Haematologica, Volume 103, Issue 12 by Haematologica

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

Ancient Greek

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

Scientific Latin

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

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

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

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

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

Managing Director Antonio Majocchi (Pavia)

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

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

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

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

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

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

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

Institutional Euro 600

Personal Euro 150

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

haematologica calendar of events

Journal of the European Hematology Association Published by the Ferrata Storti Foundation The American Society of Hematology 60th ASH Annual Meeting and Exposition American Society of Hematology (ASH) Chairs: A Thompson, M Crowther, M Sekeres, J Crispino, M Sola-Visner December 1-4, 2018 San Diego, USA EHA-SLCH Hematology Tutorial on Myeloid Malignancies and MDS Chairs: G Ossenkoppele, V Gunawardena, HW Goonasekera February 8-9, 2019 Colombo, Sri Lanka 1st European CAR T cell in Hematology Meeting Organized by EHA and EBMT February 14-16, 2019 Paris, France EHA-AORK Hematology Tutorial on Lymphoma and Multiple Myeloma Chair: D Kaidarova, Co-chairs: S Gabbasova & B Afanasyev March 14-16, 2019 Almaty, Kazakhstan

Pediatric Course 2019 Endorsed by EHA April 3-6, 2019 Sorrento, Italy EHA-ROHS-RHS Hematology Tutorial on Real World Challenges and Opportunities in Diagnostics and Management of Onco-Hematological Patients Today Chairs: I Poddubnaya, E Parovnichikova April 12-13, 2019 Moscow, Russia 24th Congress of EHA European Hematology Association June 13-16, 2019 Amsterdam, The Netherlands

Calendar of Events updated on November 5, 2018

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

Table of Contents Volume 103, Issue 12: December 2018 Cover Figure

Peripheral blood smear showing plasmacytoid lymphocytes and increased rouleaux formation in a patient with Waldenstrรถm macroglobulinemia. Courtesy of Prof. Rosangela Invernizzi.

Editorials 1937

Twisting the bone marrow stem cell niche Haixia Niu and Jose A. Cancelas

1939

Assessment of iron deficiency Chaim Hershko

1942

Is DNA a better assay for residual disease in chronic myeloid leukemia? Jerald Radich

Review Article 1945

Osteogenic niche in the regulation of normal hematopoiesis and leukemogenesis Phuong M. Le et al.

1956

TP53 aberrations in chronic lymphocytic leukemia: an overview of the clinical implications of improved diagnostics Elias Campo et al.

Articles Hematopoiesis

1969

Niche TWIST1 is critical for maintaining normal hematopoiesis and impeding leukemia progression Xiaoyan Liu et al.

1980

ASXL2 regulates hematopoiesis in mice and its deficiency promotes myeloid expansion Vikas Madan et al.

Iron Metabolism & its Disorders

1991

Optimizing diagnostic biomarkers of iron deficiency anemia in community-dwelling Indian women and preschool children Giridhar Kanuri et al.

Red Cell Biology & its Disorders

1997

Non-muscle myosin II drives vesicle loss during human reticulocyte maturation Pedro L. Moura

2008

The phenotypic spectrum of germline YARS2 variants: from isolated sideroblastic anemia to mitochondrial myopathy, lactic acidosis and sideroblastic anemia 2 Lisa G. Riley et al.

Chronic Myeloid Leukemia

2016

BCR-ABL1 mediated miR-150 downregulation through MYC contributed to myeloid differentiation block and drug resistance in chronic myeloid leukemia Klara Srutova et al.

2026

BCR-ABL1 genomic DNA PCR response kinetics during first-line imatinib treatment of chronic myeloid leukemia Ilaria S. Pagani et al.

Haematologica 2018; vol. 103 no. 12 - December 2018 http://www.haematologica.org/

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

2033

Arsenic trioxide is required in the treatment of newly diagnosed acute promyelocytic leukemia. Analysis of a randomized trial (APL 2006) by the French Belgian Swiss APL group Lionel AdĂ¨s et al.

2040

Physician uncertainty aversion impacts medical decision making for older patients with acute myeloid leukemia: results of a national survey Pierre Bories et al.

Non-Hodgkin Lymphoma

2049

Bromodomain and extra-terminal domain inhibition modulates the expression of pathologically relevant microRNAs in diffuse large B-cell lymphoma Afua A. Mensah et al.

2059

Cyclin-dependent kinase 9 as a potential specific molecular target in NK-cell leukemia/lymphoma Shiori Kinoshita et al.

Chronic Lymphocytic Leukemia

2069

Trisomy 12 chronic lymphocytic leukemia expresses a unique set of activated and targetable pathways Lynne V. Abruzzo et al.

Plasma Cell Disorders

2079

Daratumumab plus bortezomib and dexamethasone versus bortezomib and dexamethasone in relapsed or refractory multiple myeloma: updated analysis of CASTOR Andrew Spencer et al.

2088

Daratumumab plus lenalidomide and dexamethasone versus lenalidomide and dexamethasone in relapsed or refractory multiple myeloma: updated analysis of POLLUX Meletios A. Dimopoulos et al.

Platelet Biology & its disorders

2097

Inhibition of Btk by Btk-specific concentrations of ibrutinib and acalabrutinib delays but does not block platelet aggregation mediated by glycoprotein VI Phillip L.R. Nicolson et al.

Stem Cell Transplantation

2109

Diffuse alveolar hemorrhage is most often fatal and is affected by graft source, conditioning regimen toxicity, and engraftment kinetics Fatma Keklik et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/103/12.toc

e561

Recurrent heteroplasmy for the MT-ATP6 p.Ser148Asn (m.8969G>A) mutation in patients with syndromic congenital sideroblastic anemia of variable clinical severity Simon Berhe et al. http://www.haematologica.org/content/103/12/e561

e564

Sideroblastic anemia with myopathy secondary to novel, pathogenic missense variants in the YARS2 gene Frances Smith et al. http://www.haematologica.org/content/103/12/e564

Haematologica 2018; vol. 103 no. 12 - December 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation e567

Development of angiotensin II (1-7) analog as an oral therapeutic for the treatment of chemotherapy-induced myelosuppression Kevin Gaffney et al. http://www.haematologica.org/content/103/12/e567

e571

The fetal liver lymphoid-primed multipotent progenitor provides the prerequisites for the initiation of t(4;11) MLL-AF4 infant leukemia Camille Malouf and Katrin Ottersbach http://www.haematologica.org/content/103/12/e571

e575

A novel type of NPM1 mutation characterized by multiple internal tandem repeats in a case of cytogenetically normal acute myeloid leukemia Nicolas Duployez et al. http://www.haematologica.org/content/103/12/e575

e578

Acute myeloid leukemia in very old patients Vladimir Lj Lazarevic et al. http://www.haematologica.org/content/103/12/e578

e581

Relapse of acute myeloid leukemia after allogeneic stem cell transplantation is associated with gain of WT1 alterations and high mutation load Sebastian Vosberg et al. http://www.haematologica.org/content/103/12/e581

e585

A comprehensive inventory of TLX1 controlled long non-coding RNAs in T-cell acute lymphoblastic leukemia through polyA+ and total RNA sequencing Karen Verboom et al. http://www.haematologica.org/content/103/12/e585

e590

Interim PET-directed therapy in limited-stage Hodgkin lymphoma initially treated with ABVD Diego Villa et al. http://www.haematologica.org/content/103/12/e590

e594

Impact of Sox11 over-expression in Ba/F3 cells Martin Lord et al. http://www.haematologica.org/content/103/12/e594

e598

Idelalisib impairs T-cell-mediated immunity in chronic lymphocytic leukemia Silvia Martinelli et al. http://www.haematologica.org/content/103/12/e598

e602

Treatment optimization for multiple myeloma: schedule-dependent synergistic cytotoxicity of pomalidomide and carfilzomib in in vitro and ex vivo models Enrica Borsi et al. http://www.haematologica.org/content/103/12/e602

Case Reports Case Reports are available online only at www.haematologica.org/content/103/12.toc

e607

Somatic reversion events point towards RPL4 as a novel disease gene in a condition resembling Diamond-Blackfan anemia Marjolijn C. J. Jongmans et al. http://www.haematologica.org/content/103/12/e607

e610

Novel iatrogenic amyloidosis caused by peptide drug liraglutide: a clinical mimic of AL amyloidosis Carlo O. Martins et al. http://www.haematologica.org/content/103/12/e610

e613

A mutation in the gene coding for the sialic acid transporter SLC35A1 is required for platelet life span but not proplatelet formation Alexandre Kauskot et al. http://www.haematologica.org/content/103/12/e613

Haematologica 2018; vol. 103 no. 12 - December 2018 http://www.haematologica.org/

EDITORIALS Twisting the bone marrow stem cell niche Haixia Niu1 and Jose A. Cancelas1,2 1

Division of Experimental Hematology and Cancer Biology, Cancer & Blood Diseases Institute, Cincinnati Childrenâ&#x20AC;&#x2122;s Hospital Medical Center, 3333 Burnet Ave., Cincinnati, and 2Hoxworth Blood Center, University of Cincinnati Academic Health Center, OH, USA E-mail: jose.cancelas@uc.edu doi:10.3324/haematol.2018.206029

ematopoietic stem cells (HSC) are the source of blood cells over the whole lifespan of mammals. HSC locate in the bone marrow microenvironment in the so-called bone marrow stem cell niche. This niche consists of multiple cell types, such as endothelial cells, mesenchymal stem cells (MSC), CXCL12-expressing adventitial cells (CAR cells), nonmyelinating Schwann cells, pericytes, osteoblasts, adipocytes and hematopoietic cells including osteoclasts, regulatory T cells, macrophages and neutrophils.1 Our understanding of the specific contributions of each of these populations derives from the use of constitutional or inducible genetic murine models in which more-or-less specific niche cell populations are targeted for deletion, or for the introduction of loss-of-function or gain-of-function mutations. However, the complicated but precise crosstalk between HSC and specific bone marrow HSC niche cells is still poorly understood. In this issue of the journal, Liu et al. report on their discovery that inducible deficiency of Twist1 in the bone marrow HSC niche results in a broad alteration of multiple niche cell types as well as a reduced production of major HSC supportive factors. The dysregulated Twist1-deficient mesenchymal bone marrow HSC niche reduces HSC homing and retention, impairs HSC selfrenewal and induces myeloid-biased differentiation. Furthermore, Twist1 deficiency in the bone marrow HSC niche accelerates the progression of MLL-AF9-induced acute myeloid leukemia (AML), partially through upregulated Jagged2/Notch signaling.2 Three significant aspects of this study need to be highlighted. First of all, the discovery adds new mechanisms to our understanding of how the bone marrow HSC niche regulates HSC function. Secondly, the findings provide cumulative evidence that remodeling of the bone marrow HSC niche plays an important role in leukemogenesis and is, therefore, a potential drug target for therapy. Last but not least, the authors identified a new function of the transcription factor Twist1 which participates in normal and malignant hematopoiesis.

Niche Twist1 deficiency impairs hematopoietic stem cell function, homing and retention Twist proteins are members of the basic Helix-Loop-Helix transcription factor family with highly conserved sequences. There are two Twist proteins, Twist1 and Twist2. They are well-known for their essential functions in development.3 In the hematopoietic system, Twist1 is expressed in HSC and progenitors, while Twist2 is expressed in more differentiated myeloid lineage cells. However, the function of Twist1 in hematopoiesis has not been well studied. Earlier studies by the same group demonstrated that Twist1 is a regulator of HSC self-renewal and lineage commitment.4 Twist1 also participates in the regulation of MSC function and in MSC selfrenewal. Twist1 overexpression in MSC increased C-X-C motif chemokine 12 ligand (Cxcl12) production, which enhanced the capacity to maintain human HSC and progenihaematologica | 2018; 103(12)

tors in stromal-dependent long-term culture-initiating cell assays. Twist1-deficient heterozygous mice showed reduced numbers of MSC in the bone marrow.5 Twist1 also regulates the cell fate and commitment of MSC. Twist1 silencing enhanced in vitro and in vivo osteogenic differentiation of human adipose-derived MSC by triggering the activation of BMP-ERK-FGF signaling and TAZ upregulation. Silencing Twist1 in a murine MSC cell line, C3H10T1/2, enhanced osteogenic differentiation.6 Using a chimeric mouse model, Liu et al., found that Twist1 deletion in the bone marrow niche leads to decreased numbers of MSC and mature osteoblasts,2 findings which are consistent with already reported data. However, both sinusoidal and arteriolar endothelial cell numbers increased, consistent with an increase in the number of microvessels. This finding contrasts with that of earlier studies in which Twist1 upregulation, as identified in numerous carcinomas, promoted neo-angiogenesis.7 These distinct roles of Twist1 in tumor epithelial cells may be attributed to the different microenvironment found in solid tumors compared with the bone marrow HSC niche. More studies need to be performed to address this difference. Most HSC are retained in the BM niche in a quiescent state by interacting with the niche cells. The inducible deficiency of Twist1 in the bone marrow HSC niche promoted cell cycle entry of long-term HSC, impaired long-term HSC self-renewal and biased HSC differentiation to the myeloid lineage. The CXCL12-CXCR4 axis is essential for HSC retention in the bone marrow.8 Granulocyte-colony stimulating factor (G-CSF), an endogenously expressed cytokine secreted by MSC, osteolineage cells, endothelial cells and macrophages, induces mobilization of quiescent HSC through downregulation of bone marrow expression of CXCL12.9,10 Liu et al. found that the inducible deficiency of Twist1 impaired HSC homing and promoted HSC mobilization, effects associated with decreased expression of CXCL12 and other key cytokines/chemokines crucial for bone marrow HSC retention, such as VCAM, SCF, and Angpt1 in mutant bone marrow MSC, and increased GCSF expression in bone marrow osteolineage cells and macrophages.2 Together with their earlier finding,4 these data indicate that Twist1 regulates HSC function in both a cellautonomous and non-cell-autonomous manner. An earlier study by the same group showed that Twist1 regulates HSC myeloid differentiation by activating the myeloid transcription factors PU.1 and Gata1 and downregulating the lymphoid transcription factor Gata3.4 A Twist1-deficient bone marrow HSC niche promoted early myeloid differentiation.2 However, the underlying molecular mechanism was not identified. Twist2, a homolog of Twist1, has also been identified as a major negative regulator of myeloid cell development and the pro-inflammatory responses of these cells.11 Twist2 is expressed in granulocyte-macrophage progenitors and inhibits their proliferation and differentiation into macrophages, neutrophils and basophils through direct interactions and inhibi1937

Editorials Figure 1. Effect of inducible deficiency of Twist1 in the bone marrow niche. (A) Normal bone marrow (BM) niche. The normal BM niche constists of multiple cell types, including nonhematopoietic endothelial cells (EC), CXCL12-expressing adventitial cells (CAR cells), Leptin-receptor-positive cells (LepR+ cells), Nestin – green fluorescent protein high-expressing cells [Nestin-GFP(high) cells], myelinated fibers (surrounded by Schwann cells) and non-myelinated sympathetic fibers; mesenchymal stem cells (MSC); osteoblasts; and adipocytes; and hematopoietic cells that include regulatory T cells (Treg), macrophages (Mac) and osteoclasts (OC). The mechanisms used by these cells to control HSC activity include expression of short-acting hormones, cytokines and chemokines, leptin, stem cell factor (SCF); angiopoietin; thrombopoietin (TPO); C-X-C motif chemokine 12 (CXCL12); transforming growth factor-beta (TGFβ); neurotransmitters (norepinephrine and neuropeptide Y) and purine nucleotides (ATP and adenosine); extracellular proteins [osteopontin (OPN), fibronectin, and others] and direct cell-to-cell, contact-dependent interactions (e.g. CXCL12-CXCR4; integrinVCAM1) and transfer of organelles and secondary messengers. (B) The effect of a Twist1-deficient BM niche on HSC. Twist1 deficiency in BM niche causes EC expansion and angiogenesis, depletion of MSC, increased OPN production and decreased intramarrow levels of CXCL12, VCAM1 and SCF production while OPN production increases, which impairs HSC retention in the BM niche and promotes their mobilization to the periphery. (C) The effect of a Twist1-deficient BM niche on leukemic stem cells (LSC). The Twist1-deficient BM niche promotes acute myelogenous leukemia (AML) after transformation by the MLL-AF9 oncogene. The mechanism involves upregulated Notch signaling through upregulation of the production of local Jagged-2 (expressed by EC, osteoblasts and MSC) and the membrane expression of Notch receptors on LSC.

tion of Runx1 and C/EBPα. In mature myeloid cells, Twist2 negatively regulates pro-inflammatory responses by inhibiting the expression of pro-inflammatory cytokines such as interleukin-12, interferon-g, interleukin-1, tumor necrosis factor-α, interleukin-6, and macrophage inflammatory protein-1α, through the inactivation of c/EBPα and NF-kB. According to these data, Twist1 and Twist2 play opposite roles in myeloid differentiation although how the opposing functions of these two homologs in myeloid differentiation are balanced remains unclear.

Twist1 in acute myeloid leukemia Twist1 plays an important role in the process of epithelial-mesenchymal transition. The overexpression of Twist1 has been described as a poor prognostic factor in numerous epithelial-derived malignancies such as breast cancer, prostate cancer, colorectal cancer, bladder cancer, 1938

melanoma, hepatocellular carcinoma and neck carcinoma.7 Twist1 has been found to be overexpressed in mononuclear cells from the bone marrow of patients with AML and chronic myeloid leukemia,12 with a strong correlation between the expression of Twist1 and Bmi-1, an essential polycomb complex group with a fundamental role in the maintenance of leukemia stem cells. In fact, AML patients whose blasts overexpress Twist1 have a more aggressive clinical phenotype, with a good response to the cell cycle phase-specific agent cytarabine but not to the non-cell cycle phase-specific anthracycline, daunorubicin.13 It also been shown that Twist1 expression is augmented in the HSC and progenitor compartment and decreased in bone marrow stromal cells from patients with myelodysplastic syndrome.14 However, whether Twist1 in the bone marrow niche participates in AML pathogenesis is not clear. MLL-AF9 is an oncoprotein, a product of chromosome haematologica | 2018; 103(12)

Editorials

translocation t(9;11)(p22;q23), typically associated with the M4 or M5 French-American-British subtypes of human AML. An MLL-AF9-induced AML mouse model is widely used to study AML. Hanoun et al. showed that MLL-AF9induced AML cells disrupt bone marrow HSC niche function through alteration of the niche compartments and decrease of the expression of MSC-derived Cxcl12, Scf, Vcam1, and increased expression of Opn.15 These phenotypes in the MLL-AF9-induced AML bone marrow niche mirrored those in the Twist1-deficient bone marrow niche observed by Liu et al.2 They also found that a Twist1-deficient niche promoted progression of MLL-AF9-induced AML. By using RNAsequencing, the authors found that Jagged-2 is significantly increased in the stromal cells from the Twist1-deficient bone marrow niche and that Notch receptors are upregulated on leukemia cells. Pharmaceutical inhibition of Notch signaling partially inhibited the leukemia progression. These data indicate that AML progression is a closed loop, AML cells impair the normal bone marrow niche and alter the niche to support AML cell survival and progression. Targeting the leukemia cell - bone marrow niche loop could be an efficient strategy for leukemia therapy. Jagged-2/Notch signaling is a potential target.

Relevance of understanding the hematopoietic niche in human disease Bone marrow failure syndromes, including aplastic anemia and myelodysplastic syndromes result from cell- and non-cell-autonomous dysregulation. Similarly, there is increasing evidence of the role of the bone marrow niche in leukemogenesis. Non-cell-autonomous dysregulation is linked to changes in the signals from innate and adaptive immune environments, and endothelial and mesenchymal lineage cells. Big data analyses based on matrix analyses of combined large data sets from single-cell RNA sequencing, flow/mass cytometry, metabolomics and proteomics mass spectrometry, as well as sophisticated microanatomical analyses are generating multiple hypotheses on the specific dissection of the interacting signal networks that connect the HSC niche cells of the bone marrow. These hypotheses require exquisite analysis and validation in mutant animal

models and in therapeutic approaches when pharmacological tools to specific targets allow analyses of efficacy and safety in patients with hematologic diseases.

References 1. Gao X, Xu C, Asada N, Frenette PS. The hematopoietic stem cell niche: from embryo to adult. Development. 2018;145(2). pii: dev139691. 2. Liu X, Ma Y, Li R, et al. Niche TWIST1 is critical for maintaining normal hematopoiesis and impeding leukemia progression. Haematologica 2018;103(12):1969-1979. 3. Barnes RM, Firulli AB. A twist of insight - the role of Twist-family bHLH factors in development. Int J Dev Biol. 2009;53(7):909-924. 4. Dong CY, Liu XY, Wang N, et al. Twist-1, a novel regulator of hematopoietic stem cell self-renewal and myeloid lineage development. Stem Cells. 2014;32(12):3173-3182. 5. Arthur A, Cakouros D, Cooper L, et al. Twist-1 enhances bone marrow mesenchymal stromal cell support of hematopoiesis by modulating CXCL12 expression. Stem Cells. 2016;34(2):504-509. 6. Miraoui H, Severe N, Vaudin P, Pages JC, Marie PJ. Molecular silencing of Twist1 enhances osteogenic differentiation of murine mesenchymal stem cells: implication of FGFR2 signaling. J Cell Biochem. 2010;110(5):1147-1154. 7. Zhao Z, Rahman MA, Chen ZG, Shin DM. Multiple biological functions of Twist1 in various cancers. Oncotarget. 2017;8(12):2038020393. 8. Peled A, Petit I, Kollet O, et al. Dependence of human stem cell engraftment and repopulation of NOD/SCID mice on CXCR4. Science. 1999;283(5403):845-848. 9. Greenbaum A, Hsu YM, Day RB, et al. CXCL12 in early mesenchymal progenitors is required for haematopoietic stem-cell maintenance. Nature. 2013;495(7440):227-230. 10. Mendez-Ferrer S, Frenette PS. Hematopoietic stem cell trafficking: regulated adhesion and attraction to bone marrow microenvironment. Ann N Y Acad Sci. 2007;1116:392-413. 11. Sharabi AB, Aldrich M, Sosic D, et al. Twist-2 controls myeloid lineage development and function. PLoS Biol. 2008;6(12):e316. 12. Wang N, Guo D, Zhao YY, et al. TWIST-1 promotes cell growth, drug resistance and progenitor clonogenic capacities in myeloid leukemia and is a novel poor prognostic factor in acute myeloid leukemia. Oncotarget. 2015;6(25):20977-20992. 13. Chen CC, You JY, Gau JP, et al. Favorable clinical outcome and unique characteristics in association with Twist1 overexpression in de novo acute myeloid leukemia. Blood Cancer J. 2015;5:e339. 14. Li X, Marcondes AM, Gooley TA, Deeg HJ. The helix-loop-helix transcription factor TWIST is dysregulated in myelodysplastic syndromes. Blood. 2010;116(13):2304-2314. 15. Hanoun M, Zhang D, Mizoguchi T, et al. Acute myelogenous leukemia-induced sympathetic neuropathy promotes malignancy in an altered hematopoietic stem cell niche. Cell Stem Cell. 2014;15(3):365375.

Assessment of iron deficiency Chaim Hershko Professor Emeritus, Department of Hematology, Shaare Zedek Medical Center, Jerusalem, Israel E-mail: hershkoc@netvision.net.il doi:10.3324/haematol.2018.205575

ccording to a study involving 187 countries,1 the global prevalence of anemia in 2010 was 33% and it was responsible for 68 million years lived with disability . Iron deficiency was the top cause of anemia, with children below 5 years and women having the highest burden. In addition to iron deficiency, which was the most common etiology globally, other leading causes of anemia vary widely by geography, age, and sex. haematologica | 2018; 103(12)

Traditionally, the diagnosis of iron deficiency anemia (IDA) rests on simple measurements of serum iron, transferrin and ferritin in subjects with microcytic hypochromic anemia. However, iron deficiency and other conditions associated with anemia, such as the anemia of chronic disease and hemoglobinopathies often coexist, requiring further refinement of diagnostic strategies. The study reported by Kanuri et al.2 in this issue represents an effort to opti1939

Editorials Table 1. Diagnostic biomarkers of iron-restricted erythropoiesis.*

Test

Significance

Iron deficiency anemia

Anemia of chronic disease

Hemoglobin g/dL MCV MCH MCHC RDW CHr: reticulocyte Hb content HYPOr: % retics with Hb< 280 g/L ZPP/H: Zinc-protoporphyrin/heme ratio

Red blood cell indices Hb males <13 females <12 " " " Hypochromic microcytic " " " Decreased " " " increased RBC fluorescence with Zn replacing Fe Increased

Serum iron Transferrin Transferrin saturation Ferritin

Compartment of iron in transit Protein carrier of iron in transit Ratio of iron to transferrin Iron storage protein released from storage sites

Decreased Increased Decreased Decreased

Decreased Normal or decreased Normal or decreased Increased

sTfR: Soluble transferrin receptor Transferrin receptor/log ferritin Hepcidin

Protein derived from red cell precursors Ratio of sTfR to log ferritin Liver derived peptide hormone

Increased Increased Decreased

Unchanged Unchanged Increased

Hb males <13 females <12 Hypo- micro- to normo Decreased Increased Increased

*Based on reviews by Weiss and Goodnough 2005,19 Thomas et al. 2013,20 Kiss 2015,21 Camaschella 201522 and Hempel and Bollard 2016.23 MCV: mean corpuscular volume; MCH: mean corpuscular hemoglobin; MCHC: mean corpuscular hemoglobin concentration; RDW: red cell distribution width; Hb: hemoglobin; RBC: red blood cell; Zn: Zinc; Fe: iron.

mize diagnostic biomarkers of iron deficiency anemia employing erythrocyte zinc-protoporphyrin/heme ratio (ZPP/H) and serum hepcidin measurements. The study by Kanuri et al. was conducted among 4 groups (90 to 100 subjects each) of subjects selected from 2227 rural community-dwelling Indian women and preschool children. It included 90 non-anemic women, 100 non-anemic children, and 100 women and 100 children with IDA. Anemia was defined as hemoglobin less than 12 g/dL in women and less than 11 g/dL in children. All subjects in the normal groups had serum ferritin over 30 ng/mL and all subjects in the IDA group had serum ferritin less than 12 ng/mL and a soluble transferrin receptor (sTfR)/log ferritin index of >2. Subjects with low iron stores but normal hemoglobin were excluded from the study. The diagnostic performance of the biomarkers was estimated by analyzing receiver operating characteristic (ROC) curves to determine cut-off values with an optimal likelihood ratio>10 for IDA. A ZPP/H ratio cut-off >90mmol/mol heme in children and >107mmol/mol heme in women was associated with a high likelihood of IDA at diagnosis (children: likelihood ratio=20.3, sensitivity 81% specificity 96%; women: likelihood ratio=10.8 73% specificity 93% sensitivity 73%). Hepcidin cut-off values of ≤6.8ng/mL in children and ≤4.5ng/mL in women were associated with a high likelihood of IDA at diagnosis (children: likelihood ratio=14.3, sensitivity 86% specificity 94%; women: likelihood ratio=16.2, sensitivity 90%, specificity 94%). The authors conclude that erythrocyte ZPP/H ratio is a valid point-ofcare (POC) biomarker to diagnose IDA, and that the ZPP and hepcidin reference ranges and cut-off values identified in this study may guide clinicians to utilize these tests for the diagnosis of IDA in women and children. In order to appreciate the significance of the data reported by Kanuri et al., a brief overview of our current knowledge on ZPP/H and hepcidin measurements for evaluating iron deficiency is presented below. 1940

The zinc-protoporphyrin/heme ratio (ZPP/H) The use of ZPP/H in the assessment of body iron status was reviewed in a remarkable paper by Labbe and Dewangi in 2004.3 Heme biosynthesis takes place mainly in erythroid precursor cells in the bone marrow. Iron is chelated by protoporphyrin as the final reaction in the heme pathway. This reaction is catalyzed by ferrochelatase on the mitochondrial inner membrane. Iron and zinc compete for the metal binding site of ferrochelatase and when the Fe2+ substrate is insufficient, it is substituted by Zn2+, resulting in increased ZPP/H formation. Excess ZPP/H formation is a reflection of iron – zinc substrate competition for ferrochelatase in iron-deficient erythropoiesis. ZPP/H is highly responsive to iron status even in borderline deficiency. Conversely, the decrease in ZPP/H following iron supplementation in preanemic states illustrates the ability of ZPP/H to respond to marginal changes in iron status. A major advantage of ZPP/H measurement is the simplicity with which it can be performed, as it requires only a portable instrument , the direct reading of fluorescence without need for any reagents, and requires minimal professional training. The ZPP/H ratio is highly specific for iron-deficient erythropoiesis. However, it does not distinguish between absolute iron deficiency and iron-deficient erythropoiesis caused by anemia of chronic disease (ACD). Thus, a positive test result of ZPP/H should be followed by a serum ferritin determination to distinguish iron deficiency from irondeficient erythropoiesis associated with inflammation , or the toxic effect of lead exposure. Nevertheless, a ZPP/H reading within the reference range is strong evidence of adequate systemic iron supply. Indeed, as shown in a study conducted among Kenyan preschool children, when used in a screen-and-treat approach, the combination of hemoglobin concentration and whole blood ZPP/H in a single diagnostic score can be used as a rapid and convenient testing method to rule out iron deficiency in a substantial prohaematologica | 2018; 103(11)

Editorials

portion of children screened.4 Another issue of specificity is a modest increase in ZPP/H in Îą and Î˛ thalassemia trait. However, the combined use of red cell distribution width (RDW) or mean corpuscular volume (MCV) and ZPP/H allows for discrimination between IDA and thalassemia trait in the vast majority of subjects.5,6 The main advantage of erythrocyte ZPP/H measuring is the low cost, POC testing and the simplicity with which these tasks can be performed. Erythrocyte ZPP/H can be best used as a primary screening test for assessing iron status, especially in patients likely to have uncomplicated iron deficiency. In addition to its primary application, it can be useful in monitoring response to iron therapy.

Hepcidin Hepcidin, a liver-derived peptide hormone discovered in 2001, is a key regulator of systemic iron homeostasis.7 The central role of hepcidin in iron regulation has been extensively reviewed by Ganz8 and by Hentze et al.,9 and the use of serum hepcidin measurements in the diagnosis of iron disorders was reviewed in 2016 by Girelli et al.10 Hepcidin functions by inhibiting the entry to the plasma of iron acquired by intestinal absorption, the recycling of iron derived from catabolism of senescent red blood cells (RBC) in macrophages, and by mobilization of iron stored in the liver. The block of iron flow is achieved by the binding of hepcidin to the iron transporter ferroportin, followed by its internalization and degradation. Hepcidin production is increased by iron excess and by inflammation, and suppressed by both iron deficiency and increased erythropoiesis. Hepcidin production is flexible and changes within hours of introducing stimulatory or inhibitory messages such as iron administration or inflammatory stimulation. Because several opposing messages may present simultaneously, hepcidin output will depend on the relative strength of each. For example, in severe iron deficiency, hepcidin production tends to remain low, even in the presence of inflammation. Similarly, in conditions of ineffective or expanded erythropoiesis, such as in nontransfusion-dependent thalassemias, signals released by bone marrow erythroid precursors tend to override those from replete iron stores. One such erythroid signal, erythroferrone (ERFE), has been recently identified.11 ERFE is synthesized and secreted by erythroblasts in the marrow and extramedullary sites. The production of ERFE is induced by erythropoietin and is also proportional to the total number of responsive erythroblasts. ERFE acts on hepatocytes to suppress the production of hepcidin by inhibiting hepatic BMP/SMAD signaling. By suppressing hepcidin, ERFE facilitates iron delivery during stress erythropoiesis, but also contributes to iron overload in anemias with ineffective erythropoiesis.12 The measurement of hepcidin, unlike other tests used for evaluating iron status, is a direct reflection of the mechanism controlling iron homeostasis. This unique feature of hepcidin represents a major advantage in trying to elucidate the nature of disease and its optimal management. It can be used as a guide for iron therapy. For example, it allows the prediction of favorable response to oral iron treatment among children living in countries with a high prevalence of infectious diseases,13 or the design of optimal haematologica | 2018; 103(12)

oral iron dosing and timing by exploiting conditions that minimize iron-provoked hepcidin induction.14 It is also useful in the diagnosis of concomitant iron deficiency in patients with ACD in rheumatoid arthritis and inflammatory bowel disease, and in African children.15,16 It also allows for a rapid diagnosis of rare hereditary diseases, such as iron-refractory iron deficiency anemia (IRIDA) or ferroportin disease due to hepcidin resistant mutations.17,18 Although several assays have been developed, a gold standard is still lacking, and efforts toward harmonization are ongoing. Nevertheless, the unique advantages of hepcidin measurements can already be recognized , ranging from the use of hepcidin in diagnosing IRIDA to global health applications, such as guiding safe iron supplementation in developing countries with a high infectious disease burden.

Summary Table 1 lists the tests currently available for evaluating iron-restricted erythropoiesis in iron deficiency and in ACD.19-23 Important considerations in the choice of diagnostic tests should be the availability, affordability, sensitivity, specificity, and minimal time required for receiving POC results. Red cell indices described in the upper 4 rows are an excellent starting point, offering knowledge regarding the duration and severity of iron deficient erythropoiesis (IDA and ACD). Because of the low specificity of red cell indices, the next set of tests should include serum iron, transferrin and ferritin. Ideally, these tests should offer a clear distinction between IDA and ACD. However, in real life, and in particular in developing countries with populations at the highest risk of anemia, IDA and ACD often coexist and the opposing directions of lab results make diagnosis difficult. This is the point where a third set of tests should be considered: ZPP/H, sTfR and hepcidin. Two of these three are the subjects of the present study by Kanuri et al. Their results show high sensitivity for IDA, but specificity could not be determined because of the design of studies pre-selecting only subjects with IDA documented by ferritin less than 12 ng/mL and an increased sTfR/log ferritin ratio. In view of its high sensitivity and simplicity, ZPP/H is an excellent screening procedure which may deserve inclusion in the first set of POC tests of RBC indices. In particular, it is useful in excluding iron restricted erythropoiesis whether in IDA or ACD if results are within the normal range. Both sTfR24 and hepcidin measurements are able to identify IDA in the presence of ACD. They are not inexpensive, however, and both require further efforts to turn them into universally available and validated assays. Because the measurement of hepcidin is a direct reflection of the mechanism controlling iron homeostasis, its future development into a widely available diagnostic tool may offer a major advantage in our drive to understand the nature of iron deficiency diseases and their optimal management .

References 1. Kassebaum NJ, Jasrasaria R, Naghavi M, et al. A systematic analysis of global anemia burden from 1990 to 2010. Blood. 2014;123(5):615â&#x20AC;&#x201C;624. 2. Kanuri G, Chichula D, Sawhne R, et al. Optimizing diagnostic biomarkers of iron deficiency anemia in community-dwelling Indian women and preschool children. Haematologica. 2018;103(12):19911996.

1941

Editorials 3. Labbé RF, Dewanji A. Iron assessment tests: transferrin receptor vis-àvis zinc protoporphyrin. Clin Biochem. 2004;37(3):165-174. 4. Teshome EM, Prentice AM, Demir AY, Andang'o PEA, Verhoef H. Diagnostic utility of zinc protoporphyrin to detect iron deficiency in Kenyan preschool children: a community-based survey. BMC Hematol. 2017;17:11. 5. Tillyer ML, Tillyer CR. Zinc protoporphyrin assays in patients with alpha and beta thalassaemia trait. J Clin Pathol. 1994;47(3):205-208. 6 Harthoorn-Lasthuizen EJ, Lindemans J, Langenhuijsen MM. Combined use of erythrocyte zinc protoporphyrin and mean corpuscular volume in differentiation of thalassemia from iron deficiency anemia. Eur J Haematol. 1998;60(4):245-251. 7. Park CH, Valore EV, Waring AJ, Ganz T. Hepcidin, a urinary antimicrobial peptide synthesized in the liver. J Biol Chem. 2001;276(11):78067810. 8. Ganz T. Hepcidin and iron regulation, 10 years later. Blood. 2011;117(17):4425-4433. 9. Hentze MW, Muckenthaler MU, Galy B, Camaschella C. Two to tango: regulation of mammalian iron metabolism. Cell. 2010;142(1):24-38. 10. Girelli D, Nemeth E, Swinkels DW. Hepcidin in the diagnosis of iron disorders. Blood. 2016;127(23):2809-2813. 11. Kautz L, Jung G, Valore EV, Rivella S, Nemeth E, Ganz T. Identification of erythroferrone as an erythroid regulator of iron metabolism. Nat Genet. 2014;46(7):678-684. 12. Ganz T. Erythropoietic regulators of iron metabolism. Free Radic Biol Med. 2018 Jul 5. [Epub ahead of print] 13. Prentice AM, Doherty CP, Abrams SA, et al. Hepcidin is the major predictor of erythrocyte iron incorporation in anemic African children. Blood. 2012;119(8):1922-1928. 14. Moretti D, Goede JS, Zeder C, et al. Oral iron supplements increase

15. 16. 17.

18. 19. 20. 21. 22. 23. 24.

hepcidin and decrease iron absorption from daily or twice-daily doses in iron-depleted young women. Blood. 2015;126(17):19811989. Pasricha SR, Atkinson SH, Armitage AE, et al. Expression of the iron hormone hepcidin distinguishes different types of anemia in African children. Sci Transl Med. 2014;6(235):235re3. Bergamaschi G, Di Sabatino A, Albertini R, et al. Serum hepcidin in inflammatory bowel diseases: biological and clinical significance. Inflamm Bowel Dis. 2013;19(10):2166-2172. De Falco L, Silvestri L, Kannengiesser C, et al. Functional and clinical impact of novel TMPRSS6 variants in iron-refractory iron-deficiency anemia patients and genotype-phenotype studies. Hum Mutat. 2014;35(11):1321-1329. Sham RL, Phatak PD, Nemeth E, Ganz T. Hereditary hemochromatosis due to resistance to hepcidin: high hepcidin concentrations in a family with C326S ferroportin mutation. Blood. 2009;114(2):493-494. Weiss G, Goodnough LT. Anemia of chronic disease. N Engl J Med. 2005;352(10):1011-1023. Thomas DW, Hinchliffe RF, Briggs C, et al. Guideline for the laboratory diagnosis of functional iron deficiency. Br J Haematol. 2013;161(5):639648. Kiss JE. Laboratory and genetic assessment of iron deficiency in blood donors. Clin Lab Med. 2015;35(1):73-91. Camaschella C. Iron-deficiency anemia. N Engl J Med. 2015;372 (19):1832-1843. Hempel EV, Bollard ER. The evidence-based evaluation of iron deficiency anemia. Med Clin North Am. 2016;100(5):1065-1075. Skikne BS, Punnonen K, Caldron PH, et al Improved differential diagnosis of anemia of chronic disease and iron deficiency anemia: a prospective multicenter evaluation of soluble transferrin receptor and the sTfR/log ferritin index. Am J Hematol. 2011;86(11):923-927.

