C) Representative images from the proplatelet formation assay at 4 and 24 h.

Proplatelets are indicated by white arrows. (Top left and right) Results at 4 h for rescued iMK show increased proplatelet formation. White scale bars indicate 10 mm. (Bottom left and right) Results at 24 h show little proplatelet formation for the rescued iMK, and residual cells are either fragmented into platelet-like particles, or consist of bare nuclei.

(D) Metabolon mass spectrometry results for non-rescued and rescued iMK. The ion counts for KDS detection differed significantly between the non-rescued and non-rescued iMK (P=0.02). There was no difference in DHS level between the groups of samples but the levels of sphingosine and ceramide were lower and higher in the non-rescued versus rescued iMK, respectively. ns: not significant.

haematologica | 2019; 104(5)

Mechanism of KDSR-associated thrombocytopenia

tus but not from the parents, the healthy sibling, or 496 unrelated controls (Figure 3B). This finding was corroborated using a LC-MS platform for the selective measurement of specific sphingolipids, which confirmed that KDS was detectable in both patients and absent from the plasma of controls (Figure 3C and Online Supplementary Figure S5). Interestingly, there was no reduction in downstream sphingolipids in the patients using either platform (Online Supplementary Tables S2 and S3A). In fact, global profiling showed levels of the KDSR product, DHS, were higher for the propositus than controls. These findings raise the hypothesis that KDSR hypofunction during de novo sphingolipid synthesis is compensated in vivo by alternative mechanisms, for example, by the recycling of relatively abundant sphingomyelins along a pathway that normally contributes little to free DHS production.4

Depletion of kdsr in zebrafish causes thrombocytopenia We explored the role of the enzyme on thrombocyte formation in zebrafish by MO-mediated depletion of the kdsr transcript in Tg(cd41:EGFP). As expected, this led to a reduction of the Kdsr protein level (Figure 4A and B) and resulted in curved tails, which is a typical feature for embryos with thrombocytopenia (Online Supplementary Figure S6).15 The number of thrombocytes was inversely correlated with the dose of MO injected (Figure 4C and D). Targeted sphingolipid profiling showed elevated and undetectable KDS in lysates from MO and control embryos, respectively (Figure 4E). Similar to the results obtained with the propositus’s plasma, dihydroceramides, ceramides, sphingomyelins and glycosphingolipids that are downstream of Kdsr were not significantly different between Kdsr-depleted and control fish (Online Supplementary Table S3B).

Impaired proplatelet formation in patient megakaryocytes CFU-GEMM cultures differentiated from bone marrow HSC of the propositus showed hyperproliferation of myeloid cells (P=0.001, t-test) with a reduced myeloid/erythroid ratio compared to the controls (Online Supplementary Figure S7). CFU-MK numbers were comparable to those of the control, although individual MK colonies were denser for the propositus, and liquid cultures showed an increased number of MK (Online Supplementary Figure S8). MK in control cultures formed proplatelets, whilst MK derived from both the propositus and the affected sister showed a strong reduction in proplatelet formation, despite similar levels of membrane budding and a higher number of CD41 and CD42 positive cells in propositus-derived cultures when compared with control MK (Figure 5A and Online Supplementary Figures S8-S10). Patient-derived MK also showed extensive, abnormal formation of lamellipodia and reduced cell size (P=0.014, likelihood ratio test) (Figure 5B and C).

The abnormal morphological, functional, and biochemical features of the propositus’s induced pluripotent stem cells reprogrammed to megakaryocytes can be rescued To corroborate the atypical phenotypes of MK derived from the HSC, we transduced iPSC from the propositus with lentiviral vectors containing the reference KDSR ORF (Kresc) or an inert control vector (Kev), and reprogrammed haematologica | 2019; 104(5)

these cells to iMK (Online Supplementary Figure S11A).

Analysis of the iMK RNA-seq results showed similar KDSR gene expression but the majority of sequencing reads in the rescued iMK carried the reference allele at Chr18:61018270 G>A (p.Arg154Trp) (Online Supplementary Figure S11B-D). These findings are consistent with correction of the genetic defect without significant overexpression, and resulted in increased proplatelet formation compared with control iMK 4 h after seeding (P=0.047, t-test) (Figure 6A-C). The observed iMK proplatelets were shorter and less branched than those observed following directed differentiation from stem cell cells, in keeping with published reports using this protocol.17 Upon microscopic inspection, the rescued iMK seemed larger than the non-rescued ones, which was confirmed to be significant by flow cytometry (Online Supplementary Figure S11E) and the increased proplatelet formation resulted by 24 h in little residual cytoplasm for the rescued versus the non-rescued iMK (Figure 6C). At the biochemical level, the rescue resulted in a significant reduction in KDS levels (P=0.02, t-test) showing the effectiveness of the gene therapy approach in ‘curing’ the iMK from the propositus (Figure 6D). Similar to findings in plasma and in zebrafish, there was no difference in DHS levels between the iMK with and without functional KDSR transcripts, indicating that the postulated, compensatory mechanism is also present in iMK. We searched the iMK transcriptome landscape for possible differences in the levels of transcripts of other key enzymes that regulate sphingolipid synthesis and recycling (the enzymes examined are as shown in Figure 3A). This identified only ASAH1 and CERS6 transcripts to be down- and up-regulated respectively (posterior probabilities 0.610 and 0.774; log-fold change -0.67 and +0.70, respectively). These two enzymes regulate the ceramide-sphingosine ratio (Figure 3A and Online Supplementary Figure S12), and in keeping with these findings, the rescued iMK showed higher sphingosine and lower ceramide levels (Figure 6D).

Discussion Pathogenic mutations in KDSR have recently been associated with a recessively inherited syndrome of moderateto-severe skin pathology and thrombocytopenia. We have described two novel KDSR mutations causing thrombocytopenia in the propositus and his infant sister, expanding the phenotypic spectrum of this recently identified Mendelian disorder from severe skin pathology with no apparent hematologic involvement to profound thrombocytopenia and moderate anemia with spontaneous improvement across the first decade, and almost imperceptible dermatological abnormalities. In the propositus, BM studies also showed the novel phenotype of severe juvenile myelofibrosis; however, the sister was too young to allow confirmation of this phenotype. The biochemical sphingolipid signatures of the plasma and patient-derived iMK confirmed the predicted reduction in function with elevated levels of its substrate, KDS. This is, as expected, from the combination of a variant encoding a premature stop codon and a hypomorphic allele involving a missense variant in the catalytic domain. Unexpectedly, downstream metabolites in the sphingolipid pathway, including DHS, ceramide, and sphingosine-1-phosphate, were not reduced in plasma, suggesting that KDSR hypofunction 1043

T.K. Bariana et al.

during de novo sphingolipid synthesis is compensated by an alternative pathway. One possible alternative pathway is the recycling of relatively abundant sphingomyelins; a pathway previously shown to contribute to production of downstream sphingolipids such as dihydroceramide, ceramide, and sphingosine, but considered to contribute little, if at all, to DHS production under normal conditions.21 Post-translational modifications to sphingolipid enzymes may also explain the metabolic profile, and further work is required to explore this possibility.

Importantly, the profile of increased ceramide and reduced sphingosine in propositus-derived iMK compared with rescued iMK, and the consistent and potentially explanatory transcriptional dysregulation of enzymes ASAH1 and CERS6, is in contrast to the limited sphingolipid quantitation undertaken in previous studies which showed that ceramide levels were reduced in affected skin and that platelet surface exposure of ceramide was impaired in individuals with hypo-functional KDSR variants.9 Our observation that MK lacking functional KDSR are hyperproliferative is consistent with earlier reports,9 but we expand on this characterization by showing the ex vivo generated patient MK to be smaller than controls and to be less effective in proplatelet formation. Proplatelets are pseudopodial projections of megakaryocyte cytoplasm, supported at their core by microtubular bundles that carry granules and other platelet cargo from the body of the megakaryocyte to the tip of the proplatelet.22 Aberrant size and proplatelet formation were not only observed in MK obtained by differentiation of primary HSC obtained from the two patients, but are also present in iMK generated by forward programming of iPSC derived from the propositus’s fibroblasts. Taken together, we consider the ineffective platelet formation caused by the absence of KDSR to be the primary cause of the thrombocytopenia. Increased turnover because of a reduced platelet lifespan seems to be a less likely explanation because the immature platelet fraction was not significantly raised in the two patients compared with healthy controls (data not shown). We hypothesize that impaired platelet formation may, in turn, be caused by cytoskeletal disorganization and further experiments are required to explore this possibility. Pathogenic mutations in several other genes (e.g. MYH9, ACTN1, FLNA, TUBB1, DIAPH1, TPM4) encoding proteins with important functional roles in cytoskeletal reorganization and actin polymerization cause dominant forms of thrombocytopenia.23 However, these genetic disorders are characterized by enlarged platelets, and the mean volumes of the platelets of our patients are within the normal ranges for males and females, respectively.

The increased level of KDS in plasma was confirmed at the cellular level in iMK derived from the propositus. This increased level was normalized upon rescue of the propositus’s iMK with a KDSR transcript carrying the reference allele. The correction of the biochemical defect was mirrored by a recovery of iMK size and improvement of their capacity to form proplatelets. To further support the central importance of KDSR in thrombopoiesis, we show that KDSR knockdown in a zebrafish model is associated with impaired thrombocyte formation. Similar approaches have identified multiple potential regulators of thrombopoiesis,24 though in isolation zebrafish studies these are limited by inherent differences in thrombopoiesis 1044

between mammals and fish, notably that zebrafish have nucleate thrombocytes rather than MK. The marked, spontaneous improvement in the propositus’s thrombocytopenia and anemia led to a reversal of the decision to treat the condition by HSC transplantation.

The mechanism of this improvement is unclear, given the presence of progressive myelofibrosis and in the absence of clinical features to suggest significant extramedullary hemopoiesis such as splenomegaly. Recent studies have shown spontaneous improvement in blood counts in other inherited juvenile BM failure syndromes, most notably those associated with pathogenic variants in SAMD9 or SAMD9L.25 In these cases, the improvement was attributed to the acquisition of corrective somatic mutations. Further longitudinal studies of individuals affected by pathogenic KDSR variants is essential to determine whether the clinical course described is representative, and whether a careful watch-and-wait approach rather than early intervention may be more appropriate in this genetically-defined subgroup of cases with inherited thrombocytopenia accompanied by BM failure.

Funding TKB is supported by the British Society of Haematology and NHS Blood and Transplant. KF and CVG are supported by the Fund for Scientific Research-Flanders (FWO Vlaanderen, Belgium; G.0B17.13N) and by the Research Council of the University of Leuven (BOF KU Leuven‚ Belgium; OT/14/098).

C.V.G is holder of the Bayer and Norbert Heimburger (CSL Behring) Chairs. The structured illumination microscope was acquired through a CLME grant from Minister Lieten to the VIB BioImaging Core, Leuven. J.H is supported by the Fund for Scientific Research-Flanders (FWO Vlaanderen, Belgium) grant no.1S00816N. AK and BJ are funded by the National Institute for Health Research (NIHR) Biomedical Research Centre (RG64245). MF is supported by the British Heart Foundation (BHF) Cambridge Centre of Excellence (RE/13/6/30180). The Ouwehand laboratory receives support from the BHF, BristolMyers Squibb, European Commission, MRC, NHS Blood and Transplant, Rosetrees Trust, the NIHR Biomedical Research Centre based at Cambridge University Hospitals NHS Foundation Trust, and the University of Cambridge.

Acknowledgments The NIHR BioResource – Rare Disease Study is a multicenter whole-genome sequencing (WGS) study of approximately 13,000 patients. The genotype and phenotype data are being incorporated in the 100,000 Genomes Project. This study makes use of data generated by the NIHR BioResource and a full list of investigators who contributed to the generation of the data is available from https://bioresource.nihr.ac.uk/rare-diseases/consortia-lists/. Funding for the project was provided by the NIHR (grant number RG65966). The NIHR BioResource projects were approved by Research Ethics Committees in the UK and appropriate national ethics authorities in non-UK enrollment centers.

The authors are also grateful to all the research participants who donated their samples for this study and to Professor Andrew Mumford (University of Bristol, UK), Professor Michael Laffan (Imperial College London, UK), Dr Lining Guo (Metabolon Inc., Durham, NC, USA), and Dr Sergio Rodriguez-Cuenca and Professor Antonio Vidal-Puig from the University of Cambridge (UK) for their input. The Tg(cd41:EGF)11 line was a gift from Professor Leonard Zon (Hematology Division, Brigham and Women’s Hospital, Boston, MA, USA). haematologica | 2019; 104(5)

Mechanism of KDSR-associated thrombocytopenia

References 1. Kihara A, Igarashi Y. FVT-1 is a mammalian 3-ketodihydrosphingosine reductase with an active site that faces the cytosolic side of the endoplasmic reticulum membrane. J Biol Chem. 2004; 279(47):49243-49250.

2. Chen L, Kostadima M, Martens JH, et al.

Transcriptional diversity during lineage commitment of human blood progenitors.

Science. 2014;345(6204):1251033.

3. Battle A, Brown CD, Engelhardt BE, Montgomery SB. Genetic effects on gene expression across human tissues. Nature.

2017;550(7675):204-213.

4. Hannun YA, Luberto C, Argraves KM. Enzymes of sphingolipid metabolism: from modular to integrative signaling.

Biochemistry. 2001;40(16):4893-4903.

5. Munzer P, Schmid E, Walker B, et al.

Sphingosine kinase 1 (Sphk1) negatively regulates platelet activation and thrombus formation. Am J Physiol Cell Physiol.

2014;307(10):C920-927.

6. Munzer P, Borst O, Walker B, et al. Acid sphingomyelinase regulates platelet cell membrane scrambling, secretion, and thrombus formation. Arterioscler Thromb Vasc Biol. 2014;34(1):61-71.

7. Tani M, Sano T, Ito M, Igarashi Y. Mechanisms of sphingosine and sphingosine 1-phosphate generation in human platelets. J Lipid Res. 2005; 46(11):24582467.

8. Boyden LM, Vincent NG, Zhou J, et al.

Mutations in KDSR cause recessive progressive symmetric erythrokeratoderma.

Am J Hum Genet. 2017;100(6):978-984.

9. Takeichi T, Torrelo A, Lee JYW, et al.

Biallelic mutations in KDSR disrupt ceramide synthesis and result in a spectrum of keratinization disorders associated with

haematologica | 2019; 104(5)

10.

11.

12.

13.

14.

15.

16.

17.

thrombocytopenia. J Invest Dermatol.

2017;137(11):2344-2353.

Freson K, De Vos R, Wittevrongel C, et al.

The TUBB1 Q43P functional polymorphism reduces the risk of cardiovascular disease in men by modulating platelet function and structure. Blood. 2005; 106(7):2356-2362.

Long T, Hicks M, Yu HC, et al. Wholegenome sequencing identifies common-torare variants associated with human blood metabolites. Nat Genet. 2017;49(4):568578.

Koulman A, Woffendin G, Narayana VK, Welchman H, Crone C, Volmer DA. Highresolution extracted ion chromatography, a new tool for metabolomics and lipidomics using a second-generation orbitrap mass spectrometer. Rapid Commun Mass Spectrom. 2009; 23(10):1411-1418.

Lu L, Koulman A, Petry CJ, et al. An unbiased lipidomics approach Identifies Early second trimester lipids predictive of maternal glycemic traits and gestational diabetes mellitus. Diabetes Care. 2016;39(12):22322239.

Lin HF, Traver D, Zhu H, et al. Analysis of thrombocyte development in CD41-GFP transgenic zebrafish. Blood. 2005; 106(12):3803-3810.

Louwette S, Labarque V, Wittevrongel C, et al. Regulator of G-protein signaling 18 controls megakaryopoiesis and the cilia-mediated vertebrate mechanosensory system.

FASEB J. 2012;26(5):2125-2136.

Louwette S, Regal L, Wittevrongel C, et al.

NPC1 defect results in abnormal platelet formation and function: studies in Niemann-Pick disease type C1 patients and zebrafish. Hum Mol Genet.2013;22(1):6173.

Moreau T, Evans AL, Vasquez L, et al.

18.

19.

20.

21.

22.

23.

24.

25.

Large-scale production of megakaryocytes from human pluripotent stem cells by chemically defined forward programming.

Nat Commun. 2016;7(11208.

Lek M, Karczewski KJ, Minikel EV, et al.

Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016; 536(7616):285-291.

Kircher M, Witten DM, Jain P, O'Roak BJ, Cooper GM, Shendure J. A general framework for estimating the relative pathogenicity of human genetic variants. Nat Genet. 2014;46(3):310-315.

Gupta SD, Gable K, Han G, et al. Tsc10p and FVT1: topologically distinct shortchain reductases required for long-chain base synthesis in yeast and mammals. J Lipid Res. 2009;50(8):1630-1640.

Kitatani K, Idkowiak-Baldys J, Hannun YA. The sphingolipid salvage pathway in ceramide metabolism and signaling. Cell Signal. 2008;20(6):1010-1018.

Italiano JE Jr, Lecine P, Shivdasani RA, Hartwig JH. Blood platelets are assembled principally at the ends of proplatelet processes produced by differentiated megakaryocytes. J Cell Biol. 1999; 147(6):1299-1312.

Lentaigne C, Freson K, Laffan MA, Turro E, Ouwehand WH. Inherited platelet disorders: toward DNA-based diagnosis. Blood.

2016;127(23):2814-2823.

Bielczyk-Maczynska E, Serbanovic-Canic J, Ferreira L, et al. A loss of function screen of identified genome-wide association study Loci reveals new genes controlling hematopoiesis. PLoS Genet. 2014; 10(7):e1004450.

Bluteau O, Sebert M, Leblanc T, et al. A landscape of germ line mutations in a cohort of inherited bone marrow failure patients. Blood. 2018;131(7):717-732.

1045

ARTICLE Ferrata Storti Foundation

Coagulation & its Disorders

Prevention of the anti-factor VIII memory B-cell response by inhibition of Bruton tyrosine kinase in experimental hemophilia A Sandrine Delignat,1,2,3 Jules Russick,1,2,3 Bagirath Gangadharan,1,2,3 Julie Rayes,1,2,3 Mathieu Ing,1,2,3 Jan Voorberg,4 Srinivas V. Kaveri1,2,3 and Sébastien Lacroix-Desmazes1,2,3

Haematologica 2019 Volume 104(5):1046-1054

1 INSERM, UMR S 1138, Centre de Recherche des Cordeliers, Paris, France; 2Université Pierre et Marie Curie-Paris 6, UMR S 1138, Centre de Recherche des Cordeliers, Paris, France; 3Université Paris Descartes, UMR S 1138, Centre de Recherche des Cordeliers, Paris, France and 4Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory, Academic Medical Center, University of Amsterdam, the Netherlands.

ABSTRACT

H Correspondence: SÉBASTIEN LACROIX-DESMAZES sebastien.lacroix-desmazes@crc.jussieu.fr Received: June 21, 2018.

Accepted: November 22, 2018.

Pre-published: December 13, 2018. doi:10.3324/haematol.2018.200279 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1046 ©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.

1046

emophilia A is a rare hemorrhagic disorder caused by the lack of functional pro-coagulant factor VIII. Factor VIII replacement therapy in patients with severe hemophilia A results in the development of inhibitory anti-factor VIII IgG in up to 30% of cases. To date, immune tolerance induction, with daily injection of large amounts of factor VIII, is the only strategy to eradicate factor VIII inhibitors. This strategy is, however, efficient in only 60-80% of patients. We investigated whether blocking Bcell receptor signaling upon inhibition of Bruton tyrosine kinase prevents anti-factor VIII immune responses in a mouse model of severe hemophilia A. Factor VIII-naïve and factor VIII-sensitized factor VIII-deficient mice were fed with the selective inhibitor of Bruton tyrosine kinase, (R)-5amino-1-(1-cyanopiperidin-3-yl)-3-(4-[2,4-difluorophenoxyl] phenyl)-1H pyrazole-4-carboxamide (PF-06250112), to inhibit B-cell receptor signaling prior to challenge with exogenous factor VIII. The consequences on the anti-factor VIII immune response were studied. Inhibition of Bruton tyrosine kinase during the primary anti-factor VIII immune response in factor VIII-naïve mice did not prevent the development of inhibitory anti-factor VIII IgG. In contrast, the anti-factor VIII memory B-cell response was consistently reduced upon treatment of factor VIII-sensitized mice with the Bruton tyrosine kinase inhibitor. The Bruton tyrosine kinase inhibitor reduced the differentiation of memory B cells ex vivo and in vivo following adoptive transfer to factor VIII-naïve animals. Taken together, our data identify inhibition of Bruton tyrosine kinase using PF-06250112 as a strategy to limit the reactivation of factor VIII-specific memory B cells upon rechallenge with therapeutic factor VIII.

Introduction Hemophilia A is a rare X-linked hemorrhagic disorder that results from suboptimal levels of pro-coagulant factor VIII (FVIII). Treatment or prevention of bleeding is managed by replacement therapy using therapeutic FVIII, which restores coagulation. However, in up to 30% of patients with severe hemophilia A administration of exogenous FVIII is complicated by the development of anti-FVIII antibodies that neutralize FVIII pro-coagulant activity and are referred to as ‘FVIII inhibitors’.1,2 To date, the most efficient strategy to eradicate inhibitors in inhibitor-positive patients with the severe form of the disease consists in repeated injections of high doses of FVIII and is referred to as ‘immune tolerance induction’ (ITI). Proposed mechanisms of action of ITI include the induction of protective anti-idiotypic antibodies that neutralize FVIII inhibitors, as observed in hemophilia A patients,3,4 and the inhibihaematologica | 2019; 104(5)

BTK inhibition prevents anti-FVIII memory response

tion of FVIII-specific memory B cells, as suggested from experiments in FVIII-deficient mice.5 ITI is, however, prohibitively costly, requires extreme compliance from the patients and their families, and is successful in only 6080% of cases.6–8 Direct depletion of B cells with the antiCD20 antibody rituximab (Mabthera®) is also used, although with limited success and unpredictable consequences in the long-term in populations of pediatric patients.9 The development of FVIII inhibitors results from the engagement of a classical T-cell-dependent immune response10 as evidenced by the presence of class-switched, high affinity anti-FVIII antibodies. B cells play key roles in primary T-cell-dependent immune responses, by forming and sustaining germinal centers, by differentiating into antibody-secreting plasma cells and possibly, as recently suggested, as antigen-presenting marginal zone (MZ) B cells involved in the initial stages of activation of immune effectors.11 During recall responses, memory B cells can be reactivated upon antigen encounter and differentiate into plasma cells, replenish the memory B-cell pool or participate as key professional antigen-presenting cells owing to a higher prevalence of the cells and to the expression of a higher affinity antigen-specific B-cell receptor (BCR).12,13 Antigen-specific B cells are thus potential targets to prevent primary or recall antigen-specific immune responses.

Engagement of a surface-exposed BCR by its cognate antigen triggers the formation of an intracellular signaling complex which enhances downstream signaling through the phosphorylation and ubiquitination of proteins.

Bruton tyrosine kinase (BTK) is a key proximal and ratelimiting component of the signaling cascade critical for Bcell activation, proliferation and survival.14 This cytosolic Tec kinase is activated only when BCR signaling promotes its recruitment at the inner cell membrane. Activated BTK in turn phosphorylates the phospholipase Cγ2, which leads to the downstream production of inositol triphosphate and diacylglycerol, resulting in calcium flux and finally to the activation of the NF-κB and NFAT-dependent pathways.16 BTK is a strategic therapeutic target for B-cell malignancies that require BTK signaling for cell survival, and for autoimmune diseases associated with the presence of pathogenic autoantibodies such as rheumatoid arthritis15 or lupus.16 Several small-molecule inhibitors of BTK have been developed.17 (R)-5-amino-1-(1cyanopiperidin-3-yl)-3-(4-[2,4-difluorophenoxyl] phenyl)1H pyrazole-4-carboxamide, or PF-06250112, is a selective potent, orally bioavailable, small-molecule inhibitor of BTK. PF-06250112 forms a covalent but reversible adduct with BTK upon binding to the Cys481 residue that is proximal to the ATP-binding pocket.18 Using a mouse model of severe hemophilia A, we evaluated the therapeutic potential of BTK inhibition by PF-06250112 on the development of a primary anti-FVIII immune response and on the FVIII-specific memory B-cell recall response.

Methods PF-06250112 formulation For in vitro studies, PF-06250112 (Pfizer, New York, NY, USA) was solubilized at 1 mg/mL in dimethylsulfoxide (Sigma-Aldrich, St Louis, MO, USA). For per os administration, PF-06250112 was prepared in 0.5% methylcellulose, 0.5% hydroxypropylmethylcellulose acetate succinate H grade and 20 mM Tris at pH 7.4. haematologica | 2019; 104(5)

Flow cytometry BTK inhibition with PF-06250112 was evaluated on splenocytes labeled with anti-CD86-FITC (BD Pharmingen, San Jose, CA, USA), and anti-B220-PE (BioLegend, San Diego, CA, USA).

Phenotypic analyses were performed using anti-CD4-Alexa 700, anti-CD19-Pacific blue (Biolegend), anti-CD45-APC (eBiosciences, San Diego, CA, USA); anti-CD3e-FITC, anti-CD11b-PE (BD Pharmingen), anti-CD21-APC Cy7, anti-CD23-PE Cy7 and antiGL7 Alexa 488 (Biolegend). Fluorescence activated cell sorting analysis was done on live cells using BD LSRII and FACSDiva software.

Treatment of factor VIII-deficient mice Seven to 11 week-old exon 16 FVIII-deficient mice on a C57Bl/6 background (a gift from Prof. H.H. Kazazian, Department of Genetics, University of Pennsylvania School of Medicine, PA, USA) were handled in agreement with ethical authority guidelines (experimentation on mice was approved by the Animal Ethics Committee, authorization #02058.04 granted by the Direction Générale de la Recherche et de l’Innovation). PF-06250112 and analogs have half-lives of about 7 h.18,19 For preventive treatment, mice were fed for 5 consecutive days per week, during 4 weeks, with PF-06250112 (15 mg/kg) or vehicle in order to cover about 10 FVIII half-lives.20 Mice were injected with human recombinant B domain-deleted FVIII (0.5 mg BDD-FVIII, Refacto, Pfizer) once a week, 2 h after the second feed of the week. Mice were bled 5 days after the fourth FVIII injection. To investigate the effect of PF06250112 on recall immune responses to FVIII, mice were injected intravenously with 0.5 mg of BDD-FVIII or phosphate-buffered saline (PBS) once a week for 4 weeks. Ninety days after the last FVIII injection, FVIII-sensitized mice with anti-FVIII circulating antibodies were fed for 5 days a week, during 2 weeks, with PF06250112 (15 mg/kg) or vehicle. Only one challenge with FVIII was performed with 1 mg of BDD-FVIII, 2 h after the second day of the first week of feeding. Mice were bled before and 7 and 14 days after FVIII challenge.

Evaluation of anti-factor VIII immune responses Anti-FVIII IgG in mouse serum were measured by enzymelinked immunosorbent assay (ELISA) and FVIII inhibitory titers were measured with a chromogenic assay (Siemens, Marburg, Germany).21 Proliferation of splenocytes was assessed in 96-well plates (0.25x106 cells/well) with concanavalin A (Sigma) or BDDFVIII for 72 h.22 Cell proliferation was measured by incorporation of [3H]-thymidine (0.5 mCi/well) for an additional 18 h, using a βcounter (Microbeta 1450, Perkin Elmer).

Stimulation of adoptively transferred memory B cells Spleens from FVIII-sensitized mice that had developed detectable anti-FVIII IgG (typically 80% of the mice) were collected 7 days after the fourth injection with 1 mg of BDD-FVIII. After lysis of erythrocytes, splenocytes were pooled and plasma cells were depleted using a goat anti-mouse CD138 polyclonal antibody (BD Pharmingen), conjugated to anti-goat IgG-coated Dynabeads (Thermo Fisher, Waltham, MA, USA).23 FVIII-naïve FVIII-deficient mice were then injected intravenously with 107 CD138-depleted splenocytes. Feeding of mice with PF-06250112 or vehicle was initiated 24 h after adoptive transfer for 5 consecutive days a week during 2 weeks and 1 mg of BDD-FVIII was injected intravenously only once 2 h after the first feed. Mice were bled 14 days after the FVIII injection.

Ex vivo memory B-cell differentiation assay Pooled splenocytes from five FVIII-sensitized mice having developed anti-FVIII IgG were cultured at 1.5x106 cells/mL in 1047

S. Delignat et al.

RPMI-1640, with 10% fetal calf serum (Thermo Fisher), 100 U/mL penicillin, 100 mg/mL streptomycin, and 50 mM 2-mercaptoethanol without CD138 depletion for 6 days. At day 0, PF06250112 or vehicle was added at different concentrations. FVIII (1 mg/mL Advate®, Shire, Dublin, Ireland) was added to the culture 2 h later. Advate was used instead of Refacto to follow the seminal protocol established by Hausl et al. as closely as possible.23 Of note, Refacto and Advate present with the same immunogenicity in FVIII-deficient mice.21,24 After 6 days, the formation of FVIII-specific antibody-secreting plasma cells was assessed by an enzymelinked immunospot (ELISPOT) assay with FVIII-coated plates (0.5 mg Advate/well). After 2 h of blocking with 10% fetal calf serum in RPMI medium, cells were incubated on the membrane and cultured overnight at 37°C in 5% CO2. After washing with PBS, 0.1% Tween 20, antibodies were detected using a goat anti-mouse IgG alkaline phosphatase-conjugated antibody (Southern Biotech, Birmingham, Alabama, USA), and antibody-secreting plasma cells were revealed with Sigmafast 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (BCIP/NBT) tablets (Sigma). Plates were scanned on an ImmunoSpot Analyzer (CTL, Shaker Heights, OH, USA). Spots were automatically counted by ImmunoSpot software using the SmartCount™ and Autogate™ functions.