Is DNA a better assay for residual disease in chronic myeloid leukemia? Jerald Radich Fred Hutchinson Cancer Research Center, Seattle, Washington, USA E-mail: jradich@fhcrc.org doi:10.3324/haematol.2018.205583

hronic myeloid leukemia (CML) is not a public health menace. Despite its rarity, it has, and continues to be, the guiding path for the concept of genetically informed medicine (here you can choose your own favorite alternative catch phrase: bench to bedside medicine, personalized medicine, precision medicine, etc.). CML was the first disease where a specific chromosomal abnormality, the Philadelphia chromosome, was identified, and the first disease where the genetic underpinnings of this chromosome abnormality were discovered (the juxtaposition of portions of the BCR gene from chromosome 22 to the tyrosine kinase domains from chromosome 9).1 This unique BCR-ABL fusion gene drives the pathophysiology of the disease, and thus has led to the remarkable discovery of the tyrosine kinase inhibitors (TKIs), which have fundamentally changed the natural history of the disease. Only decades ago, the lifespan of a chronic phase CML patient was less than seven years while now these patients enjoy a survival roughly that of the normal population.2,3 Chronic myeloid leukemia has also been the model of disease monitoring using specific molecular markers. In this case, the BCR-ABL chimeric RNA is used to assess disease burden, and the clinical significance of BCR-ABL levels are so compelling as to drive treatment milestones based on BCR-ABL levels that are codified in European 1942

and US CML guidelines.4-6 Here, too, CML has laid the groundwork for other diseases to use so-called minimal (more recently, “measurable”) residual disease (MRD) to drive treatment decisions and measure clinical trial results. In CML, BCR-ABL is typically measured by testing peripheral blood RNA. RNA is used since the potential breakpoints between BCR and ABL cover many kilobases of DNA sequence, making an easy PCR procedure impossible. Rather, the mRNA species is predictable with only two major splicing variations, making quantitative RTPCR fairly straightforward. After considerable effort (mostly from the Adelaide group), RNA monitoring has been standardized in an International Scale, making the results comparable across more and more labs worldwide.6 The test is very sensitive, with levels of disease burden usually quantifiable to levels of four to five magnitudes from the standardized IS baseline (where a 4-log reduction of BCR-ABL on the IS would equal 0.01%IS, termed MR4) However, the RNA assay for BCR-ABL is not perfect. RNA is less stable than DNA, and thus is more susceptible to transit times, temperatures, etc.7,8 This problem is ameliorated by the use of two control housekeeping genes, but these are subject to the same influences that affect the target gene, and it is perhaps a bit of a leap to haematologica | 2018; 103(12)

Editorials

Figure 1. BCR-ABL rearrangement structure. The BCR gene contains three primary breakpoint clusters, the “minor” (m-bcr), “major” (M-bcr), and “micro” (µ-bcr). Gene rearrangement at the M-bcr site results in either of two p210 fusion chimeric mRNAs, composed of BCR exons 1-13 or exons 1-14 (orange and green) fused to ABL exons 2-11 (red). RT-PCR of BCR-ABL, amplifying only the exons, yields an amplicon product of ~200-300 bp. However, the genomic breakpoints in BCR and ABL are dispersed over intervals of 3.0 kb and 150 kb, respectively, making straightforward amplification of the DNA breakpoint impossible. The relative scale of the RNA and theoretical DNA product using the same set of BCR and ABL primers is shown at the bottom of the figure. Thus, for successful DNA amplification, multiple BCR and ABL primers must be used until a successful combination successfully amplifies the breakpoint. After this, sequencing is performed to identify sequences for patient-specific primers and probes.

believe that a series of environmental insults would affect both target and housekeeping genes in the same way. In addition, there are certainly differences in BCR-ABL levels from patient to patient, if not from cell to cell in the same patient. A DNA-based assay would be a more accurate measure of defining the approximate number of actual CML cells in any given patient. (Note that in rare cases a patient can harbor more than one copy of BCR-ABL, but this is typically only in cases of advanced phase disease.) Amplification of the BCR-ABL DNA is difficult, as the potential span of breakpoints within the BCR and ABL genes is vast (Figure 1), as opposed to the limited base pair distance once the chimeric BCR-ABL mRNA is assembled.9,10 To perform the DNA-based assay for BCRABL, an initial PCR uses multiple primer sets to first identify the possible DNA breakpoint. Once the sequence of the breakpoint is identified, patient-specific primers are constructed to make a very sensitive assay. Since the patient-specific PCR will have different kinetics from patient to patient, the drop in the disease burden must be measured against the patient's diagnostic disease burden value. This is a very similar concept to following MRD in patients with acute lymphoblastic leukemia, where patient-specific IgVDJ or TCR rearrangements must be amplified with consensus primers, then patient-specific primers and probes developed for each unique assay. [This complexity led to the development of sensitive flow cytometry and next generation sequencing (NGS) methods, the latter discussed below.] haematologica | 2018; 103(12)

The potential value of a sensitive DNA test is at least 2fold. First, one could study the differences in RNA versus DNA load, correlating this with disease response. Secondly, a DNA assay would allow the detection of CML in cases in which the RNA assay of BCR-ABL is undetectable. This would be especially interesting in the case of discontinuation in CML, where some patients with a prolonged deep molecular response stop TKI treatment and do not relapse.11,12 Are those cases that quickly relapse after discontinuation simply those in whom the disease is at a higher burden, though undetectable by RNA assays? Could patients who have no MRD by RNA or DNA assays be the lucky patients for whom discontinuation will be successful?13 In this issue of the Journal, the Adelaide group expands on their previous studies of DNA-based BCR-ABL detection and show the potential of this assay to probe basic disease and clinical issues.14 They studied 59 patients with 516 samples on which RNA and DNA assessments of disease burden were performed. Several important findings were found. First, they found that, early in disease treatment, the number of copies of RNA was generally higher than the DNA (roughly 2-fold) whereas after around six months, RNA and DNA levels were fairly similar. The biological reason for this is unclear. However, the kinetic decay of BCR-ABL with TKI therapy shows a multi-order decay, with an initial decline, followed by a slower decay. These two findings combined suggest that, at diagnosis, a pop1943

Editorials

ulation of committed progenitors, making a fair amount of BCR-ABL RNA (and protein), die quickly, followed by a population of cells less active with regards to BCR-ABL production. Second, the BCR-ABL decline appears to be more significant in the case of patients with the b14a2 transcript, rather than the shorter b13a2 transcript. This has been seen before in other studies.15,16 The reason is unclear, but the speculation is that the exon responsible for the longer version is immunogenetic. CML is well-known to be unusually susceptible to immune-mediated attack (note the effects of interferon, allogeneic transplant, and donor lymphocyte infusions), and the current speculation is that this is another manifestation of this effect.17 If so, we might expect the b14a2 cases to also enjoy more success with discontinuation. Third, in multiple cases, DNA detected residual disease whereas RNA did not. Thus, in 86 cases where BCR-ABL was undetected by RNA, the DNA assay found disease in 42 (49%). Moreover, the median level of detectable disease after 12 months of therapy was higher by DNA than by RNA. Where will this lead? First, one could imagine the study of colonial heterogeneity. It is becoming clear that in many diseases (e.g. acute myeloid leukemia) there are multiple related clones at diagnosis, and treatment may cause a Darwinian selection of resistance. Since resistance is less common in CML in the TKI era, and since our tools of defining disease are insensitive to measuring the subtle difference (the Ph and BCR-ABL RNA), DNA-based assays that identify different unique BCR and ABL breakpoints may be able to eventually detect multiple clones. Moreover, DNA-based assays may help distinguish those patients who can and those who should not undergo discontinuation. It may be especially interesting to study cases who have BCR-ABL by DNA but do not subsequently relapse after TKI is discontinued. Is this evidence of immunological control of residual CML? Is this assay detecting BCR-ABL DNA in lymphocytes that may not be involved in the disease process or relapse? In order to do some of these things, better assays will be needed. The advent of single cell technologies that can perform either genotyping and transcription analysis also need to be developed, and the search for techniques that might allow both to be performed is ongoing.18 This could allow for studies of biology, heterogeneity, and response. Assay methods to quickly genotype complex DNA rearrangement structures are now FDA approved to study immunoglobulin gene rearrangements for MRD in lymphoid malignancies, and the same approach could be used to streamline the DNA approach in CML. New sequencing methods can detect single base pair differences at a one in a million resolution, approximately 3-4 orders of magnitude better than NGS.19 Believers in the theory of the “RNA world” suggest that RNA was the key to life’s first steps from primordial ooze to cellular creatures (alas, some have made it farther than others).20 DNA followed as a more durable

1944

way to collect and store information. Perhaps the same evolutionary order is on tap for those researchers interested in the clinical importance of molecular diagnostics in CML.

References 1. Rowley JD. A new consistent chromosome abnormality in chronic myelogenous leukemia identified by quinacrine fluorescence and Giemsa staining. Nature. 1973;243(5405):209-213. 2. 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. 3. Huang X, Cortes J, Kantarjian H. Estimations of the increasing prevalence and plateau prevalence of chronic myeloid leukemia in the era of tyrosine kinase inhibitor therapy. Cancer. 2012;118(12):3123-3127. 4. Hughes TP, Kaeda J, Branford S, et al. Frequency of major molecular responses to imatinib or interferon alfa plus cytarabine in newly diagnosed chronic myeloid leukemia. N Engl J Med. 2003;349(15):14231432. 5. Hanfstein B, Muller MC, Hehlmann R, et al. Early molecular and cytogenetic response is predictive for long-term progression-free and overall survival in chronic myeloid leukemia (CML). Leukemia. 2012;26(9):2096-2102. 6. Hughes T, Deininger M, Hochhaus A, et al. Monitoring CML patients responding to treatment with tyrosine kinase inhibitors: review and recommendations for harmonizing current methodology for detecting BCR-ABL transcripts and kinase domain mutations and for expressing results. Blood. 2006;108(1):28-37. 7. Radich JP, Mao M, Stepaniants S, et al. Individual-specific variation of gene expression in peripheral blood leukocytes. Genomics. 2004;83(6):980-988. 8. Dvinge H, Ries R, Ilagan J, Stirewalt DL, Meshinchi S, Bradley RK. Sample processing obscures cancer-specific alterations in leukemic transcriptomes. PNAS. 2014;111(47):16802-16807. 9. Ross DM, O'Hely M, Bartley PA, et al. Distribution of genomic breakpoints in chronic myeloid leukemia: analysis of 308 patients. Leukemia. 2013;27(10):2105-2107. 10. Linhartova J, Hovorkova L, Soverini S, et al. Characterization of 46 patient-specific BCR-ABL1 fusions and detection of SNPs upstream and downstream the breakpoints in chronic myeloid leukemia using next generation sequencing. Mol Cancer. 2015;14:89. 11. Mahon FX, Réa D, Guilhot J, et al. Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial. Lancet Oncol. 2010;11(11):1029-1035. 12. Hughes TP, Ross DM. Moving treatment-free remission into mainstream clinical practice in CML. Blood. 2016;128(1):17-23. 13. Ross DM, Branford S, Seymour JF, et al. Patients with chronic myeloid leukemia who maintain a complete molecular response after stopping imatinib treatment have evidence of persistent leukemia by DNA PCR. Leukemia. 2010;24(10):1719-1724. 14. Pagani IS, Dang P, Kommers IO, et al. BCR-ABL1 genomic DNA PCR response kinetics during first-line imatinib treatment of chronic myeloid leukemia. Haematologica. 2018;103(12):2026-2032. 15. Jain P, Kantarjian H, Patel KP, et al. Impact of BCR-ABL transcript type on outcome in patients with chronic-phase CML treated with tyrosine kinase inhibitors. Blood. 2016;127(10):1269-1275. 16. Lucas CM, Harris RJ, Giannoudis A, et al. Chronic myeloid leukemia patients with the e13a2 BCR-ABL fusion transcript have inferior responses to imatinib compared to patients with the e14a2 transcript. Haematologica. 2009;94(10):1362-1367. 17. Clark RE, Dodi IA, Hill SC, et al. Direct evidence that leukemic cells present HLA-associated immunogenic peptides derived from the BCRABL b3a2 fusion protein. Blood. 2001;98(10):2887-2893. 18. Zheng GXY, Terry JM, Belgrader P, et al. Massively parallel digital transcriptional profiling of single cells. Nat Commun. 2017;8:14049. 19. Wu D, Emerson RO, Sherwood A, et al. Detection of minimal residual disease in B lymphoblastic leukemia by high-throughput sequencing of IGH. Clin Cancer Res. 2014;20(17):4540-4548. 20. Neveu M, Kim HJ, Benner SA. "The "strong" RNA world hypothesis: fifty years old". Astrobiology. 2013;13(4):391-403.

haematologica | 2018; 103(12)

REVIEW ARTICLE

Osteogenic niche in the regulation of normal hematopoiesis and leukemogenesis

Ferrata Storti Foundation

Phuong M. Le,1 Michael Andreeff2 and Venkata Lokesh Battula2,3

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA; 2Section of Molecular Hematology and Therapy, Leukemia Department, The University of Texas MD Anderson Cancer Center, Houston, TX and and 3Department of Breast Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

ABSTRACT

he bone marrow microenvironment, also known as the bone marrow niche, is a complex network of cell types and acellular factors that supports normal hematopoiesis. For many years, leukemia was believed to be caused by a series of genetic hits to hematopoietic stem and progenitor cells, which transform them to preleukemic, and eventually to leukemic, cells. Recent discoveries suggest that genetic alterations in bone marrow niche cells, particularly in osteogenic cells, may also cause myeloid leukemia in mouse models. The osteogenic niche, which consists of osteoprogenitors, preosteoblasts, mature osteoblasts, osteocytes and osteoclasts, has been shown to play a critical role in the maintenance and expansion of hematopoietic stem and progenitor cells as well as in their oncogenic transformation into leukemia stem/initiating cells. We have recently shown that acute myeloid leukemia cells induce osteogenic differentiation in mesenchymal stromal cells to gain a growth advantage. In this review, we discuss the role of the osteogenic niche in the maintenance of hematopoietic stem and progenitor cells, as well as in their transformation into leukemia cells. We also discuss the signaling pathways that regulate osteogenic nichehematopoietic stem and progenitor cells or osteogenic niche-leukemic stem/initiating cell interactions in the bone marrow, together with novel approaches for therapeutically targeting these interactions.

Introduction Hematopoietic stem cells (HSCs) home to specific microenvironments in the bone marrow (BM) and receive signals that drive their fate under both normal and pathological conditions. So far, two predominant niches that differentially regulate HSCs through their non-hematopoietic compartments and levels of hypoxia have been identified.1,2 The endosteal niche near the inner bone surface is populated by osteoblastic lineage cells, including osteoprogenitor cells, pre-osteoblasts, mature osteoblasts, and osteocytes, as well as mesenchymal stromal cells (MSCs) and osteoclasts, whereas the non-endosteal niche consists mainly of sinusoidal endothelial cells, pericytes, and non-myelinating Schwann cells. Both niches are highly vascularized yet associated with distinct subtypes of blood vessels that support either the bone-forming or sinusoidal domain.3 Recent work from the Adams group also revealed a strong association between the osteogenic niche and a third vessel type that made up the transition zone in the developing bone. This subset seems to function upstream of both endosteal and sinusoidal endothelium, though more functionally related to the former, and connect the two vasculatures during the early stages of specialization.4 Stromal cells in both niches share overlapping signatures; however, it has been suggested that endosteal MSCs support HSC quiescence whereas non-endosteal MSCs promote HSC proliferation.5 Acute myeloid leukemia (AML) is one of the most aggressive hematologic malignancies, characterized by increased numbers of myeloid precursors in the BM that fail to differentiate into more mature myeloid cells. Recent studies have haematologica | 2018; 103(12)

Haematologica 2018 Volume 103(12):1945-1955

Correspondence: vbattula@mdanderson.org

Received: May 2, 2018. Accepted: September 10, 2018. Pre-published: October 18, 2018. doi:10.3324/haematol.2018.197004 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/1945 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1945

P.M. Le et al.

highlighted complex tumor-host interactions within the BM during AML progression. Malignant cells compete with their normal counterparts for niche resources and occupancy, and disrupt normal hematopoiesis by inflicting a differentiation block, which often manifests itself as BM failure and pancytopenia.6,7 In these conditions, leukemic cells seem to lose sensitivity to antiproliferative cues from the niche.8 Under the expansion of leukemia, MSCs have shown signs of “reprogramming”.9-11 In particular, the role of the osteoblast-rich region of the BM has been implicated in both AML chemoresistance and relapse.12,13 Unraveling the mechanisms underlying osteogenic niche-mediated support to AML cells is key to identifying molecular targets in order to develop effective drug therapies. In this review, we focus on advances in our understanding of the osteogenic niche in the leukemic BM microenvironment and discuss the key components of this niche as therapeutic candidates in AML.

Osteolineage cells regulate normal hematopoiesis Non-random distribution of HSCs in the BM highlights the role of osteolineage cells in HSC maintenance. The physical association of HSCs with the endosteum correlates strongly with the colony formation and proliferative capacity of HSCs, and is primarily evident after BM transplantation.14,15 Anatomical evidence has provided the basis on which the functional relationships between osteolineage cells and HSCs have continued to be unraveled. Osteoblasts secrete cytokines and growth factors including granulocyte-colony stimulating factor (G-CSF),16 hepatocyte growth factor,17 and osteopontin (OPN),18 which have been shown to maintain the pool size of the CD34+ progenitor population in the BM. Osteoblasts mediate HSC migration in and out of the BM, primarily through the CXCL12/CXCR419 and VCAM-1/VLA-420 axes, and under the influence of the sympathetic nervous system.21 In a knockout mouse model lacking bone morphogenetic protein (BMP) receptor I, Zhang et al.22 reported that an increase in HSC number was associated exclusively with a cell population that lined the long bone and had an osteoblastic phenotype. Similarly, Calvi et al.23 demonstrated that increasing osteoprogenitor or preosteoblast activation by augmenting parathyroid hormone (PTH) signaling enriched Lin- Sca-1+ c-Kit+, or HSClike, cells in vivo. Interestingly, this HSC expansion occurred without substantially affecting the overall number of hematopoietic cells. These observations suggest that PTH-induced signaling in osteoprogenitor cells or pre-osteoblasts might play a selective role in maintaining HSC self-renewal but not in the proliferation of their committed progenitors. How osteoblasts regulate HSC quiescence has been rigorously investigated. Loss of ligand-receptor interactions, such as angiopoietin-1 receptor tyrosine kinase 2 (Ang-1/Tie2)24 and thrombopoietin-MPL (TPO/MPL),25 deregulates not only cell-cycle checkpoints but also coping mechanisms against extrinsic stressors, resulting in a reduction in slow-cycling hematopoietic cells. Stem-cell exhaustion and reduced self-renewal capacity after inhibition of Wingless (Wnt) signaling in osteoblasts further suggest that the mechanism underlying osteoblast-mediated regulation of HSCs does not follow a single axis.26 Surprisingly, osteoblast ablation, although associated with poorer HSC engraftment in vivo, does not lead to a massive loss of quiescent HSCs.27 It has also been shown 1946

that osteoblast deficiency in chronic inflammatory conditions, such as rheumatoid arthritis, does not affect the frequency of Lin- Sca-1+ c-Kit+ cells or their long-term repopulating potential.28 Mice with conditional deletion of CXCL1229 or stem cell factor (SCF)30 in osteoblasts do not exhibit HSC defects. It is possible that osteoblastic regulation of HSCs overlaps with other regulatory pathways and hence is easily compensated. Different osteolineage members may also share common signals while differing in the degree of impact.31 Together, these data suggest that osteolineage cells or more primitive cells such as MSCs orchestrate a diverse, though possibly non-essential, network of signals to maintain the stemness of HSCs and prompt hematopoietic activities, such as mobilization and expansion, in response to physiological needs.

Altered osteogenic niche leads to myeloid leukemia in BM It has been firmly demonstrated that mutations affecting the ability of HSCs to differentiate into mature hematopoietic cells transform HSCs into pre-leukemic cells, and ultimately to leukemic cells when additional mutations are acquired (Figure 1).32-34 However, very little is known about the influence of other cellular components in the BM microenvironment on leukemic transformation of hematopoietic cells. The Scadden group was the first to show that genetic alterations in osteolineage cells could lead to myelodysplastic syndromes (MDS) and leukemia. Deletion of Dicer1, a critical RNA processor and microRNA synthesizer, in Osterix (Osx)-expressing osteoprogenitor cells in a conditional knockout mouse model caused MDS and, on occasions, secondary AML.35 These mice first developed severe cytopenia and myelodysplasia, which transformed into monoblastic AML in 4 out of 200 cases, presenting as invasive myeloid sarcomas, anemia, and monocyte-like blast expansion in the peripheral blood, spleen, and BM. Of interest, Dicer1 was intact in the myeloblastic tumors, suggesting that dysfunctional osteoblast precusors could mediate clonal evolution in neoplastic formation. Similarly, constitutive activation of β-catenin in mouse osteoblasts resulted in a broad spectrum of dysfunctional hematopoiesis, including monocytosis, lymphocytopenia, and somatic mutations that resembled those of human AML in myeloid progenitors. Kode et al.36 noted that both wild-type mice engrafted with long-term (LT) HSCs from β-catenin-mutant mice and β-catenin-mutant mice engrafted with healthy BM cells developed AML and died shortly after transplantation. These observations suggest that an altered osteogenic niche could induce permanent damage to LT-HSCs and transform them to preleukemic and/or leukemic cells. Kousteni et al. attributed this niche-induced carcinogenesis to the oncogenic role of FoxO members involved in bone formation, which, surprisingly, are known tumor suppressors.37,38 This discovery sparks a debate about whether osteoblasts differentially regulate normal and malignant hematopoiesis. Recently, Dong et al.39 also reported that mice with a mutant allele of protein tyrosine phosphatase SHP2 (Ptpn11) in osteoprogenitors or Nestin+ MSCs could develop juvenile myelomonocytic leukemia-like myeloproliferative neoplasms (MPN). With concomitant mutations in HSCs, mice with mutated MSCs were twice as likely to progress from MPN to acute leukemia as were mice with altered endothelial cells. This study underhaematologica | 2018; 103(12)

Role of osteogenic niche in AML progression

scores cell-type-specific leukemogenic effects of various niche components. While these findings in mice offer direct evidence for osteoblast-induced leukemogenesis, emerging reports of donor cell leukemia in humans (1-5% of all post-transplant leukemia relapses), also suggest the role of an oncogenic microenvironment driving secondary malignancy.40 Collectively, it has been increasingly recognized that genetic aberrations in the endosteal compartment could be a key event in AML initiation and progression (Figure 1).

AML induces osteogenic and osteolytic activity Numerous AML studies have emphasized the toxicity of leukemic expansion to BM niches. AML cells have been shown to alter BM niches by competing with HSCs

for niche support, thereby affecting normal hematopoiesis.7,41,42 Whether the genomic landscape of non-hematopoietic components of BM niches changes has remained largely unexplored, and whether these alterations may drive AML initiation, progression, and resistance to chemotherapy is questionable. Due to inconsistencies in methodology, cytogenetic analyses from different labs have led to a debate about the existence of chromosomal aberrations in leukemia patient BM-derived MSCs.10,43-45 To explore global changes induced by AML in stromal cells, our group performed a large-scale comparison of proteomic, microRNA, and gene expression profiles between AML patient-derived (AML-MSCs) and healthy donor-derived BM MSCs. We found upregulation of multiple pro-proliferative and anti-

Figure 1. Osteogenic niche in hematopoietic stem cell (HSC) maintenance and leukemogenesis. Interactions between HSCs and osteogenic niche cells could happen in two ways. First, HSCs stay quiescent and self-renew when they are in osteogenic niche. When they acquire mutations under physiological stress, HSCs become pre-leukemic and eventually transform into leukemia blast cells. Alternatively, genetic abnormalities in osteogenic cells in the bone marrow could induce myeloid leukemia in non-mutated or in pre-leukemic HSCs. Second, leukemic cells could induce osteogenic differentiation in mesenchymal stromal cells (MSCs), which normally go through a series of differentiation steps to become fully mature osteoblasts or osteocytes. This feedback loop, involving bone remodeling, probably fuels leukemia progression. However, the extent to which acute myeloid leukemia (AML) cells induce osteogenic differentiation is not clear. BMP: bone morphogenetic protein; CHIP: clonal hematopoiesis of indeterminate potential; CTGF: connective tissue growth factor; HSC: hematopoietic stem cell; LC: AML cell; OPN: osteopontin.

haematologica | 2018; 103(12)

1947

P.M. Le et al.

apoptotic pathways and downregulation of RNA regulators previously implicated in survival and differentiation of leukemic cells.46 Of particular interest, marked underexpression was observed for IGFBP5, an insulin-like growth factor binding protein that primarily inhibits osteoblast differentiation of MSCs.47 Moreover, TP53 was increased in AML-derived MSCs, resulting in senescence. We also found that the leukemia genotype, in particular the presence of FLT3-ITD mutations and lack of p53, induce both shared and leukemia genome-specific alterations in MSCs.48 These reports suggest that AML cells alter stromal development, and potentially their functionality. Previously, Hanoun et al. had reported that AML primed MSCs to commit to osteoblastic lineage.49 The endosteal surface of mice transplanted with MLL-AFL9 leukemic cells was packed with Osx-expressing osteoprogenitor cells yet lacking Osteocalcin-positive (Osc+) mature osteoblasts. It is important to note that despite this osteogenic potential, these mice showed deficient bone mineralization and a lack of terminally differentiated osteoblasts compared with healthy controls. These observations were confirmed independently both in vitro and in vivo by our group.50 AML-MSCs displayed significantly higher alkaline phosphatase (ALP) expression and activity than did healthy donor-derived MSCs. In addition, when cultured in osteogenic differentiation medium, AML-MSCs differentiated to mature osteoblasts (alizarin red-positive) within two weeks compared with the three weeks needed for normal MSCs. Remarkably, gene expression analysis of normal MSCs co-cultured with different leukemic cell lines for five days revealed 2to 10-fold upregulation of osteogenic markers, such as Runt-related transcriptional factor (Runx2), Osx, Opn, and tissue non-specific ALP (Tnap), suggesting that this osteoprogenitor-priming pattern in MSCs resulted from AML exposure.50 To validate AML-induced osteoblast differentiation in vivo, we created a human BM implant mouse model and assessed osteogenic potential of BM MSCs after four weeks of leukemia engraftment. Human MSCs obtained from these transplanted mice showed a 5- to 7fold increase in Osx and Runx2 expression compared with control mice.50 These experimental data were consistent with OSX and RUNX2 upregulation in BM biopsies of AML patients. We also found that AML-MSCs became less multipotent since they differentiated poorly into adipocytes and chondrocytes, two mesodermal lineages that usually arise from MSCs; the Bhatia group confirmed this adipocyte suppression in the setting of AML by immunostaining within human BM and global transcriptome analysis of AML-MSCs.51 Note that in the same study, gene sets poised towards osteoblast, but not adipocyte lineage were enriched in AML-MSCs, underscoring the need to understand the role of distinct mesenchymal fractions in AML progression. We asked whether this osteolineage-specific priming provided any advantage for leukemic growth. Indeed, AML cells up-regulated connective tissue growth factor (CTGF) in MSCs and activated BMP signaling via Smad1/5 phosphorylation, both of which have been associated with persistence of tumors and poor prognosis in patients with acute leukemia.52-54 Besides, AML-induced TNAP overexpression in MSCs was implicated in osteoblast-mediated protection of leukemia blasts against 1948

apoptosis.55 By unraveling a feedback loop between stroma functionality and AML expansion, our study has highlighted the dynamics of the endosteal niche in AML pathogenesis (Figure 2). The reduced bone mineralization seen by Hanoun et al.49 could have resulted from altered osteolytic activity.56 A short-lived increase in osteoclasts and upregulation of CCL3, a pro-inflammatory cytokine with pro-osteoclastic action previously established in multiple myeloma,57 was found in a murine model of blast-crisis chronic myeloid leukemia (CML) phenotype.56 These acute leukemia-like mice showed a significant reduction in Osc+ osteoblasts and thinning of bone structures that could not be reversed completely by inhibition of osteoclasts. Bone deposition and resorption are tightly coupled processes that maintain bone homeostasis; however, this evidence suggests that the leukemic condition distorts this balance. Of interest, excessive CCL3 production does not typically lead to osteolytic lesions or bone loss,58 as seen in Ptpn11-mutated leukemic mice,39 although overexpression of this protein is common in the BM of AML patients.56 It is possible that monocyte differentiation into osteoclasts is defective in AML, yet the effects are masked by strong CCL3-driven recruitment of monocytes into the osteogenic niche.59 The extent to which osteoblasts and osteoclasts work in tandem to reconstruct an inhospitable microenvironment under AML burden needs further investigation.

Osteoprogenitors or mature osteoblasts: true 'partners-in-crime' in AML progression? The osteogenic niche comprises a variety of cell types which differ in their maturation status, ranging from very immature multipotent MSCs to mature osteoblasts and osteocytes (Figure 1). However, the differentiation status of osteogenic cells supporting normal hematopoiesis or leukemogenesis is an emerging question that remains to be resolved. Several studies have shown that, compared with less mature osteoblasts, terminally differentiated osteoblasts regulate HSC lineage commitment, such as B lymphopoiesis and erythropoiesis, while having less effect on HSC proliferation.60-62 Whether this functional stratification applies to the context of malignancy is poorly understood. Accumulating evidence has demonstrated that defective osteolineage cells are potent initiators of leukemia in the BM. These findings led to the question: which osteolineage cells, osteoprogenitors or mature osteoblasts, play a major role in promoting leukemogenesis? As previously discussed, Raaijmakers et al.35 were the first to show that hematopoiesis could go awry as a result of a genetic alteration in osteoprogenitors. The Scadden group emphasized the differential leukemogenic capacity of immature and mature osteoblasts by comparing AML phenotype in Dicer1fl/fl mice with Osx- versus Osc-driven Cre recombinase. Mice with a Dicer1 defect in mature osteoblasts did not exhibit any hematologic problem besides bone-related deformities. Similarly, Dong et al.39 confirmed the distinct role of stage-specific osteoblasts in leukemic development by generating mice with Ptpn11 mutations at various stages of MSCs: mesenchymal progenitor/stem cells, differentiated MSCs, Osx+ osteoprogenitors, and Osc+ mature osteoblasts. Of interest, the leukemogenic effect of this abnormality was observed in mice bearing the mutated form of either primitive MSCs haematologica | 2018; 103(12)

Role of osteogenic niche in AML progression

or bone progenitor cells, but not more differentiated osteoblasts. These data are consistent with our observations that AML-MSCs show characteristics of osteoprogenitors but not of mature osteoblasts. AML-MSCs express early-stage osteoblast markers, including osterix, RUNX2, and Col1a1, but not mature osteoblast markers such as osteocalcin.50 In addition, functional assays revealed that AML-MSCs stained positive for ALP enzyme activity but were negative for alizarin red S staining.50 These observations suggest that AML-MSCs can differentiate into committed osteoprogenitors, but not mature osteoblasts. These data were also validated by coculture of AML cell lines with normal BM-MSCs in vitro and by different AML mouse models.50,63 These findings

also do not contradict the observations of Frisch et al.,56 Geyh et al.64 and Krevvata et al.65 since the osteoblasts inhibited in these studies were marked by osteocalcin. Using intravital microscopy, Duarte et al. also showed a significant depletion of Col2.3 promoter-expressing mature osteoblasts in areas with a high level of AML filtration.66 Collectively, stalling the maturation of osteoblast precursors appears to be a key step in AML initiation and progression (Figure 2). This differentiation blockade could be mediated by different AML-derived factors. Kumar et al.63 reported upregulation of DKK1, a negative regulator of osteogenesis, when co-culturing AML-derived exosomes with BM MSCs. Of particular interest, a short-term dose of DKK1

Figure 2. Schematic representation of normal versus acute myeloid leukemia (AML)-bone marrow (BM) microenvironment. Normal BM consists of osteoprogenitor cells, pre-osteoblasts, mature osteoblasts, and osteocytes, mesenchymal stromal cells (MSCs) and osteoclasts at endosteal niche and endothelial cells, pericytes, and non-myelinating Schwann cell at non-endosteal niche. In addition to these cell types, adipocytes are present throughout the BM cavity. Hematopoietic stem cells (HSC) are present in both niche areas and gain support from stromal cells to stay quiescent and self-renew, whereas in AML BM, leukemic blasts displace HSCs from the protective niche area and occupy this sanctuary, thereby affecting normal hematopoiesis. In addition, AML cells create or expand the existing niche by inducing osteogenic but inhibiting adipogenic differentiation in MSCs. However, there are no reports suggesting higher bone volume in AML patients. Therefore, it is possible that induction of osteogenic differentiation is halted at the osteo-progenitor or pre-osteoblastic stage.

haematologica | 2018; 103(12)

1949

P.M. Le et al.

inhibitor promoted terminal differentiation of these osteolineage-primed MSCs in vitro and improved survival of mice engrafted with AML. This suggests a tight coupling of AML development with impaired maturation of osteoprogenitors. The fact that disruption of miRNA processing in immature, but not mature, osteoblasts could trigger AML development35 implies that deregulation of the maturation process at the post-transcriptional level might play a role in its failure. It would be interesting to further investigate the function of the miR-29 family, whose members are commonly down-regulated in AML blasts67 while engaging Wnt signaling antagonists, such as DKK1, in a feedback loop to promote osteolineage development.68 Another potential mediator is IL-1Î˛, a pleiotropic cytokine produced abundantly by AML blasts69 that has been shown

to suppress osteogenesis of MSCs in periodontal tissue at a high physiological level.70 Whether one or more pathways are involved in causing this lack of osteoblast maturation remains to be elucidated. Intriguingly, the effect of maturing osteoblasts on leukemia progression may be context-dependent and disease-specific. Schepers et al.71 showed that development of CML-like MPN induced by the BCR/ABL oncogene led to the expansion of a mixture of immature and mature osteoblasts that formed BM fibrosis and had decreased capacity to support HSCs. Krevvata et al.65 reported a strong correlation between the decrease in mature osteoblasts in mice and aggravated engraftment of different acute leukemia cell lines. In this study, over-stimulating Osc+ osteoblast production by inhibiting gut-derived

Figure 3. Key signaling pathways in the osteogenic niche that regulate the fate of hematopoietic stem cells (HSC) and leukemia stem/initiating cells (LSCs). HSCs home to bone marrow (BM) and receive maintenance signals from both endosteal and non-endosteal niches. Various cell types in the perisinusoid region, such as CXCL12-abundant reticular (CAR) cells and endothelial cells, maintain less quiescent HSCs via CXCL12 and stem cell factor (SCF). Non-myelinating Schwann cells and osteolineage cells, including mesenchymal stromal cells (MSCs), osteoprogenitors, and premature and mature osteoblasts, play a major role in retaining slowcycling HSCs near the bone surface. LSCs exploit the same cues from the osteogenic niche to hibernate and evade chemotherapy. In acute myeloid leukemia (AML), the BM niches are relatively hypoxic, whereas, in normal BM the hypoxic regions are more restricted to HSC-residing areas. Ang-1: angiopoietin 1; CXCL12: C-X-C motif chemokine 12; CXCR4: C-X-C chemokine receptor type 4; LSC: leukemic stem/initiating cell; OPN: osteopontin; SCF: stem cell factor; TGF-Î˛: transforming growth factor Î˛; TPO: thrombopoietin; VCAM-1: vascular cell adhesion protein 1; VLA-4: very late antigen-4.

1950

haematologica | 2018; 103(12)

Role of osteogenic niche in AML progression

serotonin synthesis in mice bearing MLL-AF9-induced AML attenuated disease burden.65 Conversely, using the same inducible model, Krause et al.8 showed that increasing osteoblastic activity through PTH activation augments leukemic expansion in AML while inhibiting CML-like MPN.8 These contrasting data are, at least in part, due to the heterogeneity of osteolineage cells that are characterized differently across studies. However, these results may also reflect the intrinsic differences in osteoblast-leukemia interaction between acute and chronic myeloid malignancies.

Deregulated signaling network in the osteogenic niche offers promising therapeutic targets In the past decade, huge strides forward have been made in AML induction therapy by combining chemoand targeted therapies, yet this approach has offered limited success in preventing disease recurrence. It is assumed that the osteogenic niche shields slow-cycling AML cells from cell cycle-dependent treatment just as it maintains HSC quiescence (Figure 3); however, that is just

the tip of the iceberg. The larger implication is that AML induces osteogenic dysfunction and disrupts the signaling network associated with the osteogenic niche (Table 1). This transformation of the niche could in turn fuel leukemia persistence and resistance to therapy. It is, therefore, critical to identify and target less visible threats underlying AML-osteogenic niche interactions to achieve more profound treatment efficacy. Current approaches revolve mainly around disrupting BM homing axes, notably CXCL12/CXCR4. CXCR4 overexpression is common both at diagnosis72 and after chemotherapy,73 and correlates with poor prognosis in AML patients.74,75 CXCR4 inhibition prevents AML anchorage and promotes mobilization of leukemic stem/initiating cells (LSCs) out of the endosteal niche, thereby increasing their vulnerability to chemotherapy. Pre-clinical and clinical studies of CXCR4 antagonists have shown encouraging results, demonstrating that these agents not only sensitize AML cells to chemotherapy, but also reverse stroma-mediated antiapoptotic effects.76,77 Although the first generation of CXCR4

Table 1. Dysregulation of signaling network and potential therapeutic targets associated with osteogenic niche in acute myeloid leukemia (AML) and myeloproliferative neoplasms (MPN) progression.

Role in normal hematopoiesis

Pathway

Deregulated feature

Prognostic marker?

CXCL12/CXCR4

CXCR4

Yes

HSC homing and retention in BM osteogenic niche

VCAM-1/VLA-4

VLA-4

Yes/No

OPN/CD44

OPN CD44v

Yes

HSC maintenance

Jagged-1/Notch

Jagged-1 and Notch receptor with limited autonomous Notch activation

TBI

Activation via fusion gene

TBI

TGF-β

BMP Bone and hematopoietic homeostasis

SNS

TBI

CCL3

Activation via fusion gene TBI Activation via osteogenic priming by AML Neuropathy

TBI

Proinflammatory cytokines TBI

Contribution to leukemogenesis

BM homing of AML LSCs Prosurvival/antiapoptotic pathways BM homing of AML LSCs Stroma-mediated chemoresistance via NF-kB activation BM homing of AML LSCs

Activation effect is context-dependent: AML expansion in vitro AML progression (crosstalk unspecified) AML progression in crosstalk with β-catenin/FoXO1 in osteoblasts Loss of TGF-β signaling is favorable though not required for AML initiation. Activation effect is context-dependent: LSC quiescence and chemoresistance AML progression & MPN progression via PTH activation in osteoblasts AML progression via CTGF upregulation in MSCs AML progression in crosstalk with TGF-β/CTGF in MSCs

AML progression via osteogenic priming in MSCs MPN progression via deficient sympathetic stimulation of MSCs MPN progression via HSC displacement

Ref

73-77 81-84 89-92 88,112

38,99,100,113

8,11,50,114

50,53

49,101

In AML cells; In osteolineage cells; Increase; Decrease; Mediate; TBI: to be investigated; HSC: hematopoietic stem cells; CXCL12: stromal derived factor-1; CXCR4 C-X-C: chemokine receptor 4; VCAM-1: vascular cell adhesion protein-1; VLA-4: very late antigen-4; OPN: osteopontin; TGF-β: transforming growth factor β; BMP:bone morphogenetic protein; SNS: sympathetic nervous system; CCL3: chemokine ligand 3; BM: bone marrow; LSC: leukemic stem/initiating cells; NF-kB: nuclear factor kappa-light-chain enhancer of activated B cells; PTH: parathyroid hormone; CTGF: connective tissue growth factor.

haematologica | 2018; 103(12)

1951

P.M. Le et al.

inhibitors, such as AMD3100 (plerixafor) and AMD3465, show anti-leukemic effects only synergistically with chemotherapy and their action is rather transient, a second generation that has potential as monotherapy is emerging.78,79 Of note, the clinical benefit of CXCR4 blockade could be further optimized given the role of differentiating osteoblasts in shielding AML cells from CXCL12-mediated apoptosis in hypoxia.12 Though such induction of apoptosis80 is controversial,79,8183 it cannot be excluded that the interplay between hypoxia-mediated protection of leukemic cells typically found in AML BM84,85 and niche components may tip the balance between CXCL12-mediated pro-survival and apoptosis. This possibility is noteworthy given CXCR4 is a well-established target of HIF-1α, and therefore hypoxia.86 The role of HIF-1α in survival and maintenance of CML has also been described.87 The Bonnet group found that HIF-2α, another factor that is regulated by hypoxia, plays a crucial role in regulation of the longterm re-populating ability of CD34+ umbilical cord blood cells. In addition, their data demonstrate that inhibition of HIF-2α in primary AML cells inhibits their proliferation and sensitizes them to endoplasmic reticulum stress-induced apoptosis by upregulation of reactive oxygen species.85 Similarly, chemosensitization and reduced AML engraftment could be achieved in mouse models with the use of CD44 specific antibody88 and VLA-4 blocking agents, such as natalizumab89 and AS101.90 Extramedullary BM models by Chen et al.91 and Jacamo et al.92 demonstrated that AML and stromal cells interact via VLA-4 and VCAM1 to activate downstream NFkB signaling in both cell types. Blockade of these interactions resulted in inhibition of stroma-induced chemoresistance in AML cells.91,92 Besides, antagonizing these adhesion molecules has been found to relieve differentiation block in blasts,88 a clinical benefit also seen when epigenetic modifiers93 or FLT3 and IDH1/2 inhibitors94,95 were used to treat AML. This mobilizing approach, while preferentially mobilizing AML cells, carries the risk of moving HSCs out of their protective BM niche, among other adverse effects.96-98 Further randomized trials are needed to determine whether targeting AML homing axes is safe and how to optimize this chemosensitizing approach without impairing normal hematopoiesis. There has been growing evidence to indicate another promising strategy: to target osteoblast function. Leukemia-stroma contact potentiates osteoblast differentiation in MSCs, which counteracts apoptogenic cues and promotes proliferative signals from the microenvironment to leukemic cells.12,50,55 It has also been shown that abnormal signaling pathways and crosstalks that take place specifically in osteoblasts could induce or aggravate AML phenotype.8,38 Manipulating signals from osteolineage cells would, therefore, render the osteogenic niche hostile to AML cells and abrogate the feedback loop fueling their perpetual life cycle. Indeed, modulation of mature osteoblast numbers by inhibiting gut-derived serotonin synthesis results in leukemia regression, providing a 'proof of concept' for this approach.65 The fact that stage-specific osteolineage cells have distinct functions and may differentially regulate normal and malignant hematopoiesis makes them an even more attractive target, especially with regard to their involvement in pleiotropic signaling pathways that support 1952

HSCs, such as Notch or TGF-β. For example, despite already being known for its tumor-suppressor role in AML,99,100 Notch activation has been reported to be leukemogenic when synergizing with activating βcatenin/FoxO signaling in Col1a1+ pre-osteoblasts.38 This observation suggests targeting FoxO signaling in preosteoblasts may be beneficial to patients with constitutive activating β-catenin mutation. As the effects of myeloid leukemia on cell differentiation along the osteolineage unfold, more leukemia modulators might be identified, and these will facilitate patient stratification and prevent treatment failure. Restricting osteogenic capacity of MSCs could also be a therapeutic option. This strategy potentially limits the OPN reservoir of the BM, further preventing AML cells from hiding in the osteogenic niche and evading chemotherapy. Maintaining the primitive MSC pool via β2- and β3-adrenergic agonists has shown multiple advantages in managing AML and MPN: rescuing healthy HSCs in the osteogenic niche with HSC maintenance factors and preventing LSCs from crowding out these normal residents.49,101 Studies have further shown that the bone surface and periarteriolar region are prone to inflammation during the early stage of osteogenic niche remodeling.39,56 This can be ameliorated by blocking receptors of pro-inflammatory cytokines, e.g. via CCL3 receptor antagonists. Alternatively, promoting adipocyte differentiation of MSCs has been demonstrated to be a viable strategy to improve disease management by rescuing at least the generation of myeloid-erythroid lineages.51 However, the long-term efficacy of this pro-adipogenesis therapy remains to be tested given the debatable evidence about the role of adipocytes in AML seen so far. Different groups reported on adipocyte re-programming in which AML blasts exploit these energy reservoirs through lipolysis to fuel uncontrolled expansion.102-104 On the other hand, Lu et al.105 only found a statistically significant correlation between AML patients’ poor prognosis and an increase in small adipocytes, but not the decrease in largeor medium-sized ones. This finding suggests that lipid transfer may not be the only mechanism through which adipocytes aggravate leukemia burden. It cannot be excluded that, as previously shown in acute lymphoblastic leukemia,106,107 adipocytes may acquire a chemoprotective role in the setting of AML. Recent discoveries provide evidence that mitochondria are transferred from BM stromal cells to leukemia cells which influence leukemia progression.108 These studies demonstrate that mitochondria are transferred via tunneling nanotubes (TNTs) or extracellular vesicles resulting in enhanced ATP production through increased oxidative phosphorylation (OXPHOS) which translates into higher drug resistance in AML cells and relapse after chemotherapy.109,110 Therefore, inhibition of mitochondrial transfer by targeting TNT formation or inhibiting OXPHOS is currently being considered as novel therapeutic strategies in AML therapy.

Conclusions and emerging questions Findings on BM stroma-mediated chemoprotection in AML since the early 2000s have paved the way for a wave of new insights into leukemia-BM niche interactions, hence re-defining the paradigm of leukemic develhaematologica | 2018; 103(12)

Role of osteogenic niche in AML progression

opment and response to therapy. Although its fundamental role in HSC maintenance is still debatable, the osteogenic niche stands out as a pivotal sanctuary for LSCs and the cradle of blast production. Putting this into perspective, we foresee that AML-induced genetic changes and osteogenic priming in MSCs illustrate not only the long-standing multiple-hit hypothesis of carcinogenesis but also the newly-coined microenvironmentinduced oncogenesis. Importantly, the role of the BM niches in the development of MDS and AML from clonal hematopoiesis of indeterminate potential is still completely unknown.111 Many questions emerge, for instance, regarding the differential role of stage-specific osteolineage cells in AML progression, the uncoupling between osteogenesis and osteoclastogenesis, or the net therapeutic benefits of LSC dislodgement at the expense of HSC homelessness. Recent advances in our understanding of this osteoblast-rich region in AML progression provide a convincing premise with which to build the next genera-

References 1. Morrison SJ, Scadden DT. The bone marrow niche for haematopoietic stem cells. Nature. 2014;505(7483):327-334. 2. Itkin T, Gur-Cohen S, Spencer JA, et al. Distinct bone marrow blood vessels differentially regulate haematopoiesis. Nature. 2016;532(7599):323-328. 3. Kusumbe AP, Ramasamy SK, Adams RH. Coupling of angiogenesis and osteogenesis by a specific vessel subtype in bone. Nature. 2014;507(7492):323-328. 4. Langen UH, Pitulescu ME, Kim JM, et al. Cell-matrix signals specify bone endothelial cells during developmental osteogenesis. Nat Cell Biol. 2017;19(3):189-201. 5. Asada N, Takeishi S, Frenette PS. Complexity of bone marrow hematopoietic stem cell niche. Int J Hematol. 2017;106(1):45-54. 6. Miraki-Moud F, Anjos-Afonso F, Hodby KA, et al. Acute myeloid leukemia does not deplete normal hematopoietic stem cells but induces cytopenias by impeding their differentiation. Proc Natl Acad Sci USA. 2013;110(33):13576-13581. 7. Glait-Santar C, Desmond R, Feng X, et al. Functional Niche Competition Between Normal Hematopoietic Stem and Progenitor Cells and Myeloid Leukemia Cells. Stem Cells. 2015;33(12):3635-3642. 8. Krause DS, Fulzele K, Catic A, et al. Differential regulation of myeloid leukemias by the bone marrow microenvironment. Nat Med. 2013;19(11):1513-1517. 9. Zhou HS, Carter BZ, Andreeff M. Bone marrow niche-mediated survival of leukemia stem cells in acute myeloid leukemia: Yin and Yang. Cancer Biol Med. 2016;13(2):248259. 10. Blau O, Hofmann W-K, Baldus CD, et al. Chromosomal aberrations in bone marrow mesenchymal stroma cells from patients with myelodysplastic syndrome and acute myeloblastic leukemia. Exp Hematol. 2007;35(2):221-229. 11. Tabe Y, Shi YX, Zeng Z, et al. TGF-betaNeutralizing Antibody 1D11 Enhances Cytarabine-Induced Apoptosis in AML Cells

haematologica | 2018; 103(12)

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

tion of AML therapy to target the osteogenic niche. However, further studies are needed to clarify the selfreinforcing loop between AML and the osteogenic niche, with the goal of inducing deep remissions and controlling long-term disease. Acknowledgements This work is supported by the Leukemia SPORE career development award (CA100632), Rolanette and Berdon Lawrence Research Award from Bone Disease Program of Texas, Institutional Research Grant (IRG) from MD Anderson Cancer Center, Cure Sonia Foundation and Golfers Against Cancer Foundation to VLB. In addition, this work was supported by grants from the National Institutes of Health (CA055164) and the MD Anderson Cancer Center Support Grant (CA016672), Cancer Prevention Research Institute of Texas (CPRIT, RP121010), and the Paul and Mary Haas Chair in Genetics to MA. We thank Dr. Marina Konopleva for her valuable suggestions in the preparation of this review article.

in the Bone Marrow Microenvironment. PLoS One. 2013;8(6):e62785. Kremer KN, Dudakovic A, McGeeLawrence ME, et al. Osteoblasts Protect AML Cells from SDF-1-Induced Apoptosis. J Cell Biochem. 2014;115(6):1128-1137. Ishikawa F, Yoshida S, Saito Y, et al. Chemotherapy-resistant human AML stem cells home to and engraft within the bonemarrow endosteal region. Nat Biotechnol. 2007;25(11):1315-1321. Lord BI, Testa NG, Hendry JH. The relative spatial distributions of CFUs and CFUc in the normal mouse femur. Blood. 1975;46(1):65-72. Lo Celso C, Fleming HE, Wu JW, et al. Liveanimal tracking of individual haematopoietic stem/progenitor cells in their niche. Nature. 2009;457(7225):92-96. Taichman RS, Emerson SG. Human osteoblasts support hematopoiesis through the production of granulocyte colony-stimulating factor. J Exp Med. 1994;179(5):16771682. Taichman RS, Reilly MJ, Verma RS, Ehrenman K, Emerson SG. Hepatocyte growth factor is secreted by osteoblasts and cooperatively permits the survival of haematopoietic progenitors. Br J Haematol. 2001;112(2):438-448. Stier S, Ko Y, Forkert R, et al. Osteopontin is a hematopoietic stem cell niche component that negatively regulates stem cell pool size. J Exp Med. 2005;201(11):1781-1791. Sugiyama T, Kohara H, Noda M, Nagasawa T. Maintenance of the Hematopoietic Stem Cell Pool by CXCL12-CXCR4 Chemokine Signaling in Bone Marrow Stromal Cell Niches. Immunity. 2006;25(6):977-988. Ulyanova T, Scott LM, Priestley GV, et al. VCAM-1 expression in adult hematopoietic and nonhematopoietic cells is controlled by tissue-inductive signals and reflects their developmental origin. Blood. 2005;106(1): 86-94. Katayama Y, Battista M, Kao WM, et al. Signals from the sympathetic nervous system regulate hematopoietic stem cell egress from bone marrow. Cell. 2006;124(2):407421.

22. Zhang J, Niu C, Ye L, et al. Identification of the haematopoietic stem cell niche and control of the niche size. Nature. 2003;425(6960):836-841. 23. Calvi LM, Adams GB, Weibrecht KW, et al. Osteoblastic cells regulate the haematopoietic stem cell niche. Nature. 2003;425 (6960):841-846. 24. Arai F, Hirao A, Ohmura M, et al. Tie2/angiopoietin-1 signaling regulates hematopoietic stem cell quiescence in the bone marrow niche. Cell. 2004;118(2):149161. 25. Yoshihara H, Arai F, Hosokawa K, et al. Thrombopoietin/MPL signaling regulates hematopoietic stem cell quiescence and interaction with the osteoblastic niche. Cell Stem Cell. 2007;1(6):685-697. 26. Fleming HE, Janzen V, Celso CL, et al. Wnt signaling in the niche enforces hematopoietic stem cell quiescence and is necessary to preserve self-renewal in vivo. Cell Stem Cell. 2008;2(3):274-283. 27. Bowers M, Zhang B, Ho Y, et al. Osteoblast ablation reduces normal long-term hematopoietic stem cell self-renewal but accelerates leukemia development. Blood. 2015;125(17):2678-2688. 28. Ma YD, Park C, Zhao H, et al. Defects in osteoblast function but no changes in longterm repopulating potential of hematopoietic stem cells in a mouse chronic inflammatory arthritis model. Blood. 2009;114(20):44024410. 29. Greenbaum A, Hsu Y-MS, Day RB, et al. CXCL12 Production by Early Mesenchymal Progenitors is Required for Hematopoietic Stem Cell Maintenance. Nature. 2013;495(7440):227-230. 30. Ding L, Saunders TL, Enikolopov G, Morrison SJ. Endothelial and perivascular cells maintain haematopoietic stem cells. Nature. 2012;481(7382):457-462. 31. Nakamura Y, Arai F, Iwasaki H, et al. Isolation and characterization of endosteal niche cell populations that regulate hematopoietic stem cells. Blood. 2010;116 (9):1422-1432. 32. Walter MJ, Shen D, Ding L, et al. Clonal Architecture of Secondary Acute Myeloid

1953

P.M. Le et al.

33. 34.

35.

36.

37.

38.

39.

40. 41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

1954

Leukemia. N Engl J Med. 2012;366(12):10901098. Genomic and Epigenomic Landscapes of Adult De Novo Acute Myeloid Leukemia. N Engl J Med. 2013;368(22):2059-2074. Shlush LI, Zandi S, Mitchell A, et al. Identification of pre-leukemic hematopoietic stem cells in acute leukemia. Nature. 2014;506(7488):328-333. Raaijmakers MH, Mukherjee S, Guo S, et al. Bone progenitor dysfunction induces myelodysplasia and secondary leukaemia. Nature. 2010;464(7290):852-857. Kode A, Manavalan JS, Mosialou I, et al. Leukaemogenesis induced by an activating beta-catenin mutation in osteoblasts. Nature. 2014;506(7487):240-244. Tothova Z, Gilliland DG. FoxO Transcription Factors and Stem Cell Homeostasis: Insights from the Hematopoietic System. Cell Stem Cell. 2007;1(2):140-152. Kode A, Mosialou I, Manavalan SJ, et al. FoxO1-dependent induction of acute myeloid leukemia by osteoblasts in mice. Leukemia. 2016;30(1):1-13. Dong L, Yu WM, Zheng H, et al. Leukaemogenic effects of Ptpn11 activating mutations in the stem cell microenvironment. Nature. 2016;539(7628):304-308. Wiseman DH. Donor Cell Leukemia: A Review. Biol of Blood and Marrow Transplantation. 2011;17(6):771-789. Wang Y, Krivtsov AV, Sinha AU, et al. The Wnt/beta-catenin pathway is required for the development of leukemia stem cells in AML. Science. 2010;327(5973):1650-1653. Lane SW, Wang YJ, Lo Celso C, et al. Differential niche and Wnt requirements during acute myeloid leukemia progression. Blood. 2011;118(10):2849-2856. Soenen-Cornu V, Tourino C, Bonnet M-L, et al. Mesenchymal cells generated from patients with myelodysplastic syndromes are devoid of chromosomal clonal markers and support short- and long-term hematopoiesis in vitro. Oncogene. 2005;24 (15):2441-2448. Huang JC, Basu SK, Zhao X, et al. Mesenchymal stromal cells derived from acute myeloid leukemia bone marrow exhibit aberrant cytogenetics and cytokine elaboration. Blood Cancer J. 2015;5:e302. Blau O, Baldus CD, Hofmann WK, et al. Mesenchymal stromal cells of myelodysplastic syndrome and acute myeloid leukemia patients have distinct genetic abnormalities compared with leukemic blasts. Blood. 2011;118(20):5583-5592. Kornblau SM, Ruvolo PP, Wang RY, et al. Distinct protein signatures of acute myeloid leukemia bone marrow-derived stromal cells are prognostic for patient survival. Haematologica. 2018;103(5):810-821. Mukherjee A, Rotwein P. Insulin-Like Growth Factor-Binding Protein-5 Inhibits Osteoblast Differentiation and Skeletal Growth by Blocking Insulin-Like Growth Factor Actions. Mol Endocrinol. 2008;22 (5):1238-1250. Jacamo R, Davis RE, Ling X, et al. Tumor Trp53 status and genotype affect the bone marrow microenvironment in acute myeloid leukemia. Oncotarget. 2017;8(48): 83354-83369. Hanoun M, Zhang D, Mizoguchi T, et al. Acute myelogenous leukemia-induced sympathetic neuropathy promotes malignancy in an altered hematopoietic stem cell niche. Cell Stem Cell. 2014;15(3):365-375. Battula VL, Le PM, Sun JC, et al. AMLinduced osteogenic differentiation in mes-

51.

52.

53.

54.

55.

56.

57.

58. 59. 60.

61.

62.

63.

64.

65.

66.

67.

68.

enchymal stromal cells supports leukemia growth. JCI Insight. 2017;2(13). Boyd AL, Reid JC, Salci KR, et al. Acute myeloid leukaemia disrupts endogenous myelo-erythropoiesis by compromising the adipocyte bone marrow niche. Nat Cell Biol. 2017;19(11):1336-1347. Sala-Torra O, Gundacker HM, Stirewalt DL, et al. Connective tissue growth factor (CTGF) expression and outcome in adult patients with acute lymphoblastic leukemia. Blood. 2007;109(7):3080-3083. Crispino JD, Le Beau MM. BMP meets AML: induction of BMP signaling by a novel fusion gene promotes pediatric acute leukemia. Cancer Cell. 2012;22(5):567-568. Raymond A, Liu B, Liang H, et al. A role for BMP-induced homeobox gene MIXL1 in acute myelogenous leukemia and identification of type I BMP receptor as a potential target for therapy. Oncotarget. 2014;5(24): 12675-12693. Sterner RM, Kremer KN, Dudakovic A, et al. Tissue-Nonspecific Alkaline Phosphatase Is Required for MC3T3 Osteoblastâ&#x20AC;&#x201C;Mediated Protection of Acute Myeloid Leukemia Cells from Apoptosis. J Immunol. 2018;201(3): 1086-1096. Frisch BJ, Ashton JM, Xing L, et al. Functional inhibition of osteoblastic cells in an in vivo mouse model of myeloid leukemia. Blood. 2012;119(2):540-550. Baba T, Mukaida N. Role of macrophage inflammatory protein (MIP)-1 /CCL3 in leukemogenesis. Mol Cell Oncol. 2014;1(1): e29899. Lane SW. Bad to the bone. Blood. 2012;119(2):323-325. Shi C, Pamer EG. Monocyte recruitment during infection and inflammation. Nat Rev Immunol. 2011;11(11):762-774. Wu JY, Purton LE, Rodda SJ, et al. Osteoblastic regulation of B lymphopoiesis is mediated by G(s) -dependent signaling pathways. Proc Natl Acad Sci USA. 2008;105(44):16976-16981. Rankin EB, Wu C, Khatri R, et al. The HIF signaling pathway in osteoblasts directly modulates erythropoiesis through the production of EPO. Cell. 2012;149(1):63-74. Ding L, Morrison SJ. Haematopoietic stem cells and early lymphoid progenitors occupy distinct bone marrow niches. Nature. 2013;495(7440):231-235. Kumar B, Garcia M, Weng L, et al. Acute myeloid leukemia transforms the bone marrow niche into a leukemia-permissive microenvironment through exosome secretion. Leukemia. 2017;32(3):575-587. Geyh S, Rodriguez-Paredes M, Jager P, et al. Functional inhibition of mesenchymal stromal cells in acute myeloid leukemia. Leukemia. 2016;30(3):683-691. Krevvata M, Silva BC, Manavalan JS, et al. Inhibition of leukemia cell engraftment and disease progression in mice by osteoblasts. Blood. 2014;124(18):2834-2846. Duarte D, Hawkins ED, Akinduro O, et al. Inhibition of Endosteal Vascular Niche Remodeling Rescues Hematopoietic Stem Cell Loss in AML. Cell Stem Cell. 2017;22 (1):64-77.e6. Gong JN, Yu J, Lin HS, et al. The role, mechanism and potentially therapeutic application of microRNA-29 family in acute myeloid leukemia. Cell Death Differ. 2014;21(1):100-112. Kapinas K, Kessler C, Ricks T, Gronowicz G, Delany AM. miR-29 Modulates Wnt Signaling in Human Osteoblasts through a Positive Feedback Loop. J Biol Chem. 2010;285(33):25221-25231.

69. Arranz L, Arriero MdM, Villatoro A. Interleukin-1&#x3b2; as emerging therapeutic target in hematological malignancies and potentially in their complications. Blood Rev. 2017;31(5):306-317. 70. Mao Cy, Wang Yg, Zhang X, et al. Doubleedged-sword effect of IL-1 on the osteogenesis of periodontal ligament stem cells via crosstalk between the NF- B, MAPK and BMP/Smad signaling pathways. Cell Death Dis. 2016;7(7):e2296. 71. Schepers K, Pietras EM, Reynaud D, et al. Myeloproliferative Neoplasia Remodels the Endosteal Bone Marrow Niche into a SelfReinforcing Leukemic Niche. Cell Stem Cell. 2013;13(3):285-299. 72. Voermans C, van Heese WP, de Jong I, Gerritsen WR, van Der Schoot CE. Migratory behavior of leukemic cells from acute myeloid leukemia patients. Leukemia. 2002;16(4):650-657. 73. Sison EAR, McIntyre E, Magoon D, Brown P. Dynamic chemotherapy-induced upregulation of surface CXCR4 expression as a mechanism of chemotherapy resistance in pediatric acute myeloid leukemia. Mol Cancer Res. 2013;11(9):1004-1016. 74. Konoplev S, Rassidakis GZ, Estey E, et al. Overexpression of CXCR4 predicts adverse overall and event-free survival in patients with unmutated FLT3 acute myeloid leukemia with normal karyotype. Cancer. 2007;109(6):1152-1156. 75. Spoo AC, Lubbert M, Wierda WG, Burger JA. CXCR4 is a prognostic marker in acute myelogenous leukemia. Blood. 2007;109(2): 786-791. 76. Zeng Z, Shi YX, Samudio IJ, et al. Targeting the leukemia microenvironment by CXCR4 inhibition overcomes resistance to kinase inhibitors and chemotherapy in AML. Blood. 2009;113(24):6215-6224. 77. Nervi B, Ramirez P, Rettig MP, et al. Chemosensitization of acute myeloid leukemia (AML) following mobilization by the CXCR4 antagonist AMD3100. Blood. 2009;113(24):6206-6214. 78. Kuhne MR, Mulvey T, Belanger B, et al. BMS-936564/MDX-1338: A Fully Human Anti-CXCR4 Antibody Induces Apoptosis &lt;em&gt;In Vitro&lt;/em&gt; and Shows Antitumor Activity In Vivo in Hematologic Malignancies. Clin Cancer Res. 2013;19(2): 357-366. 79. Cho B-S, Zeng Z, Mu H, et al. Antileukemia activity of the novel peptidic CXCR4 antagonist LY2510924 as monotherapy and in combination with chemotherapy. Blood. 2015;126(2):222-232. 80. Kremer KN, Peterson KL, Schneider PA, et al. CXCR4 Chemokine Receptor Signaling Induces Apoptosis in Acute Myeloid Leukemia Cells via Regulation of the Bcl-2 Family Members Bcl-XL, Noxa, and Bak. J Biol Chem. 2013;288(32):22899-22914. 81. Zhang Y, Patel S, Abdelouahab H, et al. CXCR4 inhibitors selectively eliminate CXCR4-expressing human acute myeloid leukemia cells in NOG mouse model. Cell Death Dis. 2012;3(10):e396. 82. Borthakur G, Ofran Y, Nagler A, et al. The Peptidic CXCR4 Antagonist, BL-8040, Significantly Reduces Bone Marrow Immature Leukemia Progenitors By Inducing Differentiation, Apoptosis and Mobilization: Results of the Dose Escalation Clinical Trial in Acute Myeloid Leukemia. Blood. 2015;126(23):2546. 83. Chen Y, Jacamo R, Konopleva M, et al. CXCR4 downregulation of let-7a drives chemoresistance in acute myeloid leukemia. J Clin Invest. 2013;123(6):2395-2407.

haematologica | 2018; 103(12)

Role of osteogenic niche in AML progression 84. Benito J, Ramirez M, Millward NZ, et al. Hypoxia-activated prodrug TH-302 targets hypoxic bone marrow niches in pre-clinical leukemia models. Clin Cancer Res. 2016;22(7):1687-1698. 85. Rouault-Pierre K, Lopez-Onieva L, Foster K, et al. HIF-2alpha protects human hematopoietic stem/progenitors and acute myeloid leukemic cells from apoptosis induced by endoplasmic reticulum stress. Cell Stem Cell. 2013;13(5):549-563. 86. Fiegl M, Samudio I, Clise-Dwyer K, et al. CXCR4 expression and biologic activity in acute myeloid leukemia are dependent on oxygen partial pressure. Blood. 2009;113(7): 1504-1512. 87. Zhang H, Li H, Xi HS, Li S. HIF1 is required for survival maintenance of chronic myeloid leukemia stem cells. Blood. 2012;119(11): 2595-2607. 88. Jin L, Hope KJ, Zhai Q, Smadja-Joffe F, Dick JE. Targeting of CD44 eradicates human acute myeloid leukemic stem cells. Nat Med. 2006;12(10):1167-1174. 89. Jiang E, Pham J, Kim H-N, et al. VLA4 Blockade In Acute Myeloid Leukemia. Blood. 2013;122(21):3944. 90. Layani-Bazar A, Skornick I, Berrebi A, et al. Redox Modulation of Adjacent Thiols in VLA-4 by AS101 Converts Myeloid Leukemia Cells from a Drug-Resistant to Drug-Sensitive State. Cancer Res J. 2014;74(11):3092-3103. 91. Chen Y, Jacamo R, Shi YX, et al. Human extramedullary bone marrow in mice: a novel in vivo model of genetically controlled hematopoietic microenvironment. Blood. 2012;119(21):4971-4980. 92. Jacamo R, Chen Y, Wang Z, et al. Reciprocal leukemia-stroma VCAM-1/VLA-4-dependent activation of NF-kappaB mediates chemoresistance. Blood. 2014;123(17):26912702. 93. Abdel-Wahab O, Levine RL. Mutations in epigenetic modifiers in the pathogenesis and therapy of acute myeloid leukemia. Blood. 2013;121(18):3563-3572.

haematologica | 2018; 103(12)

94. Sexauer A, Perl A, Yang X, et al. Terminal myeloid differentiation in vivo is induced by FLT3 inhibition in FLT3/ITD AML. Blood. 2012;120(20):4205-4214. 95. Boddu P, Borthakur G. Therapeutic targeting of isocitrate dehydrogenase mutant AML. Expert Opin Investig Drugs. 2017;26(5):525530. 96. Bonig H, Wundes A, Chang K-H, Lucas S, Papayannopoulou T. Increased numbers of circulating hematopoietic stem/progenitor cells are chronically maintained in patients treated with the CD49d blocking antibody natalizumab. Blood. 2008;111(7):3439-3441. 97. Becker PS, Foran JM, Altman JK, et al. Targeting the CXCR4 Pathway: Safety, Tolerability and Clinical Activity of Ulocuplumab (BMS-936564), an AntiCXCR4 Antibody, in Relapsed/Refractory Acute Myeloid Leukemia. Blood. 2014;124 (21):386. 98. Uy GL, Rettig MP, Stone RM, et al. A phase 1/2 study of chemosensitization with plerixafor plus G-CSF in relapsed or refractory acute myeloid leukemia. Blood Cancer J. 2017;7(3):e542. 99. Lobry C, Ntziachristos P, Ndiaye-Lobry D, et al. Notch pathway activation targets AML-initiating cell homeostasis and differentiation. J Exp Med. 2013;210(2):301-319. 100. Klinakis A, Lobry C, Abdel-Wahab O, et al. A novel tumor suppressor function for the Notch pathway in myeloid leukemia. Nature. 2011;473(7346):230-233. 101. Arranz L, Sánchez-Aguilera A, Martín-Pérez D, et al. Neuropathy of haematopoietic stem cell niche is essential for myeloproliferative neoplasms. Nature. 2014;512(7512):78-81. 102. Tabe Y, Yamamoto S, Saitoh K, et al. Bone marrow adipocytes facilitate fatty acid oxidation activating AMPK and a transcriptional network supporting survival of acute monocytic leukemia cells. Cancer Res. 2017;77(6):1453-1464. 103. Shafat MS, Oellerich T, Mohr S, et al. Leukemic blasts program bone marrow adipocytes to generate a pro-tumoral

microenvironment. Blood. 2017;129(10): 1320-1332. 104. Battula VL, Chen Y, da Graca Cabreira M, et al. Connective tissue growth factor regulates adipocyte differentiation of mesenchymal stromal cells and facilitates leukemia bone marrow engraftment. Blood. 2013;122(3): 357-366. 105. Lu W, Weng W, Zhu Q, et al. Small bone marrow adipocytes predict poor prognosis in acute myeloid leukemia. Haematologica. 2018;103(1):e21-e24. 106. Behan JW, Yun JP, Proektor MP, et al. Adipocytes Impair Leukemia Treatment in Mice. Cancer Res J. 2009;69(19):7867-7874. 107. Sheng X, Tucci J, Parmentier J-H, et al. Adipocytes cause leukemia cell resistance to daunorubicin via oxidative stress response. Oncotarget. 2016;7(45):73147-73159. 108. Griessinger E, Moschoi R, Biondani G, Peyron JF. Mitochondrial Transfer in the Leukemia Microenvironment. Trends Cancer. 2017;3(12):828-839. 109. Moschoi R, Imbert V, Nebout M, et al. Protective mitochondrial transfer from bone marrow stromal cells to acute myeloid leukemic cells during chemotherapy. Blood. 2016;128(2):253-264. 110. Marlein CR, Zaitseva L, Piddock RE, et al. NADPH oxidase-2 derived superoxide drives mitochondrial transfer from bone marrow stromal cells to leukemic blasts. Blood. 2017;130(14):1649-1660. 111. Steensma DP, Ebert BL. Clonal Hematopoiesis after Induction Chemotherapy for Acute Myeloid Leukemia. N Engl J Med. 2018;378(13):1244-1245. 112. Liersch R, Gerss J, Schliemann C, et al. Osteopontin is a prognostic factor for survival of acute myeloid leukemia patients. Blood. 2012;119(22):5215-5220. 113. Tohda S, Nara N. Expression of Notchl and Jaggedl Proteins in Acute Myeloid Leukemia Cells. Leuk Lymphoma. 2001;42(3):467-472. 114. Dong M, Blobe GC. Role of transforming growth factor- in hematologic malignancies. Blood. 2006;107(12):4589-4596.

1955

REVIEW ARTICLE Ferrata Storti Foundation

TP53 aberrations in chronic lymphocytic leukemia: an overview of the clinical implications of improved diagnostics

Elias Campo,1 Florence Cymbalista,2 Paolo Ghia,3 Ulrich Jäger,4 Sarka Pospisilova,5 Richard Rosenquist,6 Anna Schuh7 and Stephan Stilgenbauer8

Hospital Clinic of Barcelona, University of Barcelona, Institute of Biomedical Research August Pi i Sunyer (IDIBAPS), Barcelona, and CIBERONC, Spain; 2Hôpital Avicenne, AP-HP, UMR INSERMU978/Paris 13 University, Bobigny, France; 3Università Vita-Salute San Raffaele and IRCCS Ospedale San Raffaele, Milan, Italy; 4Medical University of Vienna, Austria; 5Center of Molecular Medicine, Central European Institute of Technology, Masaryk University, Brno, Czech Republic; 6Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden; 7University of Oxford, UK and 8Internal Medicine III, Ulm University, Germany and Innere Medizin I, Universitätsklinikum des Saarlandes, Homburg, Germany 1

Haematologica 2018 Volume 103(12):1956-1968

All authors contributed equally to this work

ABSTRACT

Correspondence: ghia.paolo@hsr.it

Received: May 23, 3018. Accepted: October 26, 2018. Pre-published: November 15, 2018.

doi:10.3324/haematol.2018.187583 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/1956 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1956

hronic lymphocytic leukemia is associated with a highly heterogeneous disease course in terms of clinical outcomes and responses to chemoimmunotherapy. This heterogeneity is partly due to genetic aberrations identified in chronic lymphocytic leukemia cells such as mutations of TP53 and/or deletions in chromosome 17p [del(17p)], resulting in loss of one TP53 allele. These aberrations are associated with markedly decreased survival and predict impaired response to chemoimmunotherapy thus being among the strongest predictive markers guiding treatment decisions in chronic lymphocytic leukemia. Clinical trials demonstrate the importance of accurately testing for TP53 aberrations [both del(17p) and TP53 mutations] before each line of treatment to allow for appropriate treatment decisions that can optimize patients’ outcomes. The current report reviews the diagnostic methods to detect TP53 disruption better, the role of TP53 aberrations in treatment decisions and current therapies available for patients with chronic lymphocytic leukemia carrying these abnormalities. The standardization in sequencing technologies for accurate identification of TP53 mutations and the importance of continued evaluation of TP53 aberrations throughout initial and subsequent lines of therapy remain unmet clinical needs as new therapeutic alternatives become available.

Introduction Chronic lymphocytic leukemia (CLL) is associated with a highly heterogeneous disease course, with some patients surviving for more than 10 years without needing treatment, and others experiencing rapid disease progression and poor outcomes despite effective chemoimmunotherapy.1-3 This heterogeneity is partly explained by the diverse genetic aberrations identified in CLL patients.4-6 In particular, deletions in chromosome 17p [del(17p)] resulting in loss of the TP53 gene, which encodes the tumor-suppressor protein p53, are associated with a poor prognosis. Furthermore, mutations of TP53 are also associated with poor prognosis independently of the presence of del(17p).7 Collectively, these deletions and mutations will be referred to as TP53 aberrations. TP53 aberrations belong to the strongest prognostic and predictive markers guiding treatment decisions in CLL, and are associated with markedly decreased surhaematologica | 2018; 103(12)

TP53 aberrations in CLL

vival and impaired response to chemoimmunotherapy.8-12 Until recently, the only effective treatments available for patients with CLL harboring TP53 aberrations were alemtuzumab and allogeneic hematopoietic stem cell transplantation.13-17 New small-molecule inhibitors that are efficacious in patients harboring TP53 aberrations are now available, including the Bruton tyrosine kinase (BTK) inhibitor ibrutinib, the phosphatidylinositol 3-kinase (PI3K) inhibitor idelalisib, and the BCL2 inhibitor venetoclax.18-26 Identifying TP53 aberrations is therefore important for determining the most appropriate course of treatment for patients with CLL.27 Several diagnostic techniques are currently in routine use for the identification of TP53 aberrations. A substantial proportion of TP53 aberrations involve TP53 mutations in the absence of del(17p).12,28-31 Therefore, while del(17p) is routinely identified by fluorescence in situ hybridization (FISH), FISH testing alone may potentially fail to identify approximately 30-40% of patients with TP53 aberrations, i.e those carrying only mutations in the gene.32,33 Thus, it is critical to test for relevant TP53 mutations, using Sanger sequencing or high-throughput sequencing technologies, in addition to FISH detection of del(17p), and both tests should be performed before each line of therapy to select appropriate treatment, as TP53 aberrations may emerge during the disease course and after previous treatment. 27,31,34 The European Research Initiative on CLL (ERIC) has implemented a certification program (known as the TP53 Network) for clinical laboratories performing analysis of TP53 aberra-

tions in order to improve the reliability of TP53 mutation analysis and to spread knowledge on testing for TP53 aberrations in routine clinical practice, with the final aim of optimizing treatment choices and patients’ outcomes.35

Genetic aberrations in chronic lymphocytic leukemia Genetic aberrations identified in CLL include genomic abnormalities and specific gene mutations.6,36 Combinations of these aberrations, along with immunoglobulin heavy variable (IGHV) mutation status, result in biological and clinical subgroups associated with varying outcomes.10,11,37,38 An overview of the genetic aberrations frequently found in CLL is provided in Table 1. Chromosomal abnormalities frequently found in CLL include del(13q), trisomy 12, del(11q), and del(17p);4 other less frequent abnormalities have also been identified such as amplifications of chromosome 2p or 8q, and deletions in chromosomes 8p and 15q.4,36 Using conventional karyotyping of stimulated lymphocytes, the presence of three or more chromosomal abnormalities, known as a complex karyotype, has been associated with worse disease outcomes.39-42 Similar results have been obtained using arrays for DNA copy number alterations to detect genomic complexity.37,43 There is a strong association of complex karyotype with TP53 aberrations leading to genetic instability, but a complex karyotype has been demonstrated to be an independent prognostic factor for poor overall survival.28,39,40,44,45 Chromothripsis-like patterns, defined by tens to hundreds of chromosomal

Table 1. Overview of genetic complexity in chronic lymphocytic leukemia.