Statistical analysis The statistical analysis was performed using the non-parametric two-tailed Mann-Whitney U test with a 95% confidence interval.

P values ≤0.05 were considered statistically significant (ns: not significant). GraphPad Prism version 5.0 (GraphPad Software, San Diego, CA, USA) was used for the statistical analysis.

A

Results PF-06250112 inhibits B-cell receptor signaling BCR engagement initiates an intracellular signaling cascade that induces the BTK-dependent activation of B cells.14 PF-06250112 has recently been described to inhibit BCR-mediated B-cell signaling, activation and proliferation.18 We confirmed the effect of PF-06250112 on splenic B cells by monitoring the upregulation of CD86 in response to B-cell triggering with anti-IgM or anti-IgD for in vitro and in vivo assays, respectively. Pre-treatment of purified splenic B cells by PF-06250112 prevented, in a dose-dependent manner, the induction of CD86 expression on B cells upon stimulation with anti-IgM F(ab’)2 fragments (Figure 1A). It did, however, marginally affect CD86 expression upon stimulation of the cells with an anti-CD40 antibody. The calculated half maximal inhibitory concentration (IC50) was 1.1±0.6 nM (Figure 1A, inset), which is in agreement with previous observations.18 The in vivo validation of the effect of PF-06250112 was performed using an anti-IgD antiserum instead of the antiIgM antibody in order to avoid quenching of the triggering antibody by endogenous circulating IgM. The treatment of FVIII-deficient mice with PF-06250112 2 h prior to the injection of the anti-IgD antiserum prevented anti-IgDmediated CD86 induction, as measured ex vivo on splenic B cells (Figure 1B), and only marginally affected antiCD40-mediated CD86 induction. The dose of 15 mg/kg of PF-06250112 was sufficient to prevent induction of CD86

B

Figure 1. Inhibition of splenic B-cell activation by PF-06250112. (A) Splenic B cells from C57Bl/6 mice, purified by negative selection, were plated in RPMI, 10% fetal calf serum and treated with PF-06250112 or vehicle (dimethylsulfoxide) for 2 h. F(ab’)2 fragments of goat anti-mouse IgM (10 mg/mL) or a monoclonal hamster anti-mouse CD40 antibody (5 mg/mL) were added. After 24 h, the expression of CD86 by live B220+ cells was analyzed by flow cytometry. Results are representative of two independent experiments performed in triplicate (mean±SD). Treatment of B cells in vitro with PF-06250112 suppressed induction of CD86 expression with an IC50 of 1.1±0.6 nM (Inset). (B) C57Bl/6 mice were fed with PF-06250112 or vehicle. After 2 h, goat anti-IgD antiserum (400 mL, eBiosciences), or rat anti-mouse CD40 IgG (100 mg, Biolegend) was injected intraperitoneally. Eighteen hours later, mice were sacrificed and the expression of CD86 by live splenic B220+ B cells was analyzed by flow cytometry. Results are representative of two independent experiments performed with three mice per group (mean±SD).

1048

haematologica | 2019; 104(5)

BTK inhibition prevents anti-FVIII memory response

expression in the mice and was used for further experiments.

Bruton tyrosine kinase inhibition does not prevent a primary anti-factor VIII immune response We then investigated the potential of PF-06250112 to prevent the development of a primary anti-FVIII immune response in FVIII-deficient mice (Figure 2A). The effect of 4 weeks of treatment with PF-06250112 was first evaluated on the splenic B- and T-cell compartments of FVIII-

injected mice. Analysis by flow cytometry revealed that chronic inhibition of BTK with PF-06250112 resulted in an increase in CD11b+ cells (P=0.040) (Figure 2B), which include monocytes, macrophages and natural killer cells, and in a significant decrease in CD4+ T cells (P=0.031) and in different B-cell subsets, including follicular and MZ B cells (P=0.038) (Figure 2C). Although statistically significant, changes in percentages of the cell populations were biologically marginal. The anti-FVIII IgG titers measured after the fourth injection of FVIII in PF-06250112-treated

A

B

E

C

F

D

G

Figure 2. Treatment with PF-06250112 does not prevent the onset of a primary anti-factor VIII immune response. (A) Experimental scheme for the preventive treatment of FVIII-deficient mice. Mice were fed with PF-06250112 (15 mg/kg) or vehicle for 5 days a week, and injected with FVIII once a week, 2 h after the second feed of the week. Mice were sacrificed 5 days after the fourth injection of FVIII (day 27). Spleens and sera were recovered. (B and C) At sacrifice, the isolated splenocytes were labeled with anti-CD45, anti-CD11b, anti-CD19, anti-CD3 and anti-CD4 antibodies (% of live CD45+ cells) (B). Follicular B cells (Fo) were identified as CD19+CD23highCD21low, marginal zone B cells (MZ) as CD19+CD21highCD23low, and germinal center B cells (GC) as CD19+GL7+ (% of live cells) (C). (D and E) Anti-FVIII IgG titers (D) and inhibitory titers (E) were measured in the sera of mice by ELISA with the mouse monoclonal anti-FVIII IgG mAb6 as a standard (expressed in arbitrary units, AU), and chromogenic assay, respectively. (F and G) Splenocytes were incubated for 72 h with FVIII (F) or with concanavalin A (ConA) (G). The incorporation of tritiated thymidine is depicted as counts per minute (CPM) after an additional 18 h of incubation (meanÂąSEM). Data are representative of two independent experiments with nine mice per group.

haematologica | 2019; 104(5)

1049

S. Delignat et al.

mice (565±170 AU, mean ± standard error of mean) (Figure 2D) were not statistically different from those in vehicle-treated mice (1133±376 AU, P=0.488). Similarly, inhibitory titers did not differ statistically between PF06250112-treated mice (123±88 AU) (Figure 2E) and control mice (457±219 AU, P=0.572). The capacity of splenic T cells to proliferate in the presence of FVIII or concanavalin A was similar in the two groups of mice (Figure 2F,G). Thus, inhibition of BTK at the time of FVIII injection does not prevent the onset of a primary anti-FVIII immune response.

Bruton tyrosine kinase inhibition alters the memory factor VIII-specific B-cell response We investigated the effect of BTK inhibition in the context of a memory anti-FVIII B-cell response, in which the FVIII-specific BCR in principle have a higher affinity for FVIII than BCR of naïve B cells. FVIII-deficient mice were injected with FVIII once a week for 4 weeks, and then left untouched for 90 days to allow the spontaneous elimina-

tion of FVIII-specific short-lived plasmocytes.25 FVIII-sensitized mice were then fed with vehicle or PF-06250112 and treated with FVIII as described in Figure 3A. AntiFVIII IgG titers measured prior to FVIII re-challenge were heterogeneous among mice (Figure 3B), as previously described.21,26,27 The levels of anti-FVIII IgG depicted in Figure 3C were, therefore, normalized with respect to the initial levels of anti-FVIII IgG measured for each individual mouse. Feeding mice with PF-06250112 significantly reduced the increase of the anti-FVIII IgG response, as compared to that in control mice (Figure 3C).

We then exploited an alternative model in order to study the effect of PF-06250112 in the absence of a potential bias provided by the presence of plasma cells and circulating anti-FVIII IgG. Splenocytes from FVIII-treated mice or -naïve mice were depleted of CD138+ plasmocytes and adoptively transferred to FVIII-naïve FVIII-deficient mice as previously described.23 Feeding of the recipient mice with PF-06250112 or vehicle was initiated on the next day, and one injection of FVIII was given 2 h after the

A

B

C

Figure 3. Treatment with PF-06250112 controls the recall response to factor VIII. (A) FVIII-deficient mice were injected with FVIII once a week for 4 weeks. After 90 days, mice were fed for 5 days a week, during 2 weeks, with PF-06250112 or vehicle, and injected with FVIII only once, 2 h after the second day feed. (B) Levels of anti-FVIII IgG were assessed by enzyme-linked immunosorbent assay using the mouse monoclonal anti-FVIII antibody mAb6 as a standard. Mice were then randomly attributed to either the PF-06250112- or the vehicle-fed groups. (C) Levels of anti-FVIII IgG were measured 7 and 14 days after the last injection of FVIII. IgG levels were normalized with respect to the initial levels of anti-FVIII IgG of each respective mouse measured at Bleed 0. Graphs depict means±SEM of three independent experiments with a total of 17 mice/group.

1050

haematologica | 2019; 104(5)

BTK inhibition prevents anti-FVIII memory response

first feed (Figure 4A). The treatment of mice with PF06250112 prevented the production of anti-FVIII IgG as compared to that in control mice (0.9±0.3 versus 124.0±115.7, P=0.011) (Figure 4B). Anti-FVIII IgG titers in PF-06250112-treated mice injected with FVIII were similar to those of control mice adoptively transferred with splenocytes from FVIII-sensitized mice injected with PBS and to those of control mice adoptively transferred with splenocytes from FVIII-naïve and FVIII-injected mice. We further performed re-stimulation of FVIII-specific memory B cells ex vivo. Splenocytes from FVIII-sensitized FVIIIdeficient mice having developed a humoral response to FVIII were isolated and pre-treated with PF-06250112 or vehicle before stimulation with FVIII. After 6 days of culture, FVIII-specific antibody-secreting plasma cells were

measured by ELISPOT. A statistically significant dosedependent reduction of antibody-secreting plasma cell formation was observed when splenocytes were pre-incubated with PF-06250112 as compared to splenocytes preincubated with vehicle, prior to stimulation with FVIII (P≤0.003) (Figure 5). Thus, inhibition of BTK prevents the activation of the FVIII-specific memory B-cell response.

Discussion The occurrence of FVIII inhibitors following FVIII replacement therapy remains a major clinical and societal challenge in hemophilia A. B cells are a key effector of the anti-FVIII immune response. Depending on their subtype

A

B

Figure 4. Treatment with PF-06250112 inhibits the anti-factor VIII memory B-cell response. (A) FVIII-deficient mice were injected with phosphate-buffered saline (PBS) or FVIII once a week for 4 weeks. Seven days after the last injection, mice were sacrificed and spleens were collected and pooled. CD138-depleted splenocytes from FVIII-treated or naïve (PBStreated) FVIII-deficient mice were adoptively transferred to naïve FVIII-deficient mice. From the next day onwards, host FVIII-deficient mice were fed during 2 weeks with 15 mg/kg of PF061250112 (18 mice) or vehicle (20 mice), daily for 5 days. Two control groups were injected with PBS or were adoptively transferred with splenocytes from naïve mice (7 mice per group).

Mice were injected with FVIII or PBS only once 2 h after the first feed. (B) Levels of anti-FVIII IgG were measured by enzyme-linked immunosorbent assay 14 days after FVIII challenge.

Means±SEM are depicted on the graph.

haematologica | 2019; 104(5)

1051

S. Delignat et al.

and localization and depending on the stage of the immune response, B cells play different roles as (i) antigenpresenting cells either in the initiation phase of an immune response (i.e., MZ B cells involved in FVIII capture and trafficking in the spleen),11 or in recall responses (higher number of FVIII-specific memory B cells with high-affinity FVIII-specific BCR), and (ii) precursors of antibodysecreting plasma cells (that produce FVIII inhibitors) localized in the spleen or in the bone marrow. Therefore, B cells represent an ideal therapeutic target either to prevent the initiation of anti-FVIII immune responses or to eradicate ongoing responses in inhibitor-positive patients.

Incidentally, the only strategies known to eradicate FVIII inhibitors in patients’ target B cells. ITI was shown to favor the development of anti-idiotypic antibodies that neutralize FVIII inhibitors3,4 and, at least in mice, eliminate FVIII-specific memory B cells.5 Likewise, the therapeutic anti-CD20 antibody rituximab specifically targets CD20expressing naïve and memory B cells, thus preventing the replenishment of the pool of plasma cells upon re-challenge with therapeutic FVIII. Both strategies fail to target long-lived plasma cells that reside in niches in the bone marrow and lack expression of CD20 and of the BCR.28 Attempts to target plasma cells in FVIII-deficient mice using either inhibitors of the proteasome29 or cocktails of immunosuppressive agents30 have met with limited success or lack of antigen specificity. Here we investigated a new approach to prevent the activation of antigen-specific B cells by inhibiting BTK, a kinase involved in upstream BCR signaling. Inhibition of BTK targets only those B cells that are activated at the time of administration of therapeutic FVIII, thus providing some degree of specificity for FVIII. We first validated that, upon BCR triggering, PF06250112 blocks BCR signaling in vitro and in vivo by measuring the induction of CD86 expression, a co-stimulation molecule involved in anti-FVIII immune responses.31 PF06250112 inhibited BCR signaling in vitro with an IC50 of about 1 nM, a value close to that reported previously.18 PF06250112 used at 15 mg/kg also inhibited the induction of CD86 in vivo following stimulation of splenic B cells with an anti-IgD serum. A marginal effect of PF-06250112 was observed on B cells stimulated by CD40 ligation, a BCRindependent pathway. This moderate alteration of CD40 signaling had also been reported in the case of PCI-32765, another BTK inhibitor; a possible interference of BTK inhibitors on CD40-induced NF-κB-signaling was evoked.32 As previously reported in an experimental model of systemic lupus erythematosus18 or using immunization of mice with the T-dependent model antigen SRBC,19 the regular per os administration of PF-06250112 to FVIII-deficient mice led to a statistically significant decrease in the percentages of MZ and follicular B cells. In contrast to the latter reports however, there was no change in percentages of splenic germinal center B cells, possibly owing to the use of a different antigen or of a different disease model.

In our hands, treatment of FVIII-deficient mice with PF06250112 failed to prevent the onset of a primary antiFVIII immune response. Exogenously administered FVIII was previously shown to closely associate with MZ macrophages and MZ B cells.11,33,34 The depletion of MZ macrophages or MZ B cells was independently shown to delay the onset of an anti-FVIII immune response.11,33 However, it is not clear whether reduced immune 1052

responses to FVIII following depletion of MZ macrophages or B cells result from the direct involvement of these cells in FVIII internalization and presentation to T cells, or trafficking to the germinal center, or from a mere disruption of the splenic architecture.35 The lack of effect of chronic BTK inhibition on the development of primary anti-FVIII humoral responses, despite an alteration of peripheral B-cell populations, including MZ B cells, suggests that MZ macrophages are the principal antigen-presenting cells at play in the onset of primary anti-FVIII immune responses, at least in mice. PF-06250112 feeding was also associated with a modest decrease in percentages of CD4+ T cells. PF-06250112 has been shown to be 10,000-fold more specific for BTK than for interleukin-2inducible T-cell kinase (ITK).18 In addition, there is no modulation of splenic T-cell populations in BTK-deficient mice.14 Hence, the decrease in the percentage of CD4+ T cells probably does not reflect a direct effect of PF06250112 on T cells in FVIII-sensitized mice, but rather a modulation of T-cell homeostasis consequent to a decrease in splenic follicular and MZ B cells. This illustrates a possible moderate off-target effect of PF-06250112 in naïve mice. Signaling through CD40 and CD40L is essential for the onset of primary immune responses, including in the case of therapeutic FVIII.36,37 The lack of effect of PF-06250112 on the primary anti-FVIII immune response presumably reflects the lack of effect of BTK inhibition on CD40 signaling.

The inhibition of BTK using PF-06250112 was able to

Figure 5. PF-06250112 inhibits the ex vivo differentiation of FVIII-specific memory B cells into antibody-secreting plasma cells. Splenocytes from FVIIIsensitized FVIII-deficient mice having developed a humoral response to FVIII were isolated 14 days after the last injection of FVIII. Pooled splenocytes were cultured for 6 days in the presence of 1 mg/mL FVIII. Splenocytes were incubated with vehicle or PF-06250112 2 h before adding FVIII. At day 6, newly differentiated anti-FVIII antibody-secreting plasma cells (ASCs) were detected by an enzyme-linked immunospot assay. The graph depicts means±SEM.

haematologica | 2019; 104(5)

BTK inhibition prevents anti-FVIII memory response

limit the memory B-cell response towards therapeutic FVIII. We first demonstrated that the inhibition of BTK in primed mice limits the increase in anti-FVIII IgG after a single re-challenge with FVIII. Plasma cells do not express BCR at their membrane and are, therefore, insensitive to BTK inhibition.19,38 The continuous endogenous production of anti-FVIII IgG by resident plasma cells may have introduced a bias in our experimental setup. Indeed, the formation of immune complexes between circulating anti-FVIII IgG and the administered FVIII neutralizes FVIII and alters its immunogenicity.39 In addition, immune complexes between FVIII and anti-FVIII IgG facilitate FVIII internalization by dendritic cells40 and could skew the target antigen-presenting cell population from memory B cells to dendritic cells or macrophages. In support of our initial observation, however, inhibition of BTK also prevented the in vivo re-activation of FVIII-specific memory B cells following adoptive transfer to FVIII-naïve mice and administration of FVIII, as well as the ex vivo differentiation of memory B cells into antibody-secreting plasma cells. Immune responses to FVIII in FVIII-deficient mice are notoriously heterogenous.21,26,27 This is illustrated by the fact that 80% of the mice developed primary immune responses to FVIII, and that memory responses were heterogenous in the three experimental models used in the present study. Of note, although the effect of the inhibition of BTK on the anti-FVIII memory response was heterogenous among mice, results from all three experimental models showed a significant reduction in the intensity of the anti-FVIII memory response. Our finding that BTK inhibition was efficient in blocking anti-FVIII recall but not primary immune responses is reminiscent of recent observations. Chronic inhibition of BTK with an analog of PF-06250112 was shown to prevent immune responses to T-independent antigens or to T-dependent antigens only provided that BCR had a strong affinity for the antigen, and provided that antigen ligation induced a strong signaling of the BCR.14,19 Memory B cells generally express BCR of high affinity for their cognate antigen owing to affinity maturation in the germinal centers. In the case of the anti-FVIII B-cell response, the rare human monoclonal anti-FVIII IgG studied to date were obtained following immortalization of memory B cells from inhibitor-positive patients. The affinity of the anti-C2 domain IgG4k BO2C11 for FVIII41 is 10-11 M-1 and that of the anti-C1

References 1. Ehrenforth S, Kreuz W, Scharrer I, Kornhuber B. Factor VIII inhibitors in haemophiliacs. Lancet. 1992;340(8813):253.

2. Lollar P. Pathogenic antibodies to coagulation factors. Part one: factor VIII and factor IX. J Thromb Haemost. 2004;2(7):1082– 1095.

3. Sakurai Y, Shima M, Tanaka I, Fukuda K, Yoshida K, Yoshioka A. Association of antiidiotypic antibodies with immune tolerance induction for the treatment of hemophilia A with inhibitors. Haematologica. 2004;89(6): 696–703.

4. Gilles JG, Desqueper B, Lenk H, Vermylen J, Saint-Remy JM. Neutralizing antiidiotypic antibodies to factor VIII inhibitors after desensitization in patients with hemophilia

haematologica | 2019; 104(5)

domain IgG4k LE2E9 is greater than the affinity of von Willebrand factor for FVIII (i.e., KD<0.1 nM).42 Likewise, the affinity of polyclonal anti-FVIII IgG from inhibitorpositive patients was described to be in the low nanomolar range.43 Indirectly, the efficacy of PF-06250112 at blocking the recall anti-FVIII immune response confirms that FVIII-specific memory B cells express high affinity BCR. A strong predictor of ITI failure is the intensity of the inhibitory titer peak that is reached within the 2 weeks that follow ITI initiation.45 The molecular and cellular mechanisms that underlie the reduced efficiency of ITI when the inhibitor peak is high after initiation of ITI are unknown. In particular, it is not understood whether a high level of circulating FVIII inhibitor reflects the quantitative or qualitative properties of the circulating anti-FVIII IgG, the affinity for FVIII of the BCR of memory B cells or the number of FVIII-specific memory B cells. Although the experimental protocols employed in the present work do not reflect the clinical situation, and although no animal model of ITI has been validated yet, we propose here that the inhibition of BTK-dependent BCR signaling as an adjunct therapy in patients starting ITI could increase the chances of the ITI being a success by preventing the reactivation of FVIII-specific memory B cells. Furthermore, the potential toxicity of PF-06250112 could be reduced by targeting the compound to FVIII-specific B cells using CD22L/FVIII-coated nanoparticles,44 thus paving the way towards their use in hemophilia A. Acknowledgments This study was supported by the Institut National de la Santé et de la Recherche Médicale (INSERM), Centre National de la Recherche Scientifique (CNRS), Université Pierre et Marie Curie (UPMC) Paris 6, and by a grant from Pfizer (Aspire Hemophilia research award 2014 WI185623). PF-06250112 was a gift from Pfizer according to a ‘Compound Transfer Program’-grant.

The monoclonal mouse FVIII heavy chain-specific IgG mAb6 was a kind gift from Prof Jean-Marie Saint-Remy (KUL, Leuven, Belgium). JR and MI were recipients of fellowships from Ministère de l'Enseignement Supérieur et de la Recherche. We would like to acknowledge Carole Lasne at the Centre d’Explorations Fonctionnelles (CEF) for technical support and the Centre d'Histologie, d'Imagerie et de Cytométrie (CHIC) (Centre de Recherche des Cordeliers, Paris, France).

A. J Clin Invest. 1996;97(6):1382–1388.

5. Hausl C, Ahmad RU, Sasgary M, et al. Highdose factor VIII inhibits factor VIII-specific memory B cells in hemophilia A with factor VIII inhibitors. Blood. 2005;106(10):3415– 3422.

6. Holstein K, Batorova A, Carvalho M, et al.

Current view and outcome of ITI therapy - a change over time? Thromb Res. 2016; 14838–14844.

7. Oldenburg J, Austin SK, Kessler CM. ITI choice for the optimal management of inhibitor patients - from a clinical and pharmacoeconomic perspective. Haemophilia.

2014;20 (Suppl) 617–626.

8. Coppola A, Di Minno MND, Santagostino E. Optimizing management of immune tolerance induction in patients with severe haemophilia A and inhibitors: towards evi-

9.

10.

11.

12.

13.

dence-based approaches. Br J Haematol.

2010;150(5):515–528.

Leissinger C, Josephson CD, Granger S, et al.

Rituximab for treatment of inhibitors in haemophilia A. A phase II study. Thromb Haemost. 2014;112(3):445–458.

Lacroix-Desmazes S, Navarrete A-M, André S, Bayry J, Kaveri SV, Dasgupta S. Dynamics of factor VIII interactions determine its immunologic fate in hemophilia A. Blood.

2008;112(2):240–249.

Zerra PE, Cox C, Baldwin WH, et al.

Marginal zone B cells are critical to factor VIII inhibitor formation in mice with hemophilia A. Blood. 2017;130(23):2559-2568.

Rajewsky K. Clonal selection and learning in the antibody system. Nature. 1996;381 (6585):751–758.

Kurosaki T, Aiba Y, Kometani K, Moriyama

1053

S. Delignat et al.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

1054

S, Takahashi Y. Unique properties of memory B cells of different isotypes. Immunol Rev. 2010;237(1):104–116.

Khan WN, Alt FW, Gerstein RM, et al.

Defective B cell development and function in Btk-deficient mice. Immunity. 1995;3(3): 283–299.

Di Paolo JA, Huang T, Balazs M, et al.

Specific Btk inhibition suppresses B cell- and myeloid cell-mediated arthritis. Nat Chem Biol. 2011;7(1):41–50.

Honigberg LA, Smith AM, Sirisawad M, et al. The Bruton tyrosine kinase inhibitor PCI32765 blocks B-cell activation and is efficacious in models of autoimmune disease and B-cell malignancy. Proc Natl Acad Sci U S A. 2010;107(29):13075–13080.

Hendriks RW, Yuvaraj S, Kil LP. Targeting Bruton’s tyrosine kinase in B cell malignancies. Nat Rev Cancer. 2014;14(4):219–232.

Rankin AL, Seth N, Keegan S, et al. Selective inhibition of BTK prevents murine lupus and antibody-mediated glomerulonephritis. J Immunol. 2013;191(9):4540–4550.

Benson MJ, Rodriguez V, von Schack D, et al. Modeling the clinical phenotype of BTK inhibition in the mature murine immune system. J Immunol. 2014;193(1):185–197.

Dumont JA, Liu T, Low SC, et al. Prolonged activity of a recombinant factor VIII-Fc fusion protein in hemophilia A mice and dogs. Blood. 2012;119(13):3024–3030.

Delignat S, Dasgupta S, André S, et al.

Comparison of the immunogenicity of different therapeutic preparations of human factor VIII in the murine model of hemophilia A. Haematologica. 2007;92(10):1423–1426.

Dimitrov JD, Dasgupta S, Navarrete A-M, et al. Induction of heme oxygenase-1 in factor VIII-deficient mice reduces the immune response to therapeutic factor VIII. Blood.

2010;115(13):2682–2685.

Hausl C, Ahmad RU, Schwarz HP, et al.

Preventing restimulation of memory B cells in hemophilia A: a potential new strategy for the treatment of antibody-dependent immune disorders. Blood. 2004;104(1):115– 122.

Delignat S, Repessé Y, Gilardin L, et al.

Predictive immunogenicity of Refacto AF. Haemophilia. 2014;20(4):486–492.

Nutt SL, Hodgkin PD, Tarlinton DM, Corcoran LM. The generation of antibody-

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

secreting plasma cells. Nat Rev Immunol.

2015;15(3):160–171.

Meeks SL, Cox CL, Healey JF, et al. A major determinant of the immunogenicity of factor VIII in a murine model is independent of its procoagulant function. Blood. 2012;120 (12):2512–2520.

Wroblewska A, van Haren SD, Herczenik E, et al. Modification of an exposed loop in the C1 domain reduces immune responses to factor VIII in hemophilia A mice. Blood.

2012;119(22):5294–5300.

Mahévas M, Michel M, Weill J-C, Reynaud C-A. Long-lived plasma cells in autoimmunity: lessons from B-cell depleting therapy.

Front Immunol. 2013; 4:494.

Meslier Y, André S, Dimitrov JD, et al.

Bortezomib delays the onset of factor VIII inhibitors in experimental hemophilia A, but fails to eliminate established anti-factor VIII IgG-producing cells. J Thromb Haemost.

2011;9(4):719–728.

Liu CL, Lyle MJ, Shin SC, Miao CH. Strategies to target long-lived plasma cells for treating hemophilia A inhibitors. Cell Immunol. 2016;30165–30173.

Qian J, Collins M, Sharpe AH, Hoyer LW. Prevention and treatment of factor VIII inhibitors in murine hemophilia A. Blood.

2000;95(4):1324–1329.

Herman SEM, Gordon AL, Hertlein E, et al.

Bruton tyrosine kinase represents a promising therapeutic target for treatment of chronic lymphocytic leukemia and is effectively targeted by PCI-32765. Blood.

2011;117(23):6287–6296.

Navarrete A, Dasgupta S, Delignat S, et al.

Splenic marginal zone antigen-presenting cells are critical for the primary allo-immune response to therapeutic factor VIII in hemophilia A. J Thromb Haemost. 2009;7(11): 1816–1823.

Lai JD, Cartier D, Hartholt RB, et al. Early cellular interactions and immune transcriptome profiles in human factor VIII-exposed hemophilia A mice. J Thromb Haemost.

2018;16(3):533-545.

Kraal G, Mebius R. New insights into the cell biology of the marginal zone of the spleen. In: International Review of Cytology. Academic Press; p175–215.

Rossi G, Sarkar J, Scandella D. Long-term induction of immune tolerance after block-

37.

38.

39.

40.

41.

42.

43.

44.

45.

ade of CD40-CD40L interaction in a mouse model of hemophilia A. Blood. 2001;97(9): 2750–2757.

Reipert BM, Sasgary M, Ahmad RU, Auer W, Turecek PL, Schwarz HP. Blockade of CD40/CD40 ligand interactions prevents induction of factor VIII inhibitors in hemophilic mice but does not induce lasting immune tolerance. Thromb Haemost.

2001;86(6):1345–1352.

Smith CI, Baskin B, Humire-Greiff P, et al.

Expression of Bruton’s agammaglobulinemia tyrosine kinase gene, BTK, is selectively down-regulated in T lymphocytes and plasma cells. J Immunol. 1994;152(2):557–565.

Meslier Y, André S, Teyssandier M, Kaveri SV, Lacroix-Desmazes S. Maternally transferred anti-factor VIII IgG reduce the antifactor VIII humoral immune response in factor VIII-deficient mice. Immunology.

2010;131(4):549–555.

Hartholt RB, Wroblewska A, Herczenik E, et al. Enhanced uptake of blood coagulation factor VIII containing immune complexes by antigen presenting cells. J Thromb Haemost.

2017;15(2):329–340.