Genetic aberration

Chromosomal abnormalities

Frequency in Time to first PFS untreated treatment (median, patients (median, months) months) del(17p) del(11q) Trisomy 12 del(13q)

4–8.5% 17–18% 12–16% 35–55%

Other (e.g. amp[2p]; 2–7% amp[8q]; del[8p]; del[15q]; and del[6q])

Gene mutation

OS (median, months)

Coexistence with other genetic aberrations

TP53 mutations (4, 8, 11, 28, 56) ATM and/or SF3B1, BIRC3 mutations (4, 11, 28, 56) NOTCH1 mutations (4, 11, 28, 56) miRNA 15a/16-1 encoded within DLEU2 (4, 11, 28, 56) intron in 13q23 (4, 11, 28, 56)

9 13 33 92

31–33a 72–79a 97–114a 113–133a

TP53

5–12%

4–58

4–23b

21–90b

NOTCH1

10–14%

5–42

18–86b

15–34b

SF3B1

9–14%

2–86

5–43b

28–90b

ATM

11–26%

Other (e.g. FAT1, MYD88, POT1, and RPS15)

Significantly 8–40b reduced independently of del(11q) RPS15: reduced PFS

26–85b RPS15: reduced OS

References

The majority of clonal mutations (5, 6, 8, 10, 28, 31, are associated with del(17p) 36, 56, 73, 110) Mostly associated with U-CLL Mostly in U-CLL (82%) Frequently associated with trisomy 12 (6, 10, 28, 31, 36, 56) Found together with TP53 mutations (5, 6, 28, 31, 36) in some studies, but not in others ATM and del(11q) occur mostly (5, 6, 28, 31, in U-CLL 36, 56) RPS15 can be exclusive of TP53 mutations

(36, 52, 54, 73)

U-CLL: IGHV unmutated CLL; aIn previously untreated patients bAcross all lines of treatment in chemoimmunotherapy studies. CLL: chronic lymphocytic leukemia; OS: overall survival; PFS: progression-free survival; WT: wild type.

haematologica | 2018; 103(12)

1957

E. Campo et al.

rearrangements in a localized region of the genome, have also been identified in some patients with CLL,46-48 usually associated with TP53 and SETD2 mutations.6,49 Apart from TP53, the most frequent mutations associated with disease outcomes in CLL are found in the ATM, BIRC3, NOTCH1, and SF3B1 genes.6,31,50-53 These and other mutations have been associated with the development of high-risk disease, with a higher incidence of these mutations being found in fludarabine-refractory CLL than in untreated CLL.6,52,54-56 The impacts of these mutations on outcomes in CLL are outlined in Table 1 but the clinical value of each of them remains to be established.57

IGHV gene status Another important CLL feature that affects prognosis is the IGHV gene mutation status. The clinical course is generally more aggressive in patients with unmutated IGHV genes than in those with mutated IGHV genes.58,59 TP53 mutations may be found in both mutated and unmutated CLL, but are usually associated with unmutated CLL.56 Immunogenetic studies have recently revealed that approximately one third of patients with CLL carry quasiidentical or stereotyped B-cell receptors (BCR) and can be grouped into subsets that share clinico-biological features and outcome.57

What is TP53? Over 50% of human cancers carry TP53 gene mutations,60 and the importance of TP53 in tumor development is highlighted by the increased incidence of cancer before the age of 30 in patients with Li-Fraumeni syndrome, which results from germline mutations in the TP53 gene.61 TP53 encodes the tumor-suppressor protein p53, which has numerous cellular activities including regulation of the cell cycle and apoptosis, and promotion of DNA repair in response to cellular stress signals such as DNA damage.60,62,63 Following DNA damage, p53 triggers either apoptosis or G1 cell-cycle arrest until the cell has completed DNA repair processes, thereby preventing replication of potentially harmful genetic abnormalities.62

What are the different types of TP53 aberration and how do they affect p53 function and pathogenicity? TP53 aberrations can arise through deletion of the TP53 locus on chromosome 17 (17p13.1) or gene mutations including missense mutations, insertions or deletions (indels), nonsense mutations or splice-site mutations. Gene mutations are heavily concentrated in the DNAbinding domain, encoded by exons 4â&#x20AC;&#x201C;8 of the TP53 gene, but mutations can also appear in the oligomerization domain or C-terminal domain.33,63-65 del(17p) and/or TP53 mutations in various combinations can result in the loss of wildtype p53 function in CLL (Figure 1).12,28,29,31,33 Six â&#x20AC;&#x2DC;hotspotâ&#x20AC;&#x2122; codons in particular (codons 175, 245, 248, 249, 273, and 282) are affected at elevated frequency.33,63,66 This is in line with a disease-specific TP53 mutational profile in CLL.66 The most commonly found mutations in TP53 are missense mutations in the coding region of TP53, which lead to an amino acid change in the p53 protein and account for approximately 75% of TP53 mutations identified.33,60,63 Missense mutations may result in expression of a mutated p53 protein that cannot activate the p53 tumor-suppressive transcriptional response, have dominant-negative effects over any remaining wildtype p53, and/or could gain oncogenic functions independent of wildtype p53,5,33,60,64 illustrating their pathogenic and prognostic impact even if occurring in one copy (mono-allelic) of TP53 with retention of a potentially functional allele.32 In contrast, del(17p), frameshift mutations, indels, nonsense mutations, and splice-site mutations result in loss of functional p53, and although functional p53 may still be expressed in the presence of a second wildtype allele, this has not been proven to diminish the adverse prognostic impact of such abnormalities (Figure 2).33 Based on data obtained from Sanger sequencing, approximately 80% of patients harboring del(17p) also carry TP53 mutations in the second allele.8,30,67 Overall, del(17p) associated with TP53 mutations is the most common abnormality affecting the TP53 gene in CLL, accounting for approximately two-thirds of cases.8,10,30,33 The

Figure 1. Loss of wildtype (wt) p53 function in chronic lymphocytic leukemia can occur as a result of del(17p) and/or TP53 mutations.12,28,29,31,33 The most common cause of TP53 aberrations is the result of a combination of TP53 mutation and del(17p), which accounts for up to two-thirds of all TP53 aberrations.

1958

haematologica | 2018; 103(12)

TP53 aberrations in CLL

remaining cases with TP53 aberration carry either gene mutation(s) or sole del[17p].28,29,31,33 A TP53 mutation can be accompanied by a copy-number neutral loss of heterozygosity of the second TP53 allele.5,6,30,31

Clonality and clonal evolution Individual cancer samples are genetically heterogeneous and contain clonal and subclonal populations.68,69 These populations may be in equilibrium, with the relative proportions of each subclone remaining stable, or may undergo evolution, with some subclones emerging as dominant.50 While most untreated CLL, and a minority of treated CLL, maintain stable clonal equilibrium, treatment may shift the architecture in favor of one or more aggressive subclones.50 This clonal evolution is a key feature of cancer progression and relapse, with tumors likely evolving through competition and interactions between genetically diverse clones (Figure 3).5 In CLL, clonal evolution after treatment or at the time of relapse has been identified as ‘the rule, not the exception’.5,70 In a study by Landau et al.,5 47 out of 49 patients with CLL had clonal evolution at the time of relapse. Importantly, chemoimmunotherapy pressure is thought to lead to clonal evolution, most prominently for TP53 aberrant subclones.71 TP53 aberrations are indeed strongly associated with clonal evolution in CLL.44,72,73 TP53 aberrations are less frequent at diagnosis (Table 1), while 40–50% of cases with advanced or therapy-refractory CLL harbor aberrations, highlighting the need to re-assess TP53 status before each line of treatment because the clones could expand at relapse and/or during disease progression.8,10,56,74 Single or multiple minor subclones harboring TP53 mutations may

be present before therapy or may develop during relapse at any stage. These TP53-mutant minor subclones are often present at very low frequencies that may be undetectable by Sanger sequencing and are highly likely to expand to dominant clones under the selective pressure of chemoimmunotherapy.12,31,51

How do we test for and report TP53 aberrations? Techniques frequently used for assessing TP53 status in CLL include FISH for del(17p), Sanger sequencing, and next-generation sequencing for TP53 mutations (Table 2).27,35,74,75 As TP53 mutations are associated with a poor prognosis independently of the presence of del(17p),7 it is important to assess for TP53 mutation status using a sequencing technique.27,35

Sequencing of the TP53 gene TP53 sequencing should cover exons 4–10 (corresponding to the DNA binding domain at codons 100–300 and the oligomerization domain at codons 323–365) at a minimum. Sequencing of the whole coding region (exons 2– 11) and adjacent splice sites is highly recommended using either bidirectional Sanger sequencing or next-generation sequencing, as studies of the latter have shown that variants can also occur in exons outside the DNA binding domain although their frequency is low (Figure 2).35 Sanger sequencing is a widely and routinely used technique to assess TP53 status in CLL in clinical practice. The technique provides a relatively simple, accessible sequencing approach, but is time-consuming and lacks sensitivity for detecting minor subclones harboring TP53 mutations, with a detection limit for mutated alleles of 10–

Figure 2. TP53 gene organization and distribution of mutations by codon.63,121,122 The TP53 gene is located at the p13.1 locus on the short arm of chromosome 17 and comprises 11 exon sequences that encode for the p53 protein. While the majority of gene mutations cluster within the DNA-binding domain (codons 100–300, exons 4–8), gene mutations have been detected in almost every codon. Sequencing should, therefore, cover the DNA-binding domain and oligomerization domain as a minimum (exons 4–10), but sequencing of the whole coding region (exons 2–11) is highly recommended.

haematologica | 2018; 103(12)

1959

E. Campo et al.

20%.27,29,35,76-78 As stated earlier, minor TP53-mutant subclones that may be missed by Sanger sequencing also appear to carry the same unfavorable prognostic impact as clonal TP53 mutations.7,12,31,51,69 Next-generation sequencing technologies include targeted next-generation sequencing, which has good correlation with Sanger sequencing in comparison studies12,28,31,35,75,78 and detects low-frequency mutations below the threshold for Sanger sequencing.38,79-81 The sensitivity threshold varies depending on a number of variables, including the hardware, methods used for testing and the analytical pipeline, and should be defined by each labora-

tory using standardized criteria or equivalent medical laboratory standards.35,75 Reports of TP53 mutational analysis should always include the type of analysis and methodology used, the exons analyzed, the limit of detection, and coverage for next-generation sequencing (median and ≥99% minimum).35 Low-level TP53 mutations occurring in <10% of DNA that may be subject to further clonal selection are also identified by next-generation sequencing. Recent recommendations on the methodological approaches for TP53 mutation analysis from The TP53 Network of ERIC35 concluded that the clinical importance of mutations in

Figure 3. An example of possible clonal evolution scenarios across the course of disease in chronic lymphocytic leukemia.28,50 Genomic diversification of CLL occurs through sequential acquisition of gene mutations, represented by clones of different colors. Treatment may reduce or eliminate the incumbent clone, shifting the clonal architecture in favor of one or more aggressive subclones. Different therapies may preferentially provide selective advantages for different mutations. For example, the red circles are TP53-mutated clones, which have been selected for by chemotherapy, whereas the turquoise clones would have acquired resistance to the targeted therapy.

Table 2. Comparison of methods for the detection of TP53 aberrations.

Method

Description

Advantages

Disadvantages

FISH

FISH uses fluorescent DNA probes to target specific chromosomal locations within the nucleus that can be detected by fluorescence microscopy

Sanger sequencing

Sanger sequencing uses selective incorporation of chain-terminating dideoxynucleotides by DNA polymerase during DNA replication, thereby creating sequences of various lengths, which are then separated by size to derive the DNA sequence NGS covers a range of technologies that allow high-throughput sequencing of millions or billions of DNA strands in parallel

• Rapid evaluation of fresh cells or paraffin-embedded interphase nuclei • Widely used in routine clinical practice • High specificity • Simple and widely available • Provides direct information on mutation type • Can produce relatively long read lengths • High specificity (~93%)

• Can only detect genetic defects (111-114) recognized by a specific probe • Cannot detect copy-neutral loss of heterozygosity • Relatively time-consuming (27, 29, 35, 76-78) • Limited sensitivity (usually approximately 10–20% of mutant alleles) • Limited throughput • Upfront cost of instrumentation, (6, 27, 29, 31, although some NGS sequencers 35, 76-78) are now cheaper than capillary sequencers (for Sanger) • High throughput needed for cost-effectiveness • High cost (43, 44, 48, • Cannot detect balanced 115-117) rearrangements i.e. translocations, balanced insertions, inversions

NGS

Genomic arrays

A technique that allows high-resolution, genome-wide screening of segmental copy number aberrations

• High and customizable sensitivity • Simultaneous analysis of large numbers of genes • No PCR with some platforms • Very high specificity (100%) • Provides high resolution, genome-wide information • Can detect genomic imbalances (deletions/amplifications) and copy-neutral loss of heterozygosity

References

FISH: fluorescence in situ hybridization; NGS: next-generation sequencing; PCR: polymerase chain reaction.

1960

haematologica | 2018; 103(12)

TP53 aberrations in CLL

<10% of alleles within the cancer cell population remains an unresolved issue and there is not enough evidence to make therapeutic decisions based on mutations undetectable by Sanger sequencing. This conclusion should be always stated when reporting variants present at a frequency of below 10%. Outside of the context of research, determination of TP53 status at diagnosis may not be required; initiation of first-line treatment can be deferred until patients have symptomatic active disease irrespective of TP53 status.82-85

Naming, reporting, and pathogenicity of mutations The consistent use of nomenclature in managing DNA sequence mutations is essential for concise communication of diagnostic testing and genetic risk assessment.60 In clinical practice, aberrations are often referred to as mutations, and are referred to as such in clinical reports. However, one must note that the more accurate technical term is â&#x20AC;&#x2DC;variantâ&#x20AC;&#x2122;. It is recommended that mutations are named according to the Human Genome Variation Society guidelines, or according to American College of Medical Genetics guidelines on mutations and mutation pathology in the case of germline mutations.86,87 Description of mutations at the DNA level using the stable Locus Reference Genomic reference sequence is recommended to enable comparison across studies and databases.88 The pathogenicity of more frequent TP53 mutations is well known, with functional analyses demonstrating that all TP53 hot-spot mutations result in a clear loss of p53 activity.5,60 The pathogenicity of some less frequently occurring TP53 mutations may be less clear, particularly in the case of missense mutations which can have varied functional consequences.5,33,60,64 A combination of factors are considered when determining whether a mutation is likely to be pathogenic, including whether the mutation results in an amino acid

change, whether the mutation is found in a conserved region of the genome or hotspot region, and whether there is a predicted functional effect of the amino acid splicing change on the protein or post-translational modification.60 Pathogenicity assessments should be performed by experienced diagnosticians, follow standardized procedures, and be documented. TP53 locus-specific databases are available and are important tools for analyzing and assessing the pathogenicity of TP53 mutations. These are the IARC TP53 database (http://p53.iarc.fr/), the TP53 website (http://p53.fr/), and the Seshat online software (http://p53.fr/tp53-database/seshat). The Seshat online software, for example, provides a quality check of the mutation nomenclature, generates a description of the mutation, and assesses the pathogenicity of each mutation with the use of specific algorithms. Structural and functional information for each mutation is also produced.35,89

Clinical implications of TP53 aberrations Patients with del(17p) and/or TP53 mutations usually respond poorly to the standard first-line chemoimmunotherapy, and have an aggressive disease course.8-12 In the CLL8 study comparing first-line treatment with fludarabine plus cyclophosphamide or fludarabine plus cyclophosphamide with rituximab, TP53 aberrations were found to be the strongest prognostic markers in multivariable analyses and were associated with markedly reduced progression-free survival and overall survival (Figure 4).10 Both in front-line and relapsed/refractory settings, treatment with bendamustine plus rituximab was also shown to be associated with low response rates and poor survival outcomes in patients with CLL harboring TP53 aberrations.90 Consequently, chemoimmunotherapy is no longer considered standard therapy for patients with TP53 aberrations. Until recently, the anti-CD52 antibody alemtuzumab was considered to be the only effective agent available for patients with TP53 aberrations, despite an

Figure 4. Progression-free and overall survival according to TP53 status in the CLL8 study.10 Re-published with permission from The American Society of Hematology, from: Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Stilgenbauer S et al. Blood. 2014;123(21):3247-3254; permission conveyed through Copyright Clearance Center, Inc. FC: fludarabine plus cyclophosphamide; FCR: fludarabine plus cyclophosphamide plus rituximab; mut: mutated; OS: overall survival; PFS: progression-free survival; WT: wild-type.

haematologica | 2018; 103(12)

1961

E. Campo et al.

Table 3. Overview of clinical evidence from phase 2/3 trials for novel treatments in patients with TP53 aberrations. Study/treatment Population TP53 aberrations Overall response PFS in del(17p)/ OS in del(17p)/ Sponsors at baseline in del(17p)/TP53 TP53 mutated TP53 mutated mutated population population population RESONATE-17: A phase 2, open-label, multicenter study of ibrutinib in patients with R/R CLL/SLL and del(17)p Ibrutinib 420 mg OD NCT01744691

Adult patients with previously treated del(17p) CLL or SLL (n=144) Median age (range): 64 (57-72) ECOG score: 0: 49 (34%) ≥1: 95 (66%)

del(17p) 144/144 (100%) TP53 mutations 107/116 (92%)

Safety (experimental arm, overall population)

Reference

ORR in del(17p) patients was 64% by independent review and 83% by investigator assessment (prespecified primary analysis, median 11.5 months follow-up)

Median PFS (investigatorassessed) not reached at a median follow-up of 11.5 months (prespecified primary analysis)

Median OS not reached at 11.5 months (prespecified primary analysis)

Grade 3–5 AE occurring (21) in >5% of patients: Neutropenia (18%) Pneumonia (13%) Hypertension (13%) Anemia (10%) Thrombocytopenia (8%) Atrial fibrillation (6%) (24-month extended analysis)

Median PFS in del(17p) and/or TP53 patients not reached at 19 months follow-up in patients treated with ibrutinib

Median OS in del(17p) or TP53 patients not reached at 19 months follow-up in patients treated with ibrutinib

Grade 3–5 AE occurred (18, 99) in 56% of patients treated with idelalisib + rituximab and 48% treated with placebo + rituximab Grade 3–5 AE occurred in >5% of patients: Idelalisib + rituximab Neutropenia (34%) Thrombocytopenia (10%) Placebo + rituximab Neutropenia (22%) Thrombocytopenia (16%) Anemia (14%) (overall study population)

Pharmacyclics LLC. Janssen Research & Development, LLC

Median prior regimens (IQR): 2 (1–3)

RESONATE: a phase 3, open-label, multicenter study of ibrutinib versus ofatumumab in patients with previously treated CLL/SLL Ibrutinib 420 mg OD versus ofatumumab

Adult patients with del(17p) R/R CLL/SLL 127/391 (32%) (n=391) Ibrutinib arm. Median age (range): 67 (30–86) ECOG score: 0: 79 (41%) 1: 116 (59%) Median prior regimens: 3 (1–12) Ofatumumab arm. Median age (range): 67 (37–88) ECOG score: 0: 80 (41%) 1:116 (59%) Median prior regimens: 2 (1–13)

ORR in del(17p) patients treated with ibrutinib: 89%

Older patients (≥65 years) with previously untreated CLL or SLL (n=64) Median age (range): 71 (65–90) ECOG score/Karnofsky status: not reported Median prior regimens: 0

ORR in either del(17p) or TP53 mutation: 100%

Median PFS in del(17p) and/or TP53 patients not reached after a median 22.4 months on treatment

Median OS in del(17p) and/or TP53 patients not reached after a median of 22.4 months on treatment

Grade 3–5 AE occurred (22) in 89.1% of patients. Grade 3–5 AE occurred in >5% of patients: Diarrhea and/or colitis (42%) Pneumonia (19%) (overall study population)

ORR in del(17p) and/or TP53 patients treated with Idelalisib plus rituximab: 77%

Median PFS in del(17p) and/or TP53 patients treated with idelalisib plus rituximab: not reached

Not reported in del(17p) and/or TP53 patients

ORR in del(17p) and/or TP53 patients treated with rituximab: 15% (second interim analysis, median exposure 5 months with idelalisib, 4 months with rituximab)

Median PFS in del(17p) and/or TP53 patients treated with rituximab: 4.0 months (second interim analysis)

Grade 3–5 AE occurred (19, 23) in 56% of patients treated with idelalisib + R and 48% treated with placebo + rituximab Grade 3–5 AE occurred in >5% of patients: Idelalisib + rituximab arm: Neutropenia (34%) Thrombocytopenia (10%) Placebo + rituximab arm: Neutropenia (22%) Thrombocytopenia (16%) Anemia (14%) (overall study population)

NCT01578707 Pharmacyclics LLC. Janssen Research & Development, LLC

Study 101-08: a phase 2 study of idelalisib plus rituximab in elderly patients with untreated CLL or SLL Idelalisib 150 mg BD plus rituximab NCT01203930 Gilead Sciences

Study 116: a randomized, double-blind, placebocontrolled study of idelalisib in combination with rituximab for previously treated CLL Idelalisib 150 mg BD plus rituximab versus placebo plus rituximab NCT01539512 Gilead Sciences

1962

del(17p) only: 2/64 (3.1%)

ORR in del(17p) patients treated with ofatumumab: 20% (median follow-up 19 months)

Median PFS in del(17p) and/or TP53 patients 5.8 months in patients treated with ofatumumab Patients with both del17p and TP53 mutation (n=38) had worse PFS compared with patients with neither of these abnormalities (n=68) (P=0.0381) at a median follow-up of 19 months

TP53 mutation only: 3/63 (4.7%) Either del(17p) or TP53 mutation: 9/64 (14.1%)

Median OS not reported in del(17p) or TP53 patients treated with ofatumumab

Both del(17p) and TP53 mutation: 4/64 (6.3%)

Adult patients with R/R del(17p) and/or CLL not eligible TP53 mutations for cytotoxic agents (n=220); Idelalisib + rituximab PD within 24 months 46/110 (42%) of last treatment Idelalisib + rituximab Rituximab: Median age (range): 50/110 (45%) 71 (48–90) ECOG score/Karnofsky status: not reported Median prior regimens: 3 (1–12) Placebo + rituximab arm. Median age (range): 71 (47–92) ECOG score/Karnofsky status: not reported Median prior regimens: 3 (1–9)

continued on the next page

haematologica | 2018; 103(12)

TP53 aberrations in CLL continued from the previous page

Study/treatment Sponsors

Population

TP53 aberrations at baseline

Overall response in del(17p)/TP53 mutated population

PFS in del(17p)/ TP53 mutated population

Study 115: a randomized, double-blind and placebo-controlled study of idelalisib in combination with bendamustine and rituximab (BR) for previously treated CLL Idelalisib 150 mg BD plus BR versus BR

Adult patients with R/R CLL (n=416); PD within 36 months of last treatment Idelalisib + BR Median age (range): 62 (56–69) ECOG score/ Karnofsky status: not reported Median prior regimens: 2 (1–4) Placebo plus BR Median age (range): 64 (56–70) ECOG score/ Karnofsky status: not reported Median prior regimens: 2 (1–4)

del(17p) and/or TP53 mutations

ORR in del(17p) patients treated with idelalisib + BR: 22/38 (58%)

Median PFS in del(17p) and/or TP53 patients treated with idelalisib + BR: 11.3 months

NCT01569295 Gilead Sciences

Study 119: a phase 3, randomized, controlled study evaluating the efficacy and safety of idelalisib (GS-1101) in combination with ofatumumab for previously treated CLL

Adult patients with R/R CLL (n=261); PD within 24 months of last treatment Idelalisib plus ofatumumab Median age (range): 68 (61–74) Karnofsky status: Idelalisib 150 mg BD + 80 (80–90) ofatumumab versus Median prior ofatumumab alone regimens: 3 (2–4) Ofatumumab alone NCT01659021 Median age (range): 67 (62–74) Gilead Sciences Karnofsky status: 80 (80–90) Median prior regimens: 3 (2–5)

A phase 2 open-label study of the efficacy of ABT-199 (GDC-0199) in subjects with R/R or previously untreated CLL harboring the 17p deletion Venetoclax 400 mg OD NCT01889186

Adult patients with R/R CLL with del(17p) (n=107) Median age (range): 67 (37–85) ECOG score n (%): 0: 42 (39%) 1: 56 (52%) 2: 9 (8%) Median prior regimens (IQR): 2 (1–4)

Idelalisib + BR: 69/207 (33%) BR: 68/209 (33%)

ORR in del(17p) patients treated with BR: (9/40) 23%

OS in del(17p)/ TP53 mutated population

Median OS in del(17p) and/or TP53 patients treated with idelalisib + BR: Median PFS in del(17p) not reached at a and/or TP53 patients median follow-up treated with BR: 8.3 months of 14 months Median OS in del(17p) and/or TP53 patients treated with BR: 20.3 months

del(17p) and/or TP53 mutations Idelalisib plus ofatumumab: 70/174 (40%)

ORR in del(17p) and/or TP53 patients treated with idelalisib plus ofatumumab: not reported

Ofatumumab: 33/87 (38%)

ORR in del(17p) and/or TP53 patients treated with ofatumumab: not reported

del(17p) 100%

ORR in del(17p) patients: 79.4% (independent review committee assessment)

TP53 mutated 60/107 (72%)

AbbVie Genentech, Inc.

Median PFS in del(17p) and/or TP53 patients treated with idelalisib plus ofatumumab: 15.5 months Median PFS in del(17p) and/or TP53 patients treated with ofatumumab: 5.8 months

Median PFS in del(17p) patients: not reached at a median follow-up of 12.1 months

Median OS in del(17p) and/or TP53 patients treated with idelalisib + ofatumumab: 25.8 months Median OS in del(17p) and/or TP53 patients treated with ofatumumab: 19.3 months

Median OS in del(17p) patients: not reached at median follow-up of 12.1 months

Safety (experimental arm, overall population)

Reference

Grade 3–5 AE (118) occurring in ≥5% of patients: Idelalisib + BR: Neutropenia (60%) Febrile neutropenia (23%) Placebo + BR: Neutropenia (47%) Thrombocytopenia (13%) (overall study population)

Grade 3–5 TEAE (20, 96) occurring in ≥5% of patients treated with idelalisib plus ofatumumab: Neutropenia (34%) Diarrhea (20%) Pneumonia (16%) Anemia (14%) Febrile neutropenia (12%) Thrombocytopenia (11%) Hypokalemia (8%) Pyrexia (7%) Dyspnea (6%) Hypertension (5%) Dehydration (5%) Fatigue (5%) Grade 3–5 TEAE occurring in ≥5% of patients treated with ofatumumab: Neutropenia (16%) Pneumonia (8%) Thrombocytopenia (7%) Anemia (6%) Fatigue (5%) (overall study population) Grade 3–5 AE in (24, 119) del(17p) patients occurring in 76% of patients Grade 3–5 AE occurring in ≥5% of patients: Neutropenia (40%) Anemia (18%) Thrombocytopenia (15%) Autoimmune hemolytic anemia (7%) Febrile neutropenia (5%) Pneumonia (5%) Immune thrombocytopenic purpura(5%) Tumor lysis syndrome (5%) Leukopenia (5%)

continued on the next page

haematologica | 2018; 103(12)

1963

E. Campo et al. continued from the previous page

Study/treatment Sponsors

Population

TP53 aberrations at baseline

Overall response in del(17p)/TP53 mutated population

PFS in del(17p)/ TP53 mutated population

OS in del(17p)/ TP53 mutated population

Safety (experimental arm, overall population)

MURANO: a randomized, open-label, phase 3 trial evaluating venetoclax plus rituximab versus bendamustine plus rituximab in R/R CLL

Adult patients with R/R CLL (n=389): Venetoclax plus rituximab (n=194) Median age (range): 64.5 (28–83) years ECOG score n (%) 0: 111 (57.2%) 1: 82 (42.3%) 2: 1 (0.5%) Number of prior therapies n (%): 1: 111 (57.2%) 2: 57 (29.4%) 3: 22 (11.3%) >3: 4 (2.1%) BR (n=195) Median age (range): 66 (22–85) years ECOG score n (%) 0: 108 (55.7%) 1: 84 (43.3%) 2: 2 (1.0%)

del(17p) only Venetoclax plus rituximab: 24 (14%) BR 18 (11.4%)

Not reported

del(17p) Median PFS not reached with venetoclax plus rituximab at 2-year follow-up Median PFS 15.4 months with BR

Not reported

Grade 3–4 AE in patients (120) receiving venetoclax plus rituximab: 82.0% Grade 3–4 AE in patients receiving BR 70.2%

NCT02005471 AbbVie Genentech, Inc.

TP53 mutation only Venetoclax plus rituximab: 19 (11.1%) BR 23 (14.6%) del(17p) and TP53 mutated Venetoclax plus rituximab: 22 (12.9%)

Reference

TP53 mutation Median PFS not reached with venetoclax plus rituximab at 2-year follow-up Median PFS 12.9 months with BR

BR 22 (13.9%)

Number of prior therapies n (%): 1: 117 (60.0%) 2: 43 (22.1%) 3: 34 (17.4%) >3: 1 (0.5%) AE: adverse events; ALT: alanine aminotransferase; AST: aspartate aminotransferase; BD: twice daily; BR: bendamustine plus rituximab; CLL: chronic lymphocytic leukemia; ECOG: Eastern Cooperative Oncology Group; HR: hazard ratio; IQR: interquartile range; OD: once daily; ORR: overall response rate; OS: overall survival; PD: progressive disease; PFS: progression-free survival; R/R: relapsed/refractory; SLL, small lymphocytic leukemia; TEAE: treatment emergent adverse events.

overall limited efficacy and a high risk of opportunistic infectious complications.16 Allogeneic hematopoietic stem cell transplantation is a potentially curative therapeutic option for patients with TP53 aberrations, but is only feasible for highly selected younger, physically fit patients and those who have obtained a good therapeutic response.13,15,17

Therapies with p53-independent mechanisms of action Recent developments in the treatment options for patients with CLL harboring TP53 aberrations include small-molecule kinase inhibitors that target the BCR pathway (ibrutinib and idelalisib)18-22,26 and the anti-apoptotic protein BCL2 (venetoclax).24,91-93 Ibrutinib is an inhibitor of Bruton tyrosine kinase,94,95 whereas idelalisib is an inhibitor of the PI3K p110δ isoform,19,96 both of which are involved in mediating intracellular signaling from several receptors including the BCR. Venetoclax is a BH3-mimetic inhibitor of BCL2, an anti-apoptotic protein with constitutively elevated expression in CLL.92,97 An overview of the clinical evidence from phase 2/3 trials for these treatments in patients with CLL harboring TP53 aberrations is shown in Table 3. The studies were carried out in varying patient populations, but overall, these novel therapies produced responses and favorable survival times in a high proportion of patients harboring TP53 aberrations and represent a significant advance for this high-risk population compared to chemoimmunotherapy regimes.18-26 It is impor1964

tant to note that such therapies achieved similar responses in patients with relapsed or refractory CLL, irrespective of risk factors that are associated with poorer responses to chemoimmunotherapy.92,98-100 Given the improvements seen with these therapies, accelerated approval programs have made the therapies available for CLL treatment in the clinic. Currently in Europe, ibrutinib is licensed as monotherapy for first-line treatment and for relapsed/refractory patients with CLL, or in combination with bendamustine plus rituximab in the relapsed/refractory setting.94 Idelalisib is indicated in combination with an anti-CD20 monoclonal antibody (rituximab or ofatumumab) for relapsed/refractory CLL therapy, and as first-line therapy in patients with del(17p)/TP53 mutations not suitable for other therapies.96 Venetoclax is currently licensed in Europe for patients with relapsed/refractory CLL in whom both chemoimmunotherapy and a BCR inhibitor have failed, or for patients with del(17p) or a TP53 mutation who are not suitable for BCR inhibitors or in whom BCR inhibitor treatment has failed.97 Although limited data are available for all these agents in the treatment-naïve setting, the approvals as first-line therapy reflect the high level of unmet need for patients with TP53 aberrations. Moreover, the development of these novel therapies has produced a change in therapeutic goals. In particular, frail patients with progressive CLL can now be treated with the aim of effectively controlling the disease, whereas previously palliative care would have been the only option.19 haematologica | 2018; 103(12)

TP53 aberrations in CLL

It has also become evident that patients may develop resistance to these targeted therapies. For example, mutations in the BTK and PLCG2 genes have been associated with resistance to ibrutinib, while upregulation of antiapoptotic BCL2 family members has been associated with resistance to venetoclax.101-104 Mechanisms of resistance to idelalisib have not yet been fully characterized; because idelalisib inhibits the PI3K p110δ isoform, resistance may theoretically involve upregulation of other PI3K isoforms.105 However, in a whole-exome sequencing analysis of 13 patients with CLL who had progressed while on idelalisib plus anti-CD20 treatment in three phase 3 trials, none of the patients had recurrent progression-associated mutations in the PI3K pathway or other related pathways.71 The optimal sequencing of these targeted therapies is currently unknown, but observational studies suggest that patients who discontinue a BCR pathway inhibitor due to toxicity may benefit from an alternative BCR pathway inhibitor. Conversely, those patients who progress under BCR inhibitor therapy fare better with venetoclax than an alternative BCR inhibitor.106,107 Following progression on one or more therapies, allogeneic hematopoietic stem cell transplantation also remains a valid option, especially because these novel therapies may render patients more fit for this procedure. It is important to note that, until recently, treatment guidelines for patients with TP53 aberrations were based on retrospective analyses and subgroup analyses. Patients with TP53 aberrations are still defined as a high-risk group, despite the development of these newer therapies, but their outcome has greatly improved in recent years. More long-term data and dedicated trials of these new therapies in this population are still needed to understand the long-term prognosis. Nevertheless, these therapies (as monotherapy or in combination) have become the mainstay of treatment in patients with CLL harboring TP53

References 1. Eichhorst B, Fink AM, Bahlo J, et al. First-line chemoimmunotherapy with bendamustine and rituximab versus fludarabine, cyclophosphamide, and rituximab in patients with advanced chronic lymphocytic leukaemia (CLL10): an international, openlabel, randomised, phase 3, non-inferiority trial. Lancet Oncol. 2016;17(7):928-942. 2. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomised, open-label, phase 3 trial. Lancet. 2010;376 (9747):1164-1174. 3. Howard DR, Munir T, McParland L, et al. Results of the randomized phase IIB ARCTIC trial of low-dose rituximab in previously untreated CLL. Leukemia. 2017;31(11): 2416-2425. 4. Dohner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med. 2000;343(26):1910-1916.

haematologica | 2018; 103(12)

mutations or del(17p), as well as in relapsed or refractory CLL and have led to recent updates in treatment guidelines.34,35,84,85,108,109

Future considerations As evidence from clinical trials demonstrates, it is important to test accurately for TP53 aberrations (both del[17p] and TP53 mutations) before each line of treatment, thus allowing for appropriate treatment decisions to optimize patients’ outcomes. Accurate identification of TP53 mutations demands standardization in sequencing technologies and pathogenicity assessments. Independent evaluation within prospective clinical trials is still required to determine the clinical impact of minor subclonal mutations (<10%). Similarly, given the continuing evolution of therapeutic agents in CLL, it is important to continue to evaluate TP53 aberrations as new therapeutic alternatives become available. While allogeneic hematopoietic stem cell transplantation remains the only curative treatment option for patients with CLL harboring TP53 aberrations, the recent approvals of ibrutinib, idelalisib, and venetoclax have provided significantly improved outcomes for this high-risk group of patients. Acknowledgments Editorial assistance was provided by Sarah Etheridge, PhD (ApotheCom, London, UK). The editorial assistance was funded by Gilead Sciences Europe, Ltd who had no input into the content of this work. EC is supported by grants from Instituto de Salud Carlos III (PMP15/00007, CIBERONC and ERA-NET TRANSCAN initiative (TRS-2015-00000143) AC15/00028. SP has been supported by the MEYS CZ project CEITEC 2020 (LQ1601) and MH CR grant AZV 15-31834A. SS was supported by the DFG SFB 1074 project B1 and B2. AS was supported by the NIHR Oxford Biomedical Research Centre. The views expressed are those of the authors and do not reflect the views of the United Kingdom’s Department of Health.

5. Landau DA, Tausch E, Taylor-Weiner AN, et al. Mutations driving CLL and their evolution in progression and relapse. Nature. 2015;526(7574):525-530. 6. Puente XS, Bea S, Valdes-Mas R, et al. Noncoding recurrent mutations in chronic lymphocytic leukaemia. Nature. 2015;526 (7574):519-524. 7. Zenz T, Kröber A, Scherer K, et al. Monoallelic TP53 inactivation is associated with poor prognosis in chronic lymphocytic leukemia: results from a detailed genetic characterization with long-term follow-up. Blood. 2008;112(8):3322-3329. 8. Zenz T, Eichhorst B, Busch R, et al. TP53 mutation and survival in chronic lymphocytic leukemia. J Clin Oncol. 2010;28 (29):4473-4479. 9. Gonzalez D, Martinez P, Wade R, et al. Mutational status of the TP53 gene as a predictor of response and survival in patients with chronic lymphocytic leukemia: results from the LRF CLL4 trial. J Clin Oncol. 2011;29(16):2223-2229. 10. Stilgenbauer S, Schnaiter A, Paschka P, et al. Gene mutations and treatment outcome in

11.

12.

13.

14.

chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):32473254. International CLL-IPI Working Group. An international prognostic index for patients with chronic lymphocytic leukaemia (CLLIPI): a meta-analysis of individual patient data. Lancet Oncol. 2016;17(6):779-790. Rossi D, Khiabanian H, Spina V, et al. Clinical impact of small TP53 mutated subclones in chronic lymphocytic leukemia. Blood. 2014;123(14):2139-2147. Sorror ML, Storer BE, Sandmaier BM, et al. Five-year follow-up of patients with advanced chronic lymphocytic leukemia treated with allogeneic hematopoietic cell transplantation after nonmyeloablative conditioning. J Clin Oncol. 2008;26(30):49124920. Stilgenbauer S, Zenz T, Winkler D, et al. Subcutaneous alemtuzumab in fludarabinerefractory chronic lymphocytic leukemia: clinical results and prognostic marker analyses from the CLL2H study of the German Chronic Lymphocytic Leukemia Study Group. J Clin Oncol. 2009;27(24):3994-4001.

1965

E. Campo et al. 15. Dreger P, Döhner H, Ritgen M, et al. Allogeneic stem cell transplantation provides durable disease control in poor-risk chronic lymphocytic leukemia: long-term clinical and MRD results of the GCLLSG CLL3X trial. Blood. 2010;116(14):2438-2447. 16. Pettitt AR, Jackson R, Carruthers S, et al. Alemtuzumab in combination with methylprednisolone is a highly effective induction regimen for patients with chronic lymphocytic leukemia and deletion of TP53: final results of the National Cancer Research Institute CLL206 trial. J Clin Oncol. 2012;30(14):1647-1655. 17. Dreger P, Schnaiter A, Zenz T, et al. TP53, SF3B1, and NOTCH1 mutations and outcome of allotransplantation for chronic lymphocytic leukemia: six-year follow-up of the GCLLSG CLL3X trial. Blood. 2013;121(16): 3284-3288. 18. Byrd JC, Brown JR, O'Brien S, et al. Ibrutinib versus ofatumumab in previously treated chronic lymphoid leukemia. N Engl J Med. 2014;371(3):213-223. 19. Furman RR, Sharman JP, Coutre SE, et al. Idelalisib and rituximab in relapsed chronic lymphocytic leukemia. N Engl J Med. 2014;370(11):997-1007. 20. Jones JA, Robak T, Brown JR, et al. Efficacy and safety of idelalisib in combination with ofatumumab for previously treated chronic lymphocytic leukaemia: an open-label, randomised phase 3 trial. Lancet Haematol. 2017;4(3):e114-e126. 21. O'Brien S, Jones JA, Coutre SE, et al. Ibrutinib for patients with relapsed or refractory chronic lymphocytic leukaemia with 17p deletion (RESONATE-17): a phase 2, open-label, multicentre study. Lancet Oncol. 2016;17(10):1409-1418. 22. O'Brien SM, Lamanna N, Kipps TJ, et al. A phase 2 study of idelalisib plus rituximab in treatment-naive older patients with chronic lymphocytic leukemia. Blood. 2015;126(25): 2686-2694. 23. Sharman JP, Coutre SE, Furman RR, et al. Second interim analysis of a phase 3 study of idelalisib (ZYDELIG®) plus rituximab (R) for relapsed chronic lymphocytic leukemia (CLL): efficacy analysis in patient subpopulations with Del (17p) and other adverse prognostic factors. Blood. 2014;124(21):330. 24. Stilgenbauer S, Eichhorst B, Schetelig J, et al. Venetoclax in relapsed or refractory chronic lymphocytic leukaemia with 17p deletion: a multicentre, open-label, phase 2 study. Lancet Oncol. 2016;17(6):768-778. 25. Thornton P, Brown J, Hillmen P, et al. Efficacy of ibrutinib versus ofatumumab by cytogenetic and clinical subgroups in a phase 3 trial in patients with previously treated CLL/SLL. Hematol Oncol. 2015;31(S1):96-150. 26. Zelenetz AD, Barrientos JC, Brown JR, et al. Idelalisib or placebo in combination with bendamustine and rituximab in patients with relapsed or refractory chronic lymphocytic leukaemia: interim results from a phase 3, randomised, double-blind, placebo-controlled trial. Lancet Oncol. 2017;18(3):297311. 27. Pospisilova S, Gonzalez D, Malcikova J, et al. ERIC recommendations on TP53 mutation analysis in chronic lymphocytic leukemia. Leukemia. 2012;26(7):1458-1461. 28. Lazarian G, Tausch E, Eclache V, et al. TP53 mutations are early events in chronic lymphocytic leukemia disease progression and precede evolution to complex karyotypes. Int J Cancer. 2016;139(8):1759-1763. 29. Malcikova J, Pavlova S, Kozubik KS,

1966

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

Pospisilova S. TP53 mutation analysis in clinical practice: lessons from chronic lymphocytic leukemia. Hum Mutat. 2014;35(6):663-671. Malcikova J, Smardova J, Rocnova L, et al. Monoallelic and biallelic inactivation of TP53 gene in chronic lymphocytic leukemia: selection, impact on survival, and response to DNA damage. Blood. 2009;114(26):53075314. Nadeu F, Delgado J, Royo C, et al. Clinical impact of clonal and subclonal TP53, SF3B1, BIRC3, NOTCH1, and ATM mutations in chronic lymphocytic leukemia. Blood. 2016;127(17):2122-2130. Baran-Marszak F, Vidal V, Hormi M, et al. A retrospective analysis of 450 TP53 mutations in a real life cohort of CLL from the French Innovative Leukemia Organization (FILO) group. Blood. 2017; 130:1722. Leroy B, Ballinger ML, Baran-Marszak F, et al. Recommended guidelines for validation, quality control, and reporting of TP53 variants in clinical practice. Cancer Res. 2017;77(6):1250-1260. Hallek M, Cheson BD, Catovsky D, et al. iwCLL guidelines for diagnosis, indications for treatment, response assessment, and supportive management of CLL. Blood. 2018;131(25):2745-2760. Malcikova J, Tausch E, Rossi D, et al. ERIC recommendations on TP53 mutation analysis in chronic lymphocytic leukemia – UPDATE on interpretation and methodologies including next-generation sequencing. Leukemia. 2018;32(5):1070-1080. Lazarian G, Guieze R, Wu CJ. Clinical implications of novel genomic discoveries in chronic lymphocytic leukemia. J Clin Oncol. 2017;35(9):984-993. Delgado J, Salaverria I, Baumann T, et al. Genomic complexity and IGHV mutational status are key predictors of outcome of chronic lymphocytic leukemia patients with TP53 disruption. Haematologica. 2014;99 (11):e231-234. Rigolin GM, Saccenti E, Bassi C, et al. Extensive next-generation sequencing analysis in chronic lymphocytic leukemia at diagnosis: clinical and biological correlations. J Hematol Oncol. 2016;9(1):88. Haferlach C, Dicker F, Schnittger S, Kern W, Haferlach T. Comprehensive genetic characterization of CLL: a study on 506 cases analysed with chromosome banding analysis, interphase FISH, IgV(H) status and immunophenotyping. Leukemia. 2007;21 (12):2442-2451. Dicker F, Herholz H, Schnittger S, et al. The detection of TP53 mutations in chronic lymphocytic leukemia independently predicts rapid disease progression and is highly correlated with a complex aberrant karyotype. Leukemia. 2009;23(1):117-124. Brejcha M, Stoklasova M, Brychtova Y, et al. Clonal evolution in chronic lymphocytic leukemia detected by fluorescence in situ hybridization and conventional cytogenetics after stimulation with CpG oligonucleotides and interleukin-2: a prospective analysis. Leuk Res. 2014;38(2):170-175. Herling CD, Klaumunzer M, Rocha CK, et al. Complex karyotypes and KRAS and POT1 mutations impact outcome in CLL after chlorambucil-based chemotherapy or chemoimmunotherapy. Blood. 2016;128(3): 395-404. Ouillette P, Collins R, Shakhan S, et al. Acquired genomic copy number aberrations and survival in chronic lymphocytic leukemia. Blood. 2011;118(11):3051-3061.

44. Knight SJ, Yau C, Clifford R, et al. Quantification of subclonal distributions of recurrent genomic aberrations in paired pretreatment and relapse samples from patients with B-cell chronic lymphocytic leukemia. Leukemia. 2012;26(7):1564-1575. 45. Baliakas P, Jeromin S, Iskas M, et al. Cytogenetic complexity in chronic lymphocytic leukemia: definitions, associations with other biomarkers and clinical impact; a retrospective study on behalf of ERIC. Haematologica. 2017;102(Suppl 2):170. 46. Stephens PJ, Greenman CD, Fu B, et al. Massive genomic rearrangement acquired in a single catastrophic event during cancer development. Cell. 2011;144(1):27-40. 47. Bassaganyas L, Bea S, Escaramis G, et al. Sporadic and reversible chromothripsis in chronic lymphocytic leukemia revealed by longitudinal genomic analysis. Leukemia. 2013;27(12):2376-2379. 48. Salaverria I, Martin-Garcia D, Lopez C, et al. Detection of chromothripsis-like patterns with a custom array platform for chronic lymphocytic leukemia. Genes Chromosom Cancer. 2015;54(11):668-680. 49. Parker H, Rose-Zerilli MJ, Larrayoz M, et al. Genomic disruption of the histone methyltransferase SETD2 in chronic lymphocytic leukaemia. Leukemia. 2016;30(11):21792186. 50. Landau DA, Carter SL, Stojanov P, et al. Evolution and impact of subclonal mutations in chronic lymphocytic leukemia. Cell. 2013;152(4):714-726. 51. Malcikova J, Stano-Kozubik K, Tichy B, et al. Detailed analysis of therapy-driven clonal evolution of TP53 mutations in chronic lymphocytic leukemia. Leukemia. 2015;29(4): 877-885. 52. Messina M, Del Giudice I, Khiabanian H, et al. Genetic lesions associated with chronic lymphocytic leukemia chemo-refractoriness. Blood. 2014;123(15):2378-2388. 53. Quesada V, Ramsay AJ, Rodriguez D, Puente XS, Campo E, Lopez-Otin C. The genomic landscape of chronic lymphocytic leukemia: clinical implications. BMC Med. 2013;11(1):124. 54. Lode L, Cymbalista F, Soussi T. Genetic profiling of CLL: a 'TP53 addict' perspective. Cell Death Dis. 2016;14(7):e2042. 55. Clifford R, Louis T, Robbe P, et al. SAMHD1 is mutated recurrently in chronic lymphocytic leukemia and is involved in response to DNA damage. Blood. 2014;123(7):10211031. 56. Guieze R, Robbe P, Clifford R, et al. Presence of multiple recurrent mutations confers poor trial outcome of relapsed/refractory CLL. Blood. 2015;126(18):2110-2117. 57. Stamatopoulos K, Agathangelidis A, Rosenquist R, Ghia P. Antigen receptor stereotypy in chronic lymphocytic leukemia. Leukemia. 2017;31(2):282-291. 58. Hamblin TJ, Davis ZA, Oscier DG. Determination of how many immunoglobulin variable region heavy chain mutations are allowable in unmutated chronic lymphocytic leukaemia – long-term follow up of patients with different percentages of mutations. Br J Haematol. 2008;140(3):320-323. 59. Stamatopoulos B, Timbs A, Bruce D, et al. Targeted deep sequencing reveals clinically relevant subclonal IgHV rearrangements in chronic lymphocytic leukemia. Leukemia. 2017;31(4):837-845. 60. Leroy B, Anderson M, Soussi T. TP53 mutations in human cancer: database reassessment and prospects for the next decade. Hum Mutat. 2014;35(6):672-688.

haematologica | 2018; 103(12)

TP53 aberrations in CLL 61. Malkin D, Li FP, Strong LC, et al. Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science. 1990;250(4985):1233-1238. 62. Bieging KT, Mello SS, Attardi LD. Unravelling mechanisms of p53-mediated tumour suppression. Nat Rev Cancer. 2014;14(5):359-370. 63. Pfister NT, Prives C. Transcriptional regulation by wild-type and cancer-related mutant forms of p53. Cold Spring Harbor Perspect Med. 2017;7(2). 64. Muller PA, Vousden KH. p53 mutations in cancer. Nat Cell Biol. 2013;15(1):2-8. 65. Soussi T, Wiman KG. TP53: an oncogene in disguise. Cell Death Differ. 2015;22(8):12391249. 66. Zenz T, Vollmer D, Trbusek M, et al. TP53 mutation profile in chronic lymphocytic leukemia: evidence for a disease specific profile from a comprehensive analysis of 268 mutations. Leukemia. 2010;24(12):20722079. 67. Rossi D, Cerri M, Deambrogi C, et al. The prognostic value of TP53 mutations in chronic lymphocytic leukemia is independent of del17p13: implications for overall survival and chemorefractoriness. Clin Cancer Res. 2009;15(3):995-1004. 68. Purroy N, Wu CJ. Coevolution of leukemia and host immune cells in chronic lymphocytic leukemia. Cold Spring Harbor Perspect Med. 2017;7(4):a026740. 69. Rossi D, Rasi S, Spina V, et al. Integrated mutational and cytogenetic analysis identifies new prognostic subgroups in chronic lymphocytic leukemia. Blood. 2013;121(8): 1403-1412. 70. Ljungstrom V, Cortese D, Young E, et al. Whole-exome sequencing in relapsing chronic lymphocytic leukemia: clinical impact of recurrent RPS15 mutations. Blood. 2016;127(8):1007-1016. 71. Ghia P, LjungstrĂśm V, Tausch E, et al. Whole-exome sequencing revealed no recurrent mutations within the PI3K pathway in relapsed chronic lymphocytic leukemia patients progressing under idelalisib treatment. Blood. 2016;128(22):1. 72. Amin NA, Seymour E, Saiya-Cork K, Parkin B, Shedden K, Malek SN. A quantitative analysis of subclonal and clonal gene mutations before and after therapy in chronic lymphocytic leukemia. Clin Cancer Res. 2016;22(17):4525-4535. 73. Baliakas P, Hadzidimitriou A, Sutton LA, et al. Recurrent mutations refine prognosis in chronic lymphocytic leukemia. Leukemia. 2015;29(2):329-336. 74. Pospisilova S, Sutton LA, Malcikova J, et al. Innovation in the prognostication of chronic lymphocytic leukemia: how far beyond TP53 gene analysis can we go? Haematologica. 2016;101(3):263-265. 75. Kantorova B, Malcikova J, Smardova J, et al. TP53 mutation analysis in chronic lymphocytic leukemia: comparison of different detection methods. Tumour Biol. 2015;36(5):3371-3380. 76. Chin EL, da Silva C, Hegde M. Assessment of clinical analytical sensitivity and specificity of next-generation sequencing for detection of simple and complex mutations. BMC Genet. 2013;14(1):6. 77. Minervini CF, Cumbo C, Orsini P, et al. TP53 gene mutation analysis in chronic lymphocytic leukemia by nanopore MinION sequencing. Diagn Pathol. 2016;11(1):96. 78. Sutton LA, Ljungstrom V, Mansouri L, et al. Targeted next-generation sequencing in chronic lymphocytic leukemia: a high-

haematologica | 2018; 103(12)

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

throughput yet tailored approach will facilitate implementation in a clinical setting. Haematologica. 2015;100(3):370-376. Domenech E, Gomez-Lopez G, Gzlez-Pena D, et al. New mutations in chronic lymphocytic leukemia identified by target enrichment and deep sequencing. PLoS One. 2012;7(6):e38158. Jeromin S, Weissmann S, Haferlach C, et al. SF3B1 mutations correlated to cytogenetics and mutations in NOTCH1, FBXW7, MYD88, XPO1 and TP53 in 1160 untreated CLL patients. Leukemia. 2014;28(1):108117. Wang J, Morrissette J, Lieberman DB, Timlin C, Schuster SJ, Mato AR. Utilization of next generation sequencing identifies potentially actionable mutations in chronic lymphocytic leukaemia. Br J Haematol. 2018;180(2): 299-301. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008;111 (12):5446-5456. Oscier D, Dearden C, Eren E, et al. Guidelines on the diagnosis, investigation and management of chronic lymphocytic leukaemia. Br J Haematol. 2012;159(5):541564. Eichhorst B, Robak T, Montserrat E, et al. Chronic lymphocytic leukaemia: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2015;26(Suppl 5):v78-84. National Comprehensive Cancer Network. Chronic lymphocytic leukemia/small lymphocytic leukemia, version 2. 21 Feb 2017 Available from: https://www.nccn.org/professionals/physician_gls/f_guidelines.asp Dunnen JT, Dalgleish R, Maglott DR, et al. HGVS recommendations for the description of sequence variants: 2016 Update. Hum Mutat. 2016;37(6):564-569. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. Soussi T, Leroy B, Taschner PE. Recommendations for analyzing and reporting TP53 gene variants in the high-throughput sequencing era. Hum Mutat. 2014;35 (6):766-778. Tikkanen T, Leroy B, Fournier JL, Risques RA, Malcikova J, Soussi T. Seshat: A Web service for accurate annotation, validation, and analysis of TP53 variants generated by conventional and next-generation sequencing. Hum Mutat. 2018;39(7):925-933. Fischer K, Cramer P, Busch R, et al. Bendamustine combined with rituximab in patients with relapsed and/or refractory chronic lymphocytic leukemia: a multicenter phase II trial of the German Chronic Lymphocytic Leukemia Study Group. J Clin Oncol. 2011;29(26):3559-3566. Del Poeta G, Postorino M, Pupo L, et al. Venetoclax: Bcl-2 inhibition for the treatment of chronic lymphocytic leukemia. Drugs Today (Barc). 2016;52(4):249-260. 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. Seymour JF, Ma S, Brander DM, et al.

Venetoclax plus rituximab in relapsed or refractory chronic lymphocytic leukaemia: a phase 1b study. Lancet Oncol. 2017;18(2):230-240. 94. Janssen-Cilag International NV. Imbruvica 140 mg hard capsules. Summary of Product Characteristics. Beerse, Belgium; 30 August 2017. 95. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):32-42. 96. Gilead Sciences International Ltd. Zydelig 100 mg film-coated tablets. Summary of Product Characteristics. Cambridge, UK; 17 August 2017. 97. AbbVie Ltd. Venclyxto 10 mg film-coated tablets. Summary of Product Characteristics. Maidenhead, UK; 8 May 2017. 98. Anderson MA, Tam C, Lew TE, et al. Clinicopathological features and outcomes of progression of CLL on the BCL2 inhibitor venetoclax. Blood. 2017;129(25):3362-3370. 99. Brown JR, Hillmen P, O'Brien S, et al. Extended follow-up and impact of high-risk prognostic factors from the phase 3 RESONATE study in patients with previously treated CLL/SLL. Leukemia. 2018;32(1):8391. 100. Huber H, Edenhofer S, Estenfelder S, Stilgenbauer S. Profile of venetoclax and its potential in the context of treatment of relapsed or refractory chronic lymphocytic leukemia. Onco Targets Ther. 2017;10:645656. 101. Oppermann S, Ylanko J, Shi Y, et al. Highcontent screening identifies kinase inhibitors that overcome venetoclax resistance in activated CLL cells. Blood. 2016;128(7):934-947. 102. Woyach JA, Furman RR, Liu TM, et al. Resistance mechanisms for the Bruton's tyrosine kinase inhibitor ibrutinib. N Engl J Med. 2014;370(24):2286-2294. 103. Burger JA, Landau DA, Taylor-Weiner A, et al. Clonal evolution in patients with chronic lymphocytic leukaemia developing resistance to BTK inhibition. Nat Commun. 2016;7:11589. 104. Woyach JA, Guinn D, Ruppert AS, et al. The development and expansion of resistant subclones precedes relapse during ibrutinib therapy in patients with CLL. Blood. 2016;128(22):55. 105. Woyach JA, Johnson AJ. Targeted therapies in CLL: mechanisms of resistance and strategies for management. Blood. 2015;126(4): 471-477. 106. Mato AR, Hill BT, Lamanna N, et al. Optimal sequencing of ibrutinib, idelalisib, and venetoclax in chronic lymphocytic leukemia: results from a multicenter study of 683 patients. Ann Oncol. 2017;28(5): 1050-1056. 107. Jones J, Choi MY, Mato AR, et al. Venetoclax (VEN) monotherapy for patients with chronic lymphocytic leukemia (CLL) who relapsed after or were refractory to ibrutinib or idelalisib. Blood. 2016;128(22):637. 108. Follows GA, Bloor A, Dearden C, et al. Interim statement from the BCSH CLL Guidelines Panel. 2015. Available from: h t t p : / / w w w. b - s - h . o r g . u k / m e d i a / 13488/interim-statement-cll-guidelines-version6.pdf 109. European Society for Medical Oncology. eUpdate â&#x20AC;&#x201C; chronic lymphocytic leukaemia treatment recommendations. 2017. Available from: http://www.esmo.org/ Guidelines/Haematological-Malignancies/ Chronic-Lymphocytic-Leukaemia/eUpdateTreatment-Recommendations

1967

E. Campo et al. 110. Oscier D, Wade R, Davis Z, et al. Prognostic factors identified three risk groups in the LRF CLL4 trial, independent of treatment allocation. Haematologica. 2010;95(10): 1705-1712. 111. Hu L, Ru K, Zhang L, et al. Fluorescence in situ hybridization (FISH): an increasingly demanded tool for biomarker research and personalized medicine. Biomark Res. 2014;2(1):3. 112. Wiktor AE, Van Dyke DL, Stupca PJ, et al. Preclinical validation of fluorescence in situ hybridization assays for clinical practice. Genet Med. 2006;8(1):16-23. 113. Zent CS, Burack WR. Mutations in chronic lymphocytic leukemia and how they affect therapy choice: focus on NOTCH1, SF3B1, and TP53. ASH Education Program Book. 2014;2014(1):119-124. 114. Kelley T, Xu X. The future is now for the laboratory evaluation of myelodysplastic syndromes. The Hematologist. 2014;11(5).

1968

115. Edelmann J, Holzmann K, Miller F, et al. High-resolution genomic profiling of chronic lymphocytic leukemia reveals new recurrent genomic alterations. Blood. 2012;120 (24):4783-4794. 116. Gunnarsson R, Mansouri L, Isaksson A, et al. Array-based genomic screening at diagnosis and during follow-up in chronic lymphocytic leukemia. Haematologica. 2011;96(8): 1161-1169. 117. Schwaenen C, Nessling M, Wessendorf S, et al. Automated array-based genomic profiling in chronic lymphocytic leukemia: development of a clinical tool and discovery of recurrent genomic alterations. Proc Natl Acad Sci USA. 2004;101(4):1039-1044. 118. Zelenetz AD, Barrientos JC, Brown JR, et al. Idelalisib or placebo in combination with bendamustine and rituximab in patients with relapsed or refractory chronic lymphocytic leukaemia: interim results from a phase 3, randomised, double-blind, placebo-con-

trolled trial. Lancet Oncol. 2017;18(3):297311. 119. Stilgenbauer S, Eichhorst B, Schetelig J, et al. Venetoclax for patients with chronic lymphocytic leukemia with 17p deletion: results from the full population of a phase II pivotal trial. J Clin Oncol. 2018;36(19): 1973-1980. 120. Seymour JF, Kipps TJ, Eichhorst B, et al. Venetoclax-rituximab in relapsed or refractory chronic lymphocytic leukemia. N Engl J Med. 2018;378(12):1107-1120. 121. Bouaoun L, Sonkin D, Ardin M, et al. TP53 variations in human cancers: new lessons from the IARC TP53 database and genomics data. Hum Mutat. 2016;37(9): 865-876. 122. Dufour A, Palermo G, Zellmeier E, et al. Inactivation of TP53 correlates with disease progression and low miR-34a expression in previously treated chronic lymphocytic leukemia patients. Blood. 2013;121(18): 3650-3657.

haematologica | 2018; 103(12)

ARTICLE

Hematopoiesis

Niche TWIST1 is critical for maintaining normal hematopoiesis and impeding leukemia progression

Ferrata Storti Foundation

Xiaoyan Liu, Yanping Ma, Rongrong Li, Dan Guo, Nan Wang, Yangyang Zhao, Jing Yin, Qian Ren, Yongmin Lin and Xiaotong Ma State Key Laboratory of Experimental Hematology, Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin, China

Haematologica 2018 Volume 103(12):1969-1979

ABSTRACT

he bone marrow microenvironment regulates normal and malignant hematopoiesis, but the underlying molecular mechanisms remain poorly defined. Using a chimeric mice model, we demonstrate that Twist1 deletion in the bone marrow microenvironment results in alteration of multiple niche cells as well as downregulated expression of major hematopoietic stem cell supportive factors. The perturbed microenvironment reduces hematopoietic stem cell homing and retention, impairs hematopoietic stem cell self-renewal and induces myeloid skewing. Nevertheless, it accelerates the progression of MLL-AF9 leukemia, which is partially mediated by Jagged-2-dependent Notch signaling. Our data provide the first demonstration of a pivotal role of TWIST1 in favoring normal hematopoiesis and hampering leukemia development. They also bring new insights into the role of the bone marrow niche in driving the development of acute myeloid leukemia, and suggest possible new avenues, exploiting the niche, to improve leukemia treatments.

Correspondence: maxt@ihcams.ac.cn

Introduction Hematopoietic stem cells (HSC) reside in a special bone marrow (BM) niche, which regulates their localization, self-renewal and differentiation. Studies have identified several major cell types of the niche, including mesenchymal stem cells (MSC), osteolineage cells (OLC), adipocytes and vascular endothelial cells (EC).1-5 Besides the key cellular components, some growth and survival factors are also indispensable components of the niche, including C-X-C motif chemokine 12 ligand (CXCL12), vascular cell adhesion molecule1 (VCAM1),6,7 stem cell factor (SCF)4 and osteopontin.8 A sophisticated network of interactions between these multiple BM cells, extracellular factors and adhesion molecules is essential to regulate different HSC properties during homeostasis and keep normal hematopoiesis in check. Acute myeloid leukemia (AML) has been considered for decades to be a disorder intrinsic to hematopoietic cells; however, evidence is accumulating that the microenvironment exerts more than a mere bystander effect. Leukemic cells can remodel the niche into a permissive environment favoring leukemic stem cell (LSC) expansion over normal HSC maintenance.9 Recently, emerging evidence even points to a role for the BM niche as a driver of disease maintenance/progression. Krause et al. showed that osteoblast-specific activation of the parathyroid hormone receptor enhances MLL-AF9 oncogene-induced AML in mouse transplantation models.10 To date, there are still few studies concerning the role of the bone marrow niche in initiating and maintaining AML and relevant mechanisms remain elusive. TWIST1 is a highly conserved transcription factor belonging to the basic helixloop-helix family and is implicated in diverse developmental systems.11-13 Studies have revealed that TWIST1 is a key regulator of MSC self-renewal, enhances their life-span, inhibits MSC osteo/chondrogenic differentiation and promotes adipogenic differentiation.14-16 Twist1 haploisufficiency leads to Saethre-Chotzen syndrome, which is characterized by alterations in osteogenic precursor cell proliferation, differentiation and survival.17 Recent studies have demonstrated that TWIST1 promotes angiogenesis by inducing EC proliferation and migration, and deregulation of this mechanism mediates pathological angiogenesis.18,19 Arthur et al. showed haematologica | 2018; 103(12)

Received: February 8, 2018. Accepted: July 17, 2018. Pre-published: July 19, 2018. doi:10.3324/haematol.2018.190652 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/1969 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1969

X. Liu et al.

that overexpression of Twist1 in MSC enhances the capacity to maintain human CD34+ cells in long-term culture-initiating cell assays through increasing Cxcl12 expression.20 However, the effects of TWIST1 on multiple niche elements and its modulation of normal HSC maintenance and leukemia progression in vivo have not been functionally characterized so far. To explore this issue, we generated a murine model of a Twist1-deficient microenvironment. We showed that the major niche cellular components and factors changed remarkably upon Twist1 deletion, causing severe dysfunction of normal HSC. Nevertheless, these alterations of the BM microenvironment promoted MLL-AF9 oncogeneinduced AML progression in mouse transplantation models, not only pointing to TWIST1 as an instructive signal modulating the stem cell niche, but also emphasizing the importance of the niche for AML development.

Methods Mice Twist1flox/flox mice were purchased from Mutant Mouse Regional Resource Centers. ER-Cre mice were a gift from Professor Weiping Yuan. C57BL/6 and B6.SJL mice were purchased from the animal facility of State Key Laboratory of Experimental Hematology. Twist1flox/flox mice were crossed with ER-Cre mice to generate ER-Cre;Twist1fl/fl and ER-Cre;Twist1+/+ mice. Eight- to 12week old mice were used. Cre expression was induced by daily intraperitoneal injection of tamoxifen (75 mg/kg of total body weight in corn oil; Sigma-Aldrich, St. Louis, MO, USA) for 5 days. All animal procedures complied with the animal care guidelines approved by the Institutional Animal Care and Use Committees of the State Key Laboratory of Experimental Hematology.

Transplantation assays For non-competitive BM transplantation, to create the chimeras described in Online Supplementary Figure S1A, 2x106 whole BM cells from B6.SJL (CD45.1) mice were transplanted into ERCre;Twist1+/+ or ER-Cre;Twist1fl/fl (CD45.2) recipients that were lethally irradiated (9.5 Gy from a Cesium source, 4-24 h before transplantation). Sixteen weeks later, tamoxifen was injected to induce Twist1 deletion. For competitive transplantation, 300 BM long-term HSC (CD45.1) from tamoxifen-treated ER-Cre;Twist1+/+ or ER-Cre;Twist1fl/fl chimeric mice were mixed with 2x105 congenic BM support cells and injected into lethally irradiated CD45.2 recipients. For the MLL-AF9 AML model, 5x105 GFP+ leukemic cells were transplanted into Twist1-deleted or control chimeric recipients.

Flow cytometry analysis and cell sorting The BM cell suspensions were flushed from femora and tibiae. Spleen cells were pestled by the plug of a 10 mL syringe. The cells were then filtered through a 74 mm nylon mesh. For flow cytometric analysis of stromal cells, BM was flushed using phosphatebuffered saline with 2% bovine serum, the bones were minced with scissors, then the plugs were digested in 1 mg/mL collagenase I (OLC) or IV (MSC and EC) (Sigma-Aldrich) dissolved in Hank’s balanced salt solution with 10% fetal bovine serum for 90 min (collagenase I) or 30 min (collagenase IV) at 37°C. The dissociated cells were collected and kept on ice. Cells were incubated with conjugated antibodies. Stained cells were analyzed with FACS LSR II or sorted with a FACS Aria II instrument (BD Biosciences, Franklin Lakes, NJ, USA). Data were analyzed by FlowJo software. 1970

Statistical analysis The significance of differences between two groups was determined using unpaired two-tailed Student t tests. Data are presented as means ± standard deviations. Overall survival curves were plotted according to the Kaplan-Meier method with the log-rank test applied for comparisons. *P<0.05, **P<0.01, ***P<0.001. Details of other experimental procedures are given in the Online Supplementary Methods.

Results Microenvironmental Twist1 deficiency leads to decreased numbers of mesenchymal stem cells and mature osteoblasts, an increased proportion of endothelial cells, and altered expression of cell factor genes To explore the role of TWIST1 in the BM niche and its regulation of HSC, we generated microenvironment Twist1-deleted and control chimeric mice according to the method described by Schreck and Saez.21,22 In brief, 2x106 BM cells from B6.SJL wild-type (WT) mice (CD45.1) were transplanted into ER-Cre;Twist1fl/fl and ER-Cre;Twist1+/+ recipients (CD45.2) (Online Supplementary Figure S1A). Sixteen weeks later, about 90% of the cells in the peripheral blood of recipients were donor-derived cells (Online Supplementary Figure S1B). Tamoxifen was then injected intraperitoneally for 5 days to induce Twist1 deletion. Two weeks after the last injection, mRNA detection demonstrated that Twist1 had been knocked out in all the MSC, OLC, and EC isolated from Twist1Δ/Δ mice with similar knockout levels (Online Supplementary Figure S1C), while the expression of Twist2 was almost unchanged (data not shown). To define components of the Twist1-deleted BM microenvironment that may be altered, stromal populations and extracellular factors were assessed in Twist1deleted and control chimeric mice. We observed that conditional deletion of Twist1 led to a significant decrease in the number of MSC (CD140a+CD51+CD45/Ter119/CD31)23 in the BM compared with that in control mice, as determined by flow cytometry (Figure 1A). The decrease in MSC number was further confirmed by a fibroblastic colony-forming unit assay (Online Supplementary Figure S2A). Furthermore, Twist1-deleted MSC showed a decrease in proliferative cells and an increase in apoptotic cells (Online Supplementary Figure S2B,C), indicating the mechanism underlying the reduced number of MSC. Twist1 deficiency resulted in a marked increase in the frequency of Sca-1-/CD166+ stromal cells (Figure 1B), which include immature and mature OLC.24 Meanwhile, the expression of osteoblastic differentiation genes Runx2, Ogn and Gpnmb14,23 was significantly upregulated in Twist1deleted MSC (Online Supplementary Figure S2D). To assess the ability of MSC to differentiate into the osteoblastic lineage, we induced osteoblast differentiation in MSC and found that Twist1 deletion clearly increased alkaline phosphatase activity and matrix mineralization (Online Supplementary Figure S2E). These results establish that Twist1 deficiency enhanced MSC commitment toward osteoblasts. However, expression of the mature osteoblast marker, Bglap, was downregulated (Online Supplementary Figure S2F). In addition, micro-computed tomography analysis also revealed a significant decrease in mature osteoblasts in Twist1-deleted mice, which was reflected by haematologica | 2018; 103(12)

TWIST1 regulates normal hematopoiesis and leukemia

a reduction of trabecular volume (Figure 1C). Collectively, Twist1 deletion promotes MSC to differentiate toward the osteoblast lineage with a block of mature osteoblast differentiation. Emerging data demonstrate the role of vascular EC in HSC maintenance, and arterioles and sinusoids exhibit dif-

ferent properties in relation to HSC distribution and quiescence.4,25,26 Ciuculescu et al. reported that Rac deletion in MSC leads to an inverted ratio of marrow arterioles and sinusoid vessels and impaired hematopoiesis.27 We observed that Twist1 deletion resulted in increased CD45Ter119-CD31+, CD45-Ter119-CD31+Sca1+ and CD45-Ter119-

Figure 1. Twist1 deficiency in the bone marrow microenvironment leads to decreased frequency of mesenchymal stem cells and mature osteoblasts, and an increased proportion of endothelial cells. (A) Flow cytometry (FACS) analysis of bone marrow (BM) msesenchymal stem cells (MSC, CD140a+CD51+CD45/Ter119/CD31-) in chimeric control (Ctrl) and knockout (KO) mice. Representative FACS profiles are shown on the left, and cell frequency is shown on the right (n=4, three independent experiments). (B) FACS analysis of BM osteolineage cells (OLC, Sca-1-CD166+CD45/Ter119/CD31-) in chimeric Ctrl and KO mice. Representative FACS profiles are shown on the left, and cell frequency is shown on the right (n=5, three independent experiments). (C) Micro-computed tomography analysis of the trabecular bone of chimeric Ctrl and KO mice. Representative images are shown on the left. Scale bars, 1 mm. Trabecular bone volume/total volume (BT/BV), trabecular number (Tb. N) and trabecular spacing (Tb. Sp) in the femoral metaphysis are shown on the right (n=4, two independent experiments). (D) FACS analysis of BM endothelial cells (EC) in chimeric Ctrl and KO mice. Representative FACS profiles of sinusoidal EC (SEC, CD45-Ter119-CD31+Sca-1-) and arteriolar EC (AEC, CD45-Ter119-CD31+Sca-1+) are shown on the left. Frequencies of BM total EC (CD45-Ter119-CD31+), AEC and SEC are shown on the right (n=6, two independent experiments). (E) Immunofluorescent images of the BM microvasculature in the femoral diaphysis of animals of each genotype are shown after staining for Sca-1 (white, arteries), Endoglin (green, sinusoids) and 4â&#x20AC;&#x2122;,6-diamidino-2-phenylindole (DAPI, blue), as detailed in the Methods. Scale bars, 40 mm. (n=3, two independent experiments). (F) Proliferation analysis of EC in chimeric Ctrl and KO mice (n=4, two independent experiments). (G) In vitro tube formation assay with EC from chimeric Ctrl and KO mice. (H) Quantification of the tube formation assay (n=3, two independent experiments). Column plots show the mean Âą standard deviation. *P<0.05; **P<0.01, ***P<0.001 (Student t test).

haematologica | 2018; 103(12)

1971

X. Liu et al.

CD31+Sca1- populations enriched for total EC, arteriolar EC and sinusoidal EC,28 respectively (Figure 1D). The increase of arteriolar and sinusoidal EC was further confirmed by the observation that arteries (Sca-1-staining) and sinusoids (Endoglin-staining) were both significantly increased in Twist1-deleted mice compared with control mice by immunofluorescence of femoral sections (Figure 1E). The increase of EC may be the result of cell proliferation as determined by increased bromodeoxyuridine incorporation into CD45-Ter119-CD31+ stromal cells in Twist1-deleted mice (Figure 1F). We next performed a tube formation assay to determine the effect of TWIST1 on new blood vessel development. Consistent with increased microvessels in knockout mice in vivo, capillary tube formation of Twist1deleted EC was also increased on matrigel (Figure 1G,H), indicating that Twist1 deletion promotes angiogenesis. To further evaluate the impact of Twist1 deficiency on cell factors, we performed quantitative real-time polymerase chain reaction to analyze the expression of key niche factors. The results showed significant decreases in the expression of Cxcl12, Vcam1, Angiopoietin-1 (Angpt1) and Scf, particularly the membrane-bound isoform Scf (m220 Scf), which was found to be extremely important for HSC maintenance, in MSC (5Ă&#x2014;104 cells) from Twist1-deleted mice as compared to control mice (Figure 2A). The expression of Opn, which negatively regulates the HSC pool, was obviously increased in both MSC and OLC (Figure 2B). Enzyme-linked immunoassay demonstrated the reduced protein levels of CXCL12, VCAM1, SCF and elevated level of osteopontin (Figure 2C) in BM super-

natants of Twist1-deleted mice as compared to those in control mice. Collectively, Twist1 deletion leads to significant alterations in various key niche components, demonstrating its functional importance in the BM microenvironment, and implying its potential regulatory role in HSC maintenance.

Microenvironmental Twist1 deficiency impairs the homing and retention of hematopoietic stem cells but promotes their mobilization Most HSC are retained in the BM niche in a quiescent, nonmotile mode by adhesion to stromal cells, which are essential for normal hematopoiesis and for protection from myelotoxic injury. Twist1 deletion resulted in decreased expression of BM CXCL12 and VCAM1, which are critical for HSC retention, homing and mobilization,29,30 implying that TWIST1 may have a functional impact on the migration of HSC. We first evaluated homing of normal HSC to the BM of Twist1-deleted mice. Freshly isolated c-Kit+ cells from B6.SJL (CD45.1) mice were injected into lethally irradiated Twist1-deleted or control mice. Sixteen hours after transplantation, the absolute number of CD45.1+ and CD45.1+ LSK (Lin-Sca-1+c-Kit+) cells was significantly decreased in Twist1-deleted mice compared to the number in controls (Figure 3A), indicating that Twist1 deficiency impairs homing of hematopoietic stem/progenitor cells (HSPC) to the BM microenvironment. We next determined the contribution of TWIST1 to HSC retention and mobilization by assessing total cells and HSPC in the BM, spleen and peripheral blood of

Figure 2. Twist1 deletion in the bone marrow microenvironment changes expression of niche factors. (A) Quantitative real-time polymerase chain reaction (qRT-PCR) analysis of the expression of C-X-C motif chemokine ligand 12 (Cxcl12), Vascular cell adhesion molecule 1 (Vcam1), Stem cell factor (Scf) and Angiopoietin-1 (Angpt1) in freshly sorted mesenchymal stem cells (MSC), osteolineage cells (OLC) and endothelial cells (EC) from chimeric control (Ctrl) and knockout (KO) mice (n=4). (B) qRT-PCR analysis of the expression of Osteopontin (Opn) in MSC and OLC from chimeric Ctrl and KO mice (n=4). (C) Enzyme-linked immunosorbent assay analysis of BM protein concentrations of CXCL12, VCAM1, SCF and OPN in chimeric Ctrl and KO mice (n=5-8). Column plots show the mean Âą standard deviation from three independent experiments. *P<0.05; **P<0.01; ***P<0.001 (Student t test).

1972

haematologica | 2018; 103(12)

TWIST1 regulates normal hematopoiesis and leukemia

Twist1-deleted mice and control mice (Figure 3B). The results revealed that Twist1 deletion led to decreases in the numbers of total cells (Figure 3C), colony-forming units (Figure 3D), and HSC-enriched SLAM LSK (CD150+CD48LSK) in the BM (Figure 3E,F) but significant increases in those in the spleen and circulation (Figure 3G-J). Granulocyte colony-stimulating factor (G-CSF) is a hematopoietic cytokine known as the prototypic mobilizing agent.31 To determine whether Twist1 deletion promot-

ed secretion of endogenous G-CSF, we examined G-CSF levels in the BM supernatants of Twist1-deleted and control mice by enzyme-linked immunosorbent assay. We found that G-CSF protein expression was significantly elevated in Twist1-deficient mice as compared to that in controls. Furthermore, quantitative real-time polymerase chain reaction results revealed that among MSC, OLC, EC and macrophagocytes, which are all producers of G-CSF, OLC and macrophagocytes were the major source of G-CSF

Figure 3. Decreased hematopoietic stem/progenitor cell homing and retention in bone marrow and increased hematopoietic stem/progenitor cell mobilization to spleen and peripheral blood in Twist1-deficient mice. (A) Experimental scheme of the hematopoietic stem/progenitor cell (HSPC) homing assay (left), and absolute number of CD45.1+ and CD45.1+LSK (Lin-Sca-1+c-Kit+) cells homing to the bone marrow (BM) (right) (n=5, two independent experiments). (B) Experimental scheme for analysis of HSPC retention and mobilization. (C-F) Analysis of BM cells of chimeric control (Ctrl) and knockout (KO) mice. Total BM cells (C) of femora and tibiae, number of BM progenitor cells (D) measured by colony-forming cell (CFC) assay in methylcellulose, and frequency and number of SLAM LSK cells (CD150+CD48-LSK) (E-F) are shown (n=4-6, three independent experiments). (G-J) Analysis of peripheral blood cells of chimeric Ctrl and KO mice. Number of total cells (G), number of colony-forming units (CFU) (H), frequency and number (I) of SLAM LSK cells in spleen, and number of CFU (J) in peripheral blood are shown (n=4-6, three independent experiments). Column plots show the mean Âą standard deviation. *P<0.05; **P<0.01; ***P<0.001 (Student t test).

haematologica | 2018; 103(12)

1973

X. Liu et al.

(Online Supplementary Figure S3A,B). We next assessed HSPC mobilization after infusion of exogenous G-CSF and found that Twist1 deletion significantly increased exogenous G-CSF-induced mobilization of total cells and colony-forming cells to the blood and spleen, and SLAM LSK to the spleen (Online Supplementary Figure S3C-H). Taken together, these results demonstrate a functional role of TWIST1 in HSC homing, retention, baseline mobilization and stress mobilization in response to G-CSF.