Jacquemin MG, Desqueper BG, Benhida A, et al. Mechanism and kinetics of factor VIII inactivation: study with an IgG4 monoclonal antibody derived from a hemophilia A patient with inhibitor. Blood. 1998;92(2): 496–506.

Jacquemin M, Benhida A, Peerlinck K, et al.

A human antibody directed to the factor VIII C1 domain inhibits factor VIII cofactor activity and binding to von Willebrand factor. Blood. 2000;95(1):156–163.

Moreau A, Lacroix-Desmazes S, Stieltjes N, et al. Antibodies to the FVIII light chain that neutralize FVIII procoagulant activity are present in plasma of nonresponder patients with severe hemophilia A and in normal polyclonal human IgG. Blood. 2000;95(11): 3435–3441.

Macauley MS, Pfrengle F, Rademacher C, et al. Antigenic liposomes displaying CD22 ligands induce antigen-specific B cell apoptosis. J Clin Invest. 2013;123(7):3074–3083.

Hay CRM, DiMichele DM, International Immune Tolerance Study. The principal results of the International Immune Tolerance Study: a randomized dose comparison. Blood. 2012;119(6):1335–1344.

haematologica | 2019; 104(5)

ARTICLE

Stem Cell Transplantation

HLA discrepancy between graft and host rather than that graft and first donor impact the second transplant outcome

Yoshinobu Maeda,1* Tomotaka Ugai,2,3* Eisei Kondo,4 Kazuhiro Ikegame,5 Makoto Murata,6 Naoyuki Uchida,7 Toshihiro Miyamoto,8 Satoshi Takahashi,9 Kazuteru Ohashi,10 Hirohisa Nakamae,11 Takahiro Fukuda,12 Makoto Onizuka,13 Tetsuya Eto,14 Shuichi Ota,15 Makoto Hirokawa,16 Tatsuo Ichinohe,17 Yoshiko Atsuta,18 Yoshinobu Kanda3,19 and Junya Kanda;20 on behalf of the HLA Working Group of the Japan Society for Hematopoietic Cell Transplantation

1 Department of Hematology and Oncology, Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences; 2Division of Cancer Epidemiology and Prevention, Department of Preventive Medicine, Aichi Cancer Center Research Institute, Nagoya; 3Division of Hematology, Saitama Medical Center, Jichi Medical University, Tochigi; 4 Division of Hematology, Department of Medicine, Kawasaki Medical School, Okayama; 5 Division of Hematology, Department of Internal Medicine, Hyogo Medical College; 6 Department of Hematology and Oncology, Nagoya University Graduate School of Medicine; 7Department of Hematology, Federation of National Public Service Personnel Mutual Aid Associations Toranomon Hospital, Tokyo; 8Hematology, Oncology & Cardiovascular medicine, Kyushu University Hospital, Fukuoka; 9Division of Molecular Therapy, The Advanced Clinical Research Center, The Institute of Medical Science, The University of Tokyo; 10Hematology Division, Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome Hospital, Tokyo; 11Department of Hematology, Osaka City University Hospital; 12Department of Hematopoietic Stem Cell Transplantation, National Cancer Center Hospital, Tokyo; 13Department of Hematology/Oncology, Tokai University School of Medicine, Kanagawa; 14Department of Hematology, Hamanomachi Hospital, Fukuoka; 15Department of Hematology, Sapporo Hokuyu Hospital, Hokkaido; 16Department of General Internal Medicine and Clinical Laboratory Medicine, Akita University Graduate School of Medicine; 17Department of Hematology and Oncology, Research Institute for Radiation Biology and Medicine (RIRBM), Hiroshima University; 18Department of Healthcare Administration, Nagoya University Graduate School of Medicine; 19Division of Hematology, Department of Medicine, Jichi Medical University, Tochigi and 20Department of Hematology and Oncology, Graduate School of Medicine, Kyoto University, Japan

*YM and TU contributed equally to this work.

ABSTRACT

S

econd allogeneic hematopoietic stem cell transplantation is a curative treatment option for patients with hematologic malignancies.

However, it is unclear whether HLA discrepancy between graft and first donor has an impact on the outcome of second transplantation. We retrospectively analyzed 646 patients receiving second transplantation after an initial HLA mismatched transplantation. With regard to graft-versus-host, the one-allele mismatch (1 mismatch) group (SHR, 1.88; 95%CI: 0.79-4.45; P=0.163) and more than one-allele mismatch group (≥ 2 mismatch) (SHR, 1.84; 95%CI, 0.75–4.51; P=0.182) had higher risks of grade III–IV acute graft-versus-host disease (GvHD) compared to the HLA-matched (0 mismatch) group. In contrast, no difference in risk of acute GvHD was found among the 0, 1, and ≥ 2 mismatch group with respect to graft-versus-first donor. With regard to graft-versus-host, the ≥ 2 mismatch group showed a significantly higher risk of treatmentrelated mortality (SHR, 1.90; 95%CI, 1.04–3.50; P=0.038) compared to the 0 mismatch group, while the risk of relapse was slightly lower in the ≥ 2 mismatch group (SHR, 068; 95%CI, 0.44–1.06; P=0.086). In contrast, with regard to graft-versus-first donor, there were no significant differences in treatment-related mortality or relapse among the three groups.

These findings suggested that HLA discrepancy between graft and host induces transplant-related immunological responses in second transplantation leading to an increase in treatment-related mortality, in contrast, the biological effects of HLA discrepancy between graft and first donor on outcome may be negligible. haematologica | 2019; 104(5)

Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):1055-1061

Correspondence: YOSHINOBU MAEDA yosmaeda@md.okayama-u.ac.jp Received: August 20, 2018.

Accepted: November 23, 2018.

Pre-published: December 6, 2018. doi:10.3324/haematol.2018.204438 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1055 ©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.

1055

Y. Maeda et al.

Introduction For patients with malignant hematologic diseases who relapse after allogeneic hematopoietic stem cell transplantation (HSCT), a second HSCT is thought to be a curative option. It is believed that use of a second donor may confer increased therapeutic potency by inducing a more potent graft-versus-leukemia (GvL) effect; however, there are no data to support this assumption.1-9 In early studies, a second HSCT after a first HLA-matched transplantation was associated with similar risks of relapse and acute graft-versus-host disease (GvHD) using a different HLAmatched donor.3 There was no significant difference in survival between the transplantations from the original donor and another donor.

Over the years, the use of HLA-mismatched (MM) transplantation for hematologic diseases has increased, including haploidentical HSCT and cord blood transplantation (CBT). Following HLA-MM transplantation, a second donor is selected due to HLA discrepancy between the graft and the host. Physicians pay little attention to HLA discrepancy between the graft and the first donor, although the impact of this discrepancy on the outcome of second HSCT is unclear. Recipient non-hematopoietic gastrointestinal cells can express MHC class II, which is critical for inducing experimental acute GvHD in cases of minor histocompatibility antigen (mHAg) MM.10,11 In contrast, hematopoietic antigen presenting cells (APCs), especially dendritic cells, induce MHC class I-dependent acute GvHD in mHAg MM cases.12 Furthermore, in the MHC MM setting, hematopoietic APCs play an important role in the induction of both MHC class I- and II-dependent acute GvHD.13-16 As hematopoietic APCs are of first donor origin, HLA discrepancy between the graft and the first donor may be related to transplant-related immunological responses of the second HSCT. To elucidate the biological effects of HLA discrepancy between the graft and the first donor that impact the outcome of the second HSCT, we compared the effects of HLA-MM between the graft and the first donor to those between the graft and the host in 646 patients receiving a second HSCT after an initial HLA-MM transplantation.

data; or 5) received more than two HSCTs. The final study population consisted of 646 patients. The study was approved by the Data Management Committee of TRUMP and the Institutional Review Board of Okayama University.

Study end points The outcomes assessed included acute GvHD, chronic GvHD, neutrophil engraftment, transplant-related mortality (TRM), relapse, and overall survival (OS). Acute and chronic GvHD were diagnosed and graded using the standard criteria.20,21 Neutrophil engraftment was considered to have occurred when the absolute neutrophil count was ≥ 0.5x109 cells/L for 3 consecutive days.

Death from any cause was the event of interest in determining OS. TRM was defined as death during remission.

Statistical analysis Descriptive statistics were generated for patients' characteristics. Differences in characteristics between groups were evaluated by the c2 test and analysis of variance. The probability of OS was estimated according to the Kaplan-Meier method, and groups were compared using the log rank test. Subsequently, the probabilities of relapse, TRM, and acute and chronic GvHD were estimated on the basis of cumulative incidence curves.22 Competing events were death without relapse for relapse, relapse for TRM, death without engraftment for engraftment, and death without GvHD for acute or chronic GvHD. The groups were compared using Gray’s test.

To evaluate the impact of HLA discrepancy on transplant outcomes, we estimated the hazard ratios (HRs) or subhazard ratios (SHRs) and 95% confidence intervals (CIs) adjusted for potential confounders. The Cox proportional hazards model was used to evaluate the impact on OS, whereas multivariable competingrisks regression was used to evaluate the impact on the other end points. Several potential confounders considered in the multivariable analyses were provided in the Online Supplementary Appendix.

In all analyses, P<0.05 was considered statistically significant.

All statistical analyses are performed with Stata (v.15,0; Stata Corp., College Station, TX, USA) and EZR software (Saitama Medical Center, Jichi Medical University, Japan).23

Results Patients' and transplantation characteristics

Methods Study population Patients who were at least 16 years of age with acute myelogenous leukemia (AML), acute lymphoblastic leukemia (ALL), myelodysplastic syndrome (MDS), chronic myelogenous leukemia (CML), malignant lymphoma (ML), or other malignant hematologic disease, and who received a second HSCT after an initial HLA-MM transplantation, were included in this study.

Furthermore, patients must have received first and second allogeneic HSCTs between 1994 and 2016, with full HLA-A, -B, and -DRB1 allele data. Hematopoietic stem cell transplantation recipient clinical data were collected by the Japan Society for Hematopoietic Cell Transplantation (JSHCT) and the Japanese Data Center for Hematopoietic Cell Transplantation (JDCHCT) using the Transplant Registry Unified Management Program (TRUMP).17-19 We excluded individuals who: 1) first received HLA-matched HSCT; 2) received a second HSCT within 30 days after the first HSCT, in a planned manner or due to rejection/engraftment failure; 3) died within 30 days and lacked data on survival status and survival date; 4) lacked accurate allele 1056

A total of 646 patients who received a second HSCT after an initial HLA-MM transplantation were analyzed.

Patients' and transplantation characteristics are presented in Table 1. With respect to the HLA discrepancy in the graft-versus-host direction (graft vs. host), HLA matching was categorized as follows: HLA -A, -B, -DRB1 match (0 MM, n=85), MM at one allele (1 MM, n=160), or mismatch at more than one allele (≥2 MM, n=401). With regard to HLA discrepancy in the graft-versus-first donor direction (graft vs. first donor), the second HSCT was categorized as follows: HLA -A, -B, -DRB1 match (0 MM, n=72), mismatch at one allele (1 MM, n=100), or mismatch at more than one allele (≥2 MM, n=474). In the graft-versus-host comparison, the ≥2 MM group received cord blood more frequently (0 MM, 20.0%; 1 MM, 21.3%; ≥2 MM, 60.2%, P<0.001), were more likely to use a reduced-intensity conditioning regimen (0 MM, 56.5%; 1 MM, 66.3%; ≥2 MM, 70.7%, P=0.012), and had a higher rate of in vivo T-cell depletion (0 MM, 10.6%; 1 MM, 18.1%; ≥2 MM, 25.4%, P=0.003). The interval between the first and second HSCT was shorter in this group (<12 haematologica | 2019; 104(5)

HLA discrepancy and outcome of second HSCT

Table 1. Patients’ and donor characteristics.

HLA mismatch for graft-versus-host Match 1 allele ≥2 allele (N=85) mismatch mismatch (N=160) (N=401) N (%) N (%) N (%) Recipient age, years Median 43 (range) (18-73) HLA mismatch A allele 0 0.0 B allele 0 0.0 DR allele 0 0.0 Recipient gender (%) Male 47 55.3 Female 38 44.7 Donor/recipient gender Match 45 52.9 Male to Female 23 27.1 Female to Male 15 17.6 Unknown 2 2.4 Diagnosis Acute myeloid leukemia 45 52.9 Acute lymphoblastic leukemia 15 17.6 Chronic myeloid leukemia 6 7.1 Myelodysplastic syndrome 6 7.1 Malignant lymphoma 10 11.8 Others 3 3.5 Disease risk at transplant Standard risk 31 36.5 High risk 54 63.5 Stem cell source Bone marrow 37 43.5 Peripheral blood 18 21.2 Cord blood 17 20.0 Conditioning regimen Myeloablative 37 43.5 Reduced intensity 48 56.5 GvHD prophylaxis Cyclosporine based 22 25.9 Tacrolimus based 63 74.1 Others 0 0.0 In vivo T-cell depletion Yes 9 10.6 No 76 89.4 Year of transplant 1994-2010 30 35.3 2011-2016 55 64.7 Interval between first and second SCT <12 months 29 34.1 ≥12-23 months 23 27.1 ≥24 months 27 31.8 Missing 6 7.1 Interval between first SCT and first relapse <12 months 34 40.0 ≥2-12 months 32 37.6 ≥12 months 8 9.4 Missing 11 12.9

42 (16-69)

42 (16-70)

P

0.138

24 35 101

15.0 21.9 63.1

276 350 340

67.3 85.4 82.9

96 64

60.0 40.0

229 172

55.9 42.0

79 30 37 14

49.4 18.8 23.1 8.8

168 78 106 49

89 30 4 15 11 11

55.6 18.8 2.5 9.4 6.9 6.9

51 109

P

HLA mismatch for graft-versus-first donor Match 1 allele ≥2 allele (N=72) mismattch mismatch (N=100) (N=474) N (%) N (%) N (%) 42 (18-69)

44 (16-73)

42 (16-71)

0.240

0 0 0

0.0 0.0 0.0

23 26 51

23.0 26.0 51.0

339 429 427

71.5 90.5 90.1

0.740

47 25

65.3 34.7

59 41

59 41

266 208

56.1 43.9

0.326

41.0 19.0 25.9 12.0

0.183

32 17 20 3

44.4 23.6 27.8 4.2

49 19 21 11

49.0 19.0 21.0 11.0

211 95 117 51

44.5 20.0 24.7 10.8

0.852

254 81 9 20 22 15

62.0 19.8 2.2 4.9 5.4 3.7

0.049

38 12 4 6 6 6

52.8 16.7 5.6 8.3 8.3 8.3

56 20 4 8 7 5

56.0 20.0 4.0 8.0 7.0 5.0

294 94 11 27 30 18

62.0 19.8 2.3 5.7 6.3 3.8

0.556

31.9 68.1

110 291

26.8 71.0

0.200

24 48

33.3 66.7

34 66

34.0 66.0

134 340

28.3 71.7

0.406

52 14 34

32.5 8.8 21.3

78 149 247

19.0 36.3 60.2

<0.001

66 13 6

91.7 18.1 8.3

85 26 49

85.0 26.0 49.0

16 142 243

3.4 30.0 51.3

0.001

54 106

33.8 66.3

111 290

27.1 70.7

0.012

26 46

36.1 63.9

44 56

44.0 56.0

132 342

27.8 72.2

0.004

27 131 2

16.9 81.9 1.3

74 322 5

18.0 78.5 1.2

0.228

17 52 3

23.6 72.2 4.2

25 74 1

25.0 74.0 1.0

81 390 3

17.1 82.3 0.6

0.088

29 131

18.1 81.9

104 297

25.4 72.4

0.003

6 66

8.3 91.7

13 87

13.0 87.0

123 351

25.9 74.1

<0.001

55 105

34.4 65.6

118 283

28.8 69.0

0.372

27 45

37.5 62.5

41 59

41.0 59.0

135 339

28.5 71.5

0.025

71 45 37 7

44.4 28.1 23.1 4.4

206 106 67 22

50.2 25.9 16.3 5.4

0.008

23 19 18 12

31.9 26.4 25.0 16.7

38 28 27 7

38.0 28.0 27.0 7.0

245 127 86 16

51.7 26.8 18.1 3.4

0.029

67 57 18 18

41.9 35.6 11.3 11.3

175 145 32 52

42.7 35.4 7.8 12.7

0.793

29 23 10 10

40.3 31.9 13.9 13.9

35 42 7 16

35.0 42.0 7.0 16.0

212 166 41 55

44.7 35.0 8.6 11.6

0.217

N: number; GvHD: graft-versus-host disease; SCT: stem cell transplantation.

haematologica | 2019; 104(5)

1057

Y. Maeda et al.

Table 2. Effect of HLA allele mismatch on acute graft-versus-host disease (GvHD), chronic GvHD and engraftment in multivariate analyses.

HLA mismatch for graft-versus-host Match 1 allele ≥2 allele (N=85) mismatch mismatch (N=160) (N=401)

HLA mismatch for graft-versus-first donor Match 1 allele ≥2 allele (N=72) mismatch mismatch (N=100) (N=474)

Grades III to IV acute GvHD SHR1 (95%CI)

1 (ref)

1.88 (0.79-4.45, P=0.163)

1.84 (0.75-4.51, P=0.182)

1 (ref)

0.84 0.91 (0.35-2.02, P=0.669) (0.43-1.93, P=0.800)

Chronic GvHD SHR1 (95%CI)

1 (ref)

1.45 (0.84-2.50, P=0.181)

1.20 (0.60-2.38, P=0.605)

1 (ref)

0.98 0.91 (0.55-1.76, P=0.956) (0.54-1.51, P=0.702)

Neutrophil engraftment SHR1 (95%CI)

1 (ref)

0.81 (0.62-1.06, P=0.126)

0.77 (0.56-1.05, P=0.097)

1 (ref)

1.06 1.23 (0.75-1.48, P=0.753) (0.92-1.66, P=0.167)

Adjusted for recipient age at transplant (continuous), recipient gender, gender mismatch (match, male to female, female to male, unknown), diagnosis (acute myeloid leukemia, acute lymphoblastic leukemia, chronic myeloid leukemia, myelodysplastic syndrome, malignant lymphoma or others), disease risk at transplant (standard or high), stem cell source (bone marrow, peripheral blood, cord blood), conditioning regimen (myeloablative or reduced intensity), graft-versus-host disease (GvHD) prophylaxis (cyclosporine based, tacrolimus based, others), in vivo T-cell depletion (Yes, No), year of transplant (1994-2010, 2011-2016), interval between first and second stem cell transplantation (SCT) (<12 months, ≥12-23 months, ≥24 months, missing) and interval between first SCT and relapse (<2 months, ≥2-12 months, ≥12 months, missing). SHR: subdistribution hazard ratios.

1

months: 0 MM, 34.1%; 1 MM, 44.4%; ≥ 2 MM, 50.2%, P=0.008). With regard to graft-versus-first donor comparison, the ≥2 MM group showed a similar trend to that for the graft-versus-host group comparison. The ≥2 MM group required more cord blood (P=0.001), used a reduced-intensity conditioning regimen (P=0.004), and had greater in vivo T-cell depletion (P<0.001) and a shorter interval between the first and second HSCT (P=0.029).

Acute graft-versus-host disease, chronic graft-versus-host disease, and engraftment With regard to the graft-versus-host results, the unadjusted cumulative incidence rates of grade III-IV acute GvHD at 100 days post transplantation were 9.5% (95%CI: 4.417.0%) in the 0 MM group, 13.8% (95%CI: 9.0-19.7%) in the 1 MM group, and 11.0% (95%CI: 8.2-14.3%) in the ≥2 MM group (Figure 1). In multivariate analysis, the 1 MM group (SHR, 1.88; 95%CI: 0.79-4.45; P=0.163) and ≥2 MM group (SHR, 1.84; 95%CI: 0.75-4.51; P=0.182) tended to have higher risk of grade III-IV acute GvHD compared to the 0 MM group, although the results were not statistically significant (Table 2). With regard to affected organ, the risk of skin, gut and liver acute GvHD increased among the 1 MM group and ≥2MM group compared to the 0 MM group (Table 3). There was no statistically significant difference in risk of chronic GvHD among the groups in multivariate analysis. The cumulative incidence rate of neutrophil engraftment at day 50 was 94.0% (95%CI: 85.6-97.6%) in the 0 MM group, 96.9% (95%CI: 92.398.7%) in the 1 MM group, and 91.0% (95%CI: 87.793.4%) in the ≥2 MM group. In multivariate analysis, the ≥2 MM group tended to show delayed engraftment compared to the 0 MM group (SHR, 0.77; 95%CI: 0.56-1.05; P=0.097).

With regard to the graft-versus-first donor results, there were no significant differences in the risk of grade III-IV acute GvHD, chronic GvHD, or neutrophil engraftment among the groups in multivariate analysis (Table 2).

Next, the association of each HLA allele MM with 1058

GvHD was evaluated (Online Supplementary Table S1).

With regard to graft-versus-host, B allele MM was associated with an increased risk of grade III-IV acute GvHD in multivariate analysis (SHR, 2.87; 95%CI: 1.42-5.79; P=0.003), and DR allele MM was associated with delayed neutrophil engraftment (SHR, 0.80; 95%CI: 0.67-0.95, P=0.011); no such associations were found for the other MM types. With regard to the graft-versus-first donor results, no HLA allele MM showed an association with grade III-IV acute GvHD, chronic GvHD, or neutrophil engraftment in multivariate analysis.

Transplant-related mortality, relapse, and overall survival With regard to the graft-versus-host results, the unadjusted cumulative incidence rates of TRM and relapse at 5 years post transplantation were 19.8% (95%CI: 11.829.2%) and 55.6% (95%CI: 43.9-65.7%) in the 0 MM group, 32.5% (95%CI:25.2-39.9%) and 45.2% (95%CI: 37.3-52.8%) in the 1 MM group, and 34.7% (95%CI: 30.039.4%) and 46.8% (95%CI: 41.8-51.7%) in the ≥2 MM group, respectively (Figures 2 and 3). Multivariate analysis indicated that the risk of TRM was marginally higher in the 1 MM group (SHR, 1.67; 95%CI: 0.94-2.98; P=0.081), and significantly higher in the ≥2 MM group (SHR, 1.90; 95%CI: 1.04-3.50; P=0.038), versus the 0 MM group. In contrast, the risk of relapse was slightly lower in both the 1 MM group (SHR, 073; 95%CI: 0.50-1.07; P=0.110) and the ≥2 MM group (SHR, 068; 95%CI: 0.44-1.06; P=0.086).

Consequently, no significant differences in OS were found among the three groups in multivariate analyses (Table 4).

Analysis of each HLA allele MM revealed that only HLA DR allele MM was significantly associated with a lower risk of relapse (SHR, 0.75; 95%CI: 0.58-0.95; P=0.018) and a higher risk of TRM (SHR, 1.44; 95%CI: 1.03-2.00; P=0.033) (Online Supplementary Table S2). The main causes of TRM differed among the three groups. The rates of interstitial pneumonia, TMA, and especially acute GvHD haematologica | 2019; 104(5)

HLA discrepancy and outcome of second HSCT

Figure 1. The unadjusted cumulative incidence of grades III to IV acute graft-versus-host disease (GvHD) by HLA mismatch (MM) for graft. With regard to the graft-versus-host results, the unadjusted cumulative incidence rates of grade III-IV acute GvHD were 9.5% (95%CI: 4.4-17.0%) in the 0 MM group, 13.8% (95%CI: 9.0%–19.7%) in the 1 MM group, and 11.0% (95%CI: 8.2%– 14.3%) in the ≥ 2 MM group.

Figure 2. The unadjusted cumulative incidence of relapse by HLA mismatch (MM) for graft-versus-host. With regard to the graft-versus-host results, the unadjusted cumulative incidence rates of relapse were 55.6% (95%CI: 43.965.7%) in the 0 MM group, 45.2% (95%CI: 37.3-52.8%) in the 1 MM group, and 46.8% (95%CI: 41.8-51.7%) in the ≥ 2 MM group.

(0 MM, 0.0%; 1 MM, 11.8%; ≥2 MM 10.9%) were increased in the 1 MM group and the ≥2 MM group (Online Supplementary Table S3). With regard to the graft-versus-first donor outcomes, there were no significant differences in TRM, relapse, or OS among the three groups (Table 4). In addition, no allele MM was associated with relapse, TRM, or OS in the analysis of each HLA allele MM (Online Supplementary Table S2).

Analyses by stem cell sources Finally, we performed analyses according to stem cell source (Online Supplementary Tables S4 and 5). We did not observe any obvious statistically heterogeneity among stem cell sources. However, the small sample size for some categories partially precluded evaluation of significance.

Discussion There have been several studies on the role of donor change in the outcome of second HSCT; however, these studies were performed mainly in HLA-matched or 1 AgMM cases and focused on procedures in which a second HSCT from the same donor was performed.1-9 In this study, we evaluated the role of HLA discrepancy between the graft and host and between the graft and the first donor on the outcome of second HSCT after HLA-MM initial HSCT. On evaluating 646 recipients of a second HSCT, it was found that graft-host HLA-match was associated with a reduced rate of TRM compared to HLAMM, while HLA discrepancy between the graft and the first donor had no impact on the outcome of second HSCT. In the largest retrospective analysis performed to date (n=1285 patients) to compare the incidence of GvHD in the same cohort, the incidence rate of grade II-IV acute GvHD in first HSCT was 26% versus 46% in second HSCT.24 In our study, the incidence of grade II-IV and haematologica | 2019; 104(5)

Figure 3. The unadjusted cumulative incidence of transplant-related mortality (TRM) by HLA mismatch (MM) for graft-versus-host. With regard to the graft-versus-host results, the unadjusted cumulative incidence rates of treatment-related mortality TRM were 19.8% (95%CI: 11.8-29.2%) in the 0 MM group, 32.5% (95%CI: 25.2-39.9%) in the 1 MM group, and 34.7% (95%CI: 30.0-39.4%) in the ≥ 2 MM group.

grade III-IV acute GvHD for first HSCT was 36.4% and 9.0% versus 34.2% and 11.2%, respectively, for second HSCT. Due to the higher rate of GvHD for second HSCT, prevention of acute GvHD represents an important, and as yet unmet, medical need.

Experimental murine studies reported that hematopoietic APCs play an important role in the induction of acute GvHD in an MHC MM setting.13-16 In the present study, HLA-MM between the graft and first donor was not associated with an increased risk of acute GvHD in HSCT recipients having hematopoietic 1059

Y. Maeda et al.

Table 3. Effect of HLA allele mismatch on Grades III to IV acute graft-versus-host disease (GvHD) by affected organ.

Skin GvHD SHR1 (95%CI) Gut GvHD SHR1 (95%CI) Liver GvHD SHR1 (95%CI)

Match (N=85)

HLA mismatch for graft-versus-host 1 allele mismatch (N=160)

≥2 allele mismatch (N=401)

1 (ref)

2.49 (0.87-7.13, P=0.088)

2.94 (0.94-9.19, P=0.063)

1 (ref)

3.33 (0.90-12.3, P=0.072)

3.14 (0.82-12.0, P=0.094)

1 (ref)

2.16 (0.53-8.85, P=0.283)

3.24 (0.73-14.4, P=0.122)

Adjusted for recipient age at transplant (continuous), recipient gender, gender mismatch (match, male to female, female to male, unknown), diagnosis (acute myeloid leukemia, acute lymphoblastic leukemia, chronic myeloid leukemia, myelodysplastic syndrome, malignant lymphoma or others), disease risk at transplant (standard or high), stem cell source (bone marrow, peripheral blood, cord blood), conditioning regimen (myeloablative or reduced intensity), graft-versus-host disease (GvHD) prophylaxis (cyclosporine based, tacrolimus based, others), in vivo T-cell depletion (Yes, No), year of transplant (1994-2010, 2011-2016), interval between first and second stem cell transplantation (SCT) (<12 months, ≥12-23 months, ≥24 months, missing) and interval between first SCT and relapse (<2 months, ≥2-12 months, ≥12 months, missing). SHR: subdistribution hazard ratios.

1

Table 4. Effect of HLA allele mismatch on transplant-related mortality, relapse and overall survival in multivariate analyses.

HLA mismatch for graft-versus-host Match 1 allele ≥2 allele (N=85) mismatch mismatch (N=160) (N=401)

HLA mismatch for graft-versus-first donor Match 1 allele ≥2 allele (N=72) mismatch mismatch N=100) (N=474)

Transplant-related mortality SHR1 (95%CI)

1 (ref)

1.67 (0.94-2.98, P=0.081)

1.90 (1.04-3.50, P=0.038)

1 (ref)

0.89 (0.52-1.52, P=0.665)

0.67 (0.42-1.07, P=0.095)

Relapse SHR1 (95%CI)

1 (ref)

0.73 (0.50-1.07, P=0.110)

0.68 (0.44-1.06, P=0.086)

1 (ref)

1.18 (0.72-1.95, P=0.516)

1.41 (0.89-2.22, P=0.143)

Overall survival HR1 (95%CI)

1 (ref)

1.00 (0.72-1.41, P=0.952)

1.21 (0.84-1.73, P=0.313)

1 (ref)

0.84 (0.57-1.21, P=0.347)

0.85 (0.61-1.17, P=0.313)

*Bold denotes statistical significance. 1Adjusted for recipient age at transplant (continuous), recipient gender, gender mismatch (match, male to female, female to male, unknown), diagnosis (acute myeloid leukemia, acute lymphoblastic leukemia, chronic myeloid leukemia, myelodysplastic syndrome, malignant lymphoma or others), disease risk at transplant (standard or high), stem cell source (bone marrow, peripheral blood, cord blood), conditioning regimen (myeloablative or reduced intensity), graft-versus-host disease (GvHD) prophylaxis (cyclosporine based, tacrolimus based, others), in vivo T-cell depletion (Yes, No), year of transplant (1994-2010, 2011-2016), interval between first and second stem cell transplantation (SCT) (<12 months, ≥12-23 months, ≥24 months, missing) and interval between first SCT and relapse (<2 months, ≥2-12 months, ≥12 months, missing). SHR: subdistribution hazard ratios; HR: hazard ratios.