Twist1 deficiency impairs hematopoietic stem cell quiescence and self-renewal, and induces enhanced early myeloid lineage differentiation Quiescence, self-renewal and committed differentiation are important properties of HSC, which could be controlled by stromal cells, extracellular matrix, cytokines and chemokines.32,33 Our study revealed that Twist1 deletion altered multiple stromal cells and the level of expression of HSC supportive factors, so we next investigated whether these HSC features were consequently changed. Immunophenotypic analysis demonstrated that Twist1 deletion resulted in a significant decrease in the number of long-term HSC (CD34-Flt3-LSK) in the BM (Online Supplementary Figure S4A). Ki67 staining revealed a significant decrease in the percentage of HSC (CD34-LSK) in G0 phase in Twist1-deleted mice, together with an increase in G1 phase (Figure 4A). Bromodeoxyuridine labeling further identified a higher frequency of proliferating cells in CD34LSK cells from Twist1-deleted mice compared to control mice (Figure 4B). These data suggest that TWIST1 in the microenvironment plays an important role in maintaining HSC quiescence, and loss of Twist1 drives aberrant proliferation of HSC. To clarify the role of TWIST1 in HSC self-renewal, we conducted serial transplantation assays (Figure 4C). Three hundred long-term HSC (CD34-Flt3-LSK, CD45.1) from Twist1-deleted or control chimeric mice were transplanted into lethally irradiated recipients (CD45.2), together with 2×105 CD45.2+ support BM cells. Then secondary transplantation assays were performed 16 weeks later. Three hundred long-term HSC (CD45.1+) from the primary recipients were transplanted into lethally irradiated secondary recipients, together with 2×105 CD45.2+ support BM cells. Donor cells from Twist1-deleted mice showed lower engraftment capacity than control cells out to 16 weeks in both primary and secondary transplantation (Figure 4D). From the above, it can be concluded that Twist1 deficiency in niche cells impairs HSC self-renewal capacity. We then performed FACS analysis to evaluate the differentiation capacity of HSC, and found increases in the numbers of common myeloid progenitors (CD34+CD16/32-Lin-Sca-1-c-Kit+, 1.5-fold, P=0.047) and granulocyte/macrophage progenitors (GMP: CD34+CD16/32+Lin-Sca-1-c-Kit+, 1.3-fold, P=0.022) in the BM of Twist1-deleted mice in comparison with those in controls, accompanied by decreases in the numbers of megakaryocyte/erythroid progenitors (MEP, CD34CD16/32-Lin-Sca-1-c-Kit+, 1.5-fold, P=0.028) and common lymphoid progenitors (Lin−Sca-1lowc-KitlowIL7R+, 2.5-fold, P=0.046) (Figure 4E,F). In accordance with the change of progenitors, the proportion of mature myeloid cells (Mac1+, 1.2-fold, P=0.007; Gr-1+, 1.2-fold, P=0.008; Mac1+Gr-1+, 1.2-fold, P=0.011) was also significantly increased, accompanied by reduced proportions of B lymphoid cells (B220+, 1.4-fold, P=0.004) and erythrocytes (Ter119+, 1.31974

fold, P=0.008) in the BM of Twist1-deleted mice (Figure 4G,H). These results suggest that Twist1 deletion in niche cells promotes HSC differentiation into the myeloid lineage. We observed that apart from the increase of HSPC in the spleen of Twist1-deleted mice, the numbers of mature myeloid cells and erythrocytes were also increased, while the number of lymphoid cells remained unchanged (Online Supplementary Figure S4B-D), indicating the occurrence of extramedullary hematopoiesis in the spleen of Twist1deleted mice. Taken together, these data suggest that TWIST1 in the BM microenvironment plays a critical role in HSC maintenance, and Twist1 deletion impairs all the fundamental features of HSC.

Twist1 deficiency promotes progression of MLL-AF9-induced acute myeloid leukemia It has been reported that the MLL-AF9 AML model exhibited multiple alterations in the niche compartments, including decreased frequencies of MSC and osteoblasts, an increased number of vascular EC, and downregulated expression of Vcam1, Cxcl12, Angpt1, and Scf, together with upregulation of Opn.34 Intriguingly, these phenotypes are quite similar to the niche alterations in our Twist1-deleted mice. Considering the emerging importance of the BM niche for leukemia maintenance and progression, we were tempted to speculate that the altered niche in Twist1-deleted mice may play a role in the development of MLL-AF9 AML. To validate this hypothesis, we transduced WT BM c-Kit+ cells with retrovirus expressing MLL-AF9, and injected these cells into lethally irradiated WT recipient mice, in which the disease was rapidly induced with massive BM and spleen infiltration of GFP+ leukemic cells. We next injected 5×105 GFP+ spleen cells from these mice into Twist1-deleted and control chimeric recipient mice (Figure 5A). Notably, the overall survival of Twist1-deleted recipients was significantly shorter than that of control recipients (Figure 5B), and Twist1-deleted recipient mice exhibited a greater infiltration of total cells and leukemic cells in the BM, peripheral blood and spleen than that of control mice (Online Supplementary Figure S5A-G). Previous studies using MLL-AF9 AML models have established that LSC are enriched in the leukemic GMP populations (IL-7R-Lin-GFP+c-KithiCD34+CD16/32hi)34,35 or cKit+Gr-1−.36 We found that the infiltration of leukemic GMP cells was significantly increased in the BM and periphery of Twist1-deleted recipient mice compared to control mice (Figure 5C,D), and so was that of GFP+c-Kit+Gr-1- cells (Online Supplementary Figure S6A-C). Additionally, cell cycle analysis showed that the proportion of LSC (GFP+cKit+Gr-1-) in the G0 phase was decreased and that in the G1 or S/G2/M phases was increased in both BM and spleen from Twist1-deleted mice, but rates of LSC apoptosis in BM and spleen did not differ between Twist1-deleted and control mice (Online Supplementary Figure S6D-G). We then performed secondary transplants using LSC (Figure 5A), injecting 5000 GFP+c-Kit+Gr-1- cells isolated from Twist1deleted and control mice into WT recipient mice. We found that the survival of the secondary recipients of LSC grafts from leukemic Twist1-deleted donors was significantly reduced compared to that of recipients of control LSC grafts (Figure 5E). Taken together, these results demonstrate that the altered niche in Twist1-deleted mice accelerates the prohaematologica | 2018; 103(12)

TWIST1 regulates normal hematopoiesis and leukemia

gression of MLL-AF9-induced AML by increasing the expansion and leukemogenic capacity of LSC.

Microenvironmental Twist1 deletion promotes acute myeloid leukemia development partially through the Notch signaling pathway To understand the mechanism underlying TWIST1 function in the development of AML, we performed RNA-

sequencing analysis on BM stromal cells isolated from Twist1-deleted or control chimeric mice. There were 6948 differentially expressed genes (4764 downregulated and 2184 upregulated; fold change of >2 and P value <0.05). Analysis of leukemia-related signaling revealed a marked increase of Jagged-2 expression in stromal cells from Twist1deleted mice (Figure 6A). Previous studies have demonstrated that TWIST1 regulates the Notch pathway in

Figure 4. Twist1 deletion causes impaired quiescence, self-renewal of hematopoietic stem cells and myeloid skewing. (A) Percentage of the cell cycle distribution of CD34-LSK (Lin-Sca-1+c-Kit+) cells in chimeric control (Ctrl) and knockout (KO) mice. Representative flow activated cell sorting profiles are shown on the left, and cell frequency is shown on the right (n=4-5, two independent experiments). (B) Proliferation analysis of CD34-LSK cells in chimeric Ctrl and KO mice (n=4-5, two independent experiments). (C) Schematic overview of the serial transplantation assay. (D) Percentages of donor-derived peripheral blood (PB) cells and bone marrow (BM) cells after the primary and secondary competitive transplants (n=5, two independent experiments). (E-F) Frequency (E) and number (F) of common myeloid progenitors (CMP, CD34+CD16/32-Lin-Sca-1-c-Kit+), granulocyte/macrophage progenitors (GMP, CD34+CD16/32+Lin-Sca-1-c-Kit+), megakaryocyte/erythroid progenitors (MEP, CD34-CD16/32-Lin-Sca-1-c-Kit+) and common lymphoid progenitors (CLP, Linâ&#x2C6;&#x2019;Sca-1lowc-KitlowIL7R+) in chimeric Ctrl and KO mice (n=4-5, three independent experiments). (G-H) Frequency (G) and number (H) of B cells (B220+), T cells (CD3+), myeloid cells (Mac-1+ and Gr-1+) and erythrocytes (Ter119+) in chimeric Ctrl and KO mice (n=4-5, three independent experiments). Column plots show the mean Âą standard deviation. *P<0.05; **P<0.01; ***P< 0.001 (Student t test).

haematologica | 2018; 103(12)

1975

X. Liu et al.

prospective coronal suture mesenchyme and osteoprogenitors.37,38 Moreover, aberrant Notch signaling is a common mechanism in niche-induced AML and pre-leukemic conditions.39-42 To investigate whether Twist1 deficiency promotes the development of MLL-AF9 AML through Notch signaling, we determined the expression of all Notch ligands (Dll1, Dll3, Dll4, Jagged-1, Jagged-2) in MSC, OLC and EC of Twist1-deleted mice. The results revealed that Jagged2 was significantly upregulated in all these cells (Figure 6BD). Additionally, the levels of expression of all four Notch receptors (Notch1-4), cleaved Notch1 and the Notch targets Dtx, Hes1, Hes5, Hey1, and Hey2 were significantly upregulated in LSC from Twist1-deleted mice compared to those of controls (Figure 6E-G), indicating increased Notch signaling in this population. Furthermore, pharmacological inhibition of Notch signaling with a g-secretase inhibitor (DBZ) (Figure 6H-K) or blockade of Notch with dominantnegative MAML1 (DNMAML1) (Online Supplementary Figure S7) partially rescued leukemic cell infiltration and LSC engraftment, and prolonged the overall survival of Twist1-deleted recipients. These data suggest that a Twist1deleted microenvironment contributes to MLL-AF9 AML development at least in part via Notch signaling.

Discussion In the current study, we demonstrated that excision of the Twist1 gene from the BM microenvironment resulted in a significant decrease in the numbers of MSC and mature osteoblasts, and an increase in the number of EC. The expression of CXCL12, VCAM1 and SCF was reduced, while that of osteopontin was increased. These changes led to a marked impairment of HSC localization, selfrenewal, quiescence and differentiation. By transplanting MLL-AF9 cells into the Twist1-deleted and control chimeric mice, we verified that Twist1 deletion resulted in accelerated development of leukemia, at least partially through Notch signaling (Figure 7). These results reveal the essential role of TWIST1 in supporting normal hematopoiesis and perturbing AML development. In our model, Twist1 deletion in the BM microenvironment leads to an increased number of EC and microvessel density, suggesting the existence of an indirect and powerful mechanism for promoting angiogenesis in vivo. Ohki et al. reported that G-CSF can markedly increase vascular endothelial growth factor (VEGF) release from G-CSFresponsive myelomonocytic cells, which promote the corecruitment of VEGFR1+ (VEGF receptor 1) cells contributing to neo-angiogenesis.43 Since we have found elevated G-

Figure 5. Twist1 deletion in the bone marrow microenvironment promotes the progression of acute myeloid leukemia. (A) Experimental scheme of the MLL-AF9 acute myeloid leukemia model and leukemic stem cell (LSC, GFP+c-Kit+Gr-1-) transplantation. (B) Kaplan–Meier survival curve of chimeric control (Ctrl) and knockout (KO) recipient mice (n=5, three independent experiments, log-rank test). (C) Representative flow cytometry profiles of L-GMP (IL-7R-Lin-GFP+c-KithiCD34+CD16/32hi). (D) Frequency and absolute number of L-GMP in the bone marrow (BM) and spleen of Ctrl and KO recipients (n=5, three independent experiments. Column plots show the mean ± standard deviation. **P<0.01; ***P<0.001, Student t test). (E) Kaplan–Meier survival curve of mice transplanted with LSC from chimeric Ctrl and KO mice (n=6, two independent experiments, log-rank test).

1976

haematologica | 2018; 103(12)

TWIST1 regulates normal hematopoiesis and leukemia

CSF secretion in the BM supernatant of Twist1-deleted mice, we then determined VEGF mRNA expression in BM cells and its protein concentration in BM supernatant. As expected, the results revealed a marked increase of VEGF production in Twist1-deleted mice compared with control mice (Online Supplementary Figure S8A,B). These observations provide a possible explanation for the EC alterations, i.e., Twist1 deficiency in the BM microenvironment leads to increased production of G-CSF, which in turn induces the secretion of VEGF, exerting a promotive effect on the proliferation of EC. This effect overrides the direct inhibitory role of Twist1 deletion on EC, and results in increased numbers of EC. As VEGF can be produced by various cell types,

the specific mechanism needs further investigation. Our understanding of niche contributions to AML has increased tremendously over the past decade. However, most studies have focused on how the leukemic cells actively shape their microenvironment to reinforce disease progression. There are a limited number of reports showing that certain niche alterations can act as a driver of AML initiation or progression, without having been educated by leukemic cells.40,44,45 Our present study demonstrates that environmental deletion of Twist1, a conserved transcriptional factor gene, results in diverse cellular and factor alterations common to the microenvironmental dysregulation exhibited by AML. These alterations appear to be predis-

Figure 6. Notch signaling is activated in leukemic stem cells and inhibition of Notch signaling partially rescues MLL-AF9-induced acute myeloid leukemia progression in Twist1-deficient mice. Expression of Jagged-2 in RNA-sequencing analysis of stromal cells (CD45-Ter119-) from chimeric control (Ctrl) and knockout (KO) mice. (B-D) Quantitative real-time polymerase chain reaction (qRT-PCR) analysis of Notch ligands (Dll1, Dll3, Dll4, Jagged-1 and Jagged-2) in freshly sorted mesenchymal stem cells (MSC) (B), osteolineage cells (OLC) (C) and endothelial cells (EC) (D) from chimeric Ctrl and KO mice. (E) qRT-PCR analysis of Notch receptors (Notch1-4) in freshly sorted GFP+c-Kit+Gr-1- from chimeric Ctrl and KO mice. (F) Western blot showed a significant increase of cleaved Notch1 expression in leukemic stem cells (LSC) from chimeric KO mice compared to Ctrl mice. (G) qRT-PCR analysis of downstream genes (Dtx, Hes1, Hes5, Hey1 and Hey2) regulated by the Notch pathway in freshly sorted GFP+c-Kit+Gr-1- from chimeric Ctrl and KO mice. (B-E, G) Data represent the mean Âą standard deviation from three independent experiments. *P<0.05, **P<0.01, ***P<0.001 (Student t test). (H-J) GFP+ leukemic cells were transplanted into chimeric Ctrl and KO recipient mice. Five days later, the mice were treated daily with vehicle (dimethylsulfoxide, DMSO) or g-secretase inhibitor (DBZ) (2 mmol per kg body weight) for 10 days. The counts of white blood cells (WBC) (H), leukemic cells in peripheral blood (PB) (I), and L-GMP (IL-7R-Lin-GFP+c-KithiCD34+CD16/32hi) cells in bone marrow (BM) (J) are shown (n=4, two independent experiments). Column plots show the mean Âą standard deviation. *P<0.05; **P<0.01, Student t test). (K) Survival curve of chimeric Ctrl and KO recipients treated with DMSO or DBZ (n=7-8, log-rank test).

haematologica | 2018; 103(12)

1977

X. Liu et al.

Figure 7. Overview of the alterations in the bone marrow niche of Twist1-deleted mice. Simplified scheme of the normal hematopoietic stem cell (HSC) niche and its alterations in the Twist1-deleted mice. The left panel shows that HSC are around arterioles, sinusoids and endosteum where factors such as C-X-C motif chemokine ligand 12 (CXCL12), vascular cell adhesion molecule1 (VCAM1), stem cell factor (SCF) and osteopontin (OPN) secreted by mesenchymal stem cells (MSC), endothelial cells and osteoblastic cells (OBC) influence their self-renewal, quiescence, retention and differentiation. The right panel summarizes alterations of the niche, HSC and leukemic stem cells (LSC) observed in Twist1-deleted mice.

posing or initiating factors for the evolution of AML and point to TWIST1 as an instructive signal that alters the function of the niche. The opposing effects of TWIST1 on normal HSC and LSC found in this study are of value. In an effort to elucidate the underlying mechanism, we performed RNAsequencing and quantitative real-time polymerase chain reaction analysis. We found that Twist1 deletion leads to increased expression of the Notch ligand Jagged-2 in all the OLC, EC and MSC. LSC from Twist1-deleted chimeric mice have robust expression of all Notch receptors and canonical downstream Notch target genes, suggesting the aberrant activation of Notch signaling. A previous report showed that Notch activation promotes expansion and self-renewal of LSC,40 consistent with our results obtained by deletion of Twist1. We also found activation of Notch receptors and target genes in normal HSC (CD34-LSK) after Twist1 deletion (Online Supplementary Figure S9). In contrast to the promoting role in LSC, Notch activation in HSC has been reported to cause loss of stem cell quiescence,46 which often correlates with impaired self-renewal capacity of HSC, in line with the observations in our mouse model. Besides the direct impact of activated Notch signaling on LSC and HSC, the augmented proliferation and infiltration of LSC compared to normal HSC could be favorable for their competition for the niche over HSC. Various studies have demonstrated that LSC could positively remodel the BM microenvironment to enhance support of LSC at the expense of HSC,47,48 and this remodeling may in turn further promote leukemia progression and impair normal hematopoiesis. In addition, the reduced expression of Cxcl12, Scf and Angpt1 in Twist1deleted mice may also account for the opposing impact of Twist1 deletion on HSC and LSC, since compared with HSC, LSC are less factor-dependent.34,45,49 In consideration of the important role of TWIST1 in reg1978

ulating MSC, osteoblasts and EC, and to exclude the interference of hematopoietic cells, which were found to express Twist1 in our previous work,50 we generated the chimeric mouse model, in which Twist1 was diffusely deleted in the BM microenvironment. The BM niche comprises multiple cell types, which not only closely connect but also communicate with each other via cell factors and adhesion molecules throughout the BM. Due to the complexity of the niche, an overall environmental knockout strategy will facilitate the detection of direct and indirect effects of TWIST1 on the niche components. Utilizing our model, we uncovered extensive cellular and factor alterations in the BM niche and the AML-like microenvironmental phenotype resulting from Twist1 deficiency, and demonstrated the essential role of TWIST1 in HSC maintenance and suppression of AML evolution. To refine the contribution of different cell populations, studies in which Twist1 is modified in specific stromal cell subsets are ongoing in our laboratory. In conclusion, we used a Twist1-deficient chimera model to obtain, for the first time in vivo, direct evidence that TWIST1 in the microenvironment plays a key role in maintaining the hematopoietic phenotype and hampering leukemia progression. These findings provide new insights into the importance of the BM niche for AML development, and lay the foundation for tackling leukemia from a different angle to improve current treatments. Acknowledgments This work was supported by grants from The National Key Research and Development Program of China (2016YFA0100603), CAMS Initiative for Innovative Medicine (2016-I2M-1-017), National Natural Science Foundation of China (81470278, 81670158, 81600138, and 81700106), and Tianjin Municipal Science and Technology Commission grants (17JCZDJC35100 and 17JCQNJC10800). haematologica | 2018; 103(12)

TWIST1 regulates normal hematopoiesis and leukemia

References 18. 1. Mendez-Ferrer S, Michurina TV, Ferraro F, et al. Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature. 2010;466(7308):829-834. 2. Zhang J, Niu C, Ye L, et al. Identification of the haematopoietic stem cell niche and control of the niche size. Nature. 2003;425(6960):836-841. 3. Arai F, Hirao A, Ohmura M, et al. Tie2/angiopoietin-1 signaling regulates hematopoietic stem cell quiescence in the bone marrow niche. Cell. 2004;118(2):149161. 4. Ding L, Saunders TL, Enikolopov G, Morrison SJ. Endothelial and perivascular cells maintain haematopoietic stem cells. Nature. 2012;481(7382):457-462. 5. Morrison SJ, Scadden DT. The bone marrow niche for haematopoietic stem cells. Nature. 2014;505(7483):327-334. 6. Sugiyama T, Kohara H, Noda M, Nagasawa T. Maintenance of the hematopoietic stem cell pool by CXCL12-CXCR4 chemokine signaling in bone marrow stromal cell niches. Immunity. 2006;25(6):977-988. 7. Greenbaum A, Hsu YM, Day RB, et al. CXCL12 in early mesenchymal progenitors is required for haematopoietic stem-cell maintenance. Nature. 2013;495(7440):227-230. 8. Stier S, Ko Y, Forkert R, et al. Osteopontin is a hematopoietic stem cell niche component that negatively regulates stem cell pool size. J Exp Med. 2005;201(11):1781-1791. 9. Cheng H, Hao S, Liu Y, et al. Leukemic marrow infiltration reveals a novel role for Egr3 as a potent inhibitor of normal hematopoietic stem cell proliferation. Blood. 2015;126(11):1302-1313. 10. Krause DS, Fulzele K, Catic A, et al. Differential regulation of myeloid leukemias by the bone marrow microenvironment. Nat Med. 2013;19(11):1513-1517. 11. Thisse B, el Messal M, Perrin-Schmitt F. The twist gene: isolation of a Drosophila zygotic gene necessary for the establishment of dorsoventral pattern. Nucleic Acids Res. 1987;15(8):3439-3453. 12. Verzi MP, Anderson JP, Dodou E, et al. Ntwist, an evolutionarily conserved bHLH protein expressed in the developing CNS, functions as a transcriptional inhibitor. Dev Biol. 2002;249(1):174-190. 13. Chen ZF, Behringer RR. Twist is required in head mesenchyme for cranial neural tube morphogenesis. Genes Dev. 1995;9(6):686699. 14. Isenmann S, Arthur A, Zannettino AC, et al. TWIST family of basic helix-loop-helix transcription factors mediate human mesenchymal stem cell growth and commitment. Stem Cells. 2009;27(10):2457-2468. 15. Goodnough LH, Chang AT, Treloar C, Yang J, Scacheri PC, Atit RP. Twist1 mediates repression of chondrogenesis by betacatenin to promote cranial bone progenitor specification. Development. 2012;139(23): 4428-4438. 16. Hjiantoniou E, Iseki S, Uney JB, Phylactou LA. DNazyme-mediated cleavage of Twist transcripts and increase in cellular apoptosis. Biochem Biophys Res Commun. 2003;300 (1):178-181. 17. Howard TD, Paznekas WA, Green ED, et al. Mutations in TWIST, a basic helix-loophelix transcription factor, in Saethre-

haematologica | 2018; 103(12)

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

Chotzen syndrome. Nat Genet. 1997;15(1) :36-41. Mahmoud MM, Kim HR, Xing R, et al. TWIST1 integrates endothelial responses to flow in vascular dysfunction and atherosclerosis. Circ Res. 2016;119(3):450-462. Mammoto T, Jiang A, Jiang E, Mammoto A. Role of Twist1 phosphorylation in angiogenesis and pulmonary fibrosis. Am J Respir Cell Mol Biol. 2016;55(5):633-644. Arthur A, Cakouros D, Cooper L, et al. Twist-1 enhances bone marrow mesenchymal stromal cell support of hematopoiesis by modulating CXCL12 expression. Stem Cells. 2016;34(2):504-509. Schreck C, Istvanffy R, Ziegenhain C, et al. Niche WNT5A regulates the actin cytoskeleton during regeneration of hematopoietic stem cells. J Exp Med. 2017;214(1):165-181. Saez B, Ferraro F, Yusuf RZ, et al. Inhibiting stromal cell heparan sulfate synthesis improves stem cell mobilization and enables engraftment without cytotoxic conditioning. Blood. 2014;124(19):2937-2947. Pinho S, Lacombe J, Hanoun M, et al. PDGFRalpha and CD51 mark human nestin+ sphere-forming mesenchymal stem cells capable of hematopoietic progenitor cell expansion. J Exp Med. 2013;210(7):13511367. Nakamura Y, Arai F, Iwasaki H, et al. Isolation and characterization of endosteal niche cell populations that regulate hematopoietic stem cells. Blood. 2010;116(9):1422-1432. Kunisaki Y, Bruns I, Scheiermann C, et al. Arteriolar niches maintain haematopoietic stem cell quiescence. Nature. 2013;502(7473):637-643. Acar M, Kocherlakota KS, Murphy MM, et al. Deep imaging of bone marrow shows non-dividing stem cells are mainly perisinusoidal. Nature. 2015;526(7571):126-130. Ciuculescu MF, Park SY, Canty K, Mathieu R, Silberstein LE, Williams DA. Perivascular deletion of murine Rac reverses the ratio of marrow arterioles and sinusoid vessels and alters hematopoiesis in vivo. Blood. 2015;125(20):3105-3113. Balderman SR, Li AJ, Hoffman CM, et al. Targeting of the bone marrow microenvironment improves outcome in a murine model of myelodysplastic syndrome. Blood. 2016;127(5):616-625. Ratajczak MZ. A novel view of the adult bone marrow stem cell hierarchy and stem cell trafficking. Leukemia. 2015;29(4):776782. Ara T, Tokoyoda K, Sugiyama T, Egawa T, Kawabata K, Nagasawa T. Long-term hematopoietic stem cells require stromal cell-derived factor-1 for colonizing bone marrow during ontogeny. Immunity. 2003;19(2):257-267. Taichman RS, Emerson SG. Human osteoblasts support hematopoiesis through the production of granulocyte colony-stimulating factor. J Exp Med. 1994;179(5):16771682. Mayani H. A glance into somatic stem cell biology: basic principles, new concepts, and clinical relevance. Arch Med Res. 2003;34(1):3-15. Bowers M, Zhang B, Ho Y, Agarwal P, Chen CC, Bhatia R. Osteoblast ablation reduces normal long-term hematopoietic stem cell self-renewal but accelerates leukemia development. Blood. 2015;125(17):2678-2688.

34. Hanoun M, Zhang D, Mizoguchi T, et al. Acute myelogenous leukemia-induced sympathetic neuropathy promotes malignancy in an altered hematopoietic stem cell niche. Cell Stem Cell. 2014;15(3):365-375. 35. Krivtsov AV, Twomey D, Feng Z, et al. Transformation from committed progenitor to leukaemia stem cell initiated by MLLAF9. Nature. 2006;442(7104):818-822. 36. Wang Y, Krivtsov AV, Sinha AU, et al. The Wnt/beta-catenin pathway is required for the development of leukemia stem cells in AML. Science. 2010;327(5973):1650-1653. 37. Yen HY, Ting MC, Maxson RE. Jagged1 functions downstream of Twist1 in the specification of the coronal suture and the formation of a boundary between osteogenic and non-osteogenic cells. Dev Biol. 2010;347(2):258-270. 38. Kamath BM, Stolle C, Bason L, et al. Craniosynostosis in Alagille syndrome. Am J Med Genet. 2002;112(2):176-180. 39. Kode A, Mosialou I, Manavalan SJ, et al. Foxo1-dependent induction of acute myeloid leukemia by osteoblasts in mice. Leukemia. 2016;30(1):1-13. 40. Kode A, Manavalan JS, Mosialou I, et al. Leukaemogenesis induced by an activating beta-catenin mutation in osteoblasts. Nature. 2014;506(7487):240-244. 41. Wang L, Zhang H, Rodriguez S, et al. Notchdependent repression of miR-155 in the bone marrow niche regulates hematopoiesis in an NF-kappaB-dependent manner. Cell Stem Cell. 2014;15(1):51-65. 42. Kim YW, Koo BK, Jeong HW, et al. Defective Notch activation in microenvironment leads to myeloproliferative disease. Blood. 2008;112(12):4628-4638. 43. Ohki Y, Heissig B, Sato Y, et al. Granulocyte colony-stimulating factor promotes neovascularization by releasing vascular endothelial growth factor from neutrophils. FASEB J. 2005;19(14):2005-2007. 44. Raaijmakers MH, Mukherjee S, Guo S, et al. Bone progenitor dysfunction induces myelodysplasia and secondary leukaemia. Nature. 2010;464(7290):852-857. 45. Zambetti NA, Ping Z, Chen S, et al. Mesenchymal inflammation drives genotoxic stress in hematopoietic stem cells and predicts disease evolution in human preleukemia. Cell Stem Cell. 2016;19(5):613627. 46. Chiang MY, Shestova O, Xu L, Aster JC, Pear WS. Divergent effects of supraphysiologic Notch signals on leukemia stem cells and hematopoietic stem cells. Blood. 2013;121(6):905-917. 47. Bernasconi P, Farina M, Boni M, Dambruoso I, Calvello C. Therapeutically targeting SELF-reinforcing leukemic niches in acute myeloid leukemia: a worthy endeavor? Am J Hematol. 2016;91(5):507-517. 48. Huang MM, Zhu J. The regulation of normal and leukemic hematopoietic stem cells by niches. Cancer Microenviron. 2012;5(3):295305. 49. Chen S, Zambetti NA, Bindels EM, et al. Massive parallel RNA sequencing of highly purified mesenchymal elements in low-risk MDS reveals tissue-context-dependent activation of inflammatory programs. Leukemia. 2016;30(9):1938-1942. 50. Dong CY, Liu XY, Wang N, et al. Twist-1, a novel regulator of hematopoietic stem cell self-renewal and myeloid lineage development. Stem Cells. 2014;32(12):3173-3182.

1979

ARTICLE

Hematopoiesis

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):1980-1990

ASXL2 regulates hematopoiesis in mice and its deficiency promotes myeloid expansion Vikas Madan,1 Lin Han,1,2 Norimichi Hattori,1,3 Weoi Woon Teoh,1 Anand Mayakonda,1 Qiao-Yang Sun,1 Ling-Wen Ding,1 Hazimah Binte Mohd Nordin,1 Su Lin Lim,1 Pavithra Shyamsunder,1 Pushkar Dakle,1 Janani Sundaresan,1 Ngan B. Doan,4 Masashi Sanada,5,6 Aiko Sato-Otsubo,6 Manja Meggendorfer,7 Henry Yang,1 Jonathan W. Said,4 Seishi Ogawa,6 Torsten Haferlach,7 Der-Cherng Liang,8 Lee-Yung Shih,9 Tsuyoshi Nakamaki,3 Q. Tian Wang10 and H. Phillip Koeffler1,11,12

1 Cancer Science Institute of Singapore, National University of Singapore; 2Department of Medicine, Yong Loo Lin School of Medicine, National University of Singapore; 3 Division of Hematology, Department of Medicine, School of Medicine, Showa University, Shinagawa-Ku, Tokyo, Japan; 4Department of Pathology and Laboratory Medicine, Santa Monica-University of California-Los Angeles Medical Center, Los Angeles, CA, USA; 5 Department of Advanced Diagnosis, Clinical Research Center, National Hospital Organization Nagoya Medical Center, Japan; 6Department of Pathology and Tumor Biology, Graduate School of Medicine, Kyoto University, Japan; 7MLL Munich Leukemia Laboratory, Germany; 8Division of Pediatric Hematology-Oncology, Mackay Memorial Hospital and Mackay Medical College, Taipei, Taiwan; 9Division of HematologyOncology, Department of Internal Medicine, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan; 10Department of Biological Sciences, University of Illinois at Chicago, IL, USA; 11Cedars-Sinai Medical Center, Division of Hematology/Oncology, UCLA School of Medicine, Los Angeles, CA, USA and 12 Department of Hematology-Oncology, National University Cancer Institute of Singapore (NCIS), National University Hospital, Singapore

LH and NH contributed equally to this work

ABSTRACT

Correspondence: csivm@nus.edu.sg or nhattor@med.showa-u.ac.jp or sly7012@cgmh.org.tw Received: January 31, 2018. Accepted: July 26, 2018. Pre-published: August 9, 2018. doi:10.3324/haematol.2018.189928 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/1980

hromosomal translocation t(8;21)(q22;q22) which leads to the generation of oncogenic RUNX1-RUNX1T1 (AML1-ETO) fusion is observed in approximately 10% of acute myelogenous leukemia (AML). To identify somatic mutations that co-operate with t(8;21)-driven leukemia, we performed whole and targeted exome sequencing of an Asian cohort at diagnosis and relapse. We identified high frequency of truncating alterations in ASXL2 along with recurrent mutations of KIT, TET2, MGA, FLT3, and DHX15 in this subtype of AML. To investigate in depth the role of ASXL2 in normal hematopoiesis, we utilized a mouse model of ASXL2 deficiency. Loss of ASXL2 caused progressive hematopoietic defects characterized by myeloid hyperplasia, splenomegaly, extramedullary hematopoiesis, and poor reconstitution ability in transplantation models. Parallel analyses of young and >1-year old Asxl2-deficient mice revealed age-dependent perturbations affecting, not only myeloid and erythroid differentiation, but also maturation of lymphoid cells. Overall, these findings establish a critical role for ASXL2 in maintaining steady state hematopoiesis, and provide insights into how its loss primes the expansion of myeloid cells.

ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1980

Introduction Chromosomal translocation t(8;21) (q22;q22) is a frequent cytogenetic abnormality observed in approximately 10% of acute myelogenous leukemia (AML). This rearrangement involves RUNX1 gene on chromosome 21 and RUNX1T1 gene on chromosome 8 and results in generation of oncogenic RUNX1-RUNX1T1 fusion protein (also known as AML1-ETO).1 However, expression of RUNX1-RUNX1T1 alone, either as a transgene or using viral transduction in hematopoietic cells, is insufficient to cause leukemia in mice.2-4 Loss of sex chromosome and other chromosomal aberrations, as well as recurrent somatic mutations of KIT, FLT3, NRAS and KRAS, are associated with t(8;21) AML.2,4 Recent high-throughput sequencing haematologica | 2018; 103(12)

Progressive hematopoietic defects in ASXL2 KO mice

approach has identified additional mutagenic events involving ASXL2, ASXL1 and DHX15 genes in t(8;21) AML.5-10 Several secondary genetic events that co-operate with RUNX1-RUNX1T1 fusion in inducing leukemia have been demonstrated using mouse models.2-4,11 ASXL1, ASXL2 and ASXL3 are human homologs of Drosophila Asx (Additional sex combs) and function as epigenetic regulators through recruitment of Polycomb group repressor complexes (PRC) and Trithorax group activator complexes.12,13 ASXL proteins can interact with BAP1, NCOA1, EZH2, WTIP and nuclear receptors, which suggests diverse functions of ASXL family members in epigenetic and transcriptional regulation.12,13 Somatic mutations of ASXL1 occur in a range of hematologic disorders,14-20 and its deletion in mouse hematopoietic cells results in multilineage cytopenias and dysplasia, which are associated with global reduction of H3K27 trimethylation.21 In contrast, somatic mutations of ASXL2 are observed exclusively in t(8;21) AML,5,8 while mutations of ASXL3 are not reported in hematologic malignancies.13 Silencing of ASXL2 is partially embryonic lethal depending on the genetic background, and leads to congenital heart defects. Surviving homozygous Asxl2 mutant mice are smaller than wild-type littermates, demonstrate skeletal homeotic transformations consistent with disruption of Polycomb/Trithorax complex functions, develop cardiac dysfunction, and exhibit decreased bone mineral density.22-26 More recently, its role in normal hematopoietic development and leukemic development has been investigated.27,28 In the present study, we used exome sequencing to identify somatic mutations co-operating with RUNX1RUNX1T1 rearrangement in AML. Frequent truncating mutations of ASXL2 were identified in our t(8;21) AML cohort. Using an Asxl2-deficient mouse model, we aimed to clarify the function of ASXL2 in hematopoiesis. We demonstrate that ASXL2 is crucial to maintain hematopoietic stem cell (HSC) number and function. Loss of ASXL2 leads to myeloproliferation and extramedullary hematopoiesis, as well as to progressive defects in multilineage differentiation. These results establish ASXL2 as a key component of the epigenetic machinery involved in maintaining hematopoietic development.

Methods Acute myelogenous leukemia samples t(8;21) AML was diagnosed by karyotypic analysis and/or reverse transcriptase PCR assay for the detection of fusion transcript of RUNX1-RUNX1T1. Samples were collected with informed consent at diagnosis, during complete remission, and at time of frank relapse in accordance with the Declaration of Helsinki. Bone marrow (BM) mononuclear cells were obtained by Ficoll-Hypaque density gradient centrifugation (1.077 g/mL; Amersham Pharmacia, Sweden). The study was approved by the institutional review boards of the respective institutes.

Enrichment System for Illumina Multiplexed Sequencing was used. RNA baits were designed to capture exons of 530 genes (Online Supplementary Table S1) and libraries were sequenced on HiSeq 2000 (Illumina). 100 bp paired-end reads were aligned to human reference genome (reference build: hg19) using bwa-mem aligner with default parameters. PCR duplicate reads were marked with samblaster.30 Resulting BAM files were further processed according to GTAK best practices including INDEL re-alignment and base quality recalibration (https://software.broadinstitute.org/gatk/best-practices/bp_3step.php?case=GermShortWGS). Somatic variants were detected using Varscan2 somatic command.31 Raw variants were processed with processSomatic command with the P-value set to 0.05 to obtain high confident variants. Variants were further processed using fpFilter per script to remove potential false positives (https://github.com/ckandoth/variant-filter). Resulting variants were annotated using Variant Effect Predictor and filtered against germline variants in both dbSNP and ExAC, while keeping deleterious and clinically significant variants.32 For samples without germline controls, variants were called using MuTect using a panel of normal derived from in-house cohort of AML remission samples, as described previously.29,33 Oncoplots were drawn using maftools Bioconductor package.34 All the somatic mutations for whole exome sequencing and about 90% mutations reported for targeted exome sequencing were validated using PCR amplification and Sanger sequencing.

Mice Mice with Asxl2 gene-trap allele [referred to as knockout (KO) allele in this study] have been described previously.24 Asxl2 heterozygous KO mice were maintained on C57BL/6J (B6) and 129Sv (129) genetic backgrounds. Asxl2 homozygous null mice [and corresponding wild-type (WT) controls] used in all experiments were B6 x 129 F1 obtained by crossing Asxl2 heterozygous B6 and 129 mice. Mouse colonies were housed and maintained at the animal facility of Comparative Medicine Centre, National University of Singapore (NUS). All mice experiments were approved by Institutional Animal Care and Use Committee, NUS, Singapore.

Flow cytometry Cells were incubated with fluorochrome-conjugated antibodies for 30 min on ice, washed and resuspended in SYTOX Blue Dead Cell Stain (ThermoFisher Scientific) before acquisition on FACS LSR II flow cytometer (BD Biosciences). Sorting of cells was performed on FACSAria cell sorter (BD Biosciences) and data were analyzed using FACSDIVA software (BD Biosciences). See Online Supplementary Table S2 for the list of antibodies used for flow cytometric staining.

Statistical analysis All statistical analyses were performed using GraphPad Prism 7 software.

Accession numbers The accession numbers for the sequencing data reported in this study are SRP122878 (exome sequencing) and GSE106798 (RNA sequencing).