APCs originating from the first donor. The antigen-presenting function of the first-donor hematopoietic cells may be insufficiently strong to induce GvHD. An alternative explanation is that recipient hematopoietic APCs have a limited capacity to induce acute GvHD, possibly owing to their predisposition to induce donor T-cell death.11 In contrast, HLA discrepancy between the graft and host may impact the risk of acute GvHD during the second transplant. In this study, HLA-MM between the graft and host showed increased risk of grade III-IV acute GvHD, although the results were not significant. In addition, B allele MM was significantly associated with an increased risk of grade III-IV acute GvHD in the analysis of each HLA allele mismatch [relative risk (RR) 2.87, 95%CI: 1.42-5.79; P=0.003]. Several experimental studies showed that non-hematopoietic gastrointestinal cells are able to express MHC class II and induce CD4+ T-cell-dependent acute GvHD.10,11 As the antigen-presenting function of epithelial cells is enhanced in the presence of an inflammatory environment, epithelial cells after the first HSCT could play a major role in inducing GvHD following second HSCT, although further studies are needed to validate this.

The length of remission after first HSCT and the disease 1060

status at second HSCT, are two main independent prognostic factors for predicting the outcome of a second HSCT.2,3,5 Despite a significant increase in the proportion of patients of advanced age, having an advanced disease stage, and receiving alternative donor transplants, there has been a continual decrease in TRM, reflecting the impact of advances in supportive care and more widespread use of reduced-intensity conditioning regimens.

However, the reduction in rate of TRM has been less obvious in patients following a second remission or refractory disease.25 Due to more advanced disease and accumulating toxicity, second transplants are more problematic than first transplants, and often result in an increase in TRM and overall mortality rates. Attempted enhancement of the GvT effect by switching donor may be affected by the toxicity of the second HSCT. Reducing TRM remains one of the most significant challenges in second HSCT. Our analysis showed that HLA-MM between the graft and first donor had no influence on GvHD, relapse, TRM, or OS. In contrast, with regard to graft-versus-host, the risk of TRM was significantly higher in the ≥2 MM group versus the 0 MM group (RR, 1.90; 95%CI: 1.04-3.50; P=0.038).

Analysis of each HLA allele MM revealed that the DR allele MM was significantly associated with a lower rate haematologica | 2019; 104(5)

HLA discrepancy and outcome of second HSCT

of relapse versus the 0 MM group (RR, 0.75; 95%CI: 0.580.95; P=0.018), but this was offset by a higher rate of TRM (RR, 1.44; 95%CI: 1.03-2.00; P=0.033). Our data suggested that use of an HLA-MM donor may induce a more potent GvL effect, but also increases the allogeneic responses of the second HSCT and provokes an increase in TRM events. These effects tended to cancel each other out in respect to OS. This is the first study to focus on patients after initial HLA-MM transplantation and identify risk factors for a poor second HSCT outcome. However, several limitations of the study should be mentioned. First, although this was a relatively large-scale study on second transplant, the sample size was still modest, and therefore further studies with larger sample sizes are required. Second, it used a retrospective design and included a heterogeneous patient group. Moreover, the strategies of the different treatment centers with respect to donor change are unknown, and any heterogeneity in transplantation procedure, year of transplant, and patients' characteristics may have biased the results, although we attempted to reduce bias by adjusting for these factors in multivariate analyses. Third, we did not adjust for multiple comparisons and therefore caution is required when interpreting the results, in partic-

References 1. Bosi A, Laszlo D, Labopin M, et al. Second allogeneic bone marrow transplantation in acute leukemia: results of a survey by the European Cooperative Group for Blood and Marrow Transplantation. J Clin Oncol.

2001;19(16):3675-3684.

2. Shaw BE, Mufti GJ, Mackinnon S, et al.

Outcome of second allogeneic transplants using reduced-intensity conditioning following relapse of haematological malignancy after an initial allogeneic transplant. Bone Marrow Transplant. 2008;42(12):783-789.

3. Eapen M, Giralt SA, Horowitz MM, et al.

Second transplant for acute and chronic leukemia relapsing after first HLA-identical sibling transplant. Bone Marrow Transplant.

2004;34(8):721-727.

4. Christopeit M, Kuss O, Finke J, et al. Second allograft for hematologic relapse of acute leukemia after first allogeneic stem-cell transplantation from related and unrelated donors: the role of donor change. J Clin Oncol. 2013; 31(26):3259-3271.

5. Ruutu T, de Wreede LC, van Biezen A, et al.

Second allogeneic transplantation for relapse of malignant disease: retrospective analysis of outcome and predictive factors by the EBMT. Bone Marrow Transplant. 2015; 50(12):1542-1550.

6. Vrhovac R, Labopin M, Ciceri F, et al. Second reduced intensity conditioning allogeneic transplant as a rescue strategy for acute leukaemia patients who relapse after an initial RIC allogeneic transplantation: analysis of risk factors and treatment outcomes. Bone Marrow Transplant. 2016;51(2):186-193.

7. Orti G, Sanz J, Bermudez A, et al. Outcome of Second Allogeneic hematopoietic cell transplantation after relapse of myeloid malignancies following allogeneic hematopoietic cell transplantation: a Retrospective Cohort on Behalf of the Grupo Espanol de Trasplante Hematopoyetico. Biol Blood Marrow

haematologica | 2019; 104(5)

ular those of the stratified analyses. In addition, HLA-C typing and high-resolution DNA typing were either rarely, or not routinely, performed on the donors. Finally, donor chimerism was not systematically analyzed and cell subset chimerism data were not available for most patients.

In conclusion, HLA-MM donor is an option after initial HLA-MM transplantation. However, TRM remains a challenge, particularly with a â&#x2030;Ľ2 MM donor regarding graft-versus-host. In this study, the biological effects of HLA discrepancy between the graft and the first donor on the outcome appeared negligible, and our findings shed light on the role of non-hematopoietic APCs on transplant-related immunological responses.

Funding This work was supported in part by the Practical Research Project for Allergic Diseases and Immunology (Research Technology of Medical Transplantation) from Japan Agency for Medical Research and Development, AMED under Grant Number 18ek0510023h0002. The authors are grateful to all physicians and data managers at the centers who contributed valuable data on transplantation to the JMDP and TRUMP. The authors also thank the members of the data management committees of JDMP and TRUMP for their assistance.

Transplant. 2016;22(3):584-588.

8. Andreola G, Labopin M, Beelen D, et al.

Long-term outcome and prognostic factors of second allogeneic hematopoietic stem cell transplant for acute leukemia in patients with a median follow-up of 10 years. Bone Marrow Transplant. 2015;50(12):1508-1512.

9. Horstmann K, Boumendil A, Finke J, et al.

Second allo-SCT in patients with lymphoma relapse after a first allogeneic transplantation.

A retrospective study of the EBMT Lymphoma Working Party. Bone Marrow Transplant. 2015;50(6):790-794.

10. Jones SC, Murphy GF, Friedman TM, Korngold R. Importance of minor histocompatibility antigen expression by nonhematopoietic tissues in a CD4+ T cell-mediated graft-versus-host disease model. J Clin Invest. 2003;112(12):1880-1886.

11. Koyama M, Kuns RD, Olver SD, et al.

Recipient nonhematopoietic antigen-presenting cells are sufficient to induce lethal acute graft-versus-host disease. Nat Med. 2011; 18(1):135-142.

12. Shlomchik WD, Couzens MS, Tang CB, et al.

Prevention of graft versus host disease by inactivation of host antigen-presenting cells.

Science. 1999;285(5426):412-415.

13. Teshima T, Ordemann R, Reddy P, et al.

Acute graft-versus-host disease does not require alloantigen expression on host epithelium. Nat Med. 2002;8(6):575-581.

14. Duffner UA, Maeda Y, Cooke KR, et al. Host dendritic cells alone are sufficient to initiate acute graft-versus-host disease. J Immunol.

2004;172(12):7393-7398.

15. Zhang Y, Louboutin JP, Zhu J, Rivera AJ, Emerson SG. Preterminal host dendritic cells in irradiated mice prime CD8+ T cell-mediated acute graft-versus-host disease. J Clin Invest. 2002;109(10):1335-1344.

16. Reddy P, Maeda Y, Liu C, Krijanovski OI, Korngold R, Ferrara JL. A crucial role for antigen-presenting cells and alloantigen expression in graft-versus-leukemia responses. Nat

Med. 2005;11(11):1244-1249.

17. Atsuta Y, Suzuki R, Yoshimi A, et al.

Unification of hematopoietic stem cell transplantation registries in Japan and establishment of the TRUMP System. Int J Hematol.

2007;86(3):269-274.

18. Atsuta Y. Introduction of Transplant Registry Unified Management Program 2 (TRUMP2): scripts for TRUMP data analyses, part I (variables other than HLA-related data). Int J Hematol. 2016;103(1):3-10.

19. Kanda J. Scripts for TRUMP data analyses.

Part II (HLA-related data): statistical analyses specific for hematopoietic stem cell transplantation. Int J Hematol. 2016;103(1):11-19.

20. Przepiorka D, Weisdorf D, Martin P, et al.

1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant.

1995;15(6):825-828.

21. Sullivan KM, Agura E, Anasetti C, et al.

Chronic graft-versus-host disease and other late complications of bone marrow transplantation. Semin Hematol. 1991;28(3):250259.

22. 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.

23. Kanda Y. Investigation of the freely available easy-to-use software 'EZR' for medical statistics. Bone Marrow Transplant. 2013; 48(3):452-458.

24. Duncan CN, Majhail NS, Brazauskas R, et al.

Long-term survival and late effects among one-year survivors of second allogeneic hematopoietic cell transplantation for relapsed acute leukemia and myelodysplastic syndromes. Biol Blood Marrow Transplant.

2015;21(1):151-158.

25. Bacigalupo A, Sormani MP, Lamparelli T, et al. Reducing transplant-related mortality after allogeneic hematopoietic stem cell transplantation. Haematologica. 2004; 89(10):12381247.

1061

ARTICLE Ferrata Storti Foundation

Cell Therapy & Immunotherapy

Human stem cells transplanted into the rat stroke brain migrate to the spleen via lymphatic and inflammation pathways

Kaya Xu,1,2 Jea-Young Lee,1 Yuji Kaneko,1 Julian P. Tuazon,1 Fernando Vale,1 Harry van Loveren1 and Cesario V. Borlongan1

Center of Excellence for Aging and Brain Repair, Department of Neurosurgery and Brain Repair, University of South Florida College of Medicine, Tampa, FL, USA and 2Department of Neurosurgery, The Affiliated Hospital of Guizhou Medical University, China

1

Haematologica 2019 Volume 104(5):1062-1073

ABSTRACT

D

Correspondence: CESARIO V. BORLONGAN cborlong@health.usf.edu Received: September 11, 2018.

Accepted: November 30, 2018.

Pre-published: December 4, 2018. doi:10.3324/haematol.2018.206581 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1062 ©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.

1062

espite mounting evidence of a massive peripheral inflammatory response accompanying stroke, the ability of intracerebrally transplanted cells to migrate to the periphery and sequester systemic inflammation remains unexamined. Here, we tested the hypothesis that human bone marrow mesenchymal stromal cells intracerebrally transplanted in the brain of adult rats subjected to experimental stroke can migrate to the spleen, a vital organ that confers peripheral inflammation after stroke. Sham or experimental stroke was induced in adult SpragueDawley rats by a 1 hour middle cerebral artery occlusion model. One hour after surgery, rats were intracerebrally injected with human bone marrow mesenchymal stromal cells (3×105/9 mL), then euthanized on day 1, 3, or 7 for immunohistochemical assays. Cell migration assays were performed for human bone marrow mesenchymal stromal cells using Boyden chambers with the bottom plate consisting of microglia, lymphatic endothelial cells, or both, and treated with different doses of tumor necrosis factor-a. Plates were processed in a fluorescence reader at different time points. Immunofluorescence microscopy on different days after the stroke revealed that stem cells engrafted in the stroke brain but, interestingly, homed to the spleen via lymphatic vessels, and were propelled by inflammatory signals. Experiments using human bone marrow mesenchymal stromal cells co-cultured with lymphatic endothelial cells or microglia, and treated with tumor necrosis factor-a, further indicated the key roles of the lymphatic system and inflammation in directing stem cell migration. This study is the first to demonstrate brain-to-periphery migration of stem cells, advancing the novel concept of harnessing the lymphatic system in mobilizing stem cells to sequester peripheral inflammation as a brain repair strategy.

Introduction Ischemic stroke continues to stand as a leading cause of death and disability worldwide, with an ongoing need for effective therapies.1 Cell-based therapies have emerged as a promising modality for stroke treatment, yet a complete understanding of their mechanisms remains elusive.2-4 The study of stem cell therapy for stroke has focused primarily on the effects of the grafted cells within the local brain tissue, despite the recognition of a peripheral inflammatory response exacerbating the pathological outcomes in the stroke brain.5,6 Following stroke, a compromised blood-brain barrier (BBB) allows peripheral major histocompatibility complex class II (MHC-II)-positive immune cells – including neutrophils, T cells, and monocytes/macrophages7 – to infiltrate the brain parenchyma, perpetuating a state of cerebral inflammation.8-10 Pharmacological and cell-based anti-inflammatory methods which attenuate cerebral and systemic inflammation have been shown to improve stroke outcomes.11,12 Thus, an understanding of how stem cells sequester and modulate peripheral inflammation is key for furthering the application of stem haematologica | 2019; 104(5)

Brain to spleen stem cell migration

cell therapies in stroke and other neurological disorders with pathologies characterized by aberrant inflammation.

The spleen is a major contributor to the peripheral inflammatory response observed following stroke.13,14 Acting as a reservoir for leukocytes, the spleen is the primary disseminator of inflammatory cells in response to injury.15 This splenic response, paired with the compromised BBB following stroke, contributes to the infiltration of pro-inflammatory mediators into the brain and worsened outcomes.16-18 We have previously reported that human bone marrow mesenchymal stromal cells (hBMSC) delivered intravenously preferentially migrate to the spleen, dampening systemic inflammation.19 These findings support the therapeutic potential of targeting the peripheral inflammatory response via the spleen to abrogate neuroinflammation, in addition to implicating stem cells as inflammation-homing biologics.

In light of the spleen and peripheral inflammation being principal culprits in neuroinflammatory-induced cell death processes20,21 the recently characterized cerebral lymphatic system opens a new avenue of research in stem cell therapies for neurological disorders.22 Cognizant that the spleen is a major destination for lymphatic drainage, the cerebral lymphatic system could serve as an efficient route for brain-to-spleen stem cell migration. To date, this notion of intracerebrally transplanted stem cells migrating remotely away from the implantation sites in ischemic regions, albeit outside the brain, has not been investigated.

Here, we report for the first time that stem cells can migrate from the cerebrum to the periphery via lymphatic vessels, likely amplified by stroke-induced local and peripheral inflammation. This line of investigation advances the concept of targeting the source of the peripheral inflammatory response by harnessing lymphatic vessel-directed migration of stem cells. The present study also provides valuable data toward a novel understanding of how intracerebral transplantation of stem cells functions to repair the damaged brain through peripheral effectors.

carotid artery for 30 min. In contrast, mild stroke was induced by 60 min occlusion of the right internal carotid artery, without ligating the left common carotid artery.

Cell preparation hBMSC and bEnd.3-expressing lymphatic vessel endothelial hyaluronan receptor-1 (LYVE-1) were purchased from the American Type Culture Collection (ATCC, VA, USA). The immortalized BV-2 murine microglial cells24 were maintained in Dulbecco modified Eagle medium (Gibco, MA, USA). Immortalized bEnd.3 and BV-2 cells were used to better manage the growth of these cells in culture over longer periods of time.25 For transplantation preparation, hBMSC density was adjusted to 7.5×106 cells in 216 mL of phosphate-buffered saline. For cell migration, the cell density was adjusted to 1×106 cells in 5 mL fluorescent medium (ThermoFisher, MA, USA). Thereafter, cells were incubated with 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindodicarbocyanine perchlorate (DiD, Invitrogen, OR, USA) for 30 min, to aid the visualization of hBMSC migration.

Transplantation One hour after middle cerebral artery occlusion surgery, rats were anesthetized and hBMSC were injected intracerebrally into the striatum and cortex of the ischemic hemisphere over 10 min as previously described.26 Within a single needle pass, two deposits of hBMSC were made in the striatum (DV=5.0 mm and DV=4.0 mm) and one deposit of hBMSC was made in the cortex (DV=3.0 mm) for a total of three separate deposits (AP=+0.5 mm, ML=+2.8 mm, DV=5.0 mm, 4.0 mm, and 3.0 mm).26 Each deposit contained 1×105 cells/3 mL phosphate-buffered saline for a total of 3×105 cells suspended in 9 mL phosphate-buffered saline for the entire transplant regimen per animal.

Brain and organ harvesting, fixation, and sectioning Rats were euthanized under deep anesthesia on day 1, day 3, or day 7 after transplantation for ex vivo imaging analysis, as described in our past study.21

Measurement of infarct area Methods

Hematoxylin and eosin staining was performed to confirm the core infarct injury of our stroke model as shown in our previous studies.24,27

Animals and housing All experiments were approved by the Institutional Animal Care and Use Committee of the University of South Florida, Morsani College of Medicine and were conducted in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the United States Public Health Service's Policy on Humane Care and Use of Laboratory Animals.

All experiments were carried out on 2-month old Sprague– Dawley male rats (Harlan Laboratories, Indianapolis, IN, USA) and rats were either exposed to sham (n=6) or stroke surgery, with the latter further classified as mild (n=9) or severe (n=9) based on the severity of the stroke as evidenced by pathological outcomes.

There were six animals in the sham-treated group, nine in the mild stroke group, and nine in the severe stroke group across all treatments, and all animals were treated with hBMSC.

Stroke surgery Animals underwent middle cerebral artery occlusion surgery as described in our previous study.23 Sham surgery involved exposing and isolating the common carotid and internal carotid arteries before closing the incision. Severe stroke was induced by 60 min intraluminal filament occlusion of the right internal carotid artery with simultaneous ligation of the contralateral (left) common haematologica | 2019; 104(5)

Immunohistochemistry assays Human nuclei (HuNu) and OX6 staining was performed as described in our previous study.21 Details of the immunohistochemistry assays are described in the Online Supplementary Material.

Cell migration assay The cell migration assay was performed using a FalconTM FlouroBlokTM 96-well HTS insert system with 3.0 mm pores (Life Science, NC, USA). BV2 and bEnd.3 cells were fed with fresh growth medium at the bottom of the lower chamber in a 96-well plate. There were three groups in the bottom cell: BV2, bEnd.3, and BV2+bEnd.3. The cell density was adjusted to 2×104 cells in 200 mL growth medium/well overnight. Cells were treated with different doses of tumor necrosis factor-alpha (TNF-a; 0 ng/mL, 25 ng/mL, 50 ng/mL, or 100 ng/mL) in an incubator overnight. The details of the cell migration assay are described in the Online Supplementary Material.

Statistical analysis All data are expressed as the mean ± standard error of mean and statistically evaluated using one-way or two-way analysis of vari1063

K. Xu et al.

ance (ANOVA) followed by a Bonferroni test (GraphPad version 5.01). Comparisons with a P value <0.05 were considered statistically significant.

between the groups with mild stroke and between the groups with severe stroke (P<0.001) (Figure 2C,D). The detection of surviving hBMSC in the brain and spleen was most robust on day 3, with the highest number of cells identified in the groups with severe stroke (Figure 2C,D).

Results Infarct sizes Hematoxylin and eosin staining indicated the presence of infarcts in the animals with mild and severe stroke.

Percentages of infarct lesion area were rated accordingly: mild ≤10%, and severe >10% (Figure 1B). The groups with severe stroke had significantly higher percentages of infarct area than those with mild stroke at all time points (P<0.001; P<0.01). Furthermore, transplanting hBMSC grafts did not reduce infarct sizes and no significant outliers were detected across the treatment groups (Figure 1A).

Survival of human bone marrow mesenchymal stromal cells in the brain and spleen Confocal microscopy with HuNu staining was used to analyze hBMSC survival and detected positive HuNu expression in brains transplanted with hBMSC (Figure 2A). Additionally, HuNu-positive cells in the spleen indicated that intracerebrally injected stem cells migrated from the brain to the spleen (Figure 2B). On day 1, the number of HuNu-positive cells in the brains of the group with severe stroke was significantly higher than that of the group with mild stroke and the sham-treated group (P<0.001). On day 3, the numbers of HuNu-positive cells in the brains of the groups with mild and severe stroke were significantly higher than the number in the shamtreated group P<0.01) (Figure 2C). On all days, the groups with mild and severe stroke had significantly higher numbers of hBMSC in the spleen compared to the numbers in the spleen of the sham-treated group (P<0.001) (Figure 2D). When comparing between days and within experimental groups, there were no significant differences in either the brain or the spleen, between the sham-treated groups (P>0.05) but there were significant differences

A

Human bone marrow mesenchymal stromal cells were visualized within lymphatic vessels in the brain and spleen To further elucidate whether transplanted stem cells could migrate from the brain to the peripheral immune organs through the lymphatic system, we co-stained for HuNu to visualize hBMSC, and LYVE-1 to visualize lymphatic endothelial cells. In all animals, few HuNu-positive cells were found within brain lymphatic vessels, which expressed LYVE-1, at any time point (Figure 3A).

Additionally, there were no significant differences in the quantity of HuNu and LYVE-1 co-localization in the brain at any time point between sham-treated animals and those with mild or severe stroke (P>0.05) (Figure 3C). In contrast, many hBMSC were visualized within all transplanted rats’ splenic lymphatic vessels on all days (Figure 3B). Animals with mild and severe strokes displayed significantly more hBMSC within splenic lymphatic vessels than did sham-treated animals on any day on which measurements were made (P<0.05) (Figure 3D). On day 3, the group with severe stroke expressed the greatest amount of co-localized cells in the spleen (P<0.001).

Stroke induces neuroinflammation in the brain and spleen To reveal any upregulation of stroke-induced neuroinflammation, we stained brain and spleen sections with OX6 to label microglia possessing the proinflammatory M1 phenotype (Figure 4A,B). OX6 expression in the brain and spleen was significantly higher in the groups with mild and severe stroke than in the sham-treated group at all time points (P<0.01; P<0.001) (Figure 4C,D). On day 3, the brains and spleens of animals groups with mild and severe stroke, but not those of the sham-treated group, exhibited significantly more OX6-positive cells than they

B

Figure 1. Infarct sizes in a rat model of stroke. (A) Representative hematoxylin and eosin–stained brain sections from rats that had undergone sham surgery (sham), surgery to induce a mild stroke (mild) or surgery to induce a severe stroke (severe). One hour after sham, mild or severe stroke surgery, rats were intracerebrally injected with human bone marrow mesenchymal stromal cells. Brains from animals with mild or severe stroke show ischemic lesions. Red circles indicate infarct areas. (B) Bar graph depicting infarct size in stroke animals. The infarct area in the ipsilateral hemisphere is expressed as a percentage of the area of the contralateral hemisphere. Values are indicated as means ± standard error of mean. Significance bars: *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

1064

haematologica | 2019; 104(5)

Brain to spleen stem cell migration

did on other days, with the highest concentration found in the group with severe stroke (P<0.001) (Figure 4C,D).

Human specific phagocytic marker CD68, anti-apoptosis inhibitor 5 and neuronal marker triple staining in the brain and spleen

OX6-positive cells localized near or within lymphatic vessels in the brain and spleen

Stroke-induced ischemic and apoptotic neurons result in inflammation. To test whether transplanted hBMSC can phagocytose ischemic neuronal cells in the brain and migrate to the periphery, we stained brain and spleen sections with human specific phagocytic marker CD68, antiapoptosis inhibitor 5, and anti-160kD neurofilament medium antibody-rat specific neuronal marker. The human specific phagocytic marker CD68 was employed to label the transplanted hBMSC demonstrating phagocytic activity, the anti-apoptosis inhibitor 5 was utilized to mark rat neurons undergoing apoptosis, and the anti160kD neurofilament medium antibody-rat specific neuronal marker was used to label rat neurons. CD68 expression was negligible for these hBMSC in vitro, suggesting that phagocytic expression may be triggered by the ischemic microenvironment of the transplant site. In the brain and in the spleen, apoptotic neuronal cells were found inside phagocytic human cells (Figure 6A,B). The amounts of cells positive for all three staining labels were

OX6-positive cells localized near or within lymphatic vessels in the brain (Figure 5A) and spleen (Figure 5B) and their quantity was higher in animals with mild or severe stroke than in sham-treated animals. Groups with mild or severe stroke had more OX6-positive cells close to or within lymphatic vessels on day 3 than at other time points. OX6 expression near or within lymphatic vessels in the brain and spleen was significantly higher in the groups with mild and severe stroke than in the shamtreated group at all time points (P<0.05; P<0.01; P<0.001) (Figure 5C,D). On day 3, the brains and spleens of the groups with mild and severe stroke, but not sham-treated group brains and spleens, exhibited significantly more OX6-positive cells next to or within lymphatic vessels than they did on other days, with the highest concentration found in the group with severe stroke (P<0.01) (Figure 5C,D).

A

C

B

D

Figure 2. Survival of human bone marrow mesenchymal stromal cells in the brain and spleen. (A) In the brain, transplanted human bone marrow mesenchymal stromal cells (hBMSC) express antigens for human nuclei (HuNu). The hBMSC grafts in the brain were positively stained with HuNu. (B) Representative merged images showing co-localization of HuNu-positive cells and 4,6-diaminodino-2-phenylindole (DAPI) for hBMSC in the spleen. (A and B) Arrow heads indicate HuNu-positive cells. Scale bars = 100 mm. Green: HuNu; blue: DAPI. (C and D) Quantitative analyses of the estimated number of HuNu-positive hBMSC in the brain (C) and in the spleen (D) of stroke and sham-treated animals revealed more HuNu-positive cells survived in the brain and migrated to the spleen on day 3, especially in the groups with severe stroke. Significance bars: **P<0.01; ***P<0.001. (C) a: The group with mild stroke had significantly more HuNu-positive cells in the brain on day 3 than on other days (P<0.01); b: the group with severe stroke had significantly more HuNu-positive cells in the brain on day 3 than on other days (P<0.01). (D) a: The group with mild stroke had significantly more HuNu-positive cells in the spleen on day 3 than on other days (P<0.001); b: the group with severe stroke had significantly more HuNu-positive cells in the spleen on day 3 than on other days (P<0.001).

haematologica | 2019; 104(5)

1065

K. Xu et al.

significantly higher in brains and spleens of the groups with mild and severe stroke than in sham-treated group brains and spleens (P<0.001) (Figure 6C,D). Animals with mild or severe stroke, but not sham-treated animals, had significantly more triple stain-positive cells in both organs on day 3, relative to other days, especially the group with severe stroke (P<0.05) (Figure 6C,D).

Human bone marrow mesenchymal stromal cells migrate toward bEnd.3, BV2, and bEnd.3+BV2 cells, a process enhanced by tumor necrosis factor-a treatment hBMSC migrated toward lymphatic endothelial cells (bEnd.3, expressing LYVE-1), microglia (BV2), and a combination of bEnd.3+BV2 cells in in vitro cell migration assays (Figure 7A-F). Administering TNF-a escalated hBMSC migration in a dose-dependent manner. At different time points (0 h - 72 h), the density of hBMSC migrating toward bEnd.3 (Figure 7A,B), BV2 (Figure 7C,D), and bEnd.3+BV2 (Figure 7E,F) increased significantly with pro-

gressively higher concentrations of TNF-a (P<0.05).

Additionally, the density of hBMSC migration to bEnd.3, BV2, and bEnd.3+BV2 cells increased over time (Figure 7B,D,F). Cell migration to bEnd.3+BV2 peaked at 48 h with the highest TNF-a dose (Figure 7F), which is consistent with the in vivo results showing that the number of hBMSC was highest in the group with severe stroke on day 3 (Figure 2D). BV2 cells had the highest density of hBMSC migration relative to bEnd.3 and bEnd.3+BV2 groups at all time points, demonstrating that hBMSC preferentially migrated to BV2 cells (P<0.01) (Figure 7G-J).