Exome sequencing and somatic variant discovery DNA was sheared using a Covaris instrument and assessed on a 2100 Bioanalyzer. Library preparation and exome sequencing were performed as described previously.29 For whole exome sequencing, DNA were captured using SureSelect Human All Exon 50Mb kit (Agilent), according to the manufacturerâ&#x20AC;&#x2122;s instructions. For targeted capture library, Agilentâ&#x20AC;&#x2122;s SureSelect XT2 Target haematologica | 2018; 103(12)

Results Mutational profile of RUNX1-RUNX1T1 AML We performed whole-exome sequencing of 10 paired samples of newly-diagnosed and relapsed AML with RUNX1-RUNX1T1 rearrangement from an Asian cohort, 1981

V. Madan et al.

Figure 1. Mutational landscape of t(8;21) acute myelogenous leukemia (AML). Matrix displays individual somatic mutations detected in diagnosis and relapse AML with RUNX1-RUNX1T1 fusion. Mutational frequencies are illustrated on the left and bar graphs on the right depict absolute number and type of mutations. The annotation bars at the bottom display patient information including the cytogenetic aberrations detected in karyotype analysis. NA: information not available.

along with matched germline (complete remission) DNA (Online Supplementary Table S3). We achieved a mean depth of 100x (range 64-271x); an average 68% of nucleotides were covered by at least 20 reads (Online Supplementary Figure S1). We identified 55 somatic mutations (in 52 genes) in newly-diagnosed cases and 76 mutations (in 69 genes) at relapse, which included recurrent mutations in KIT, TET2, DHX15 and MGA (Online Supplementary Table S4). We assessed the stability of mutations in diagnosis and relapse samples analyzed using whole exome sequencing. Overall, the disease evolution followed the pattern reported previously for AML,35 as the founding clone or a subclone at diagnosis survived the therapy, gained additional mutations, and expanded at relapse (Online Supplementary Figure S2). Interestingly, all four mutations of TET2 (in 3 cases) were acquired at relapse (Online Supplementary Table S4). Further to uncovering the repertoire of co-operating mutations in t(8;21) AML, we analyzed mutational status of 530 genes (Online Supplementary Table S1) in 76 newlydiagnosed and 19 relapse t(8;21) AML cases using targeted-exome sequencing (Online Supplementary Table S3). In this cohort, the mean sequencing coverage across targeted bases was 96x (range 76-319x), with 70% of bases covered greater than 20x (Online Supplementary Figure S1). Overall, we observed that the mutations of KIT were most frequent at both diagnosis and relapse in our cohort, along with recurring alterations in ASXL2, MGA, DHX15, TET2 and FLT3 genes (Figure 1, Online Supplementary Table 1982

S5 and Online Supplementary Figure S3). Somatic mutations of ASXL2 were detected in 20% of newly-diagnosed (17 of 86) and 10% of relapsed (3 of 29) cases while ASXL1 was mutated at a much lower frequency (2% of newlydiagnosed cases and 7% of relapsed cases) (Figure 1, Online Supplementary Table S5 and Online Supplementary Figure S3). Mutations of DHX15, a RNA helicase, occurred exclusively at Arg222 residue in 7 cases. MGA and TET2 genes predominantly harbored nonsense and frameshift mutations spread throughout the transcript (Figure 1, Online Supplementary Figure S3 and Online Supplementary Table S5). ASXL2 was the second most frequently mutated gene, and interestingly, all alterations were truncating mutations located in exons 11 and 12 (Figure 1 and Online Supplementary Figure S3). Unlike alterations of ASXL1, which are recurrent in several hematologic diseases, recent reports5,8,9 have highlighted the incidence of ASXL2 mutations specifically in the t(8;21) subtype of AML. In this study, we aimed to investigate the consequences of deficiency of ASXL2 on hematopoietic development using a mouse model.

Impaired hematopoiesis in Asxl2-deficient mice Our RT-PCR analysis showed that Asxl2 is expressed in a broad range of murine hematopoietic cell types (Online Supplementary Figure S4A). To investigate the physiological role of ASXL2 in steady-state hematopoiesis, we used genetrap mice (referred to as Asxl2 KO mice) described previoushaematologica | 2018; 103(12)

Progressive hematopoietic defects in ASXL2 KO mice

Figure 2 Myeloproliferation and extramedullary hematopoiesis in Asxl2-deficient mice. (A) White blood cell counts in peripheral blood of 8-10 week old wild-type (WT) and Asxl2-deficient mice. Whiskers extend from the minimum to the maximum values. (B) Proportion of B cells (CD19+), T cells (CD3+), and myeloid cells (CD11b+) in the peripheral blood of >1-year old WT and Asxl2 knockout (KO) mice determined using flow cytometry (B and T cells, n=3; myeloid cells, n=11). (C) Representative photograph of spleens isolated from young (11-week old) and old (>1-year old) WT and Asxl2 KO mice. (D) Total leukocyte counts in spleens from >1-year old mice. (E and F) Frequencies (E) and absolute numbers (F) of lymphoid and myeloid cells in spleens of old WT and KO mice. Flow-cytometric analysis on splenocytes was similar to (B) (n=12-13). (G) Frequency of LSK cells in spleen and peripheral blood of old WT and KO mice. (H) Proportion of erythroid progenitors detected in flow cytometric analysis (staining with CD71 and TER119 antibodies) of spleens from old WT and KO mice (n=10). Bars represent meanÂąStandard Error of Mean. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns: not significant.

ly.24 As reported,24 Asxl2 KO mice had a significantly shorter life span compared with the WT mice (Online Supplementary Figure S4B). We observed a significant reduction in Asxl2 transcript and protein levels in the BM, spleen and thymus of Asxl2 KO mice compared to WT mice (Online Supplementary Figure S5A-F), whereas the expression of Asxl1 was not affected (Online Supplementary Figure S5G-I). We cultured the BM cells from WT and KO mice for two weeks in myeloid differentiation-promoting conditions and observed an increase in the proportion of CD11b+ cells with a concomitant decrease in Linâ&#x20AC;&#x201C;Kit+ cells in the cultures from Asxl2-deficient BM, suggesting altered myeloid differentiation (Online Supplementary Figure S6A and B). Moreover, in re-plating assays, we observed a sustained ability of Asxl2 KO BM cells to generate myeloid colonies compared with WT cells (Online Supplementary Figure S6C). To gain initial insights into the effect of ASXL2 deficiency on hematopoiesis in vivo, we periodically analyzed the haematologica | 2018; 103(12)

peripheral blood (PB) from Asxl2 KO and WT mice beginning at eight weeks of age. We observed an age-dependent increase in white leukocyte count in the PB of Asxl2deficient mice, while the numbers were unchanged in the WT mice (Figure 2A). This was associated with reduced red blood cell (RBC) counts and hemoglobin in peripheral blood of Asxl2 KO mice with increasing age (Online Supplementary Figure S7). This observation suggested progressive defects in the hematopoietic compartment in Asxl2 KO mice, and therefore we investigated systematically hematopoietic development in >1-year old mice. Flow cytometric analysis of blood leukocytes from these mice showed a higher proportion of myeloid (CD11b+) and a reduced proportion of lymphoid (CD19+ and CD3+) cells in KO mice (Figure 2B and Online Supplementary Figure S8A). Old Asxl2 KO mice exhibited extensive splenomegaly (Figure 2C and D) which was associated with marked myeloproliferation evident by increased fre1983

V. Madan et al. A

Figure 3. ASXL2 deficiency perturbs frequency and function of hematopoietic stem cells (HSCs.) (A) Representative photograph of tibias and femurs, respectively, from >1-year old wild-type (WT) and Asxl2 knockout (KO) mice. (B) Bone marrow (BM) cellularity (two femurs and two tibias) in old WT and Asxl2 KO mice. (C) Proportion of LSK cells in the BM of old mice (n=14). (D) Frequency of BrdU+ LSK cells in the BM of 10-20-weeks old WT and Asxl2 deficient mice (n=5). (E) Proportions of CMP, GMP and MEP in the BM of >1-year old WT and KO mice (n=12). (F) Frequency of erythroid precursors (proE: CD71+TER119lo, EryA: CD71+TER119+FSChi, EryB: CD71+TER119+FSClo and EryC: CD71–TER119+FSClo) in BM of >1-year old WT and Asxl2 KO mice (n=8). (G-J) Competitive repopulation assay: 4000 (black circles) or 2000 (gray circles) LSK cells were injected in recipient mice and proportion of donor-derived cells (CD45.2+) in peripheral blood was determined at 4, 8, 12 and 16 weeks post transplantation. Ability of WT and Asxl2 KO LSK cells to reconstitute B cells (CD19+) (G), T cells (CD3+) (H), granulocytes (CD11b+Gr1+), (I) and monocytes (CD11b+Gr1–F4/80+) (J) was analyzed using flow cytometry. Bars represent mean±Standard Error of Mean. *P<0.05, ***P<0.001, ****P<0.0001, ns: not significant.

1984

haematologica | 2018; 103(12)

Progressive hematopoietic defects in ASXL2 KO mice

Figure 4. ASXL2 is essential for normal thymocyte maturation. (A) Thymocyte count in >1-year old wild-type (WT) and Asxl2 knockout (KO) mice. (B) Representative FACS plots depict the proportion of DN (CD4–CD8–), DP (CD4+CD8+), CD4+ single positive (SP) and CD8+ SP cells in the thymus of old WT and Asxl2 KO mice. (C) Frequencies of DP, DN, CD4+ SP and CD8+ SP populations in old WT (n=17) and Asxl2 KO (n=18) mice. (D) FACS plots show representative flow cytometric staining for surface expression of CD44 and CD25 within the DN population in the thymus of old mice. (E) Proportions of DN1 (CD44+CD25– DN), DN2 (CD44+CD25+ DN), DN3 (CD44–CD25+ DN) and DN4 (CD44–CD25– DN) sub-populations within the DN compartment of thymus of old WT and KO mice. (F and G) Frequencies of myeloid cells (CD11b+) (F) and B cells (CD19+) (G) in the thymus of >1-year old WT and KO mice (n=5-7). Data are represented as mean±Standard Error of Mean. ***P<0.001, ****P<0.0001.

quency of CD11b+ cells and myeloperoxidase-positive cells (Figure 2E and Online Supplementary Figure S8B and C). The absolute numbers of CD11b+ myeloid cells were significantly increased (20-fold) while the number of lymphoid cells were 2-fold higher in the spleens of KO mice compared with the WT mice (Figure 2F). Histological examination of spleens demonstrated loss of normal architecture in KO mice compared with the WT mice (data not shown). Moreover, extramedullary hematopoiesis in the KO spleens was evident by an elevated proportion and numbers of total Lin–Kit+ and Lin–Kit+Sca1+ (LSK) cells (Figure 2G and Online Supplementary Figure S9A-D) as well as significantly higher frequency of erythroid progenitors (populations proE, EryA and EryB) (Figure 2H and Online Supplementary Figure S9E). We also detected markedly increased frequency of stem/progenitor cells in the peripheral blood of old KO mice (Figure 2G and Online Supplementary Figure S9F). Parallel analyses of spleens from young KO mice (8-14 weeks old) revealed modestly increased spleen size with a trend towards elevated proportion and number of LSK cells (Online Supplementary Figure S10A-C). A propensity for increased frequencies of myeloid cells (CD11b+) and erythroid progenitors was also noted in the spleens of young KO mice (Online Supplementary Figure S10D and E). This indicates an onset of extramedullary hematopoiesis in young mice, which manifests by marked myeloid and erythroid cell expansion as the mice grow older.

ASXL2 deficiency results in defective differentiation and function of hematopoietic stem cells ASXL2 deficiency resulted in paler bones and decreased marrow cellularity in old mice (Figure 3A and B). Notably, BM of old Asxl2 KO mice had higher frequency and absolute number of LSK cells compared with the WT mice haematologica | 2018; 103(12)

(Figure 3C and Online Supplementary Figure S11A and B). The proportion of LT-HSCs (CD34–Flt3–LSK), ST-HSCs (CD34+Flt3–LSK) and MPPs (CD34+Flt3+LSK) within the LSK compartment was largely unaltered (Online Supplementary Figure S11C), albeit an overall increase in the absolute number of these subpopulations occurred in the old KO mice (Online Supplementary Figure S11D and E). We also observed reduced BM cellularity in young KO mice; and although the proportion of LSK cells was not significantly altered, the frequency and absolute numbers of LT-HSCs were significantly higher, suggesting early defects in maintaining HSC frequency (Online Supplementary Figure S11F-J). In vivo BrdU incorporation assay uncovered a significantly higher frequency of BrdU+ LSK cells in the BM of Asxl2 KO mice, indicating increased cycling of stem/progenitor cells lacking ASXL2 (Figure 3D). These results illustrate that ASXL2 is required for maintaining the number and self-renewal of HSCs during steady-state hematopoiesis. Flow cytometric analyses also demonstrated decreased proportion and number of common myeloid precursors (CMP; Lin–Kit+Sca1–CD34+FcgRII/IIIlo) and granulocyte monocyte precursors (GMP; Lin–Kit+Sca1– CD34+FcgRII/IIIhi) in >1-year old Asxl2 KO mice (Figure 3E and Online Supplementary Figure S12A and B). Furthermore, a marked reduction in the frequencies of erythroid precursors was noted in the old KO mice (Figure 3F and Online Supplementary Figure S12C), indicating impaired erythropoiesis. Reduction in erythroid precursors as well as CMP and GMP populations was also apparent in the BM of the young KO mice, signifying an early inception of defects in erythroid and myeloid differentiation because of ASXL2 deficiency (Online Supplementary Figure S12D-F). Further, multi-lineage reconstitution ability of Asxl2-deficient HSCs was assessed in a competitive repop1985

V. Madan et al.

Figure 5. Impaired B-cell lymphopoiesis in Asxl2-deficient mice. (A and B) Proportion (A) and absolute number (B) of cells at different stages of B-cell development in the bone marrow of >1-year old WT and Asxl2 KO mice (n=4-6). proB: CD43+B220+, preB: CD43–B220+IgM–, immature B (Imm. B): CD43–B220+IgM+, mature B (Mat. B): CD43–B220hiIgM+, Fraction A (Fr. A): CD24–BP1– proB cells, Fraction B (Fr. B): CD24+BP1– proB cells, Fraction C (Fr. C): CD24+BP1+ proB cells. Data are represented as mean±Standard Error of Mean. *P<0.05, **P<0.01, ***P<0.001, ns: not significant.

ulation assay. ASXL2 deficiency resulted in poor reconstitution of lymphoid lineage (both B and T cells) compared with the WT cells (Figure 3G and H). The reconstitution of granulocytes was also impaired in mice transplanted with Asxl2-deficient LSK cells (Figure 3I), while the proportion of donor-derived monocytes was similar to the recipients transplanted with WT cells (Figure 3J). These repopulation assays demonstrated that ASXL2 is essential for differentiation into multiple hematopoietic lineages including the lymphoid lineage, prompting us to investigate the development of lymphoid lineage in the Asxl2 KO mice.

Impaired thymopoiesis in mice lacking ASXL2 A marked reduction of thymus size and cellularity was a consistent feature in >1-year old Asxl2 KO mice compared with the WT mice (Figure 4A). Flow cytometric analyses revealed a striking reduction in CD4+CD8+ double positive (DP) cells and a concomitant increase in the proportion and number of CD4–CD8– double negative (DN) cells in the thymi of Asxl2 KO mice (Figure 4B and C and Online Supplementary Figure S13A). Within the DN compartment, the proportion of DN1 (CD44+CD25–) subpopulation was unchanged, whereas a significant reduction was detected in the proportion of DN2 (CD44+CD25+) and DN3 (CD44–CD25+) subpopulations, indicating a partial block in differentiation of earliest thymocytes (Figure 4D and E). A large number of DN cells in the thymi of KO mice consisted primarily of CD25– cells, 1986

present in the DN1 and DN4 gates (Figure 4D and Online Supplementary Figure S13B). Further characterization showed that the vast majority of cells in the DN gate were CD19+ (which contributed to 40% of total thymus-residing hematopoietic cells), and the frequency and number of CD11b+ myeloid cells were also noticeably increased in the thymus of KO mice compared with the WT mice (Figure 4F and G and Online Supplementary Figure S13C-F). Young Asxl2 KO mice exhibited decreased thymus cellularity, but displayed less pronounced effect on major thymic compartments compared with the >1-year old mice (Online Supplementary Figure S13G-I). These findings illustrate that ASXL2 is essential for maintaining normal thymopoiesis in mice and its deficiency leads to progressive impairment in thymocyte maturation, which is manifested by a partial block from the DN to the DP stage, and an accumulation of myeloid and B cells in the thymus of the old mice.

Defects in B-cell development in Asxl2 knockout mice We also examined the consequences of ASXL2 deficiency on B-cell development in the BM by examining the expression of several surface antigens, which define the advancing stages of B-cell maturation. While the young Asxl2 KO mice showed largely conserved subsets during B-cell lymphopoiesis, a notable decrease in the proportion of mature B cells (CD43–IgM+B220hi) compared with the WT mice was observed (Online Supplementary Figure S14). haematologica | 2018; 103(12)

Progressive hematopoietic defects in ASXL2 KO mice

Figure 6. Cell-intrinsic effects of Asxl2 deficiency on lymphoid and myeloid lineages. Lethally irradiated mice transplanted with either wild-type (WT) or Asxl2 knockout (KO) bone marrow (BM) cells were analyzed after one year. (A) Number of white blood cells (WBCs) in the peripheral blood of recipient mice. (B) Number of thymocytes in recipient mice. (C) Proportion of donor-derived (CD45.2+) DN, DP, CD4+SP and CD8+SP cells in thymi of recipient mice. (D) Spleen cellularity. (E) Percentages of CD45.2+ B, T and myeloid cells in the spleen of mice transplanted with either WT or KO cells. (F) Frequencies of erythroid precursors in spleens of recipient mice. Bars represent mean±Standard Error of Mean. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

However, the BM of old KO mice showed a substantial reduction in the frequencies and absolute number of proB (CD43+B220+), preB (CD43–IgM–B220+), immature B (CD43–IgM+B220+), and mature B cells (Figure 5A and B), indicating an age-dependent paucity in B-cell development. Further fractionation of proB cells using the scheme proposed previously,36 revealed a lower proportion of CD24–BP1– (Fraction A) and CD24+BP1– (Fraction B) proB cells in the old Asxl2 KO mice (Figure 5A and B), demonstrating a partial arrest of B-cell maturation.

Cell-intrinsic role of ASXL2 in hematopoiesis To test the cell-autonomous function of ASXL2 in hematopoietic development, we transplanted lethally irradiated mice with either WT or Asxl2 KO BM cells and analyzed the hematopoietic compartment one year later. We observed that the recipient mice transplanted with Asxl2 KO BM cells tended to have higher WBC counts in peripheral blood (Figure 6A), similar to the age-dependent increase in WBC observed in KO mice (Figure 2A). We also noted decreased thymocyte cellularity and accumulation of donor-derived DN cells in the thymi of mice transplanted with Asxl2 KO BM compared with those transplanted with WT BM (Figure 6B and C). Spleens of Asxl2 KO BM-recipients were enlarged and exhibited elevated proportion of myeloid cells (Figure 6D and E). Moreover, our analysis also indicated a higher proportion of erythroid precursors in spleens of mice reconstituted with Asxl2 KO BM cells compared with the WT cells (Figure 6F), similar to the phenotype observed in old Asxl2 KO mice. Overall, the recipient mice recapitulated hematopoihaematologica | 2018; 103(12)

etic features of old Asxl2 KO mice, suggesting a hematopoietic cell-intrinsic function of ASXL2.

Identification of altered gene expression in ASXL2-deficient hematopoietic stem cells To determine the molecular basis of the defects observed in Asxl2 KO HSCs, we performed global gene expression profiling (RNA-Sequencing) of sorted LSK cells from old and young WT and Asxl2 KO mice. Comparison of transcriptomic profiles of LSK cells from old WT and Asxl2 KO identified more than 2500 genes differentially expressed (983 genes up-regulated and 1653 genes downregulated in the KO cells; FDR<0.1) (Online Supplementary Table S6), including those involved in myeloid differentiation such as Csf1, Gfi1b, Gata2, Hoxa9, Hoxa5, Mpl, Cdk6, Ccr1 and Ets1 (Online Supplementary Figure S15A). Asxl2-deficient LSK cells from old mice exhibited more pronounced changes in gene expression compared with the young mice (Figure 7A and B and Online Supplementary Figure S15B). However, a significant overlap of genes commonly up-regulated and down-regulated in the KO LSK cells was noted within the two age groups (Figure 7C and Online Supplementary Table S6). Interestingly, GSEA analysis revealed that the expression of RUNX1-RUNX1T1 targets37 inversely correlated in the KO cells (Figure 7D and Online Supplementary Figure S15C). Also, genes down-regulated in immature BM progenitors upon silencing of CBFA2T3 (also called ETO2)38 were also suppressed in the LSK cells from both old and young Asxl2 KO mice (Figure 7E and Online Supplementary Figure S15D). CBFA2T3 is involved in the translocation t(16;21)(q24;q22) with 1987

V. Madan et al.

RUNX1 in AML, a disease with phenotypic and transcriptional profile reminiscent of t(8;21) AML.39,40

Discussion This study, along with recent reports,5-10 provides a comprehensive landscape of the mutational spectrum of t(8;21) AML, which is distinctive from other subtypes of AML and characterized by high frequency of alterations of KIT, ASXL2, DHX15 and MGA genes; and rare mutations

of other commonly altered genes in myeloid leukemia (DNMT3A, RUNX1, IDH2, NPM1). Truncating mutations of ASXL2 occurred in 20% of newly-diagnosed t(8;21) AML cases, and were localized to exons 11 and 12, a pattern highly reminiscent of the mutational profile for ASXL1 in hematologic malignancies.41,42 However, unlike ASXL1, which is altered in a diverse range of hematologic diseases, mutations of ASXL2 are specific to the t(8;21) AML subtype. We also noted that the mutational frequency of RAS pathway genes (NRAS and KRAS) and ZBTB7A in our Asian cohort of t(8;21) AML were notably lower

1988

Figure 7. Gene expression changes in Asxl2-deficient LSK cells. (A and B) Changes in transcript levels in LSK cells from >1year old (A) and young (8-12week old) (B) Asxl2 knockout (KO) versus wild-type (WT) littermates. For each age group and genotype, cells from 2 mice were analyzed. Genes significantly upregulated in KO cells are represented in red circles while downregulated genes are shown in blue (FDR<0.1). (C) Venn diagrams show overlap of genes either up-regulated or down-regulated in young and old Asxl2 KO LSK cells compared with the WT cells from young and old mice, respectively. (D and E) GSEA comparing expression of genes in LSK cells from >1-year old WT and KO littermates for the selected gene sets (FDR<0.1): genes up-regulated in normal hematopoietic progenitors by RUNX1-RUNX1T1 fusion (D), and genes down-regulated in hematopoietic progenitors in CBFA2T3 KO mice (E). NES: normalized enrichment score.

haematologica | 2018; 103(12)

Progressive hematopoietic defects in ASXL2 KO mice

than previous reports from Western populations.5,7-9,40,43 We obtained a mutational rate of 6% for NRAS and 2% for KRAS compared to 13-26% for NRAS and 4-7% for KRAS observed in previous studies.5,7-9,43 Similarly, ZBTB7A mutations that are frequent in Caucasian t(8;21) AML (923%)9,40,43 were observed in only 3% of our Asian patient cohort. Although this might reflect a true difference between genetic backgrounds, this requires further confirmation in independent t(8;21) AML cohorts from different genetic backgrounds. In this study, we focused on functional characterization of Asxl2 in hematopoietic differentiation using a mouse model of Asxl2 deficiency. Mice used in this study were C57BL/6 x 129Sv F1 as the animals in either C57BL/6 or 129Sv background die perinatally,23 which precludes investigation of Asxl2 knockout on an inbred genetic background. Moreover, this mouse model exhibits constitutive loss of Asxl2, which is distinct from the truncating somatic mutations of ASXL2 observed in t(8;21) AML that may possibly generate a C-terminal truncated protein. Despite these limitations, our study strengthens the findings concerning the key role of ASXL2 in hematopoiesis reported in two recent studies.27,28 Furthermore, our study reports several additional features of Asxl2 deficient mice, which had not been described previously. While the significance of ASXL2 in maintaining normal hematopoiesis is evidenced from these studies, we performed a parallel indepth analysis in young (8-14 weeks old) versus old (>1year old) mice, which helped establish the progressive phenotype associated with Asxl2 deficiency, including uncovering the crucial function of ASXL2 in development of the lymphoid lineage. Age-dependent defects in hematopoiesis in the knockout mice were characterized by an increased proportion of LSK cells and reduced frequencies of CMP and GMP cells in the BM of >1-year old KO mice. Asxl2-deficient mice also exhibited a progressive myeloproliferative phenotype, accompanied by increased peripheral WBC counts, splenomegaly and extramedullary hematopoiesis, indicating perturbed hematopoiesis. In addition, erythroid maturation was impaired in the old mice lacking ASXL2, signifying agedependent defects in erythroid development. Both previous studies showed that ASXL2 deficiency led to poor reconstitution ability of HSCs in transplantation models, consistent with our findings. We observed that Asxl2-deficient LSK cells exhibited poor reconstitution ability of the lymphoid lineage and flow cytometric analyses of >1-year old mice demonstrated defects in both T-cell maturation in the thymus and B-cell development in the BM of the old KO mice. Apart from the intrinsic effect of ASXL2 on lymphoid differentiation, chronic overproduction of myeloid cells and accompanying inflammatory signals in the mice lacking ASXL2 may also negatively affect the lymphoid output. These findings establish ASXL2 as an essential

References 1. Downing JR. The AML1-ETO chimaeric transcription factor in acute myeloid leukaemia: biology and clinical significance. Br J Haematol. 1999;106(2):296-308. 2. Peterson LF, Boyapati A, Ahn EY, et al. Acute myeloid leukemia with the 8q22;21q22 translocation: secondary muta-

haematologica | 2018; 103(12)

component of hematopoietic development in mice. Gene expression analysis of sorted LSK cells from WT and Asxl2 KO BM identified several key regulators of hematopoiesis as downstream targets of ASXL2 in HSCs. Prominently, GSEA analysis revealed that the expression of genes regulated by RUNX1-RUNX1T1 correlated with the expression of ASXL2, indicating ASXL2 is possibly required for transcriptional activity of RUNX1-RUNX1T1. An identical finding was described recently for Asxl2-deficient LSK cells in an independent mouse model. Here the authors also detected binding of ASXL2 to the genomic loci occupied by RUNX1 and RUNX1-RUNX1T1, although no direct interaction between ASXL2 and either RUNX1 or RUNX1-RUNX1T1 was observed.27 Further studies are needed to investigate whether ASXL2 acts as a likely co-regulator of RUNX1-RUNX1T1 during leukemogenesis. GSEA also indicated a correlation between Asxl2 deficiency and the expression levels of reported targets of BMI1 (a PRC1 member)44 and SUZ12 (a PRC2 member)45 (Online Supplementary Figure S15E and F), consistent with a role for ASXL2 in transcriptional regulation mediated through the PRC complexes.12 In summary, the current study identified frequent ASXL2 mutations in t(8;21) AML and characterized its role in hematopoietic development in mice. This work demonstrates that ASXL2 plays a critical role in multi-lineage differentiation and highlights how its loss leads to progressive hematopoietic defects and promotes myeloid expansion, thereby advancing our understanding of epigenetic machinery that regulates hematopoiesis. Acknowledgments We would like to thank the staff of Comparative Medicine, NUS for their support in mice maintenance and experiments. We would also like to acknowledge expert help and support from the FACS facility at CSI, Singapore. We also appreciate the help of Dr. Motomi Osato, CSI, Singapore, for providing reagents and Dr. Maya Jeitany for critical reading of the manuscript and useful discussions. We thank the Melamed Family and Reuben Yeroushalmi for their generous support. Funding This work was funded by the Leukemia and Lymphoma Society, the Singapore Ministry of Healthâ&#x20AC;&#x2122;s National Medical Research Council (NMRC) under its Singapore Translational Research (STaR) Investigator Award to HPK (NMRC/STaR/0021/2014), Singapore Ministry of Education Academic Research Fund Tier 2 (MOE2013-T2-2-150), the NMRC Centre Grant awarded to National University Cancer Institute of Singapore (NMRC/CG/012/2013) and the National Research Foundation Singapore and the Singapore Ministry of Education under its Research Centres of Excellence initiatives. The study was supported by grant MOHW104-TDU-B-212124-006, OMRPG3C0021 awarded to L-YS.

tional events and alternative t(8;21) transcripts. Blood. 2007;110(3):799-805. 3. Hatlen MA, Wang L, Nimer SD. AML1ETO driven acute leukemia: insights into pathogenesis and potential therapeutic approaches. Front Med. 2012;6(3):248-262. 4. Muller AM, Duque J, Shizuru JA, Lubbert M. Complementing mutations in core binding factor leukemias: from mouse

models to clinical applications. Oncogene. 2008;27(44):5759-5773. 5. Micol JB, Duployez N, Boissel N, et al. Frequent ASXL2 mutations in acute myeloid leukemia patients with t(8;21)/RUNX1-RUNX1T1 chromosomal translocations. Blood. 2014;124(9):14451449. 6. Sood R, Hansen NF, Donovan FX, et al.

1989

V. Madan et al.

10.

11.

12. 13.

14.

15.

16.

17. 18.

19.

1990

Somatic mutational landscape of AML with inv(16) or t(8;21) identifies patterns of clonal evolution in relapse leukemia. Leukemia. 2016;30(2):501-504. Krauth MT, Eder C, Alpermann T, et al. High number of additional genetic lesions in acute myeloid leukemia with t(8;21)/RUNX1-RUNX1T1: frequency and impact on clinical outcome. Leukemia. 2014;28(7):1449-1458. Duployez N, Marceau-Renaut A, Boissel N, et al. Comprehensive mutational profiling of core binding factor acute myeloid leukemia. Blood. 2016;127(20):2451-2459. Faber ZJ, Chen X, Gedman AL, et al. The genomic landscape of core-binding factor acute myeloid leukemias. Nat Genet. 2016; 48(12):1551-1556. Yamato G, Shiba N, Yoshida K, et al. ASXL2 Mutations are Frequently Found in Pediatric AML Patients with t(8;21)/RUNX1-RUNX1T1 and Associated with a Better Prognosis. Genes Chromosomes Cancer. 2017;56(5):382-393. Hatlen MA, Arora K, Vacic V, et al. Integrative genetic analysis of mouse and human AML identifies cooperating disease alleles. J Exp Med. 2016;213(1):25-34. Katoh M. Functional and cancer genomics of ASXL family members. Br J Cancer. 2013;109(2):299-306. Katoh M. Functional proteomics of the epigenetic regulators ASXL1, ASXL2 and ASXL3: a convergence of proteomics and epigenetics for translational medicine. Expert Rev Proteomics. 2015;12(3):317-328. Haferlach T, Nagata Y, Grossmann V, et al. Landscape of genetic lesions in 944 patients with myelodysplastic syndromes. Leukemia. 2014;28(2):241-247. Carbuccia N, Murati A, Trouplin V, et al. Mutations of ASXL1 gene in myeloproliferative neoplasms. Leukemia. 2009; 23(11):2183-2186. Quesada V, Conde L, Villamor N, et al. Exome sequencing identifies recurrent mutations of the splicing factor SF3B1 gene in chronic lymphocytic leukemia. Nat Genet. 2012;44(1):47-52. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. Yoshizato T, Dumitriu B, Hosokawa K, et al. Somatic Mutations and Clonal Hematopoiesis in Aplastic Anemia. N Engl J Med. 2015;373(1):35-47. Abdel-Wahab O, Pardanani A, Patel J, et al. Concomitant analysis of EZH2 and ASXL1 mutations in myelofibrosis, chronic myelomonocytic leukemia and blast-phase myeloproliferative neoplasms. Leukemia. 2011;25(7):1200-1202.

20. Boultwood J, Perry J, Zaman R, et al. Highdensity single nucleotide polymorphism array analysis and ASXL1 gene mutation screening in chronic myeloid leukemia during disease progression. Leukemia. 2010;24(6):1139-1145. 21. Abdel-Wahab O, Gao J, Adli M, et al. Deletion of Asxl1 results in myelodysplasia and severe developmental defects in vivo. J Exp Med. 2013;210(12):2641-2659. 22. Farber CR, Bennett BJ, Orozco L, et al. Mouse genome-wide association and systems genetics identify Asxl2 as a regulator of bone mineral density and osteoclastogenesis. PLoS Genet. 2011;7(4):e1002038. 23. Lai HL, Grachoff M, McGinley AL, et al. Maintenance of adult cardiac function requires the chromatin factor Asxl2. J Mol Cell Cardiol. 2012;53(5):734-741. 24. Baskind HA, Na L, Ma Q, Patel MP, Geenen DL, Wang QT. Functional conservation of Asxl2, a murine homolog for the Drosophila enhancer of trithorax and polycomb group gene Asx. PLoS One. 2009;4(3):e4750. 25. McGinley AL, Li Y, Deliu Z, Wang QT. Additional sex combs-like family genes are required for normal cardiovascular development. Genesis. 2014;52(7):671-686. 26. Lai HL, Wang QT. Additional sex combslike 2 is required for polycomb repressive complex 2 binding at select targets. PLoS One. 2013;8(9):e73983. 27. Micol JB, Pastore A, Inoue D, et al. ASXL2 is essential for haematopoiesis and acts as a haploinsufficient tumour suppressor in leukemia. Nat Commun. 2017;8(15429. 28. Li J, He F, Zhang P, et al. Loss of Asxl2 leads to myeloid malignancies in mice. Nat Commun. 2017;8:15456. 29. Madan V, Shyamsunder P, Han L, et al. Comprehensive mutational analysis of primary and relapse acute promyelocytic leukemia. Leukemia. 2016;30(12):2430. 30. Faust GG, Hall IM. SAMBLASTER: fast duplicate marking and structural variant read extraction. Bioinformatics. 2014; 30(17):2503-2505. 31. Koboldt DC, Zhang Q, Larson DE, et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 2012; 22(3):568-576. 32. McLaren W, Gil L, Hunt SE, et al. The Ensembl Variant Effect Predictor. Genome Biol. 2016;17(1):122. 33. Cibulskis K, Lawrence MS, Carter SL, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol. 2013;31(3):213219. 34. Mayakonda A, Koeffler HP. Maftools:

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

Efficient analysis, visualization and summarization of MAF files from large-scale cohort based cancer studies. bioRxiv. 2016 May 11. Available from: http://doi.org/ 10.1101/052662 Ding L, Ley TJ, Larson DE, et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. Nature. 2012;481(7382):506510. Hardy RR, Carmack CE, Shinton SA, Kemp JD, Hayakawa K. Resolution and characterization of pro-B and pre-pro-B cell stages in normal mouse bone marrow. J Exp Med. 1991;173(5):1213-1225. Tonks A, Pearn L, Musson M, et al. Transcriptional dysregulation mediated by RUNX1-RUNX1T1 in normal human progenitor cells and in acute myeloid leukaemia. Leukemia. 2007;21(12):24952505. Chyla BJ, Moreno-Miralles I, Steapleton MA, et al. Deletion of Mtg16, a target of t(16;21), alters hematopoietic progenitor cell proliferation and lineage allocation. Mol Cell Biol. 2008;28(20):6234-6247. Athanasiadou A, Stalika E, Sidi V, Papaioannou M, Gaitatzi M, Anagnostopoulos A. RUNX1-MTG16 fusion gene in de novo acute myeloblastic leukemia with t(16;21)(q24;q22). Leuk Lymphoma. 2011;52(1):145-147. Lavallee VP, Lemieux S, Boucher G, et al. RNA-sequencing analysis of core binding factor AML identifies recurrent ZBTB7A mutations and defines RUNX1-CBFA2T3 fusion signature. Blood. 2016;127(20):24982501. Gelsi-Boyer V, Brecqueville M, Devillier R, Murati A, Mozziconacci MJ, Birnbaum D. Mutations in ASXL1 are associated with poor prognosis across the spectrum of malignant myeloid diseases. J Hematol Oncol. 2012;5:12. Schnittger S, Eder C, Jeromin S, et al. ASXL1 exon 12 mutations are frequent in AML with intermediate risk karyotype and are independently associated with an adverse outcome. Leukemia. 2013;27(1):82-91. Hartmann L, Dutta S, Opatz S, et al. ZBTB7A mutations in acute myeloid leukaemia with t(8;21) translocation. Nat Commun. 2016;7:11733. Douglas D, Hsu JH, Hung L, et al. BMI-1 promotes ewing sarcoma tumorigenicity independent of CDKN2A repression. Cancer Res. 2008;68(16):6507-6515. Pasini D, Bracken AP, Hansen JB, Capillo M, Helin K. The polycomb group protein Suz12 is required for embryonic stem cell differentiation. Mol Cell Biol. 2007; 27(10):3769-3779.

haematologica | 2018; 103(12)

ARTICLE

Iron Metabolism & its Disorders

Optimizing diagnostic biomarkers of iron deficiency anemia in community-dwelling Indian women and preschool children

Ferrata Storti Foundation

Giridhar Kanuri,1,2 Deepti Chichula,1 Ritica Sawhney,1 Kevin Kuriakose,1 Sherwin De’Souza,1 Faye Pais,1 Karthika Arumugam1 and Arun S. Shet1,3

Wellcome Trust- DBT Hematology Research Unit, St. Johns Research Institute, Bangalore, Karnataka, India; 2Department of Biotechnology, KLEF, Greenfields, Vaddeswaram, Andhra Pradesh, India and 3National Heart Lung and Blood Institute, National Institutes of Health, Bethesda, MD, USA

ABSTRACT

Haematologica 2018 Volume 103(12):1991-1996

he detection of iron deficiency anemia is challenged by the paucity of diagnostic tests demonstrating high sensitivity and specificity. Using two biomarkers, zinc-protoporphyrin/heme and hepcidin, we established the diagnostic cut-off values for iron deficiency anemia in preschool children and women. We randomly selected non-anemic individuals (n=190; women=90, children=100) and individuals with iron deficiency anemia (n=200; women=100, children=100) from a preexisting cohort of healthy preschool children and their mothers. The diagnostic performance of these biomarkers was estimated by analyzing receiver operating characteristic curves. Diagnostic cut-offs with a high predictive value for iron deficiency anemia were selected. Median zinc-protoporphyrin/heme and hepcidin values in non-anemic children were 49 mmol/mol heme and 42 ng/mL, respectively, and in non-anemic women these values were 66 mmol/mol heme and 17.7ng/mL, respectively. Children and women with iron deficiency anemia had higher zinc-protoporphyrin/heme ratios (children=151 mmol/mol heme and women=155 mmol/mol heme) and lower hepcidin levels (children=1.2ng/mL and women=0.6ng/mL). A zinc-protoporphyrin/ heme ratio cut-off >90 mmole/mole heme in children and >107 mmole/mole heme in women was associated with a high diagnostic likelihood for iron deficiency anemia (children, likelihood ratio=20.2: women, likelihood ratio=10.8). Hepcidin cut-off values of ≤6.8ng/mL in children and ≤4.5ng/mL in women were associated with a high diagnostic likelihood for iron deficiency anemia (children, likelihood ratio=14.3: women, likelihood ratio=16.2). The reference ranges and cut-off values identified in this study provide clinicians with guidance for applying these tests to detect iron deficiency anemia. Erythrocyte zinc-protoporphyrin/heme ratio is a valid point-of-care biomarker to diagnose iron deficiency anemia. Introduction Iron deficiency anemia (IDA) is the leading cause of anemia worldwide1 with well established guidelines for diagnosis and treatment.2,3 Typically, the diagnosis of IDA is made when the plasma hemoglobin (Hb) falls below normal (<11.0g/dL in children and <12g/dL in women) and the serum ferritin is <12 mg/L.4 Unfortunately, the frequent coexistence of inflammation/infection confounds serum ferritin, which is an acute phase protein, mandating the performance of additional tests e.g., C-reactive protein (CRP) and serum transferrin receptor (sTfR). As a result, the diagnosis of IDA often requires a battery of diagnostic tests, trained technicians, and the use of expensive laboratory equipment, which increases costs and delays results. Clearly, developing biomarkers that quickly, easily and reliably detect IDA would be beneficial. One such biomarker, zinc-protoporphyrin (ZPP), is formed in erythrocytes during iron-deficient erythropoiesis when the protoporphyrin ring incorporates an atom of zinc rather than iron. The ratio of zinc-protoporphyrin/heme (ZPP/H) can haematologica | 2018; 103(12)

Correspondence: arun.shet@nih.gov

Received: March 13, 2018. Accepted: August 6, 2018. Pre-published: August 9, 2018. doi:10.3324/haematol.2018.193243 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/1991 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1991

G. Kanuri et al.

Table 1. Demographic, hematological and biochemical parameters of study participants.