Discussion The brain is traditionally known to play an essential role in governing and coordinating systemic homeostasis. In recent years, it has become increasingly clear that brain health and neurological diseases are intimately associated with other physiological systems.28,29 Growing evidence

A

B

C

D

Figure 3. Visualization of human bone marrow mesenchymal stromal cells within lymphatic vessels in the brain and spleen. (A and B) Human bone marrow mesenchymal stromal cells (hBMSC) were stained with human nuclei (HuNu) and lymphatic endothelial cells were stained with lymphatic vessel endothelial hyaluronan receptor-1 (LYVE-1) to reveal co-localization of hBMSC and lymphatic endothelial cells in the brain (A) and spleen (B). Arrow heads indicate co-localization of HuNupositive and LYVE-1-positive cells and show that hBMSC localized within lymphatic vessels in the brain and spleen. Small boxes show 40x magnification. Scale bars = 100 mm. Red: LYVE-1; green: HuNu; blue: DAPI. (A) Images were taken close to the dural sinuses in the brain. (B) Images were taken near to the gate of the spleen, close to the white pulp in the spleen. (C and D) Quantitative analyses of the estimated number of co-localized HuNu-positive and LYVE-1-positive cells in the brain (C) and in the spleen (D) of animals with mild or severe stroke and sham-treated animals. Significance bars: *P<0.05; **P<0.01; ***P<0.001. (C) No significant differences were found in the magnitude of co-localization in the brain between sham-treated groups and those with mild or severe stroke on all days that the assessments were made (P>0.05). (D) Groups with mild and severe stroke displayed significantly higher quantities of co-localization in the spleen relative to the sham-treated group on all days the measurements were made (*P<0.05). Staining overlap was most prominent on day 3, especially in the groups with severe stroke. a: The group with severe stroke had significantly more co-localized cells in the spleen on day 3 than on other days (P<0.001).

1066

haematologica | 2019; 104(5)

Brain to spleen stem cell migration

suggests that an immense peripheral inflammatory response follows stroke, yet investigations into stem cell therapy have not probed whether stem cells transplanted intracerebrally are capable of migrating to the periphery and alleviating systemic inflammation. Previously, we administered hBMSC intravenously in experimental rat models of stroke and the transplanted cells preferentially migrated to the spleen, demonstrated greater survival in the spleen than in the brain, and ameliorated strokeinduced neurostructural deficits and chronic inflammation.19 Here, we showed that lymphatic pathways and inflammatory signals enable stem cells to migrate to the spleen after intracerebral transplantation in a stroke brain (Figure 8).

Preferential migration toward the site of pathological signals is critical for stem cell transplantation to achieve its therapeutic potential. Such deposition of systemically delivered stem cells into the spleen after stroke is accompanied by reduced necrotic and apoptotic cell death in the brain, decreased motor and cognitive deficits, and a dampened splenic inflammatory response.12,30-33 This sequestration of neurodegeneration by suppressing systemic inflammation originating from the spleen was previously

demonstrated in stroke animals that had their spleen removed or were transplanted with human umbilical cord cells in the acute stage of stroke.12,30 The present in vivo imaging revealed the migration of intracerebrally transplanted hBMSC to the spleen in the acute stage. The highest numbers of these cells were found on day 3 in both brain and spleen. Transient blockage of the contralateral common carotid artery caused more intense ischemia in the brain tissue of groups with severe stroke, producing much higher levels of inflammatory factors. This increased inflammatory response heightens the migratory action of the transplanted stem cells. In animal models of focal cerebral ischemic stroke, recruitment of multiple inflammatory cell types such as neutrophils within the ischemic brain occurs within 30 minutes to a few hours after the stroke, peaking within the first 3 days.34,35 hBMSC have immunomodulatory capacities, are multipotent, and tend to migrate to sites of tissue injury/inflammation, making them promising effectors for tissue regeneration.36-40 Lymphatic vessels, in addition to draining interstitial fluids, allow cells to travel from tissues to draining lymph nodes.41 Therefore, we examined whether the cerebral

A

C

B

D

Figure 4. Immunofluorescent staining of inflammatory cells. (A and B) OX6-stained microglia/macrophages were frequently found in the periphery of the site of stem cell injection site in the brain (A) and around the blood vessels in the spleen (B), and the number of OX6-positive cells in the stroke groups was higher than in the sham-treated groups. Nuclei were stained with DAPI. Arrow heads indicate OX6-positive cells. Scale bars = 100 mm. Green: OX6; blue: DAPI. (C and D) In quantitative analyses of the brain (C) and the spleen (D), the stroke groups exhibited a higher number of OX6-positive cells than did the sham-treated group (**P<0.01).

Significance bars: **P<0.01; ***P<0.001. (C) a: The group with mild stroke had significantly more OX6-positive cells in the brain on day 3 than on other days (P<0.01); b: the group with severe stroke had significantly more OX6-positive cells in the brain on day 3 than on other days (P<0.001). (D) a: The group with mild stroke had significantly more OX6-positive cells in the spleen on day 3 than on other days (P<0.01); b: the group with severe stroke had significantly more OX6-positive cells in the spleen on day 3 than on other days (P<0.001).

haematologica | 2019; 104(5)

1067

K. Xu et al.

lymphatic vessels were capable of carrying hBMSC to the spleen. Indeed, the most novel finding in this study is the discovery that hBMSC were able to migrate directly to the spleen via lymphatic vessels; immunofluorescence analysis of the brain and spleen revealed that hBMSC were found within these vessels.

Microglial cells – the resident macrophages of the brain – are activated rapidly in response to brain injury.42,43 Experimental data have shown that resident microglia are activated within minutes of the onset of ischemia and produce a plethora of proinflammatory mediators including interleukin-1β and TNF-a, which exacerbate tissue damage44 yet may also protect the brain against ischemic and excitotoxic injury.45,46 Post-ischemic microglial proliferation peaks at 48–72 h after the onset of cerebral ischemia and may last for several weeks after the initial injury.47,48 Given the preferential migration of hBMSC toward microglia under inflammatory conditions seen in vitro, it was hypothesized that the microglia may play a key role in facilitating the transplanted stem cells’ journey into the spleen. Elevated levels of OX6-positive cells were found in both the brain and spleen after stroke, especially in animals with severe stroke. The density of inflammatory cells in the spleen was much higher than in the brain.

Greater stroke severity was correlated with a rise in OX6positive cells and their increased co-localization with

LYVE1 in both the brain and the spleen. The data reveal a more robust migration of OX6-positive cells from the brain to the spleen in response to a more severe stroke.

Thus, the present results suggest that elevated inflammation accompanying more severe strokes may also account for the greater successful migration of transplanted stem cells to the spleen. Hence, it is likely that hBMSC possess biodistribution patterns in which they localize around major sites of inflammation, such as the infarct area in the brain and the white pulp in the spleen, which contains numerous inflammatory cells and lymphatic cells that could attract these hBMSC.49 Indeed, high concentrations of hBMSC were deposited in the white pulp of the spleen.

We advance the premise that tracking the biodistribution of intracerebrally transplanted hBMSC beyond the spleen, and to other peripheral organs (e.g., thymus) which also mount strong inflammation in response to central nervous system insults, including stroke, represents an innovative paradigm-shifting future investigation (e.g., for brain disorders with robust peripheral pathological components, thereby requiring peripheral sequestration of systemic inflammation).

Normally, the BBB functions as an impermeable barricade to defend against detrimental pathogens and substances. However, inflammation induced by an ischemic stroke can lead to impaired endothelial activity and com-

A

C

B

D

Figure 5. Localization of OX6-positive cells near or within lymphatic vessels in the brain and spleen. (A and B) Staining with OX6, lymphatic vessel endothelial hyaluronan receptor-1 (LYVE-1), and DAPI, was performed to visualize microglia/macrophages, lymphatic endothelial cells, and nuclei, respectively, in the brain (A) and spleen (B). Arrow heads indicate co-localization of OX6-positive and LYVE-1-positive cells and show that microglia/macrophages localized near or within lymphatic vessels in the brain and spleen. Pictures taken under 20x and 40x magnification. Scale bars = 100 mm. Red: LYVE-1; green: OX6; blue: DAPI. (A) Images were taken close to the dural sinuses in the brain. (B) Images were taken near the gate of the spleen, close to the white pulp in the spleen. (C and D) In quantitative analyses of the brain (C) and the spleen (D), the stroke groups exhibited higher numbers of co-localized cells than did the sham-treated group (**P<0.01). Significance bars: *P<0.05; **P<0.01; ***P<0.001. (C) a: The group with mild stroke had significantly more co-localization in the brain on day 3 than on other days (P<0.05); b: the group with severe stroke had significantly more co-localized cells in the brain on day 3 than on other days (P<0.01). (D) a: The group with mild stroke had significantly more co-localization in the spleen on day 3 than on other days (P<0.05); b: the group with severe stroke had significantly more co-localized cells in the spleen on day 3 than on other days (P<0.01).

1068

haematologica | 2019; 104(5)

Brain to spleen stem cell migration

promise BBB function, enabling cells such as macrophages to pass the barrier.7 Thus, while our data demonstrate that hBMSC likely utilized lymphatic vessels to migrate to the spleen, it is also possible that hBMSC in the brain were able to cross the now permeable BBB and enter the bloodstream, enabling them to migrate to the spleen via systemic circulatory pathways. As the groups with severe stroke had more severe inflammation, this may have resulted in more damage to the BBB, and could also explain why more transplanted cells migrated from the brain to the periphery (i.e., spleen) in these groups. Future studies could further probe this concept by measuring stem cell levels in the blood, as well as graft deposition in other peripheral organs.

The in vitro results revealed an increase in hBMSC migration toward lymphatic endothelial cells, microglia, or a

A

B

combination of both when treated with TNF-a. The increasing number of migratory hBMSC was dependent on the dose of TNF-a in every experimental group.

Interestingly, the greatest migratory activity was seen in the cultures of TNF-a-treated microglia cells alone versus either lymphatic endothelial cells or the co-culture of both cell types with TNF-a. Since bEnd.3 cells are a brain endothelioma cell line, it is possible that the hBMSC also migrated to brain parenchymal endothelial cells, in addition to LYVE-1 lymphatic endothelial cells. However, adding TNF-a increases the ability of bEnd.3 cells to form LYVE-1-expressing lymphatic tubes.50 Thus, it is likely that in our cell migration assays involving TNF-a, the majority of hBMSC migrated to the LYVE-1 lymphatic endothelial cells, which were likely present in a higher proportion.

Prior studies have demonstrated that the injection of

C

D

Figure 6. Transplanted human bone marrow mesenchymal stromal cells phagocytose ischemic neurons in the brain and transport them to the spleen. (A and B) Triple immunofluorescent staining for human specific phagocytic marker CD68 (CD68), anti-apoptosis inhibitor 5 (API5), and anti-160kD neurofilament medium antibody-rat specific neuronal marker (neuronal marker) was performed in the brain (A) and spleen (B). Groups with mild and severe stroke demonstrated higher frequencies of staining overlap compared to the sham-treated group. Arrow heads indicate co-localization of CD68-positive, anti-apoptosis inhibitor 5-positive, and neuronal marker-positive cells. The small boxes show 40x magnification. Scale bars = 100 mm. Red: anti-apoptosis inhibitor 5; green: CD68; blue: neuronal marker. (C and D) Quantitative analyses of the estimated number of co-localized cells exhibiting overlap for all three stains in the brain (C) and in the spleen (D) of stroke and sham-treated animals. Significance bars: **P<0.01; ***P<0.001. Groups with mild and severe stroke displayed significantly higher quantities of co-localization in the brain and spleen relative to the sham-treated group on all days that measurements were made (**P<0.01). Co-localization levels peaked on day 3 in all organs in the groups with mild and severe stroke, especially in the groups with severe stroke (**P<0.01). (C) a: The group with mild stroke had significantly more co-localized cells in the brain on day 3 than on other days (P<0.05); b: the group with severe stroke had significantly more co-localized cells in the brain on day 3 than on other days (P<0.01). (D) a: The group with mild stroke had significantly more co-localized cells in the spleen on day 3 than on other days (P<0.05); b: the group with severe stroke had significantly more co-localized cells in the spleen on day 3 than on other days (P<0.05).

haematologica | 2019; 104(5)

1069

K. Xu et al.

hBMSC into immunocompetent rats19 has neuroprotective benefits similar to those obtained by transplanting murine-derived stem cells into rats.51 hBMSC have advantages over other cell types, such as circumventing ethical concerns and host rejection of transplanted stem cell

grafts, enabling allogeneic transplantation.52 Moreover, utilizing human-derived cells in preclinical studies may be more clinically relevant, as future cell transplantation therapies in the clinic will likely employ human-derived stem cells to treat neurological disorders. Despite these logisti-

A

C

E

B

D

F

G

H

I

J

Figure 7. Migration of human bone marrow mesenchymal stromal cells to bEnd.3, BV2, and bEnd.3+BV2 cells in cell migration assays. (A-F) Human bone marrow mesenchymal stromal cells (hBMSC) migrate to three experimental groups: lymphoid endothelial cells (bEnd.3) (A and B), microglia (BV2) (C and D), and a combination of BV2+bEnd.3 cells (E and F). Cell density was measured over time (0-72 h) and in response to varying doses of tumor necrosis factor alpha (TNF-α: 0 ng/mL, 25 ng/mL, 50 ng/mL, or 100 ng/mL). Compared with the non-treated control group, TNF-α significantly increased the number of migrating hBMSC, indicating that TNF-a plays a critical role in the migration of hBMSC in vitro. (A, C, E) Visualization of the density of hBMSC migrating to bEnd.3 (A), BV2 (C), and bEnd.3+BV2 (E) over time and in response to increasing TNF-α doses. (B, D, F) Quantitative analyses of the density of hBMSC migrating to bEnd.3 (B), BV2 (D), and bEnd.3+BV2 (F) over time and in response to increasing TNF-α doses. Significance values: (A) a: 0 ng/mL vs. 25 ng/mL (P<0.05); 0 ng/mL vs. 50 ng/mL (P<0.05). b: 0 ng/mL vs.

25 ng/mL (P<0.05); 0 ng/mL vs. 50 ng/mL (P<0.05). c: 50 ng/mL vs. 100 ng/mL (P<0.05). d: 25 ng/mL vs. 50 ng/mL (P<0.01); 50 ng/mL vs. 100 ng/mL (P<0.01).

(B) a: 0 ng/mL vs. 50 ng/mL (P<0.05); 25 ng/mL vs. 100 ng/mL (P<0.05). b: 25 ng/mL vs. 100 ng/mL (P<0.05); 50 ng/mL vs. 100 ng/mL (P<0.05). c: 0 ng/mL vs.

50 ng/mL (P<0.05); 25 ng/mL vs. 50 ng/mL (P<0.05); 25 ng/mL vs. 100 ng/mL (P<0.05). d: 25 ng/mL vs. 100 ng/mL (P<0.05). (C) a: 25 ng/mL vs. 100 ng/mL (P<0.05). b: 0 ng/mL vs. 100 ng/mL (P<0.05); 25 ng/mL vs. 50 ng/mL (P<0.05). c: 0 ng/mL vs. 50 ng/mL (P<0.05); 0 ng/mL vs. 100 ng/mL (P<0.05); 25 ng/mL vs. 50 ng/mL (P<0.05). d: 0 ng/mL vs. 50 ng/mL (P<0.01); 25 ng/mL vs. 50 ng/mL (P<0.01); 50 ng/mL vs. 100 ng/mL (P<0.01). (G-J) hBMSC preferentially migrate to BV2 microglial cells in cell migration assays. Quantitative comparisons between the amount of hBMSC migrating to BV2, bEnd.3, and BV2+bEnd.3 over time and within specific doses of TNF-α [0 ng/mL (G), 25 ng/mL (H), 50 ng/mL (I), and 100 ng/mL (J)]. The number of migrating hBMSC was highest for BV2 for all doses of TNF-α, indicating how hBMSC preferentially migrate to BV2. The density of migrating hBMSC was significantly higher for BV2 than for bEnd.3 and BV2+bEnd.3 for all TNF-α doses. There were significantly more migrating hBMSC for bEnd.3+BV2 than for bEnd.3 for all TNF-α doses. Significance values: (G) a: b.End3 vs. b.End3+BV2 (P<0.05). b: BV2 vs. b.End3+BV2 (P<0.05). c: BV2 vs. b.End3+BV2 (P<0.05). d: BV2 vs. b.End3+BV2 (P<0.05). (H) a: b.End3 vs. BV2 (P<0.001); b.End3 vs. b.End3+BV2 (P<0.001); BV2 vs. b.End3+BV2 (P<0.001). b: BV2 vs. b.End3+BV2 (P<0.05). c: BV2 vs. b.End3+BV2 (P<0.05). d: BV2 vs. b.End3+BV2 (P<0.05). (I) a: b.End3 vs. BV2 (P<0.001); b.End3 vs. b.End3+BV2 (P<0.001); BV2 vs. b.End3+BV2 (P<0.001). b: b.End3 vs. BV2 (P<0.01); b.End3 vs. b.End3+BV2 (P<0.01); BV2 vs. b.End3+BV2 (P<0.01). c: BV2 vs. b.End3+BV2 (P<0.01). d: b.End3 vs. b.End3+BV2 (P<0.05). (J) a: b.End3 vs. b.End3+BV2 (P<0.05). b: b.End3 vs. BV2 (P<0.001); b.End3 vs. b.End3+BV2 (P<0.001); BV2 vs. b.End3+BV2 (P<0.001). c: b.End3 vs. b.End3+BV2 (P<0.01). d: b.End3 vs. b.End3+BV2 (P<0.05).

1070

haematologica | 2019; 104(5)

Brain to spleen stem cell migration

cal and translational advantages of using human cells in an immunocompetent rat model, we recognize the possibility that the present results may be due to the phenomenal cross-species immunomodulation platform created by this model. We also concede that a same-species transplant paradigm may need to confirm the present observation of inflammation- and lymphatic-mediated brain-to-spleen migration of grafted stem cells. Notwithstanding this limitation, we submit that the current xeno-transplantation paradigm exaggerates the role of inflammation and/or the immune system on the lymphatic systems and stem cell fate, i.e., migration. As inhibiting lymphatic vessel contraction is associated with immunosuppression, it is possible that suppressing the immune system in an immunosuppressed rat model would hinder the lymphatic systemâ&#x20AC;&#x2122;s ability to coordinate lymph flow and thus obstruct the migration of hBMSC from the brain to the spleen.53 Despite being a subject of debate, the optimal time point for the application of stem cells in the clinical setting exists, both in terms of stem cell tropism toward the brain and overall therapeutic effectiveness. The undecided nature of this ideal time point is, however, a major hurdle to progress in stem cell therapies in stroke. In our study, rats were reperfused 1 hour after the middle cerebral artery occlusion surgery and received hBMSC at this point. Preclinical studies reveal that at this acute phase of stroke, the levels of chemokines and trophic factors increase markedly in the infarcted brain and subsequently decrease over time.54 This large release of inflammatory factors, oxygen free radicals, and excitatory neurotransmission toxins (e.g., glutamate and glycine) may pose a threat to the survival of the recently transplanted stem cells. The application of stem cells during the acute phase of stroke may be necessary to offer a range of

paracrine and immunomodulatory effects significant enough to reduce secondary injury processes and stimulate brain repair after stroke.55 Admittedly, 1 hour after a stroke is not the most clinically feasible time. However, later deliveries are plagued by the existence of a nonconducive microenvironment after cerebral infarction which may interfere with the transplanted stem cellsâ&#x20AC;&#x2122; ability to survive. Thus, it is important to determine the ideal, realistic time point, when the transplanted cells can not only survive in the infarcted area and salvage affected brain tissue, but also, as this study shows, effectively modulate peripheral immune responses to reduce the harmful inflammation. This requires more in-depth research. Here, we demonstrated a phenomenon whereby intracerebrally transplanted stem cells can migrate to the spleen via the lymphatic system, propelled by inflammatory signals. However, characterizing the function of the stem cells once within the spleen, how this in turn affects global inflammatory factors, and the ultimate effects on brain health, are mechanisms which require deeper study.

This study supports an alternative mechanism essential to cell therapy for stroke, advancing the notion that while the brain is the ultimate therapeutic target of stem cells, achieving functional recovery may occur as a systematic event. The intuitive mechanism whereby stem cells replace dead and dying neurons, integrating into functional circuits, only occurs to a slight extent.56 Thus, a complete explanation for the functional recovery associated with stem cell therapies in stroke requires an understanding of the non-cell replacement mechanisms of stem cells.

Intracerebrally transplanted hBMSC phagocytose apoptotic neurons, depart from the ischemic tissue of the brain, and travel toward the spleen via lymphatic vessels. From

Figure 8. Illustration demonstrating how stem cells migrate from the brain to the spleen via lymphatic vessels and with guidance from inflammatory cells. Stem cells engrafted in the brain enter lymphatic vessels in the brain and inflammatory cues direct the transplanted stem cells toward the spleen.

haematologica | 2019; 104(5)

1071

K. Xu et al.

the standpoint of basic science, the observation of transplanted human stem cells within the lymphatic vessels and adjacent to inflammatory cells in the brain, but surprisingly also in the spleen, reveals for the first time the migration of intracerebrally infused stem cells to the periphery. From a translational research view, the visualization of transplanted stem cells within lymphatic vessels and CD68-positive stem cells phagocytosing apoptotic neurons in the brain and spleen may correspond to a novel clean-up machinery of stem cells designed to dampen the stroke brainâ&#x20AC;&#x2122;s overload of cell death components and signals, scuttling them from the central nervous system to the periphery. Altogether, our observations indicate that stem cells are capable of phagocytic activity centrally and peripherally, likely designed to sequester and remove inflammatory cells and dying neurons from the brain and dump them in the periphery, and suggest a therapeutic mechanism involving central and peripheral sequestration of stroke inflammation. Moreover, the observed hBMSC

References 1. Kim AS, Cahill E, Cheng NT. Global stroke belt: geographic variation in stroke burden worldwide. Stroke. 2015;46(12):3564-3570.

2. Gopurappilly R, Pal R, Mamidi MK, Dey S, Bhonde R, Das AK. Stem cells in stroke repair: current success and future prospects.

CNS Neurol Disord Drug Targets.

2011;10(6):741-756.

3. Liska MG, Crowley MG, Borlongan CV. Regulated and unregulated clinical trials of stem cell therapies for stroke. Transl Stroke Res. 2017;8(2):93-103.

4. Moskowitz MA, Lo EH, Iadecola C. The science of stroke: mechanisms in search of treatments. Neuron. 2010;67(2):181-198.

5. Macrez R, Ali C, Toutirais O, et al. Stroke and the immune system: from pathophysiology to new therapeutic strategies. Lancet Neurol. 2011;10(5):471-480.

6. Yong VW, Rivest S. Taking advantage of the systemic immune system to cure brain diseases. Neuron. 2009;64(1):55-60.

7. Jin R, Yang G, Li G. Inflammatory mechanisms in ischemic stroke: role of inflammatory cells. J Leukoc Biol. 2010;87(5):779-789.

8. Denes A, Ferenczi S, Kovacs KJ. Systemic inflammatory challenges compromise survival after experimental stroke via augmenting brain inflammation, blood-brain barrier damage and brain oedema independently of infarct size. J Neuroinflammation. 2011; 8:164.

9. Kassner A, Merali Z. Assessment of bloodbrain barrier disruption in stroke. Stroke.

2015;46(11):3310-3315.

10. McColl BW, Rothwell NJ, Allan SM. Systemic inflammation alters the kinetics of cerebrovascular tight junction disruption after experimental stroke in mice. J Neurosci. 2008;28(38):9451-9462.

11. Yepes M, Brown SA, Moore EG, Smith EP, Lawrence DA, Winkles JA. A soluble fn14-fc decoy receptor reduces infarct volume in a murine model of cerebral ischemia. Am J Pathol. 2005;166(2):511-520.

12. Vendrame M, Gemma C, Pennypacker KR, et al. Cord blood rescues stroke-induced changes in splenocyte phenotype and function. Exp Neurol. 2006;199(1):191-200.

13. Crowley MG, Liska MG, Borlongan CV.

1072

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

migration patterns indicate the central role of the spleen in stroke pathology and reaffirm the importance of inflammatory signaling in stem cell migration. This study resolves the apparent paradox of robust functional recovery seen after stem cell transplantation in stroke despite minimal graft survival rates.

In summary, we demonstrate here that intracerebrally transplanted stem cells exhibit the ability to migrate from the brain parenchyma to the spleen via lymphatic vessels, led by inflammatory signals. Describing the migratory patterns and biodistribution of stem cells following transplantation furthers our understanding of how these cells offer their therapeutic effects as well as enhancing our knowledge of the spleen and lymphatic systemâ&#x20AC;&#x2122;s involvement in stroke pathology.

Acknowledgments This work was supported by NIH grants 5R01-NS09096201 to CVB and NIH 1RONS102395-0 to CVB.

Stem cell therapy for sequestering neuroinflammation in traumatic brain injury: an update on exosome-targeting to the spleen. J Neurosurg Sci. 2017;61(3):291-302 Vahidy FS, Parsha KN, Rahbar MH, et al.

Acute splenic responses in patients with ischemic stroke and intracerebral hemorrhage. J Cereb Blood Flow Metab.

2016;36(6):1012-1021.

Swirski FK, Nahrendorf M, Etzrodt M, et al.

Identification of splenic reservoir monocytes and their deployment to inflammatory sites.

Science. 2009;325(5940):612-616.

Ajmo CT Jr, Vernon DO, Collier L, et al. The spleen contributes to stroke-induced neurodegeneration. J Neurosci Res. 2008;86(10): 2227-2234.

Chiba T, Umegaki K. Pivotal roles of monocytes/macrophages in stroke. Mediators Inflamm. 2013;2013:759103.

Offner H, Subramanian S, Parker SM, Afentoulis ME, Vandenbark AA, Hurn PD. Experimental stroke induces massive, rapid activation of the peripheral immune system.

J Cereb Blood Flow Metab. 2006;26(5):654665.

Acosta SA, Tajiri N, Hoover J, Kaneko Y, Borlongan CV. Intravenous bone marrow stem cell grafts preferentially migrate to spleen and abrogate chronic inflammation in stroke. Stroke. 2015;46(9):2616-2627.

Mashkouri S, Crowley MG, Liska MG, Corey S, Borlongan CV. Utilizing pharmacotherapy and mesenchymal stem cell therapy to reduce inflammation following traumatic brain injury. Neural Regen Res. 2016;11(9):1379-1384.

Tajiri N, Acosta SA, Shahaduzzaman M, et al. Intravenous transplants of human adipose-derived stem cell protect the brain from traumatic brain injury-induced neurodegeneration and motor and cognitive impairments: cell graft biodistribution and soluble factors in young and aged rats. J Neurosci. 2014;34(1):313-326.

Louveau A, Smirnov I, Keyes TJ, et al.

Structural and functional features of central nervous system lymphatic vessels. Nature.

2015;523(7560):337-341.

Ishikawa H, Tajiri N, Shinozuka K, et al.

Vasculogenesis in experimental stroke after human cerebral endothelial cell transplanta-

tion. Stroke. 2013;44(12):3473-3481.

24. Yu SJ, Soncini M, Kaneko Y, Hess DC, Parolini O, Borlongan CV. Amnion: a potent graft source for cell therapy in stroke. Cell Transplant. 2009;18(2):111-118.

25. Trakarnsanga K, Griffiths RE, Wilson MC, et al. An immortalized adult human erythroid line facilitates sustainable and scalable generation of functional red cells. Nat Commun.

2017;8:14750.

26. Tajiri N, Quach D, Kaneko Y, et al.

Behavioral and histopathological assessment of adult ischemic rat brains after intracerebral transplantation of NSI-566RSC cell lines. PLoS One. 2014;9(3):e91408.

27. Glover LE, Tajiri N, Lau T, Kaneko Y, van Loveren H, Borlongan CV. Immediate, but not delayed, microsurgical skull reconstruction exacerbates brain damage in experimental traumatic brain injury model. PLoS One. 2012;7(3):e33646.

28. Dantzer R, O'Connor JC, Freund GG, Johnson RW, Kelley KW. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat Rev Neurosci. 2008;9(1):46-56.

29. Qureshi IA, Mehler MF. Towards a 'systems'-level understanding of the nervous system and its disorders. Trends Neurosci.

2013;36(11):674-684.

30. Vendrame M, Cassady J, Newcomb J, et al.

Infusion of human umbilical cord blood cells in a rat model of stroke dose-dependently rescues behavioral deficits and reduces infarct volume. Stroke. 2004;35(10):23902395.

31. Schwarting S, Litwak S, Hao W, Bahr M, Weise J, Neumann H. Hematopoietic stem cells reduce postischemic inflammation and ameliorate ischemic brain injury. Stroke.

2008;39(10):2867-2875.

32. Barbosa da Fonseca LM, Gutfilen B, Rosado de Castro PH, et al. Migration and homing of bone-marrow mononuclear cells in chronic ischemic stroke after intra-arterial injection. Exp Neurol. 2010;221(1):122-128.

33. Yang B, Hamilton JA, Valenzuela KS, et al.

Multipotent adult progenitor cells enhance recovery after stroke by modulating the immune response from the spleen. Stem Cells. 2017;35(5):1290-1302.

34. Koh HS, Chang CY, Jeon SB, et al. The hif-

haematologica | 2019; 104(5)

Brain to spleen stem cell migration

35.

36.

37.

38.

39.

40.

41.

1/glial tim-3 axis controls inflammationassociated brain damage under hypoxia. Nat Commun. 2015;6:6340.

Li D, Lang W, Zhou C, et al. Upregulation of microglial zeb1 ameliorates brain damage after acute ischemic stroke. Cell Rep. 2018;22(13):3574-3586.

Barbash IM, Chouraqui P, Baron J, et al.

Systemic delivery of bone marrow-derived mesenchymal stem cells to the infarcted myocardium: feasibility, cell migration, and body distribution. Circulation. 2003;108(7): 863-868.

Chamberlain G, Fox J, Ashton B, Middleton J. Concise review: mesenchymal stem cells: Their phenotype, differentiation capacity, immunological features, and potential for homing. Stem Cells. 2007;25(11):27392749.

Karp JM, Leng Teo GS. Mesenchymal stem cell homing: the devil is in the details. Cell Stem Cell. 2009;4(3):206-216.

Ponte AL, Marais E, Gallay N, et al. The in vitro migration capacity of human bone marrow mesenchymal stem cells: comparison of chemokine and growth factor chemotactic activities. Stem Cells. 2007;25 (7):1737-1745.

Takatera A, Takeuchi A, Saiki K, Morioka I, Yokoyama N, Matsuo M. Blood lysophosphatidylcholine (LPC) levels and characteristic molecular species in neonates: prolonged low blood LPC levels in very low birth weight infants. Pediatr Res. 2007;62(4):477482.

Girard JP, Moussion C, Forster R. Hevs, lym-

haematologica | 2019; 104(5)

42.

43.

44.

45.

46.

47.

48.

49.

50.

phatics and homeostatic immune cell trafficking in lymph nodes. Nat Rev Immunol.

2012;12(11):762-773.

Aloisi F. Immune function of microglia. Glia.

2001;36(2):165-179.

Nakajima K, Kohsaka S. Microglia: activation and their significance in the central nervous system. J Biochem. 2001;130 (2):169-175.

Barone FC, Arvin B, White RF, et al. Tumor necrosis factor-alpha. A mediator of focal ischemic brain injury. Stroke. 1997;28(6): 1233-1244.

Hallenbeck JM. The many faces of tumor necrosis factor in stroke. Nat Med. 2002;8(12):1363-1368.

Raivich G, Bohatschek M, Kloss CU, Werner A, Jones LL, Kreutzberg GW. Neuroglial activation repertoire in the injured brain: graded response, molecular mechanisms and cues to physiological function. Brain Res Brain Res Rev. 1999;30(1):77-105.

Denes A, Vidyasagar R, Feng J, et al.

Proliferating resident microglia after focal cerebral ischaemia in mice. J Cereb Blood Flow Metab. 2007;27(12):1941-1953.

Lalancette-Hebert M, Gowing G, Simard A, Weng YC, Kriz J. Selective ablation of proliferating microglial cells exacerbates ischemic injury in the brain. J Neurosci. 2007;27(10): 2596-2605.

Aoshi T, Zinselmeyer BH, Konjufca V, et al.

Bacterial entry to the splenic white pulp initiates antigen presentation to CD8+ T cells.

Immunity. 2008;29(3):476-486.

Yuen D, Leu R, Tse J, Wang S, Chen LL,

51.

52.

53.

54.

55.

56.

Chen L. Novel characterization of bEnd.3 cells that express lymphatic vessel endothelial hyaluronan receptor-1. Lymphology.

2014;47(2):73-81.

Wei L, Fraser JL, Lu ZY, Hu X, Yu SP. Transplantation of hypoxia preconditioned bone marrow mesenchymal stem cells enhances angiogenesis and neurogenesis after cerebral ischemia in rats. Neurobiol Dis. 2012;46(3):635-645.

Mitkari B, Nitzsche F, Kerkela E, et al.

Human bone marrow mesenchymal stem/stromal cells produce efficient localization in the brain and enhanced angiogenesis after intra-arterial delivery in rats with cerebral ischemia, but this is not translated to behavioral recovery. Behav Brain Res. 2014;259:50-59.

Liao S, Cheng G, Conner DA, et al. Impaired lymphatic contraction associated with immunosuppression. Proc Natl Acad Sci U S A. 2011;108(46):18784-18789.

Hill WD, Hess DC, Martin-Studdard A, et al.

Sdf-1 (CXCL12) is upregulated in the ischemic penumbra following stroke: association with bone marrow cell homing to injury. J Neuropathol Exp Neurol.

2004;63(1):84-96.

Savitz SI. Developing cellular therapies for stroke. Stroke. 2015;46(7):2026-2031.

Stonesifer C, Corey S, Ghanekar S, Diamandis Z, Acosta SA, Borlongan CV. Stem cell therapy for abrogating strokeinduced neuroinflammation and relevant secondary cell death mechanisms. Prog Neurobiol. 2017;158:94-131.

1073

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):1074-1082

Cell Therapy & Immunotherapy

Combination peptide immunotherapy suppresses antibody and helper T-cell responses to the major human platelet autoantigen glycoprotein IIb/IIIa in HLA-transgenic mice

Lindsay S. Hall,1,2 Charlotte S. Lennon,1 Andrew M. Hall,1 Stanislaw J. Urbaniak,1,2 Mark A. Vickers1,2* and Robert N. Barker1* 1 2

Institute of Medical Sciences, Ashgrove Road West, University of Aberdeen and Scottish National Blood Transfusion Service, Foresterhill Road, Aberdeen, UK

*MAV and RNB are joint senior authors

ABSTRACT

P

Correspondence: ROBERT N. BARKER r.n.barker@abdn.ac.uk Received: September 4, 2017.

Accepted: November 29, 2018.

Pre-published: December 4, 2018. doi:10.3324/haematol.2017.179424 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1074 Š2019 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.

1074

latelet destruction in immune thrombocytopenia is caused by autoreactive antibody and T-cell responses, most commonly directed against platelet glycoprotein IIb/IIIa. Loss of self-tolerance in the disease is also associated with deficient activity of regulatory T cells.

Having previously mapped seven major epitopes on platelet glycoprotein IIIa that are recognized by helper T cells from patients with immune thrombocytopenia, the aim was to test whether peptide therapy with any of these sequences, alone or in combination, could inhibit responses to the antigen in humanized mice expressing HLA-DR15. None of the individual peptides, delivered by a putative tolerogenic regimen, consistently suppressed the antibody response to subsequent immunization with human platelet glycoprotein IIb/IIIa. However, the combination of glycoprotein IIIa peptides aa6-20 and aa711-725, which contain the predominant helper epitopes in patients and elicited the strongest trends to suppress when used individually, did abrogate this response. The peptide combination also blunted, but did not reverse, the ongoing antibody response when given after immunization. Suppression of antibody was associated with reduced splenocyte T-cell responsiveness to the antigen, and with the induction of a regulatory T-cell population that is more responsive to the peptides than to purified platelet glycoprotein IIb/IIIa. Overall, these data demonstrate that combinations of peptides containing helper epitopes, such as platelet glycoprotein IIIa aa6-20 and aa711725, can promote in vivo suppression of responses to the major antigen implicated in immune thrombocytopenia. The approach offers a promising therapeutic option to boost T-cell regulation, which should be taken forward to clinical trials.

Introduction Immune thrombocytopenia (ITP) is a bleeding disorder defined by an isolated thrombocytopenia and caused by immune responses against self-antigens expressed on platelets and/or megakaryocytes.1-4 The most commonly targeted autoantigen is platelet glycoprotein (GP) IIb/IIIa, with responses against other glycoproteins such as GPIb/IX or GPIV seen in a minority of cases.5-7 GPIIb/IIIa, also known as integrin aIIbβ3, functions as the main fibrinogen and von Willebrand factor receptor and is highly expressed on the plasma membranes of platelets. Characterization of this major autoantigen is a key step in the development of specific immunotherapy that could selectively inhibit the pathogenic responses against platelets.7 Platelet destruction in ITP is believed to be mediated by IgG autoantibodies, which opsonize platelet and/or megakaryocyte surfaces,8,9 leading to clearance via Fcγ-receptors on macrophages, predominantly in bone marrow, spleen and liver. Thaematologica | 2019; 104(5)

GPIIb/IIIa peptide immunotherapy in HLA-transgenic mice

cell cytotoxicity has also been demonstrated against platelets and megakaryocytes.10,11 Class switching and somatic mutation of the autoantibodies12 indicate that their production is T-dependent, in line with the finding that autoreactive CD4+ helper T (Th) cells specific for the corresponding platelet antigens are activated in the disease.7,13-15 Evidence has accumulated that the loss of peripheral tolerance in ITP reflects an imbalance between regulatory and effector CD4+ T-cell activity, with, in particular, impairment of the suppressive function of the CD25+FoxP3+ regulatory T (Treg) subset.2,16-18 ITP therefore fits the paradigm emerging in other autoantibody-mediated diseases, in which pathogenic responses are helper-dependent and arise and/or are sustained due to inadequate Treg suppression.19-21 The current first-line treatment for ITP, steroids, inhibits autoantibody production and platelet degradation by inducing generalized immunosuppression.23,24 However, the numerous side effects of this treatment include infection and cardiovascular disease, which are more frequent causes of death than bleeding in ITP patients.25 Second-line treatments, for instance rituximab to deplete B cells, immunoglobulin infusions or splenectomy, are also nonspecific and typically further intensify immunosuppression. Third-line thrombopoietic agents are expensive, and although they can induce lasting remission in some patients26 and induce immune tolerance to platelet autoantigens in a murine model of ITP,27 these effects are not fully understood, and longer term side effects have yet to be fully reported.23,24 In contrast to previous untargeted approaches, antigen-specific immunotherapy offers the prospect of rebalancing pathogenic CD4+ Th and Treg responses, without compromising the rest of the immune system.28-32 Precedents in models of other immune-mediated diseases demonstrate that this can be achieved by appropriate administration of short, synthetic peptides containing CD4+ T-cell epitopes from the respective target antigens, with peptide products being tested in clinical trials for the treatment of a number of conditions, including type I diabetes, multiple sclerosis and rheumatoid arthritis.28-32 The approach induces immune regulation, which, in addition to suppressing inflammatory T-cell responses, also has the potential to inhibit helper-dependent antibody production, as demonstrated by the success of peptide therapy in blocking IgG responses to the RhD blood group protein in a humanized murine model.33,34 Once peptides that span major T-cell epitopes have been mapped, key

requirements for effective immunotherapy include delivery of these sequences in soluble form, in the absence of any adjuvant material.28,29,34,35 Despite an initial focus on mucosal administration of peptides, a variety of routes are now known to be effective, including subcutaneous.29,31,34 The fine specificity of autoreactive Th cells in ITP has been characterized with a view to developing specific peptide therapy that restores helper tolerance and avoids longterm immunosuppressive steroid treatment or splenectomy.7 When a panel of 15-mer peptides spanning the entire length of GPIIIa was screened for helper epitopes, particular sequences were found to elicit recall responses by CD4+ T cells from patients with ITP. Despite variation between different cases, seven GPIIIa peptides (aa6-20, aa331-345, aa361-375, aa421-435, aa591-605, aa661-675 and aa711725) were commonly recognized, with one or more eliciting responses by T cells from 84% of patients. Of these peptides, two predominated (aa6-20 and aa711-725), with either or both being stimulatory in 65% of patients.

The aim of the current work, and the next step in developing specific immunotherapy for ITP patients, was to test and compare the efficacy of the seven dominant GPIIIa peptides, particularly aa6-20 and aa711-725, in suppressing responses to GPIIb/IIIa in a pre-clinical animal model. In order to replicate restriction and peptide-binding preferences of the human major histocompatibility complex (MHC), the model used was a humanized transgenic mouse expressing HLA-DR rather than murine class II molecules. A strain transgenic for HLA-DR15 was chosen for these proof-of-principle studies, since it is an established model for investigating responses restricted by human class II,33,34 and, although there is no consensus as to whether certain class II alleles predispose to, or protect against, ITP,36,37 DRB1*1501 was well represented in our cohort of patients7 and is a common allele in many populations. We demonstrate that sequences aa6-20 and aa711725 in combination are effective in suppressing Th and IgG antibody responses to immunization of the mice with GPIIIa, and that this effect is associated with induction of a Treg population.

Methods Purification of glycoprotein IIb/IIIa GPIIb/IIIa was purified as previously reported,38 with minor modifications. In brief, apheresed platelets (provided by the

Table 1. GPIIIa immunodominant peptides. Peptides, previously reported to contain Th epitopes that are immunodominant in patients with immune thrombocytopenia,7 were manufactured to their wild-type sequence or extended using an arginine-lysine wrapper to improve manufacturability and solubility.

Peptide nomenclature P2 P44 P47 P53 P70 P77 P82

GPIIIa sequence

Amino acids

TTRGVSSCQQCLAVS-acid KRGVLSMDSSNVLQLIVRK-acid DLPEELSLSFNATCL-acid FKDSLIVQVTFDCDC-acid PGSYGDTCEKCPTCP-acid DDCVVRFQYYEDSSG-acid KRALLIWKLLITIHDRKRK-acid

aa6-20 aa331-345 aa361-375 aa421-435 aa591-605 aa661-675 aa711-725

Modified to improve manufacture No Yes* No No No No Yes*

*Modified by inclusion of an RK wrapper â&#x20AC;&#x201C; added in bold.

haematologica | 2019; 104(5)

1075

L.S. Hall et al.

Scottish National Blood Transfusion Service) were incubated in 50 mM octylglucopyranoside (Abcam, Cambridge, UK). GPIIb/IIIa was extracted from the supernatant onto sepharose beads displaying GPIIIa-specific peptide GRGDSPK (Cambridge Research Biochemicals, Billingham, UK), and eluted with peptide GRGDSP (Merck Chemicals, Nottingham, UK). The purity of the eluted GPIIb/IIIa preparation was confirmed by western blotting.

Peptides Peptides containing GPIIIa epitopes (Table 1) were manufactured to >95% purity (Cambridge Research Biochemicals). To achieve solubility in aqueous media, which is a key property for efficient production and purification, as well as for efficacy in immunotherapy,28,29,34,35 two sequences, P44 (aa331-345) and P82 (aa711-725), were extended by an arginine-lysine wrapper.

Mouse immunization and peptide treatment Animal work was approved by the University of Aberdeen Ethical Review Committee and authorized by the UK Government Home Office (project license 70/8744). Mice transgenic for HLA-DRA1*1010 and HLA-DRB1*1501, which express HLA-DR15 but not murine MHC class II, were originally supplied

by Professor Daniel Altman (Imperial College London, UK)39 and maintained as a genotyped and phenotyped colony at the University of Aberdeen.

Anti-platelet GPIIb/IIIa responses were induced by subcutaneous injection of mice with 10 mg purified GPIIb/IIIa antigen, repeated 2 weeks later. Mice immunized with control antigen received injections of ovalbumin (Sigma, Poole, UK) instead of GPIIb/IIIa. Mice were treated with GPIIIa peptides by subcutaneous injection, receiving 100 mg of each peptide selected, either alone or in combination.

(Miltenyi Biotec, Bisley, UK). Where required, Treg were depleted from splenocytes using the Miltenyi CD4+ CD25+ regulatory T-cell isolation kit.

Cell culture As previously described,33,34 splenocytes were cultured at 1.25x106 cells/mL in aMEM (Sigma, Poole, UK), supplemented with 0.5% syngeneic serum. Cultures were stimulated with purified GPIIb/IIIa antigen at a final concentration of 1 mg/mL, or with the control antigen ovalbumin at a final concentrations of 10 mg/mL, or with Dynabeadsâ&#x201E;˘ Mouse T-Activator CD3/CD28 (Invitrogen, Fisher Scientific, UK), or with peptides, either individually or in combination, at final concentrations of 10 mg/mL each.

Flow cytometry After 5 days of culture, splenocytes were stained with fluorescein isothiocyanate-conjugated anti-CD4 (eBioscience, Fisher Scientific, Loughborough, UK) and Pacific blue-conjugated antiCD25 (Biolegend, London, UK) antibodies, permeabilized and stained with phycoerythrin-conjugated anti-FoxP3 (eBioscience).

Samples were analyzed on an LSR FortessaTM flow cytometer (BD Bioscience, Oxford, UK) using FACSDivaTM software. Regulatory T cells were defined as CD4+CD25+FoxP3+.40

Proliferation assays and cytokine enzyme-linked immunosorbent assays As described elsewhere, after 5 days of culture,33-35 splenocyte proliferation was estimated from 3H-thymidine incorporation and presented as a stimulation index, representing the ratio of counts per minute (CPM) in stimulated versus unstimulated wells (with a ratio >3 taken as being significant41). Interleukin-10 (IL-10) production in cultures was measured by enzyme-linked immunosorbent assay (ELISA) (antibody pairs from BD Bioscience).

Blood sampling and tissue preparation Blood samples were collected into heparinized tubes. Single cell suspensions of splenocytes were prepared using the Miltenyi spleen dissociation kitTM and gentleMacs octo-dissociatorTM

A

Measurement of serum antibody specific for glycoprotein IIb/IIIa or control antigen ELISA to measure murine IgG antibodies specific for GPIIb/IIIa,

B

Figure 1. Treatment of HLA-DR15 transgenic mice with individual glycoprotein IIIa peptides containing Th epitopes neither elicits antibody specific for glycoprotein IIb/IIIa, nor consistently blocks the antibody response to immunization with purified glycoprotein IIb/IIIa. (A) Purified human GPIIb/IIIa or individual peptides were administered by subcutaneous injection to HLA-DR15 transgenic mice on days 0 and 14, with the levels of plasma IgG antibodies reactive to purified GPIIb/IIIa measured by enzyme-linked immunosorbent assay (ELISA) on day 28. (B) Individual peptides were administered to mice on days 0 and 14, followed by purified GPIIb/IIIa on days 28 and 42, with IgG antibodies measured by ELISA on day 70. Data points represent results from individual mice (n=3 per group). ***P<0.0001, one way analysis of variance. Ab: antibody.

1076

haematologica | 2019; 104(5)

GPIIb/IIIa peptide immunotherapy in HLA-transgenic mice

Statistical tests were performed using SigmaPlot (SyStat Software).

ensure responses to immunization with human GPIIb/IIIa, but in order to recapitulate in vivo the presentation of particular GPIIIa peptides by human MHC class II molecules.

GPIIIa peptides containing the T-cell epitopes that had been mapped in ITP patients would not necessarily be presented by murine class II molecules to modify responses of wild-type mice to GPIIb/IIIa, and any effects would not be of direct relevance to human disease. The administration of the peptides, given in soluble form via the subcutaneous route in the absence of adjuvant signals, was designed to favor induction of regulatory versus effector responses.28,29,34,35

Results

Stimulation of antibody responses to glycoprotein IIb/IIIa

Seven peptides (Table 1) containing GPIIIa epitopes that are commonly recognized by Th cells from patients with ITP and are, therefore, candidates for inclusion in an immunotherapeutic product7 were screened for their effects on immunity to GPIIb/IIIa in HLA-DR15 transgenic mice. The genetically modified strain was used, not to

It was first necessary to confirm that HLA-DR15 transgenic mice could mount an antibody response to GPIIb/IIIa after immunization with the purified full-length antigen, and that no such response was elicited by of any of the seven candidate peptides considered as putative suppressive treatment. Purified GPIIb/IIIa or individual GPIIIa peptides were administered to mice, with IgG anti-

or for the control antigen ovalbumin, were adapted from a published method.38 Microtiter plates (Nunc, Fisher Scientific, Loughborough, UK) were coated with either purified GPIIb/IIIa or ovalbumin, incubated with plasma samples diluted 1:20 in phosphate-buffered saline, probed with anti-mouse IgG conjugated to alkaline phosphatase (Invitrogen, Fisher Scientific), and developed with p-nitrophenyl phosphate substrate (Sigma). The absorbance was measured at 405 nm.

Statistical analyses

A

B

haematologica | 2019; 104(5)

Figure 2. Treatment of HLA-DR15 transgenic mice with a combination of glycoprotein IIIa peptides containing Th epitopes inhibits antibody responses to immunization with purified glycoprotein IIb/IIIa, but not the control antigen ovalbumin. Levels of plasma IgG antibodies reactive to (A) the antigen purified GPIIb/IIIa or (B) ovalbumin were measured by enzyme-linked immunosorbent assay in serial plasma samples from mice only immunized with GPIIb/IIIa or ovalbumin (immunized only); or pre-treated with a mixture of GPIIb/IIIa peptides 2 (aa6-20) and 82 (aa711-725) before immunization (prevention); or given the mixture of peptides after immunization (inhibition). Arrows indicate timing of peptide treatment (P2&82) and immunizations with GPIIb/IIIa (GPIIIa) or ovabumin (OVA).

Lines connect serial data points from individual mice: (A) immunized only (n=14), prevention (n=9), inhibition (n=8), tolerized (n=3); (B) immunized only (n=3), prevention (n=3). ***P<0.0005, **P<0.005, *P<0.05, paired ttest, ns indicates not significant.

1077

L.S. Hall et al.

Prevention and inhibition of antibody responses to glycoprotein IIb/IIIa by a peptide combination

bodies reactive to purified GPIIb/IIIa subsequently measured by ELISA (Figure 1A). It can be seen that only purified GPIIb/IIIa, and none of the peptides, elicited a significant antibody response.

It has been demonstrated that peptides used in combination can be more effective than individual sequences in preventing or reversing allo- or auto-antibody production.32,34,42,43 Peptides 2 (aa6-20) and 82 (aa711-725) were selected to be tested in combination here, since epitopes they contain were previously identified as the most immunodominant,7 capable of eliciting Th responses in 65% of ITP patients tested, and they had also each demonstrated some, albeit variable, inhibition of antibody generation in the individual peptide experiments. It was determined whether prior treatment with an equimolar mixture of peptides 2 (aa6-20) and 82 (aa711-725) prevented induction of the antibody response to immunization with purified GPIIb/IIIa. Given that the antibody levels are measured as relative optical density (OD) values in

Prevention of antibody responses to glycoprotein IIb/IIIa by individual peptides The seven peptides were next tested individually for their ability to prevent the IgG antibody response to GPIIb/IIIa immunization. Different groups of mice were each given one of the peptides, or left untreated, before immunization, and the development of anti-GPIIb/IIIa plasma antibodies compared (Figure 1B). None of the peptides consistently blocked the development of antibody, although there were non-significant trends for lower responses after delivery of peptides 2 (aa6-20), 70 (aa591605) or 82 (aa711-725).

Figure 3. Treatment of HLA-DR15 transgenic mice with glycoprotein IIIa peptides containing Th epitopes modulates splenocyte proliferative responses in vitro. Splenocytes were obtained from mice that were untreated controls (nonimmunized); immunized with GPIIb/IIIa only (immunized only); pre-treated with a mixture of GPIIb/IIIa peptides 2 (aa6-20) and 82 (aa711725) before GPIIb/IIIa immunization (prevention); given the mixture of peptides after GPIIb/IIIa immunization (inhibition); or given the mixture of peptides only (tolerized). The panels show proliferation of splenocytes in response to stimulation in vitro with (A) purified GPIIb/IIIa, (B) peptide 2 or (C) peptide 82. Data points represent results from individual mice: (A) nonimmunized (n=10), immunized only (n=9, prevention (n=9), inhibition (n=12), tolerized (n=3); (B and C) non-immunized (n=7), immunized only (n=6), prevention (n=6), inhibition (n=5). *P<0.05, **P<0.001, t-test. CPM: counts per minute.

A

B

1078

C

haematologica | 2019; 104(5)

GPIIb/IIIa peptide immunotherapy in HLA-transgenic mice

ELISA, and not absolute concentrations, the results were designed to be interpreted as longitudinal studies of mice in each group. In contrast to control mice that received only the immunization, no significant antibody response against GPIIb/IIIa could be detected in animals pre-treated with the peptide combination (Figure 2A, left and middle panels). Extending the time after GPIIb/IIIa immunization

by a further 14 days in another group of pre-treated mice (n=3) saw no increase in antibody levels, consistent with a sustained loss, rather than delay, in responsiveness (mean ± standard deviation OD: 2.92±0.09 at day 42 versus 3.03±0.1 at day 56, n=3).

Having demonstrated that, unlike either individual peptide, pre-treatment with the combination of peptides 2

A

C

B

Figure 4. Treatment of HLA-DR15 transgenic mice with glycoprotein IIIa peptides containing Th epitopes modulates splenocyte populations with a CD4+CD25+FoxP3+ Treg phenotype and suppressive function. Splenocytes were obtained from mice that were untreated controls (non-immunized); immunized with GPIIb/IIIa only (immunized only); pre-treated with a mixture of GPIIb/IIIa peptides 2 (aa6-20) and 82 (aa711-725) before GPIIb/IIIa immunization (prevention); given the mixture of peptides after GPIIb/IIIa immunization (inhibition). The cells were left as unstimulated controls (none; no stimulus), or stimulated in vitro with either purified GPIIb/IIIa (GPIIIa; GPIIIa stimulation) or a mixture of all seven GPIIIa peptides (Pmix; Pmix stimulation). (A) Representative examples of flow cytometric analyses of CD25 and FoxP3 expression after gating on CD4+ cells. (B) Proportions of CD25+FoxP3+ cells within the splenic CD4+ populations from all the mice. Data points represent results from individual mice (non-immunized) (n=7), immunized only (n=6), prevention (n=6), inhibition (n=5). (C) The effect of depleting CD25 +CD4+ cells (CD25-) on the ability of splenocytes from mice in the prevention group (n=3) to make proliferative responses when stimulated with GPIIb/IIIa. **P<0.005, ttest. CPM: counts per minute.

haematologica | 2019; 104(5)

1079

L.S. Hall et al.

Figure 5. Treatment of HLA-DR15 transgenic mice with glycoprotein IIIa peptides containing Th epitopes does not modulate splenocyte interleukin-10 responses in vitro. Splenocytes were obtained from mice that were untreated controls (non-immunized); immunized with GPIIb/IIIa only (immunized only); pre-treated with a mixture of GPIIb/IIIa peptides 2 (aa6-20) and 82 (aa711-725) before GPIIb/IIIa immunization (prevention); or given the mixture of peptides after GPIIb/IIIa immunization (inhibition). The cells were left as unstimulated controls (no stimulation; left panel), or stimulated in vitro with either purified GPIIb/IIIa (GPIIIa stimulation; middle panel) or the mixture of GPIIIa peptides (Pmix stimulation; right panel) and the production of interleukin-10 in culture measured by enzyme-linked immunosorbent assay. Data points represent results from individual mice (non-immunized) (n=9), immunized only (n=9), prevention (n=9), inhibition n=8). In each panel, there are no significant differences in interleukin-10 production between the unimmunized control group and any treated groups (t-test).

(aa6-20) and 82 (aa711-725) could prevent anti-GPIIb/IIIa antibody generation, we next determined whether the mixture could also suppress responses that had already been induced, by administering the peptide product after immunization with GPIIb/IIIa (Figure 2A, right panel). As expected, a significant response to immunization was detectable by the time of the peptide administration, but there was a slight, non-significant fall in mean antibody levels over the next 14 days after treatment (mean ± standard deviation OD: 0.55±0.1 at peptide treatment on day 28 versus 0.54±0.1 on day 42, n=8; P=0.8). By contrast, mean antibody levels continued to rise, modestly but significantly, in the group of control mice that received only immunization (Figure 2A, left panel) (mean ± standard deviation OD: 0.96±0.53 on day 28 versus 1.1±0.54 on day 42, n=14; P<0.05). Thus, the peptide combination blunted, but did not significantly reverse, ongoing responses to GPIIb/IIIa. The suppressive effect of treatment of transgenic mice with the peptide combination was specific to GPIIb/IIIa responses, since there was no effect on their ability to make antibody responses to immunization with the unrelated control antigen ovalbumin (Figure 2B).

either before or after immunization, and compared their ability to proliferate in vitro when stimulated with purified GPIIb/IIIa (Figure 3A). As expected, splenocytes from mice that been immunized with GPIIb/IIIa proliferated in response to the antigen, whereas there were no responses in control, unimmunized animals. Flow cytometric analyses demonstrated that T cells with the CD4+ helper phenotype predominated in the proliferating cultures (>70% in all cases, n=6). Proliferation against GPIIb/IIIa was significantly reduced when immunized mice had been pretreated with the combination of peptides 2 (aa6-20) and 82 (aa711-725), and there was also a non-significant trend for responses to be inhibited when mice were treated with the peptides after immunization.

Splenocyte responsiveness to the individual GPIIIa peptides 2 (aa6-20) or 82 (aa711-725) was also assayed (Figure 3B). Peptide 2 (aa6-20) elicited no consistent proliferative responses in any group, but peptide 82 (aa711-725) demonstrated significant stimulatory ability in both groups of animals that had received the peptide combination in addition to immunization.

Prevention and inhibition of T-cell responses to glycoprotein IIb/IIIa by the peptide combination

Having demonstrated that treatment with a combination of GPIIIa peptides can suppress both antibody and effector T-cell responses to GPIIb/IIIa immunization, we determined whether this was associated with induction of Treg cells. The proportions of cells exhibiting the CD25+FoxP3+ Treg phenotype within the CD4+ splenocyte population were compared in control, unimmunized and GPIIb/IIIa-immunized mice, and in animals that had received the peptide mixture either before or after immunization. Treg cells were enumerated both when spleno-

Class-switched antibody responses classically depend on antigen-specific Th cells,19,20 which peptide immunotherapy is intended to target and render unresponsive.28-31 To determine the effect of treatment with the combination of GPIIIa peptides 2 (aa6-20) and 82 (aa711725) on Th effector function, we obtained splenocytes from control unimmunized or GPIIb/IIIa-immunized mice, or from animals that had received the peptides 1080

Induction of regulatory T cells by the peptide combination

haematologica | 2019; 104(5)

GPIIb/IIIa peptide immunotherapy in HLA-transgenic mice

cytes had received no further stimulus in vitro, and after incubation with either purified GPIIb/IIIa, or the peptide mix. Figure 4A illustrates examples of contour plots obtained from individual animals from each group, and Figure 4B summarizes all the data. Without stimulation in culture, the proportions of cells with the Treg phenotype were low, typically <5% of CD4+ cells, irrespectively of whether the mice had been immunized or received peptide. However, after in vitro stimulation with purified GPIIb/IIIa the Treg population increased, typically to >10% of CD4+ cells, in all the immunized groups, both peptide-treated and -untreated, reaching proportions of Treg that were significantly higher than those seen in nonimmunized controls. When stimulated in vitro with the peptide mix, increases in Treg were seen in mice that had received peptide either before or after immunization, but not in animals given only the immunization. Overall, the observed inhibition of antibody and Th effector function by peptide therapy is therefore associated with recruitment of cells with a Treg phenotype, which differ in responsiveness from those induced by the purified fulllength antigen. This is consistent with induction of suppression to GPIIb/IIIa by peptide-specific Treg. It was confirmed that Treg cells contribute to the suppressive effect in peptide-treated mice since splenocyte proliferative responsiveness to GPIIb/IIIa was restored by depletion of the CD25+ T-cell population that contains the Treg cells (Figure 4C).

Production of the regulatory cytokine IL-10 is not necessarily associated with the classic CD25+FoxP3+ Treg population, but represents another mechanism by which effector responses can be downregulated following peptide immunotherapy.19-23,28,29,44 When we measured production of IL-10 by splenocytes (Figure 5), there were no differences between levels of the cytokine in cultures from mice that had been non-immunized, immunized, or given immunization and peptides, although across all groups there was a trend for IL-10 production to be higher when splenocytes were stimulated with the peptide mix compared to unstimulated or GPIIb/IIIa-stimulated cultures.

Therefore, in this model of peptide therapy, there is little evidence for IL-10 having a role in suppression.

Discussion As part of a strategy to develop and evaluate specific peptide therapy for ITP, this study determined the ability of sequences containing immunodominant helper epitopes to suppress responses to the major human platelet autoantigen GPIIb/IIIa by humanized mice expressing HLA-DR15. None of seven GPIIIa peptides, when tested individually, was able reliably to inhibit antibody responses by the mice to immunization with purified GPIIb/IIIa, but the combination of peptides 2 (aa6-20) and 82 (aa711725) did suppress the development of antibodies and proliferative effector T-cell responses. These inhibitory effects were associated with induction of a specific population of Treg cells, and raise the prospect that the peptide combination may be effective in ameliorating ITP in patients.

Previous studies in other diseases have suggested that combinations of peptides are markedly more effective than individual sequences in suppressing pathogenic immune responses in murine models,32,34,42,43 and several of these combinations have progressed into the clinic, with haematologica | 2019; 104(5)

promising results recently reported from human trials.

These include phase I trials of a combination of three HLA-DQ2-restricted immunodominant peptides in the ImmusanT product Nexvax2 to treat celiac disease,45 phase IIa trials of a mixture of four myelin peptides in the Apitope product ATX-MS-1467 in patients with multiple sclerosis (ClinicalTrials.gov Identifier: NCT01973491), and a phase I trial in type 1 diabetes of the product MultiPepT1De containing proinsulin sequences.46 Combinations of peptides may be more effective than individual sequences in triggering therapeutic suppression for a number of reasons. First, different class II molecules vary in their peptide binding preferences,47 and so mixtures of peptides can maximize the likelihood of effective presentation to CD4+ T cells in HLA-disparate patient populations.29-31 However, this does not account for the superior efficacy of the combination of peptides 2 (aa6-20) and 82 (aa711-725) versus the individual sequences in the current murine model, in which only HLA-DR15 is available. Here, a second explanation, that additional peptides interact with a wider pool of T-cell specificities and are therefore more likely to tip the balance of the response towards regulation, is more relevant.29-31,34 In vivo studies such as these provide the opportunity to catalogue the complex immune properties that therapeutic peptides can exhibit alone or in combination. Individual peptides can induce different balances between effector and regulatory T cells,21,22 yet can synergize in suppression, as illustrated here by the ability of peptide 2 (aa6-20), but not 82 (aa711725), to stimulate the proliferation of splenocytes, when mice need to be given both peptides to inhibit GPIIb/IIIa responses. The combination of peptides 2 (aa6-20) and 82 (aa711-725) is attractive as a candidate product to treat ITP, not only because of the current demonstration of efficacy in a pre-clinical model, but also because the sequences appear promiscuous in their ability to be presented by different HLA-DR molecules, and are together recognized by T cells from the majority of patients.7 The finding that the inhibitory effects of the peptides are limited to responses to GPIIb/IIIa, and not control antigen, demonstrates specificity of suppression and suggests that immunity to infection would not be compromised by treatment.

In many animal models of immune-mediated disease, and clinical investigations, the efficacy of peptide immunotherapy is associated with suppression of effector T cells and induction of Treg populations.29-32,43-46 The approach is attractive for development as a novel treatment for ITP since the disease is associated with an impairment of CD4+CD25+FoxP3+ Treg cells, with their frequency being reduced in the circulation, spleen and bone marrow.2,16-18 Dysregulation of Treg cells is also restored in line with platelet counts after treatment with dexamethasone, rituximab or thrombopoietin.2,4 Analysis of a SCID murine model of ITP, which is induced by adoptive transfer of splenocytes from CD61-/- mice that have been immunized with CD61+ platelets, further demonstrated a dysregulation of CD4+CD25+FoxP3+ Treg cells.48 These mice were successfully treated with intravenous immunoglobulins, which normalized both platelet and Treg counts.48 The suppression of antibody responses to GPIIb/IIIa in HLA-DR transgenic mice reported here now raises the prospect that Treg can be boosted therapeutically in ITP, with populations of antigen-responsive cells with a CD4+CD25+FoxP3+ Treg phenotype induced in the 1081

L.S. Hall et al.

spleens after specific peptide therapy. Treg cells can suppress antibody responses not only by blocking Th activity, but also by direct effects on B cells.40 Unlike some other examples of successful peptide therapy,28-31,44 suppression in our model was not associated with IL-10 responses, suggesting it is mediated by the CD4+CD25+FoxP3+ Treg we observed, which are classically cytokine independent, rather than by other cells such as the T regulatory 1 (Tr1) type. One notable feature of our study is that, while both immunization of mice with purified GPIIb/IIIa, and treatment with the peptide combination, induced Treg phenotype cells, the populations differed in specificity. This difference is demonstrated by the ability of the splenic Treg phenotype population induced in immunized mice to respond to GPIIb/IIIa, but not the peptides, and suggests an explanation for the success of immunization, and the ability of splenocytes to proliferate, in the face of such a population. Thus, the cells with the Treg phenotype induced as part of the response to GPIIb/IIIa are unable to block immunization, at least within the timeframe of our experiments, but the peptides tip the balance to suppression by recruiting more effective, or additional, Treg cells not induced by the antigen alone. It is recognized that antigens and peptides can induce both effector and regulatory cells, and that an evolving balance between these populations, particularly inter-conversion of Teff and Treg, determines the functional outcome.21,22 Our approach to peptide therapy demonstrated high efficacy in preventing antibody and T-cell responses to GPIIb/IIIa when the peptides were given prior to immunization, and although developing responses were blunted when mice were treated after immunization, antibody levels did not fall. This may reflect a relatively slow decay of established responses in the face of regulation, which, if necessary, could be augmented by additional peptide doses,43 since our regimen for combination peptide therapy was a single subcutaneous injection, for easy translation to clinical practice.

Alternatively, peptide therapy could also be combined with existing, more rapidly acting approaches, such as use

References 1. Semple JW. Immune pathophysiology of autoimmune thrombocytopenic purpura.

Blood Rev. 2002;16(1):9-12.

2. McKenzie CG, Guo L, Freedman J, Semple JW. Cellular immune dysfunction in immune thrombocytopenia (ITP). Br J Haematol. 2013;163(1):10-23.

3. Cines DB, Cuker A, Semple JW. Pathogenesis of immune thrombocytopenia. Presse Med. 2014;43(4 Pt 2):49-59.

4. Zufferey A, Kapur R, Semple JW. Pathogenesis and therapeutic mechanisms in immune thrombocytopenia (ITP). J Clin Med. 2017;6(2).pii:E16.

5. HĂźrlimann-Forster M, Steiner B, von Felten A. Quantitation of platelet-specific autoantibodies in platelet eluates of ITP patients measured by a novel ELISA using the purified glycoprotein complexes GPIIb/IIIa and GPIb/IX as antigens. Br J Haematol.

1997;98(2):328-335.

6. McMillan R. Autoantibodies and autoanti-

1082

7.

8.

9.

10.

11.

of the biologic rituximab to target B cells directly, while peptide-induced Treg cells block longer-term recruitment to the immune response. Solubility is a key feature in the efficient manufacture of pure peptides, and in their ability to induce regulation.28,29,34,35 Although peptide 2 (aa6-20) needed no modification to meet this criterion, we extended the sequence of peptide 82 (aa711-725) with an arginine-lysine wrapper in order to ensure adequate solubility, so that the combined peptides were appropriate for further product development.

The current results add to the body of work indicating that the manipulation of Treg cells offers an effective strategy for ameliorating ITP and other autoimmune antibodymediated diseases,16-22,48 and the data demonstrate that this goal could be achieved by peptide immunotherapy. The ability of a combination of peptides to suppress antibody and T-cell responses to GPIIb/IIIa, and to induce Treg, in our pre-clinical model provides further justification for human clinical trials of this approach in ITP. Differences between species inevitably present a risk that positive results from murine studies do not translate to human patients, but we have reduced this by testing mice with partially humanized immune systems, and by asking highly defined questions as to the ability of GPIIIa peptides presented in vivo by human MHC class II molecules to induce suppression. A product containing GPIIIa peptides 2 (aa6-20) and 82 (aa711-725) therefore represents the basis for development, with the initial indication being those ITP patients in whom conventional immunosuppressive treatments have failed.

Funding The study was funded by grants from the Medical Research Council (UK) Confidence in Concept, the Scottish National Blood Transfusion Service and the Wellcome Trust (UK).

Acknowledgments We would like to acknowledge the assistance of Iain Fraser Cytometry Centre at the University of Aberdeen.

gens in chronic immune thrombocytopenic purpura. Semin Hematol. 2000;37(3):239248.

Sukati H, Watson HG, Urbaniak SJ, Barker RN. Mapping helper T-cell epitopes on platelet membrane glycoprotein IIIa in chronic autoimmune thrombocytopenic purpura. Blood. 2007;109(10):4528-4538.

Chang M, Nakagawa PA, Williams SA, et al.

Immune thrombocytopenic purpura (ITP) plasma and purified ITP monoclonal autoantibodies inhibit megakaryocytopoiesis in vitro. Blood. 2003;102(3):887-895.

McMillan R, Wang L, Tomer A, Nichol J, Pistillo J. Suppression of in vitro megakaryocyte production by antiplatelet autoantibodies from adult patients with chronic ITP. Blood. 2004;103(4):1364-1369.

Olsson B, Andersson PO, Jernas M, et al. Tcell-mediated cytotoxicity toward platelets in chronic idiopathic thrombocytopenic purpura. Nat Med. 2003;9(9):1123-1124.

Zhang F, Chu X, Wang L, et al. Cell-mediated lysis of autologous platelets in chronic idiopathic thrombocytopenic purpura. Eur J

Haematol. 2006;76(5):427-431.

12. Roark JH, Bussel JB, Cines DB, Siegel DL. Genetic analysis of autoantibodies in idiopathic thrombocytopenic purpura reveals evidence of clonal expansion and somatic mutation. Blood. 2002;100(4):1388-1398.

13. Semple JW, Freedman J. Increased antiplatelet T-helper lymphocyte reactivity in patients with autoimmune thrombocytopenia. Blood. 1991;78(10):2619-2625.

14. Kuwana M, Kaburaki J, Ikeda Y. Autoreactive T cells to platelet GPIIb-IIIa in thrombocytopenic purpura. J Clin Invest.

1998;102(7):1393-1402.

15. Kuwana M, Kaburaki J, Kitasato H, Miyako K. Immunodominant epitopes on glycoprotein IIb-IIIa recognized by autoreactive T cells in patients with immune thrombocytopenic purpura. Blood. 2001; 98(1):130-139.

16. Yazdanbakhsh K. Imbalanced immune homeostasis in immune thrombocytopenia.

Semin Hematol. 2016;53(Suppl 1):S16â&#x20AC;&#x201C;S19.

17. Semple JW. 2003. T cell and cytokine abnormalities in patients with autoimmune

haematologica | 2019; 104(5)

GPIIb/IIIa peptide immunotherapy in HLA-transgenic mice

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

thrombocytopenic purpura. Transfus Apher Sci. 2003;28(3):237-242.

Aslam R, Hu Y, Gebremeskel S, Segel GB, et al.

Thymic retention of CD4+CD25+FoxP3+ T regulatory cells is associated with their peripheral deficiency and thrombocytopenia in a murine model of immune thrombocytopenia.

Blood.

2012;120(10):2127-2132.

Elson CJ, Barker RN. Helper T cells in antibody-mediated, organ-specific autoimmunity. Curr Opin Immunol. 2000;12(6):664-669.

Barker RN, Vickers MA, Ward FJ. Controlling autoimmunity--lessons from the study of red blood cells as model antigens.

Immunol Lett. 2007;108(1):20-26.

Hall AM, Ward FJ, Vickers MA, Stott LM, Urbaniak SJ, Barker RN. Interleukin-10mediated regulatory T-cell responses to epitopes on a human red blood cell autoantigen. Blood. 2002;100(13):4529-4536.

Ward FJ, Hall AM, Cairns LS, et al. Clonal regulatory T cells specific for a red blood cell autoantigen in human autoimmune hemolytic anemia. Blood. 2008;111(2):680687.

Neunert C, Lim W, Crowther M, Cohen A, Solberg L Jr, Crowther MA. The American Society of Hematology 2011 evidence-based practice guideline for immune thrombocytopenia. Blood. 2011;117(16):4190-4207.

Lozano ML, Revilla N, Gonzalez-Lopez TJ, et al. Real-life management of primary immune thrombocytopenia (ITP) in adult patients and adherence to practice guidelines. Ann Hematol. 2016;95(7):1089-1098.

Frederiksen H, Maegbaek ML, Nørgaard M. Twenty-year mortality of adult patients with primary immune thrombocytopenia: a Danish population-based cohort study. Br J Haematol. 2014;166(2):260-267.

Ghadaki B, Nazi I, Kelton JG, Arnold DM. Sustained remissions of immune thrombocytopenia associated with the use of thrombopoietin receptor agonists. Transfusion.

2013;53(11):2807-2812.

Nishimoto T, Numajiri M, Nakazaki H, Okazaki Y, Kuwana M. Induction of immune tolerance to platelet antigen by short-term thrombopoietin treatment in a mouse model of immune thrombocytopenia. Int J Hematol. 2014;100(4):341-344.

Larché M, Wraith DC. Peptide-based thera-

haematologica | 2019; 104(5)

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

peutic vaccines for allergic and autoimmune diseases. Nat Med. 2005;11(4 Suppl):S69-76.

Sabatos-Peyton CA, Verhagen J, Wraith DC. Antigen-specific immunotherapy of autoimmune and allergic diseases. Curr Opin Immunol. 2010;22(5):609-615.

Hoffmann HJ, Valovirta E, Pfaar O, et al.

Novel approaches and perspectives in allergen immunotherapy. Allergy. 2017;72(7): 1022-1034.

Larché M. Mechanisms of peptide immunotherapy in allergic airways disease.

Ann Am Thorac Soc. 2014;11(Suppl 5):S292296.

Hirsch DL, Ponda P. Antigen-based immunotherapy for autoimmune disease: current status. Immunotargets Ther.

2014;4:1-11 Hall AM, Cairns LS, Altmann DM, Barker RN, Urbaniak SJ. Immune responses and tolerance to the RhD blood group protein in HLA-transgenic mice. Blood. 2005;105(5): 2175-2179.

Hall LS, Hall AM, Pickford W, Vickers MA, Urbaniak SJ, Barker RN. Combination peptide immunotherapy suppresses antibody and helper T-cell responses to the RhD protein in HLA-transgenic mice.

Haematologica. 2014;99(3):588-596.

Shen CR, Youssef AR, Devine A, et al.

Peptides containing a dominant T-cell epitope from red cell band 3 have in vivo immunomodulatory properties in NZB mice with autoimmune hemolytic anemia. Blood.

2003;102(10):3800-3806.

Negi RR, Bhoria P, Pahuja A, et al.

Investigation of the possible association between the HLA antigens and idiopathic thrombocytopenic purpura (ITP). Immunol Invest. 2012;41(2):117-128.

Nomura S, Matsuzaki T, Ozaki Y, et al.

Clinical significance of HLA-DRB1*0410 in Japanese patients with idiopathic thrombocytopenic purpura. Blood. 1998;91(10):36163622.

Bessos H, Perez S, Armstrong-Fisher S, Urbaniak S, Turner M. The development of a quantitative ELISA for antibodies against human platelet antigen type 1a. Transfusion.

2003;43(3):350-356.

Ellmerich S, Takacs K, Mycko M, et al.

Disease-related epitope spread in a humanized T cell receptor transgenic model of mul-

40.

41.

42.

43.

44.

45.

46.

47.

48.

tiple sclerosis.

Eur J Immunol.

2004;34(7):1839-1848.

Nishimura E, Sakihama T, Setoguchi R, Tanaka K, Sakaguchi S. Induction of antigenspecific immunologic tolerance by in vivo and in vitro antigen-specific expansion of naturally arising Foxp3+CD25+CD4+ regulatory T cells. Int Immunol. 2004;16(8):11891201.

Devereux G, Hall AM, Barker RN. Measurement of T-helper cytokines secreted by cord blood mononuclear cells in response to allergens. J Immunol Methods.

2000;234(1-2):13-22.

Mackenzie KJ, Fitch PM, Leech MD, et al.

Combination peptide immunotherapy based on T cell epitope mapping reduces allergen-specific IgE and eosinophilia in allergic airway inflammation. Immunology.

2013;138(3):258-268.

Gibson VB, Nikolic T, Pearce VQ, Demengeot J, Roep BO, Peakman M. Proinsulin multi-peptide immunotherapy induces antigen-specific regulatory T cells and limits autoimmunity in a humanized model. Clin Exp Immunol. 2015;182(3):251260.

Nicolson KS, O'Neill EJ, Sundstedt A, Streeter HB, Minaee S, Wraith DC. Antigeninduced IL-10+ regulatory T cells are independent of CD25+ regulatory cells for their growth, differentiation, and function. J Immunol. 2006;176(9):5329-5337.

Goel G, King T, Daveson AJ, et al. Epitopespecific immunotherapy targeting CD4-positive T cells in coeliac disease: two randomised, double-blind, placebo-controlled phase 1 studies. Lancet Gastroenterol Hepatol. 2017;2(7):479-493.

Alhadj Ali M, Liu YF, Arif S, et al. Metabolic and immune effects of immunotherapy with proinsulin peptide in human newonset type 1 diabetes. Sci Transl Med. 2017;9(402).pii:eaaf7779.

Singh H, Raghava GPS. ProPred: prediction of HLA-DR binding sites. Bioinformatics.

2001;17(12):1236-1237.

Aslam R, Hu Y, Gebremeskel S, et al.

Thymic retention of CD4+CD25+FoxP3+ T regulatory cells is associated with their peripheral deficiency and thrombocytopenia in a murine model of immune thrombocytopenia. Blood. 2012;120(10):2127-2132.

1083

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(5):1084-1092

Quality of Life

Randomized controlled trial of individualized treatment summary and survivorship care plans for hematopoietic cell transplantation survivors Navneet S. Majhail,1 Elizabeth Murphy,2 Purushottam Laud,3 Jaime M. Preussler,2,4 Ellen M. Denzen,2,4 Beatrice Abetti,5 Alexia Adams,4 RaeAnne Besser,4 Linda J. Burns,2,4 Jan Cerny,6 Rebecca Drexler,4 Theresa Hahn,7 Lensa Idossa,2 Balkrishna Jahagirdar,8 Naynesh Kamani,9 Alison Loren,10 Deborah Mattila,4 Joseph McGuirk,11 Heather Moore,2 Jana Reynolds,12 Wael Saber,3,13 Lizette Salazar,14 Barry Schatz,15 Patrick Stiff,15 John R. Wingard,16 Karen L Syrjala17 and K. Scott Baker17

Blood and Marrow Transplant Program, Cleveland Clinic, OH; 2National Marrow Donor Program/Be The Match, Minneapolis, MN; 3Medical College of Wisconsin, Milwaukee, WI; 4Center for International Blood and Marrow Transplant Research, Minneapolis, MN; 5 Leukemia and Lymphoma Society, White Plains, NY; 6UMass Memorial Medical Center, Worcester, MA; 7Roswell Park Comprehensive Cancer Center, Buffalo, NY; 8Regions Hospital, St Paul, MN; 9AABB, Bethesda, MD; 10University of Pennsylvania, Philadelphia, PA; 11University of Kansas Medical Center, Kansas City, KS; 12Baylor University Medical Center, Dallas, TX; 13Center for International Blood and Marrow Transplant Research, Milwaukee, WI; 14Haledon, NJ; 15Loyola University Medical Center, Chicago, IL; 16 University of Florida, Gainesville, FL and 17Fred Hutchinson Cancer Research Center, Seattle, WA, USA 1

ABSTRACT

Correspondence: NAVNEET S. MAJHAIL majhain@ccf.org Received: August 6, 2018.

Accepted: November 23, 2018.

Pre-published: December 4, 2018. doi:10.3324/haematol.2018.203919 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/5/1084 Š2019 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.

1084

S

urvivorship Care Plans (SCPs) may facilitate long-term care for cancer survivors, but their effectiveness has not been established in hematopoietic cell transplantation recipients. We evaluated the impact of individualized SCPs on patient-reported outcomes among transplant survivors. Adult (â&#x2030;Ľ18 years at transplant) survivors who were 1-5 years post transplantation, proficient in English, and without relapse or secondary cancers were eligible for this multicenter randomized trial.

SCPs were developed based on risk-factors and treatment exposures using patient data routinely submitted by transplant centers to the Center for International Blood and Marrow Transplant Research and published guidelines for long-term follow up of transplant survivors.

Phone surveys assessing patient-reported outcomes were conducted at baseline and at 6 months. The primary end point was confidence in survivorship information, and secondary end points included cancer and treatment distress, knowledge of transplant exposures, health care utilization, and health-related quality of life. Of 495 patients enrolled, 458 completed a baseline survey and were randomized (care plan=231, standard care=227); 200 (87%) and 199 (88%) completed the 6-month assessments, respectively. Patients' characteristics were similar in the two arms. Participants on the care plan arm reported significantly lower distress scores at 6 months and an increase in the Mental Component Summary quality of life score assessed by the Short Form 12 (SF-12) instrument. No effect was observed on the end point of confidence in survivorship information or other secondary outcomes. Provision of individualized SCPs generated using registry data was associated with reduced distress and improved mental domain of quality of life among 1-5 year hematopoietic cell transplantation survivors. Trial registered at clinicaltrials.gov 02200133. haematologica | 2019; 104(5)

Individualized care plans in HCT survivors

Introduction It is estimated that there will be 250,000 hematopoietic cell transplantation (HCT) survivors in the US by 2020.1 Patients who survive the period of early complications and disease relapse (generally 1-2 years after transplantation) can expect a high probability of subsequent longterm survival.2-7 Although potentially cured of their underlying disease, HCT survivors continue to be at risk for late complications that can cause substantial morbidity, mortality, and functional deficits, and contribute to psychosocial and quality of life impairments.8-23 Established survivorship guidelines provide a pragmatic approach to the long-term follow up of autologous and allogeneic HCT survivors by recommending a minimum set of screening and preventive evaluations that need to be performed periodically post-transplantation.24,25 Hematopoietic cell transplantation survivors frequently do not receive or adhere to preventive care guidelines.26-28

Many barriers contribute to the inadequate provision of co-ordinated patient-centric survivorship care in this patient population.29-31 Among these, a lack of awareness of exposures and risks by patients is strongly associated with a lower likelihood of adherence to preventive care recommendations.26 In addition, both transplant and nontransplant providers identify lack of knowledge of risks of late complications and of awareness of guidelines as barriers to providing adequate preventive care.32 Finally, capacity limitations at transplant centers may impede provision and co-ordination of preventive care to HCT survivors.29,33-35 Interventions to enhance patient awareness of preventive care could potentially enhance appropriate healthcare utilization and adherence to survivorship guidelines, although this approach has not been previously tested.

A treatment summary and Survivorship Care Plan (SCP) is a tool that provides cancer survivors with information on their cancer type, treatments and potential conse-

Figure 1. Study schema. CIBMTR: Center for International Blood and Marrow Transplant Research; SCP: Survivorship Care Plan.

haematologica | 2019; 104(5)

1085

N.S. Majhail et al.

Table 1. Baseline characteristics of patients enrolled on the study.

Characteristic

SCP (N=231)

Routine care (N=227)

Age at HCT, years; Median (range) Time from HCT to enrollment (months); Median (range) Age group at baseline survey, years; N (%) 18-29 30-39 40-49 50-59 60-69 â&#x2030;Ľ70 Gender; N (%) Male Female Ethnicity; N (%) Hispanic/Latino Non-Hispanic/Latino Declined Race; N (%) White African-American Asian Pacific Islander Declined Diagnosis; N (%) Acute myeloid leukemia Acute lymphoblastic leukemia Myelodysplastic/myeloproliferative disorders Chronic myeloid leukemia Hodgkin lymphoma Non-Hodgkin lymphoma Plasma cell disorder/multiple myeloma Other Time from diagnosis to HCT, months; Median (range) Year of transplant; N (%) 2010 2011 2012 2013 2014 Transplant type; N (%) Allogeneic Autologous Donor type; N (%) Allogeneic, related Allogeneic, unrelated/umbilical cord blood Autologous Graft type; N (%) Bone marrow Peripheral blood Umbilical cord blood

59 (19-81) 42 (16-66)

59 (20-77) 45 (16-66)

7 (3) 10 (4) 28 (12) 55 (24) 83 (36) 48 (21)

9 (4) 14 (6) 18 (8) 58 (26) 92 (41) 36 (16)

112 (49) 119 (52)

136 (60) 91 (40)

8 (3) 216 (94) 7 (3)

7 (3) 216 (95) 4 (2)

222 (96) 5 (2) 2 (1) 1 (<1) 1 (<1)

208 (92) 15 (7) 3 (1) 0 (0) 1 (<1)

52 (23) 10 (4) 19 (8) 2 (1) 13 (6) 49 (21) 78 (34) 8 (3) 7 (1-266)

46 (20) 8 (4) 23 (10) 3 (1) 10 (4) 47 (21) 80 (35) 10 (4) 8 (1- 327)

11 (5) 67 (29) 48 (21) 81 (35) 24 (10)

22 (10) 61 (27) 53 (23) 64 (28) 27 (12)

111 (48) 120 (52)

100 (44) 127 (56)

47 (20) 64 (28) 120 (52)

36 (16) 64 (28) 127 (56)

15 (7) 207 (90) 9 (4)

16 (7) 203 (89) 8 (4) continued on next page

1086

haematologica | 2019; 104(5)

Individualized care plans in HCT survivors continued from previous page

Number of transplants; N (%) 1 ≥2 Conditioning regimen intensity; N (%) Myeloablative (including autologous regimens) Non-myeloablative/reduced-intensity Missing TBI as part of conditioning regimen; N (%) Yes No History of acute GvHD; N (%)* Yes No History of chronic GvHD; N (%)* Yes No Health literacy; N (%)† Adequate literacy Possibility of limited literacy High likelihood of limited literacy

206 (89) 25 (11)

191 (84) 36 (16)

168 (73) 62 (27) 1 (<1)

176 (78) 50 (22) 1 (<1)

49 (21) 182 (79)

46 (20) 181 (80)

70 (63) 41 (37)

67 (67) 33 (33)

67 (60) 44 (40)

66 (66) 34 (34)

154 (74) 36 (17) 18 (9)

172 (83) 27 (13) 9 (4)

N: number; SCP: Survivorship Care Plan; HCT: hematopoietic cell transplantation; TBI: total body irradiation; GvHD: graft-versus-host disease. *Allogeneic HCT only. †Assessed by Newest Vital Sign instrument; n=208 for SCP arm and n=208 for routine care arm.

quences, and recommendations regarding follow up and preventive care. SCPs are generally accepted as an important component of cancer survivorship care.36 Randomized trials of SCPs in cancer patients have primarily focused on providing information through in-person visits with patients or educating primary care providers, and evidence of their efficacy in enhancing various aspects of cancer survivorship care is generally negative.37 They are also frequently underused. This may be due to a variety of reasons, including insufficient resources for their generation and implementation, and a paucity of evidence regarding an impact on patient outcomes.38-42 The use and the dissemination of SCPs in HCT survivors are hampered by similar challenges, and many transplant centers do not routinely provide patients with this tool. Furthermore, these barriers may be accentuated because of the highly complex nature and unique exposures associated with the transplant procedure and the challenges involved in providing co-ordinated survivorship care.29 We hypothesized that a patient-centered approach with a personalized SCP based on published guidelines for the prevention of HCT-related late complications,24,25 and generated using patient data routinely submitted by US transplant centers to an international clinical outcomes registry [Center for International Blood and Marrow Transplant Research (CIBMTR)], would increase patient awareness of recommended preventive care, which in turn would reduce distress, promote healthy behaviors, enhance healthcare utilization for appropriate preventive care, and improve health-related quality of life (HRQOL). By using existing CIBMTR data, this approach would overcome several system-level barriers to providing survivorship care through transplant centers. Furthermore, it could serve as a template for a general, efficient mechanism for providing a patient-centric SCP to long-term HCT survivors who frequently are no longer under the care of haematologica | 2019; 104(5)

transplant centers and are particularly vulnerable to gaps in preventive care. In a multicenter randomized controlled trial (RCT), we evaluated the efficacy of such an individualized SCP instrument generated using registry data and mailed to patients versus standard care on improving patient-reported outcomes in adult HCT survivors 1-5 years after their transplant.

Methods Potentially eligible patients from 17 participating US centers were identified from the CIBMTR database, and paper-based SCPs personalized to HCT specific exposures were generated using CIBMTR data for patients who consented and enrolled on the study (see Online Supplementary Methods).43 Patient eligibility criteria were kept broad and included patients who were 1-5 years post transplant, ≥18 years at the time of HCT, with no evidence of disease relapse/progression or second cancers, and fluency in English; patients were eligible irrespective of transplant type (autologous or allogeneic), diagnosis, donor source or conditioning regimen. None of the participating centers had an existing mechanism for routinely providing SCPs to their patients. The RCT used a multi-center design with patient-level randomization to treatment (Figure 1), and was approved by Institutional Review Boards at the National Marrow Donor Program (NMDP) and each participating site. A random order list of survivors was generated and released in blocks to centers, who confirmed patient survival and accuracy of SCP-related data. Centers contacted patients and obtained their consent to the study, and then informed the CIBMTR, who proceeded with the rest of the study procedures.

The CIBMTR Survey Research Group (SRG) conducted a phone assessment within 30 days of the patient receiving the participant enrollment form. Patients were randomized 1:1 to the SCP or control arm (with delayed SCP). Patients randomized to the SCP arm received an informative letter by express post and their printed 1087

N.S. Majhail et al.

Table 2. Analysis for primary and secondary end points.

End point* Confidence in survivorship information† Cancer and treatment distress‡ Knowledge of transplant exposures† Health care utilization† SF12: physical component summary† SF12: mental component summary†

Mean (Standard Deviation) Baseline 6-months SCP (N=199) Routine care (N=199) SCP (N=199) Routine care (N=198) SCP (N=200) Routine care (N=198) SCP (N=200) Routine care (N=198) SCP (N=200) Routine care (N=198) SCP (N=200) Routine care (N=198)

1.44 (0.34) 1.40 (0.38) 0.91 (0.61) 0.91 (0.64) 0.86 (0.18) 0.88 (0.15) 0.80 (0.14) 0.80 (0.14) 46.1 (10.3) 46.0 (9.8) 53.9 (7.6) 53.9 (7.9)

1.50 (0.34) 1.44 (0.39) 0.78 (0.59) 0.91 (0.69) 0.87 (0.16) 0.86 (0.16) 0.80 (0.15) 0.82 (0.13) 46.2 (10.6) 45.8 (10.1) 54.7 (7.0) 53.4 (8.8)

Estimate (Standard Error)#

P#

-0.034 (0.028)

0.223

0.123 (0.042)

0.004

-0.018 (0.013)

0.182

0.014 (0.010)

0.149

-0.368 (0.638)

0.565

-8.907 (3.009)

0.003

SCP: Survivorship Care Plan; N: number; SF12: Short Form 12. *N: number who completed both baseline and 6-month assessments. #Estimate and P-value based on analysis of covariance model with center-level random effects where any differences between the treatment groups were measured after adjustment for patients’ baseline measurement; where applicable, estimates were adjusted for demographic variables and/or interactions (see Methods section). †Higher score better. ‡Lower score better.

SCP while patients on the control arm only received an informative letter. SRG then contacted all enrolled patients by phone between 7-28 days of mailing study materials to conduct a health literacy assessment using the Newest Vital Sign.44 During this contact, patients on the SCP arm were given the opportunity to address any questions about the content or use of their SCP. No further contact was made till the 6-month phone survey. The Confidence in Survivorship Information (CSI) was the primary end point (Online Supplementary Table S1).45 Secondary end points focused on Cancer and Treatment Distress (CTXD),20,46 as well as measures of Knowledge of Transplant Exposures, Health Care Utilization,26 and HRQOL using the SF-12.47 Patients on the intervention arm also received a 12-item assessment for qualitative feedback on SCP utilization. Sample size calculations were performed using a standard error formula that allowed for possible variability in treatment effect across centers and considered dropouts from baseline to 6 months. Our enrollment goal was 495 patients, which yielded adequate power to detect standardized effect sizes of ≥0.3, which are considered to be clinically meaningful, and anticipated a 10% drop-off from baseline to 6 months. An intention-to-treat approach was followed for analysis. A mixed model with center-level random effects and a fixed treatment effect was used to test whether there was a change in baseline and 6-month response between the treatment and control groups for the primary and secondary end points. The 6-month assessment was used as a response variable and the baseline assessment was used as an explanatory variable in the regression models. If a treatment effect was observed, we further evaluated whether the effect was modified by demographic variables or any interactions between variables.

Further details are available in the Online Supplementary Appendix.

Results Patients' characteristics Among the 495 patients enrolled, 458 completed the baseline survey and were randomized (SCP=231, control=227); 200 (87%) and 199 (88%) completed 6-month assessments, respectively (Figure 2). The main reasons for dropout were loss to follow up or patients not eligible for follow-up assessment due to interim disease relapse or 1088

progression. A greater proportion of patients who completed the 6-month assessment were White and reported higher health literacy scores; otherwise there were no significant differences in the demographic characteristics between patients who did and those who did not complete the 6-month assessments (Online Supplementary Table S2). Patients' and transplant characteristics (including health literacy scores) were well balanced between the two arms, except for gender (49% males in SCP compared to 60% in controls; P=0.01) (Table 1). Median age was 59 years in both arms and enrolled patients were predominantly White (96% SCP and 92% controls). In the SCP and control arms, 48% and 44% had received allogeneic HCT; among allogeneic HCT recipients 63% and 67% had a history of acute GvHD, and 60% and 66% had a history of chronic GvHD, respectively.

Analyses of primary and secondary end points Of the 458 patients randomized to the two arms, 399 completed 6-month assessments, including 398 who completed pre- and post-measurements for the primary end point (Table 2). We did not find any association between the SCP intervention and change in CSI scores from baseline to 6-months (P=0.223), even after assessing for the effect of demographic factors and interactions. However, we did observe a significant decrease in CTXD scores (P=0.004) and an increase in HRQOL Mental Component Summary (MCS) scores as assessed by SF-12 (P=0.003) among patients randomized to the SCP arm. There was no association between the SCP intervention and other secondary end points.

We further assessed the effect of demographic variables and interactions for the end points of CTXD and SF-12 MCS, where a significant treatment effect was observed.

Age was significantly associated with CTXD scores (regression estimate -0.006, standard error 0.002; P=0.001), with lower distress among older patients.

However, there was no significant interaction between age and SCP intervention and adjustment for age did not modify the treatment effect. The decrease in CTXD score for the SCP arm was independent of gender, health literacy, diagnosis, transplant type, and GvHD status (including acute and chronic GvHD). We also found a similar effect haematologica | 2019; 104(5)

Individualized care plans in HCT survivors

Figure 2. CONSORT diagram. N: number; SCP: Survivorship Care Plan.

of age on MCS scores, with older patients reporting significantly higher scores (estimate 0.03, standard error 0.034; P<0.001), and there was a significant interaction between age and SCP intervention (P=0.012). However, increase in mean MCS score in the SCP arm was independent of gender, health literacy, diagnosis, transplant type, and GvHD status.

Utilization of Survivorship Care Plans At their 6-month end-of-study assessments, patients on the intervention arm were asked questions about the usefulness of the SCP for their survivorship care (Figure 3). A relatively large proportion of survivors reported that they found the SCP somewhat or very useful for helping them better understand their HCT and related treatments (70%), side effects of HCT (65%), and managing their health (69%). The SCP helped survivors better communicate about HCT and its side effects with their medical providers. The 6-month interview included an openended question about patients' experience with the SCP; dominant themes identified on qualitative analyses included patients reporting that the SCP helped survivors focus on their overall health, supported them in making care decisions with providers, and supported emotional health and coping.

Discussion In this large multicenter RCT of HCT survivors 1-5 years post transplantation, we demonstrate that SCPs generated using a centralized clinical registry (CIBMTR), individualized to patient exposures, and without clinician haematologica | 2019; 104(5)

contact to interpret or personalize their content and recommendations, are feasible and have desirable outcomes, including lower treatment-related distress and improved mental health domain of HRQOL. Our results support further research towards broader implementation of our SCP instrument to facilitate care of HCT survivors, and provides evidence to support a patient-centered approach towards administration of SCPs. SCPs have been endorsed as a tool for facilitating the care of cancer survivors with the goal of improving patient outcomes by promoting coordination of care, shared-decision making, self-management, and adherence to treatment recommendations.36,48 Evidence on their efficacy in impacting patientsâ&#x20AC;&#x2122; outcomes is mixed, and SCPs have not been universally adopted due to other barriers, such as the lack of standardized templates, the need for extensive resources and time for their generation, and the lack of reimbursement for their implementation.42,48-50 Transplant centers face similar challenges, and many programs have capacity limitations that frequently prevent provision of personalized comprehensive SCPs to their patients. Our SCP procedure provides several advantages to patients and transplant centers. It uses data that centers routinely submit electronically to the CIBMTR and will provide a resource-effective mechanism for centers to generate the SCP for their recipients. Instead of receiving a generic SCP, patients can receive one that is specific to their treatment exposures. Our approach of facilitating patient ownership of survivorship care is different from the prevalent non-transplant cancer literature where SCPs have largely been tested in a context in which clinicians provide them to their patients.37 Our SCP instrument was in a paper-based format and was mailed to patients; more general dissemination would require its 1089

N.S. Majhail et al.

Figure 3. Patient-reported assessment of usefulness of Survivorship Care Plan (SCP) intervention. N=201 respondents on SCP arm who completed 6-month end-ofstudy assessments.

translation into an electronic format. Hence, further research is still needed to guide its implementation. An ongoing project funded by the National Cancer Institute is investigating its use in combination with an online health informatics platform to facilitate a self-management program for selected late complications among HCT survivors (clinicaltrials.gov identifier: 03125070; Syrjala, Baker and Majhail).

Of note, we did not observe any impact of the SCP intervention on our primary end point of CSI. Our study population consisted of HCT survivors who had been transplanted relatively recently (1-5 years) and enrolled by centers with an interest in providing survivorship care to their transplant recipients; it is possible our instrument may be more effective in enhancing knowledge and confidence about follow-up care among patients who underwent transplantation among patients further out from transplantation or those who are not followed primarily or closely by their transplant centers. The CSI instrument has been validated in cancer survivors but not among HCT recipients, and it is also possible that it did not adequately measure the underlying construct in our patient population. The 6-month pre- and post-intervention follow-up period was most likely too short to detect any significant associations with changes in healthcare adherence or utilization. We did not observe any interaction of 1090

GvHD with the intervention or study outcomes. This was most likely due to our study population being relatively further out from transplantation and the short duration of the intervention. Furthermore, it is likely that patients with GvHD were under the active care of transplant centers and this may have impacted patient-reported outcomes assessed in our study (e.g. greater confidence in recommended care, less distress, etc.). These same factors were probably responsible for some patients not finding the SCP tool to be useful for various aspects of survivorship care (see Figure 3; “I have not done this” and “Not at all useful” responses on SCP utilization survey administered as part of end-of-study assessments for the intervention arm).

The concordant findings of a reduction in CTXD scores and an improvement in SF-12 MCS scores cross-validate the overall effect of SCP on reducing distress and improving HRQOL in our study population of HCT survivors. It is important to note that these effects occurred over a relatively short period of time and did not require any additional clinical contact or intervention to facilitate the use of the SCP. Interestingly, we found an independent association between older age and lower CTXD scores, which is consistent with other literature where older adults are less distressed about cancer and survivorship.51-54 The SCP provided concise information on previous treatments and haematologica | 2019; 104(5)

Individualized care plans in HCT survivors

potential late effects, and practical guidance regarding recommended preventive care that survivors could easily understand and share, which may have empowered them in the CTXD domains (e.g. uncertainty, health burden and medical demands) and MCS domains (e.g. mental health, social functioning, role-emotional), leading to the improvement in those areas.55 Some limitations of our study must be acknowledged.

First, the treatment summary portion of our SCP primarily included HCT-related and post-transplant events and did not have detailed information on pre-transplant exposures as those data are not captured comprehensively by the CIBMTR. However, transplant centers have the option to add information about those exposures to the basic template of the SCP. Participants who completed 6-month assessments were more likely White and had higher health literacy, which may limit the extent to which our findings can be generalized. However, this is reflective of the prevailing healthcare disparities in HCT, and research on other interventions to facilitate SCP use in this underserved population is needed.56 Notwithstanding these limitations, the pragmatic nature of our study eligibility criteria and schema will make our results broadly applicable to transplant centers in the US. An ideal mechanism to provide SCPs to HCT survivors would involve a dynamic, adaptable, and patient-specific shared-decision making approach between patients, their transplant centers, and other providers. However, several challenges prevent centers from providing this tool to facilitate survivorship care for their patients, and SCPs that can be generated efficiently and without requiring significant center resources would have an impact on patient care. Our study supports further implementation of an individualized SCP generated using CIBMTR data in a population of HCT survivors that is at significant risk for late morbidity and mortality. Future research will examine the role of the SCP instrument in preventing specific late complications, in facilitating co-ordination of care, and will serve as a platform for investigating novel methods for survivorship care delivery and implementation.

Acknowledgments Key to the success of this patient-centered outcomes research study were the many contributions and engagement of our more than 40 stakeholder group representatives including: transplant recipients, caregivers, community hematology/oncology physi-

References 1. Majhail NS, Tao L, Bredeson C, et al.

Prevalence of hematopoietic cell transplant survivors in the United States. Biol Blood Marrow Transplant. 2013;19(10):1498-1501.

2. Wingard JR, Majhail NS, Brazauskas R, et al.

Long-term survival and late deaths after allogeneic hematopoietic cell transplantation. J Clin Oncol. 2011;29(16):2230-2239.

3. Majhail NS, Bajorunaite R, Lazarus HM, et al. High probability of long-term survival in 2-year survivors of autologous hematopoietic cell transplantation for AML in first or second CR. Bone Marrow Transplant.

2011;46(3):385-392.

4. Majhail NS, Bajorunaite R, Lazarus HM, et al. Long-term survival and late relapse in 2year survivors of autologous haematopoiet-

haematologica | 2019; 104(5)

cians, and transplant physicians, nurses, social workers, patient health educators, and patient advocates. We are deeply indebted to the National Marrow Donor Programâ&#x20AC;&#x2122;s Patient Services Advisory Group, which includes representation from many of these stakeholder groups, of whom several members participated in the protocol team for this study or provided feedback and guidance on study design and interpretation of results. The study investigators would like to the following transplant centers for enrolling patients on this study (site investigators are listed in parentheses): Baylor University, Dallas, TX (Jana Reynolds); Cleveland Clinic, Cleveland, OH (Navneet Majhail); Fred Hutchinson Cancer Research Center, Seattle, WA (K Scott Baker), Karmanos Cancer Institute, Detroit, MI (Abhinav Deol); Loyola University, Chicago, IL (Patrick Stiff); Mayo Clinic, Rochester, MN (Shahrukh Hashmi); Mayo Clinic, Scottsdale, AZ (Nandita Khera); Ohio State University, Columbus, OH (Samantha Jaglowski); Roswell Park Comprehensive Cancer Center, Buffalo, NY (Theresa Hahn); UMass Memorial Medical Center, Worcester, MA (Jan Cerny); University of Florida, Gainesville, FL (John R. Wingard); University of Kansas, Kansas City, KS (Joseph McGuirk); University of Minnesota, Minneapolis, MN (Shernan Holtan); University of North Carolina, Chapel Hill, NC (William Wood); University of Pennsylvania, Philadelphia, PA (Alison Loren); Vanderbilt University, Nashville, TN (Bipin Savani). We are especially grateful to study coordinators at all sites who worked tirelessly and often outside regular working hours to contact and enroll patients on our trial. We would also like to thank members of CIBMTRâ&#x20AC;&#x2122;s Health Services Research (HSR) Program and Resource for Clinical Investigation in Blood and Marrow Transplantation (RCI BMT), including their Survey Research Group, who helped with study conduct, data collection and management, and performed patient interviews.

Funding This study was funded through a Patient-Centered Outcomes Research Institute (PCORI) Award (CD-12-11-4062). The views in this publication are solely the responsibility of the authors and do not necessarily represent the views of the PatientCentered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee. K. Scott Baker, Navneet Majhail and Karen Syrjala are partially supported by a grant from the National Cancer Institute (R01-CA215134). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

ic cell transplantation for Hodgkin and nonHodgkin lymphoma. Br J Haematol.

2009;147(1):129-139.

5. Martin PJ, Counts GW Jr, Appelbaum FR, et al. Life expectancy in patients surviving more than 5 years after hematopoietic cell transplantation. J Clin Oncol. 2010; 28(6):1011-1016.

6. Bhatia S, Francisco L, Carter A, et al. Late mortality after allogeneic hematopoietic cell transplantation and functional status of long-term survivors: report from the Bone Marrow Transplant Survivor Study. Blood.

2007;110(10):3784-3792.

7. Bhatia S, Robison LL, Francisco L, et al. Late mortality in survivors of autologous hematopoietic-cell transplantation: report from the Bone Marrow Transplant Survivor Study. Blood. 2005;105(11):4215-4222.

8. Sun CL, Francisco L, Baker KS, Weisdorf DJ, Forman SJ, Bhatia S. Adverse psychological outcomes in long-term survivors of hematopoietic cell transplantation: a report from the Bone Marrow Transplant Survivor Study (BMTSS). Blood. 2011;118(17):47234731.

9. Armenian SH, Sun CL, Kawashima T, et al.

Long-term health-related outcomes in survivors of childhood cancer treated with HSCT versus conventional therapy: a report from the Bone Marrow Transplant Survivor Study (BMTSS) and Childhood Cancer Survivor Study (CCSS).

Blood.

2011;118(5):1413-1420.

10. Sun CL, Francisco L, Kawashima T, et al.

Prevalence and predictors of chronic health conditions after hematopoietic cell transplantation: a report from the Bone Marrow

1091

N.S. Majhail et al.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

1092

Transplant Survivor Study. Blood. 2010; 116(17):3129-3139; quiz 3377.

Baker KS, Ness KK, Weisdorf D, et al. Late effects in survivors of acute leukemia treated with hematopoietic cell transplantation: a report from the Bone Marrow Transplant Survivor Study.

Leukemia.

2010; 24(12):2039-2047.

Majhail NS, Ness KK, Burns LJ, et al. Late effects in survivors of Hodgkin and nonHodgkin lymphoma treated with autologous hematopoietic cell transplantation: a report from the bone marrow transplant survivor study. Biol Blood Marrow Transplant.

2007;13(10):1153-1159.

Majhail NS, Douglas Rizzo J. Surviving the cure: long term followup of hematopoietic cell transplant recipients. Bone Marrow Transplant. 2013;48(9):1145-1151.

Battiwalla M, Hashmi S, Majhail N, Pavletic S, Savani BN, Shelburne N. National Institutes of Health Hematopoietic Cell Transplantation Late Effects Initiative: Developing Recommendations to Improve Survivorship and Long-Term Outcomes. Biol Blood Marrow Transplant. 2017;23(1):6-9.

Battiwalla M, Tichelli A, Majhail NS. LongTerm Survivorship after Hematopoietic Cell Transplantation: Roadmap for Research and Care. Biol Blood Marrow Transplant.

2017;23(2):184-192.

Majhail NS. Long-term complications after hematopoietic cell transplantation. Hematol Oncol Stem Cell Ther. 2017;10(4):220-227.

Syrjala KL, Artherholt SB, Kurland BF, et al.

Prospective neurocognitive function over 5 years after allogeneic hematopoietic cell transplantation for cancer survivors compared with matched controls at 5 years. J Clin Oncol. 2011;29(17):2397-2404.

Kirchhoff AC, Leisenring W, Syrjala KL. Prospective predictors of return to work in the 5 years after hematopoietic cell transplantation. J Cancer Surviv. 2010;4(1):33-44.

Syrjala KL, Langer SL, Abrams JR, Storer BE, Martin PJ. Late effects of hematopoietic cell transplantation among 10-year adult survivors compared with case-matched controls. J Clin Oncol. 2005;23(27):6596-6606.

Syrjala KL, Langer SL, Abrams JR, et al.

Recovery and long-term function after hematopoietic cell transplantation for leukemia or lymphoma.

JAMA. 2004;291(19):2335-2343.

Majhail NS, Brazauskas R, Rizzo JD, et al.

Secondary solid cancers after allogeneic hematopoietic cell transplantation using busulfan-cyclophosphamide conditioning.

Blood. 2011;117(1):316-322.

Majhail NS. Secondary cancers following allogeneic haematopoietic cell transplantation in adults. Br J Haematol.

2011;154(3):301-310.

Sun CL, Kersey JH, Francisco L, et al. Burden of morbidity in 10+ year survivors of hematopoietic cell transplantation: report from the bone marrow transplantation survivor study. Biol Blood Marrow Transplant.

2013;19(7):1073-1080.

Majhail NS, Rizzo JD, Lee SJ, et al.

Recommended screening and preventive practices for long-term survivors after hematopoietic cell transplantation. Bone Marrow Transplant. 2012;47(3):337-341.

Majhail NS, Rizzo JD, Lee SJ, et al.

Recommended screening and preventive practices for long-term survivors after hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2012;18(3):348371.

26. Khera N, Chow EJ, Leisenring WM, et al.

Factors associated with adherence to preventive care practices among hematopoietic cell transplantation survivors. Biol Blood Marrow Transplant. 2011;17(7):995-1003.

27. Bishop MM, Lee SJ, Beaumont JL, et al. The preventive health behaviors of long-term survivors of cancer and hematopoietic stem cell transplantation compared with matched controls. Biol Blood Marrow Transplant.

2010;16(2):207-214.

28. Armenian SH, Sun CL, Francisco L, et al.

Health behaviors and cancer screening practices in long-term survivors of hematopoietic cell transplantation (HCT): a report from the BMT Survivor Study. Bone Marrow Transplant. 2012;47(2):283-290.

29. Hashmi SK, Bredeson C, Duarte RF, et al.

National Institutes of Health Blood and Marrow Transplant Late Effects Initiative: The Healthcare Delivery Working Group Report. Biol Blood Marrow Transplant.

2017;23(5):717-725.

30. Majhail NS. Optimizing Quality and Efficiency of Healthcare Delivery in Hematopoietic Cell Transplantation. Curr Hematol Malig Rep. 2015;10(3):199-204.

31. Khera N, Martin P, Edsall K, et al. Patientcentered care coordination in hematopoietic cell transplantation. Blood Adv. 2017; 1(19):1617-1627.

32. Mani S, Rybicki L, Carraway H, Moore H, Vakharia N, majhail NS. Primary care physician preferences and perspectives on longterm care of survivors of hematologic malignancies and hematopoietic cell transplantation. Biol Blood Marrow Transplant.

2016;22(3 (Suppl)):S85-86.

33. Denzen EM, Majhail NS, Stickney Ferguson S, et al. Hematopoietic cell transplantation in 2020: summary of year 2 recommendations of the National Marrow Donor Program's System Capacity Initiative. Biol Blood Marrow Transplant. 2013;19(1):4-11.

34. Majhail NS, Mau LW, Chitphakdithai P, et al.

National Survey of Hematopoietic Cell Transplantation Center Personnel, Infrastructure, and Models of Care Delivery.

Biol Blood Marrow Transplant.

2015;21(7):1308-1314.

35. Majhail NS, Murphy EA, Denzen EM, et al.

The National Marrow Donor Program's Symposium on Hematopoietic Cell Transplantation in 2020: a health care resource and infrastructure assessment. Biol Blood Marrow Transplant. 2012; 18(2):172182.

36. McCabe MS, Bhatia S, Oeffinger KC, et al.

American Society of Clinical Oncology statement: achieving high-quality cancer survivorship care. J Clin Oncol. 2013; 31(5):631-640.

37. Jacobsen PB, DeRosa AP, Henderson TO, et al. Systematic Review of the Impact of Cancer Survivorship Care Plans on Health Outcomes and Health Care Delivery. J Clin Oncol. 2018;36(20):2088-2100.

38. Salz T, Oeffinger KC, McCabe MS, Layne TM, Bach PB. Survivorship care plans in research and practice. CA Cancer J Clin.

2012;62(2):101-117.

39. Stricker CT, Jacobs LA, Risendal B, et al.

Survivorship care planning after the institute of medicine recommendations: how are we faring? J Cancer Surviv. 2011; 5(4):358-370.

40. Jacobs LA, Shulman LN. Follow-up care of cancer survivors: challenges and solutions.

Lancet Oncol. 2017;18(1):e19-e29.

41. Klemanski DL, Browning KK, Kue J. Survivorship care plan preferences of cancer

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

survivors and health care providers: a systematic review and quality appraisal of the evidence. J Cancer Surviv. 2016; 10(1):71-86.

Brennan ME, Gormally JF, Butow P, Boyle FM, Spillane AJ. Survivorship care plans in cancer: a systematic review of care plan outcomes. Br J Cancer. 2014;111(10):1899-1908.

Denzen EM, Preussler JM, Murphy EA, et al.

Tailoring a survivorship care plan: Patient and provider preferences for recipients of hematopoietic cell transplantation Biol Blood Marrow Transplant. 2018; S10838791(18)30614-1.

Weiss BD, Mays MZ, Martz W, et al. Quick assessment of literacy in primary care: the newest vital sign. Ann Fam Med. 2005;3(6):514-522.

Palmer SC, Jacobs LA, Mao J, Stricker CT. Are cancer survivors confident in their knowledge about their disease and care?

Poster presented at the 6th Biennial Cancer Survivorship Research: Translating Science to Care, June 14, 2012. Available from: https://smhs.gwu.edu/cancercontroltap/site s/cancercontroltap/files/CSI%20description.pdf?src=TAPResource.

Syrjala KL, Chapko ME. Evidence for a biopsychosocial model of cancer treatmentrelated pain. Pain. 1995;61(1):69-79.

Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996; 34(3):220-233.

Mayer DK, Nekhlyudov L, Snyder CF, Merrill JK, Wollins DS, Shulman LN. American Society of Clinical Oncology clinical expert statement on cancer survivorship care planning. J Oncol Pract. 2014;10(6):345351.

Blanch-Hartigan D, Forsythe LP, Alfano CM, et al. Provision and discussion of survivorship care plans among cancer survivors: results of a nationally representative survey of oncologists and primary care physicians. J Clin Oncol. 2014;32(15):1578-1585.

Hewitt ME, Bamundo A, Day R, Harvey C. Perspectives on post-treatment cancer care: qualitative research with survivors, nurses, and physicians. J Clin Oncol. 2007; 25(16):2270-2273.

Acquati C, Kayser K. Predictors of psychological distress among cancer patients receiving care at a safety-net institution: the role of younger age and psychosocial problems.

Support Care Cancer. 2017; 25(7):23052312.

Head BA, Schapmire TJ, Keeney CE, et al.

Use of the Distress Thermometer to discern clinically relevant quality of life differences in women with breast cancer. Qual Life Res. 2012;21(2):215-223.

Jacobsen PB, Shibata D, Siegel EM, et al.

Evaluating the quality of psychosocial care in outpatient medical oncology settings using performance indicators.

Psychooncology. 2011;20(11):1221-1227.

Zebrack B, Kayser K, Sundstrom L, et al.

Psychosocial distress screening implementation in cancer care: an analysis of adherence, responsiveness, and acceptability. J Clin Oncol. 2015;33(10):1165-1170.

Syrjala KL, Sutton SK, Jim HS, et al. Cancer and treatment distress psychometric evaluation over time: A BMT CTN 0902 secondary analysis. Cancer. 2017;123(8):1416-1423.

Majhail NS, Nayyar S, Santibanez ME, Murphy EA, Denzen EM. Racial disparities in hematopoietic cell transplantation in the United States. Bone Marrow Transplant.

2012;47(11):1385-1390.

haematologica | 2019; 104(5)