Non-anemic

Age ± SD Sex M:F Hemoglobin ± SD(g/dl) MCV ± SD (fL) WBC ± SD x 103/mL Platelet count ± SD x 105/mL

Women (n=90)

Children (n=100)

Iron deficiency anemia (IDA) Women Children (n=100) n=100)

26±3.7 na 13.2±0.7 86±4.5 8.7±2.0 3±0.7

3.7±0.9 49:51 11.9±0.6 77±3.7 9.8±2.4 3.9±1

25±3.8 na 10±1.2a 74±7.6a 7.8±2.0 3.4±0.7

2.4±0.8a 56:44 9.6±0.8a 65±5.9a 10.3±2.8 4.6±1.2

Biochemical parameters Serum Ferritin*(ng/mL) Serum sTfR* (mg/L) sTfR/logferritin index CRP* (mg/L)

39 (33, 48) 1.3(1.1, 1.5) 0.8 (0.7, 0.9) 1.4 (0.4, 3.4)

35 (32, 45) 1.7 (1.5, 2.0) 1.1 (0.9, 1.2) 1.1 (0.4, 2.2)

3.7a (2.9, 5.3) 3a (2.2, 4.0) 5.1a (3.5, 8) 0.6a (0.1, 2)

3.8a (2.6, 5.1) 3.9a (3, 5.1) 6.9a (4.5, 11.2) 0.5a (0.06, 1.6)

155a (101, 243) 0.6a (0.2, 1.3)

151a (104, 263) 1.2a (0.5, 3.6)

Biomarkers ZPP/H* ( mol/mol heme) Hepcidin* (ng/mL)

66 (53, 83) 17.7 (9, 38)

49 (39, 60) 42.6 (25, 62)

*Data represented as median (inter quartile range). aP<0.05 compared with the appropriate non-anemic control group. SD: standard deviation; MCV: mean corpuscular volume; WBC: white blood cells; Stfr: serum transferrin receptor; CRP: C-reactive protein; ZPP/H: zinc protoporphyrin/heme.

be determined rapidly at the point-of-care (POC) by a hematofluorometer.5,6 Serum hepcidin, a key regulator of iron homeostasis, is an important biomarker because its levels determine how well oral iron is absorbed, with low hepcidin levels indicating both a requirement for iron and an ability to utilize it if provided.7,8 Although ZPP/H reference values are available for iron deficient pregnant women5 and children,6 cut-off values that establish a diagnosis of IDA with acceptable sensitivity and specificity are lacking. Moreover, reference ranges for hepcidin in healthy rural Indian women and children are not defined. Indian women and preschool age children account for one-third of the global burden of anemia.3,9 Establishing reference ranges for these two biomarkers among healthy individuals and determining cut-off values for the diagnosis of IDA could facilitate their use and the development of novel POC assays. Therefore, we sought to define the median erythrocyte ZPP/H and serum hepcidin levels and select optimal cut-off values for the diagnosis of IDA in healthy rural preschool children and their mothers.

Methods Definition of study groups and sample selection In total, there were 2227 samples that were divided into three groups (non-anemic individuals, and those with IDA or iron deficiency without anemia) (Figure 1). Women and children with anemia (WHO recommended Hb concentrations anemia<11g/dL for children and <12g/dL for women2) with absent body iron stores (serum ferritin <12ng/mL4), were categorized into the IDA group (women n=334 and children n=560). Women and children without anemia (Hb≥11g/dL for children and ≥12g/dL for women) having normal iron stores (serum ferritin≥30ng/mL) were categorized into the non-anemic 1992

group (women n=99 and children n=173). Women and children with normal Hb but low body iron stores (ferritin<30ng/mL) were categorized as having iron deficiency without anemia and excluded from the study (n=1061). Subsequently, using a computerized random number generator we selected 200 samples from the IDA group (100 each from both women and children) and 190 samples from the non-anemic group (Children=100; Women=90) and performed biomarker measurements.10 Using a sTfR/log ferritin index >2, we diagnosed nutritional IDA without coexisting anemia of inflammation.11,12 Study location, sample size, and processing are detailed in Online Supplementary Methods.

Ethics The study was approved by the St. Johns National Academy of Health Sciences Institutional Ethical Committee (IEC115/2012, IEC119/2013, and IEC121/2015).

Biomarkers, ferritin, sTfR and Inflammation assays Serum hepcidin was quantified using an enzyme-linked immunosorbent assay (Peninsula Labs, San Carlos, CA, USA). The hepcidin concentration was extrapolated from a standard curve generated by four parametric logistic regression in accordance with manufacturer instructions. ZPP/H was measured using a hematofluorometer (Aviv Biomedical, Lakewood, NJ, USA) calibrated with commercially available standards. Samples were measured in triplicate and the average value obtained was expressed in µmole/mole heme. Due to a high baseline prevalence of inflammation in this population, we measured sTfR, an indicator of iron status that is not affected by the acute phase response. We calculated the sTfR/log ferritin index to accurately distinguish IDA from anemia of inflammation.13 Serum ferritin and sTfR levels were both measured by paramagnetic particle chemiluminescent immunoassay haematologica | 2018; 103(12)

ZPP/H for iron deficiency anemia diagnosis

Non anemic group Children n=173 Women n=99

Non anemic group Children n=100 Women n=90

Figure 1. Flow diagram of participants and categorization of study groups. Schematic representation of the study design and sample selection. Using age and gender-adjusted WHO definitions for Hb and serum ferritin, samples were divided into a non-anemic group and a group with iron deficiency anemia. Those individuals with normal Hb values but low serum ferritin were defined as having iron deficiency and excluded. Using a computer random number generator, 200 samples from the iron deficiency anemia group (children = 100; women = 100) and 190 samples from the non-anemic group (Children = 100; Women = 90) were randomly selected for biomarker measurements. IDA: iron deficiency anemia; Hb: hemoglobin.

(Access 2, Beckman Coulter). Serum high sensitivity (hs)CRP, a biochemical measure of inflammation, was determined by quantitative sandwich enzyme-linked immunosorbent assay ([ELISA] R&D systems, Minneapolis, MN, USA).

statements about statistical significance. All statistical analyses were done using SPSS 16.0 software (SPSS Inc., Chicago, IL, USA) and ROC curves were performed using MedCalc software (MedCalc, Ostend, Belgium).

Statistical analysis

Results

Variables were aggregated into mean±SD (continuous variables with a Gaussian distribution, t-test for departure from no difference) and median with the interquartile range ([IQR] continuous variables with non-Gaussian distribution, Mann-Whitney test for departure from no difference). To determine ZPP/H and serum hepcidin cut-off values for IDA diagnosis, we used receiver operating characteristic (ROC) curves with IDA defined as Hb below the normal range combined with serum ferritin<12ng/mL. Youden index (J=sensitivity+specificity-1) and likelihood ratios [LR+=sensitivity/(1-specificity)] [LR-=(1-sensitivity)/specificity] were calculated for each individual cut-off value of ZPP/H and hepcidin. This iterative process began with the selection of the highest Youden indices corresponding with the highest positive likelihood ratio for IDA.14 The conventional threshold of <0.05 was used for haematologica | 2018; 103(12)

Comparison of baseline characteristics between the two groups (non-anemic and IDA) for children and women are presented as mean±SD for variables with a normal distribution and as the median ± interquartile range for variables that are not normally distributed. Age, gender, hematological and biochemical parameters are presented in Table 1 and ROC estimates, Youden indices, and likelihood ratios are presented in Table 2.

Baseline characteristics of the anemic and non-anemic groups Non-anemic children were slightly older than children with IDA (3.7±0.9 vs. 2.4±0.8 years) (Table 1) while the ages of non-anemic women and women with IDA did not differ (26±3.7 years vs. 25±3.8 years) (Table 1). Women and children with IDA had significantly lower mean Hb 1993

G. Kanuri et al. Table 2. Properties of selected hepcidin and ZPP/H cut-off values for iron deficiency anemia diagnosis.

Study group

Children Women Children Women

Biomarker Cut-off Value ZPP/H mmol/mol heme >90 >107 Hepcidin ng/mL ≤6.85 ≤4.52

Sensitivity

95% CI Specificity

95% CI

LR+

LR–

Youden Index

Positive Negative Predictive Predictive

81 73

71.9 - 88.2 63.2 - 81.4

96 93

90.1 - 98.9 85.9 - 97.5

20.25 10.83

0.2 0.29

0.77 0.66

95.29 91.25

83.48 77.5

86 90

77.6 - 92.1 82.4 - 95.1

94 94

87.4 - 97.8 87.5 - 98.2

14.33 16.2

0.15 0.11

0.80 0.84

93.48 93.75

85.45 90.38

CI: confidence interval; +LR: positive likelihood ratio; -LR: negative likelihood ratio; ZPP/H: zinc protoporphyrin/heme.

Figure 2. Receiver operating characteristic curves for ferritin, hepcidin and ZPP/H. Pairwise comparison of area under ROC curves of hepcidin and ZPP/H with ferritin as a gold standard for IDA diagnosis in (A) children and in (B) women. As noted, the AUC is similar between the gold standard (ferritin) and either hepcidin or ZPP/H indicating the inherent ability of these two biomarker tests to discriminate between the non-anemic and IDA groups. ZPP: zinc protoporphyrin.

(Children: 9.6 ± 0.8g/dL vs. 11.9 ± 0.6g/dL; Women: 10 ± 1.2g/dL vs. 13.2 ± 0.7 g/dL), mean corpuscular volume (Children: 65 vs. 77fl, Women: 74 vs. 86fl) and median ferritin (Children: 3.8 vs. 35ng/mL, Women: 3.7 vs. 39ng/mL) compared with non-anemic women and children, respectively (Table 1). There were no differences in white blood cell and platelet counts between the two study groups (Table 1). Serum CRP values were low in both study groups (Table 1). As expected, sTfR was significantly higher in children and women with IDA compared with their non-anemic counterparts (Children: 3.9 vs. 1.7mg/L; Mothers: 3 vs. 1.3mg/L) (Table 1). Moreover, sTfR/log ferritin index values >2 in women and children with IDA confirmed a diagnosis of IDA and ruled out coexisting anemia of inflammation (Table 1).

1.2ng/mL [IQR: 0.5, 3.6] and women: 17.7ng/mL [IQR: 9, 38] vs. 0.6ng/mL [IQR: 0.2, 1.3], P<0.05 for both) (Table 1).

Diagnostic cut-off for erythrocyte ZPP/H ZPP/H >90 µmole/mole heme resulted in IDA diagnosis in children with 81% sensitivity and 96% specificity (Table 2). This cut-off had a positive likelihood ratio of 20.25 and a positive predictive value of 95.2% (Table 2). In women, a higher ZPP/H >107 µmole/mole heme resulted in IDA diagnosis with 73% sensitivity and 93% specificity. This cut-off in women yielded a positive likelihood ratio of 10.8 and a positive predictive value of 91.2% (Table 2). ROC curves for ZPP/H to diagnose IDA revealed an area under the curve (AUC)ROC of 0.94 in children (Figure 2A) and 0.89 (P<0.0001) in women (Figure 2B).

Diagnostic cut-off for serum hepcidin

Erythrocyte ZPP/H and serum hepcidin values in anemic and non-anemic groups The median ZPP/H in children and women with IDA was higher when compared with their non-anemic counterparts (children: 151 mmol/mol heme [IQR: 104, 263] vs. 49 mmol/mol heme [IQR: 39, 60] and women: 155 mmol/mol heme [IQR: 101, 243] vs. 66 mmol/mol heme [IQR: 53, 83], P<0.001 for both) (Table 1). Median hepcidin concentration was markedly higher in non-anemic children and women when compared with their counterparts who had IDA (children: 42ng/mL [IQR: 25, 62] vs. 1994

Serum hepcidin value ≤6.85ng/mL yielded an IDA diagnosis in children with 86% sensitivity and 94% specificity (Table 2). This cut-off had a likelihood ratio of 14.3 with a positive predictive value of 93.4% (Table 2). In women, a hepcidin value of ≤4.5ng/mL resulted in an IDA diagnosis with 90% sensitivity and 94% specificity (Table 2). This cut-off corresponded with a likelihood ratio of 16.2 and a positive predictive value of 93.7%. ROC curves for hepcidin to diagnose IDA revealed an AUCROC of 0.97 in children (Figure 2A) and 0.96 (P<0.0001) in women (Figure 2B). haematologica | 2018; 103(12)

ZPP/H for iron deficiency anemia diagnosis

Discussion In this study of healthy rural community-dwelling nonanemic Indian women and children and their counterparts with biochemically defined IDA, we 1) report the median values for the iron biomarkers erythrocyte ZPP/H and serum hepcidin, 2) analyze ROC curves for erythrocyte ZPP/H and serum hepcidin, and 3) define the ZPP/H ratio and serum hepcidin cut-off values for IDA diagnosis and estimate the post-test probability of IDA for these cut-off values. Overall, these findings demonstrate the utility of erythrocyte ZPP/H as a POC biomarker for IDA diagnosis, particularly in women and children from low-middle income settings. We found similar median ZPP/H levels in non-anemic children to those reported previously (47.5 and 58 mmol/mol heme).15,16 Although evaluated systematically in children and non-anemic pregnant women with iron deficiency,5,17 ZPP/H levels have not been studied either in women or preschool children using rigorous criteria for nutritional IDA. Only one large Indian study of tribal adults and children (<18 years) previously used ZPP/H to detect IDA in a subset (n=100) of anemic individuals (mean Hb 8.4) with normal Hb phenotype.18 The authors reported a higher mean ZPP/H value (214.9 ± 120.1) than in our study. This discrepancy may be explained by either the difference in the two study populations (severity of anemia or undetected Hb disorders) or methodological differences (i.e., whole blood vs. washed erythrocytes).19 The median ZPP/H ratios reported in our study probably reflect values encountered in healthy women and children residing in rural Indian communities. Serum hepcidin values in non-anemic children in our study are concordant with reports in European children,20,21 but higher than values reported in Asian22 and African children.23,24 The inclusion of <12-month-old non-anemic children in the latter studies explain these differences, since hepcidin concentrations are decreased between three and six months of age.24 Children with IDA in our study had serum hepcidin levels comparable with those reported in anemic children from Asia25 and Africa.23 Non-anemic women in our study had variable levels compared with those previously reported in European studies,26,27 discrepancies that are possibly explained by socioeconomic and dietary differences between these populations. Some of the inter-study variability is also possibly attributable to differences in the methodological assays used to estimate hepcidin.28 Women with IDA in our study had very low median hepcidin levels concordant with previously published studies.29 Using ROC analysis, we selected cut-off values for ZPP/H that detected IDA in both women and children with >90% specificity. Cut-off values selected in recent studies were lower (>40 mmole/mole30,31 heme and >70 mmol/mole heme6) and lacked specificity (56% and 60%, respectively), perhaps because they were selected to detect iron deficiency, not IDA. Another recent study utilized a ZPP/H cut-off value of 70 mmole/mole to diagnose IDA in a pediatric population, but this value had a low specificity (75%).32 The large Indian study referenced previously used a cut-off value of >80 µmole/mole heme to define IDA in a mixed population of healthy individuals and those with sickle cell trait or sickle cell anemia.18 The scientific rationale for this cut-off value and the validity haematologica | 2018; 103(12)

of ZPP/H as a stand-alone diagnostic assay for IDA in individuals with sickle cell anemia, and possibly α-thalassaemia,33 is uncertain. In contrast, our study used rigorous biochemical criteria to define IDA in a representative sample of healthy community-dwelling women and children. Consequently, this is the first study to demonstrate the utility of ZPP/H as a biomarker of IDA and define cutoff values with which to establish an IDA diagnosis in healthy rural women and children. The hepcidin cut-off values selected to diagnose IDA in children in our study were higher than those selected previously in Korean children (≤6.85ng/mL vs. ≤2.735ng/mL) and yielded higher AUCs (0.97 vs. 0.90).25 The higher AUC value indicates a better discriminative power of hepcidin in detecting IDA. However, these cut-off values were similar to values that detect IDA in six to 60-month-old Gambian and Tanzanian children (5 and 8 ng/ml, respectively).7 The proposed cut-off values, with their high sensitivity and specificity, increase the probability of a diagnosis of IDA. However, we also determined the predictive values of these tests by estimating their likelihood ratios. The likelihood ratio indicates how many times more likely a particular test result is, in a patient with that particular condition, with a likelihood (LR) ratio value >10 providing robust diagnostic evidence.34 Assuming a pre-test probability of IDA of 50% in any given population, the selected ZPP/H and hepcidin cut-off values had likelihood ratios of ~10, which according to the Fagan nomogram corresponded with a 90% post-test probability of having an IDA diagnosis.35 Thus, even in populations with lower pre-test probabilities of having IDA, these diagnostic cutoffs are valid. Measuring erythrocyte ZPP/H is procedurally simple, technically feasible by field health workers possessing <12th-grade education, and could provide rapid results at the primary health center. Rapid diagnosis would facilitate therapeutic decision-making in a single visit and favor patient convenience, an important consideration in lowmiddle income settings. Although not a prior study objective, informal assay cost estimates in our laboratory indicate that ZPP/H measurement is cheaper than hepcidin. Thus, our findings suggest that ZPP/H has greater utility as a POC diagnostic test to detect IDA in women and children. Although hepcidin is extremely useful in predicting iron absorption and incorporation into erythrocytes,8 potential limitations of its use in this setting include higher costs, lack of hepcidin standardization and the requirement to convert the immunological assay into a POC assay.36 The strengths of this study are its inclusion of a large representative sample of healthy community dwelling women and children, a random selection of blood samples for biomarker measurements, and rigorous definition of IDA using multiple biomarkers.4,37 Using a combination of low Hb with ferritin as a gold standard for IDA instead of bone marrow aspiration with perls staining raises potential concerns regarding diagnostic accuracy. Reassurance against this concern is provided by the sTfR/log ferritin index >2 and near normalization of Hb in response to iron therapy in children with IDA after six months treatment (baseline Hb for children in the IDA group = 9.6±0.8g/dL; six-month post-treatment Hb = 10.5±1.3g/dL). Finally, estimates of the diagnostic accuracy for the proposed cutoff values highlight the clinical applicability of these find1995

G. Kanuri et al. ings for IDA diagnosis.38 The results of this study are generalizable to women and children from similar agrarian parts of India and possibly to other similar low-middle income settings around the world. In conclusion, the findings of this study provide a scientific rationale for the use of ZPP/H as a POC biomarker to establish the diagnosis of IDA in women and children from low-middle income settings. The diagnostic cut-off values and their accompanying likelihood ratio’s provide clinicians with guidance for using these biomarkers to diagnose IDA.

References 1. FAO, IFAD, UNICEF W and W. The state of food security and nutrition in the world. 1-109 p. 2. WHO. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. Geneva 2011. 3. McLean E, Cogswell M, Egli I, Wojdyla D, de Benoist B. Worldwide prevalence of anaemia, WHO Vitamin and Mineral Nutrition Information System, 1993-2005. Public Heal Nutr. 2009;12(4):444-454. 4. Who. Serum ferritin concentrations for the assessment of iron status and iron deficiency in populations. Vitamin and Mineral Nutrition Information System. In: Who p1–5. 5. Mwangi MN, Maskey S, Andang o PEA, et al. Diagnostic utility of zinc protoporphyrin to detect iron deficiency in Kenyan pregnant women. BMC Med. 2014; 12(1):229. 6. Yu KH. Effectiveness of zinc protoporphyrin/heme ratio for screening iron deficiency in preschool-aged children. Nutr Res Pract. 2011;5(1):40-45. 7. Pasricha S-R, Atkinson SH, Armitage AE, et al. Expression of the iron hormone hepcidin distinguishes different types of anemia in African children. Sci Transl Med. 2014;6 (235):235re3:1-7. 8. Prentice AM, Doherty CP, Abrams SA, et al. Hepcidin is the major predictor of erythrocyte iron incorporation in anemic African children. Blood. 2012;119(8):1922-1928. 9. 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. 10. Urbaniak GC PS. Research Randomizer v4.0. 11. Weiss G, Goodnough LT. Anemia of chronic disease. N Engl J Med. 2005; 352(10):10111023. 12. Camaschella C. Iron-Deficiency Anemia. N Engl J Med. 2015;372(19):1832–1843. 13. Suominen P, Punnonen K, Rajamäki a, Irjala K. Serum transferrin receptor and transferrin receptor-ferritin index identify healthy subjects with subclinical iron deficits. Blood. 1998;92(8):2934-2939. 14. Deeks JJ, Altman DG. Diagnostic tests 4: likelihood ratios. BMJ. 2004;329(7458): 168-169.

1996

Acknowledgments The study samples were a part of a cluster randomized controlled trial which was registered with ISRCTN.com (identifier: ISRCTN68413407) on 17 September 2013. The authors gratefully acknowledge the women and children who participated in the Karnataka Anaemia Project 2 study. Funding The study was funded by the Wellcome Trust/DBT India Alliance through a Senior Fellowship Award to Dr. Arun Shet [grant reference number IA/SF/2013/AS/1].

15. Soldin OP, Miller M, Soldin SJ. Pediatric reference ranges for zinc protoporphyrin. Clin Biochem. 2003;36(1):21-25. 16. Crowell R, Ferris AM, Wood RJ, Joyce P, Slivka H. Comparative effectiveness of zinc protoporphyrin and hemoglobin concentrations in identifying iron deficiency in a group of low-income, preschool-aged children: practical implications of recent illness. Pediatrics. 2006;118(1):224-232. 17. Abioye AI, Aboud S, Premji Z, et al. Iron supplementation affects hematologic biomarker concentrations and pregnancy outcomes among iron-deficient Tanzanian women. J Nutr. 2016;146(6):1162-1171. 18. Mohanty D, Mukherjee MB, Colah RB, et al. Iron deficiency anaemia in sickle cell disorders in India. Indian J Med Res. 2008; 127(4):366-369. 19. Labbe RF. Clinical utility of zinc protoporphyrin. Clin Chem. 1992;38(11):2167-2168. 20. Sdogou T, Tsentidis C, Gourgiotis D, et al. Immunoassay-based serum hepcidin reference range measurements in healthy children : differences among age groups. J Clin Lab Anal. 2015;29(1):10-14. 21. Cangemi G, Pistorio A, Miano M, et al. Diagnostic potential of hepcidin testing in pediatrics. Eur J Haematol. 2013;90(4):323330. 22. Bhatia P, Marathe R, Hegde A, Bhardwaj D, Jain R. Plasma hepcidin levels in healthy children from Chandigarh, Northern India. Indian Pediatr. 2017;54(7):599-600. 23. Jaeggi T, Moretti D, Kvalsvig J, et al. Iron status and systemic inflammation, but not gut inflammation, strongly predict gender-specific concentrations of serum hepcidin in infants in rural Kenya. PLoS One. 2013;8(2):e57513. 24. Mupfudze TG, Stoltzfus RJ, Rukobo S, Moulton LH, Humphrey JH, Prendergast AJ. Hepcidin decreases over the first year of life in healthy African infants. Br J Haematol. 2014;164(1):150-153. 25. Choi HS, Song SH, Lee JH, Kim H-J, Yang HR. Serum hepcidin levels and iron parameters in children with iron deficiency. Korean J Hematol. 2012;47286–92. 26. Ganz T, Olbina G, Girelli D, Nemeth E, Westerman M. Immunoassay for human serum hepcidin Immunoassay for human serum hepcidin. Blood. 2008; 112(10): 4292-4297.

27. Galesloot TE, Vermeulen SH, GeurtsMoespot AJ, et al. Serum hepcidin: reference ranges and biochemical correlates in the general population. Blood. 2011; 117(25): 218-226. 28. Uijterschout L, Swinkels DW, Domellöf M, et al. Serum hepcidin measured by immunochemical and mass-spectrometric methods and their correlation with iron status indicators in healthy children aged 0.5-3 y. Pediatr Res. 2014;76(4):409-414. 29. van Santen S, van Dongen-Lases EC, de Vegt F, et al. Hepcidin and hemoglobin content parameters in the diagnosis of iron deficiency in rheumatoid arthritis patients with anemia. Arthritis Rheum. 2011; 63(12):3672-3680. 30. Zimmermann MB, Molinari L, StaubliAsobayire F, et al. Serum transferrin receptor and zinc protoporphyrin as indicators of iron status in African children. Am J Clin Nutr. 2005;81(3):615-623. 31. Teshome EM, Prentice AM, Demir AY, Andang’o PEA, Verhoef H. Diagnostic utility of zinc protoporphyrin to detect iron deficiency in Kenyan preschool children: a community-based survey. BMC Hematol. 2017;17(1):11. 32. Serdar MA, Ümit Sarici S, Kurt I, et al. The role of erythrocyte protoporphyrin in the diagnosis of iron deficiency anemia of children. J Trop Pediatr. 2000;46(6):323-326. 33. Graham EA, Felgenhauer J, Detter JC, Labbe RF. Elevated zinc protoporphyrin associated with thalassemia trait and hemoglobin E. J Pediatr. 1996;129(1):105-110. 34. Hayden SR, Brown MD. Likelihood ratio: a powerful tool for incorporating the results of a diagnostic test into clinical decisionmaking. Ann Emerg Med. 1999;33(5):575-580. 35. McGee S. Simplifying likelihood ratios. J Gen Intern Med. 2002;17(8):647-650. 36. Kroot JJC, van Herwaarden AE, Tjalsma H, Jansen RTP, Hendriks JCM, Swinkels DW. Second round robin for plasma hepcidin methods: First steps toward harmonization. Am J Hemato.l 2012;87(10):977-983. 37. WHO. Iron deficiency anaemia: assessment, prevention, and control. A guide for programme managers. World Heal Organ. 2001;114. 38. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and,accurate reporting of studies of diagnostic accuracy: the STARD initiative. Br Med J. 2003; 326(7379):41-44.

haematologica | 2018; 103(12)

ARTICLE

Red Cell Biology & its Disorders

Non-muscle myosin II drives vesicle loss during human reticulocyte maturation

Ferrata Storti Foundation

Pedro L. Moura,1 Bethan R. Hawley,1 Tosti J. Mankelow,2,3 Rebecca E. Griffiths,2,3,4 Johannes G.G. Dobbe,5 Geert J. Streekstra,5 David J. Anstee,2,3 Timothy J. Satchwell1,2,3* and Ashley M. Toye1,2,3*

School of Biochemistry, University of Bristol, UK; 2Bristol Institute for Transfusion Sciences, National Health Service Blood and Transplant (NHSBT), UK; 3NIHR Blood and Transplant Research Unit, University of Bristol, UK; 4UQ-StemCARE, Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Australia and 5 Department of Biomedical Engineering and Physics, Academic Medical Center, University of Amsterdam, the Netherlands 1

*These authors contributed equally to this work

Haematologica 2018 Volume 103(12):1997-2007

ABSTRACT

he process of maturation of reticulocytes into fully mature erythrocytes that occurs in the circulation is known to be characterized by a complex interplay between loss of cell surface area and volume, removal of remnant cell organelles and redundant proteins, and highly selective membrane and cytoskeletal remodeling. However, the mechanisms that underlie and drive these maturational processes in vivo are currently poorly understood and, at present, reticulocytes derived through in vitro culture fail to undergo the final transition to erythrocytes. Here, we used high-throughput proteomic methods to highlight differences between erythrocytes, cultured reticulocytes and endogenous reticulocytes. We identify a cytoskeletal protein, non-muscle myosin IIA (NMIIA) whose abundance and phosphorylation status differs between reticulocytes and erythrocytes and localized it in the proximity of autophagosomal vesicles. An ex vivo circulation system was developed to simulate the mechanical shear component of circulation and demonstrated that mechanical stimulus is necessary, but insufficient for reticulocyte maturation. Using this system in concurrence with non-muscle myosin II inhibition, we demonstrate the involvement of non-muscle myosin IIA in reticulocyte remodeling and propose a previously undescribed mechanism of shear stress-responsive vesicle clearance that is crucial for reticulocyte maturation.

Correspondence: t.satchwell@bristol.ac.uk or ash.m.toye@bristol.ac.uk Received: June 1, 2018. Accepted: July 26, 2018. Pre-published: August 3, 2018.

Introduction

doi:10.3324/haematol.2018.199083

Reticulocytes are anucleate erythroid cells which undergo maturation to form the biconcave erythrocyte.1 The mechanism leading to maturation is relatively undefined, but it is known to occur in the peripheral circulation within 1-2 days after cells have egressed from the bone marrow.2 As the current endpoint of existing in vitro erythroid culture systems,3,4 interest in this cell type and the mechanisms and factors that may underlie and drive their maturation to erythrocytes has received renewed interest in recent years. The phenotypic differences between reticulocytes and erythrocytes have been studied in detail. Reticulocyte maturation involves extensive membrane and cytoskeletal remodeling, with loss of approximately 20% of cell surface area during this process5-9 which allows the initially amorphous reticulocyte to acquire the characteristic erythrocyte biconcave morphology and accompanying increased resistance to shear stress. This remodeling is a highly selective process, characterized by proteasomal degradation and exocytosis of specific components (e.g. actin, myosin, talin) with preferential retention of the remainder (e.g. α or β-spectrin).7 Other notable hallmarks of maturation include a progressive loss of RNA content, an increase in cell deformability, and a decrease in both cytoplasmic and surface protein content through exocytosis, membrane shedding and autophagy-mediated pathways.6,10-12 The most well-described pathway relating to loss of protein during reticulocyte

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/1997

haematologica | 2018; 103(12)

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

1997

P.L. Moura et al.

maturation concerns the removal of the transferrin receptor CD71.10 This mechanism of plasma membrane protein loss, which involves the internalization of receptors within endocytic vesicles, multivesicular body formation and the release of exosomes has been extensively studied and shown to also mediate the loss of other proteins.12 Other major mechanisms involved in the removal of proteins during reticulocyte maturation are encompassed by the process of autophagy. Autophagy occurs through the sorting of redundant, damaged or leftover organelles and other cytoplasmic content into autophagosomes,13 which in reticulocytes have been reported to fuse with lysosomes and form autophagolysosomes that are later expelled from the cell.4,14 Parts of this process are highly specific: for instance, NIX-/- mice undergo autophagy and maturation, but not mitophagy, causing retention of mitochondria in reticulocytes.15 Using reticulocytes cultured in vitro, the autophagosome was recently shown to be involved in the process of remodeling of the reticulocyte membrane which involves the release of an inside-out vesicle.4 It is also known that, in other mammalian cells, mechanical stress upregulates autophagy, and the removal of autophagic vesicles in the reticulocyte is triggered by passage through the sinusoidal walls of the spleen.16,17 However, the process of autophagic vesicle transport and the eventual release of these vesicles from reticulocytes is poorly understood. In vitro-derived reticulocytes expanded and differentiated under conditions compatible with clinical use do not currently emulate the final stages of maturation, which occur in vivo after egress from the bone marrow and within the circulation to generate definitive erythrocytes. However, transfusion of in vitro-derived reticulocytes into mouse models induced these final stages of maturation,3,18 indicating the involvement of as of yet undefined factors or stimuli not recapitulated in the in vitro culture process. During their time in the peripheral circulation, reticulocytes are exposed to a variety of new stimuli, including shear stress, dynamic pressure changes, contact with other cell types (endothelial cells, residing spleen and liver macrophages) and a varying pH, pO2 and pCO2. We hypothesized that shear stress may be a driver for maturation and demonstrate here that it is possible to simulate the shear stress component of in vivo circulation using a simple ex vivo circulation mechanism, leading to loss of cell surface area and selective loss of protein content and of mitochondrial content in cultured reticulocytes. Finally, we delineate a novel role for non-muscle myosin IIA (NMIIA) in shear-responsive reticulocyte vesicle transport and maturation. We demonstrate its specific phosphorylation and localization in the proximity of autophagic vesicle markers in reticulocytes and show that chemical inhibition of NMIIA leads to an inability to lose cell volume, as well as a reduction in mitochondrial clearance.

Methods

ration of healthy donor platelet apheresis waste blood using CD71 MicroBead (Miltenyi Biotec) isolation according to the manufacturerâ&#x20AC;&#x2122;s instructions. In vitro-cultured reticulocytes were differentiated from CD34+ cells isolated from the mononuclear cell fraction according to previously published protocols.4 All source material was provided with written informed consent for research use given in accordance with the Declaration of Helsinki (NHSBT, Filton, Bristol). The research into the mechanisms of erythropoiesis was reviewed and approved by the Bristol Research Ethics Committee (REC Number 12/SW/0199).

Proteomics experimental design, data acquisition and analysis Two experiments were performed: (i) a comparison of erythrocytes, endogenous reticulocytes and in vitro-derived reticulocytes with three biological repeats per cell type, for a total of nine individual samples which were processed through both quantitative tandem mass tag (TMT) proteomics and qualitative TiO2-enriched phosphoproteomics; and (ii) a comparison of circulated and noncirculated in vitro-derived reticulocytes with three biological repeats per condition, for a total of six individual samples, which were processed through quantitative TMT proteomics. Cells were washed three times with phosphate-buffered saline containing 1 mg/mL bovine serum albumin and 2 mg/mL glucose; 2x106 cells were counted and used per sample for quantitative TMT proteomics, and 10x106 cells for qualitative phosphoproteomics. The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD009015, PXD009023 and PXD009024. Extended details on the methods for sample preparation, data acquisition and analysis are provided in the Online Supplementary Methods.

Immunofluorescence and automated image processing Details on the methods for fixed and live-cell immunofluorescence are provided in the Online Supplementary Methods. For quantitative analysis, tile scans composed of 10x10 images taken at 1024x1024 resolution of cells labeled with Calcein Blue and Mitotracker Deep Red were generated using confocal imaging and analyzed with arivis Vision4D (arivis, Germany). Further details on the analysis are provided in the Online Supplementary Methods.

Sample preparation for rheoscopy and cell analysis

Two million cells were diluted in 200 mL of a polyvinylpyrrolidone solution (polyvinylpyrrolidone viscosity 28.1; Mechatronics Instruments). Samples were assessed in an Automated Rheoscope and Cell Analyzer (ARCA)19 consisting of a plateâ&#x20AC;&#x201C;plate optical shearing stage (model CSS450) mounted on a Linkam imaging station assembly and temperature controlled using Linksys32 software (Linkam Scientific Instruments). The microscope was equipped with an LMPlanFl 50x with a 10.6 mm working distance objective (Olympus) illuminated by an X-1500 stroboscope (Vision Light Tech) through a band-pass interference filter (CWL 420 nm, FWHM 10 nm; Edmund Optics). Images were acquired using a uEye camera (UI-2140SE-M-GL; IDS GmbH). At least 1500 valid cells per sample were analyzed using bespoke ARCA analysis software.

Antibodies A list of antibodies used in this study is provided in the Online Supplementary Methods.

Native reticulocyte isolation and in vitro erythroid culture Native CD71+ reticulocytes and CD71- erythrocytes were isolated from the red blood cell fraction obtained by Histopaque sepa1998

Ex vivo cell circulation Details on the construction of the circulation system are provided in the Online Supplementary Methods. Cells were packed and resuspended in a mixture of culture medium and Sanquin reticulocyte stabilizer solution (Sanquin Blood Supply, the Netherlands) in a 4:1 ratio. The suspension was then circulated overnight at 37ÂşC in 5% CO2, with the 5 rpm setting. A matched control samhaematologica | 2018; 103(12)

NMII participates in reticulocyte maturation

Figure 1. Reticulocyte culture and maturation lead to differences in phosphorylation status and protein abundance. (A) Experimental design for proteomic comparison – red blood cells, native reticulocytes (CD71+ cells) and hematopoietic precursors (CD34+ cells) were separated with magnetic bead isolation. Cultured reticulocytes were leukofiltered after 21 days of culture post-CD34+ cell isolation. (B) Heatmap visualization of the tandem mass tag (TMT) proteome dataset generated from log2 fold-change values of expression in native and cultured reticulocytes with erythrocytes as a baseline. Individual samples are visualized as separate rows. Blue denotes lower expression, and red denotes higher expression. Proteins were clustered through the average link method and a corresponding dendrogram was generated, shown to the left of the heatmap (n. of proteins analyzed: 2140; false discovery rate = 5%). (C) Heatmap visualization of TMT proteome dataset generated from log2 fold-change values of expression in cultured reticulocytes with native reticulocytes as a baseline. The processing was the same as that for Figure 1B. (D) Venn diagram of differentially expressed proteins (absolute log2 fold-change >1) in endogenous and cultured reticulocytes using red blood cells as a baseline for comparison. A total of 1667 proteins were differentially expressed in reticulocytes. (E) Heatmap visualization of the “Actin Cytoskeleton and Remodeling” Gene Ontology category present in the TMT proteome dataset, comparing cultured reticulocytes with native reticulocytes as the baseline. The processing was the same as that for Figure 1B. (F) Venn diagram of the phosphorylated proteins detected through qualitative phosphoproteomics comparison of red blood cells, native reticulocytes (NatRet) and cultured reticulocytes (CultRet). The dataset was filtered at a 5% false discovery rate for this analysis, with no further constraints. A total of 521 phosphoproteins were detected. (G) High-confidence peptides (false discovery rate = 1%) detected through qualitative phosphoproteomics. The table was filtered to include only peptides which were present in at least two samples in the same condition. The ID column denotes the UniProt protein identities. Common protein names and protein symbols (GN) are defined in the Protein column. The phosphorylation site of the respective peptides is in the column Site and was acquired through comparison with the PhosphoSitePlus® database.53 The number of occurrences per condition of each phosphorylation is written in the rightmost columns.

haematologica | 2018; 103(12)

1999

P.L. Moura et al.

ple was kept in culture without circulation. For drug treatment to test NMIIA inhibition, cells were exposed during circulation to either 20 mM blebbistatin(-) (Insight Biotechnology), or 20 mM blebbistatin(+) (Source BioScience) as a negative control.

Results Quantitative proteomics data analysis reveals differences between erythrocytes, native reticulocytes and cultured reticulocytes In order to explore possible changes in protein abundance and function underlying reticulocyte maturation, we produced a TMT-based quantitative proteomics dataset and a qualitative phosphoproteomics dataset comparing donor-matched native reticulocytes, cultured reticulocytes and erythrocytes. The CD71+ population (native reticulocytes) was isolated through magnetic bead isolation. While this strategy does not distinguish between subsets of reticulocytes as compared to alternative fluorescence activated cell sorting-based methods,6 it does provide a convenient method to obtain the required number of cells from each separate donor for biochemical analysis. The CD71- population (erythrocytes) was collected from the flow-through of the CD71+ cell isolation, and CD34+ precursors were isolated, expanded and differentiated to generate in vitro-derived (cultured) reticulocytes. The design of this experiment is summarized in Figure 1A. All samples submitted to proteomics were characterized regarding established reticulocyte markers (Online

Supplementary Figure S1). Data resulting from the experiment are presented in Online Supplementary Table S1. These proteomic datasets were analyzed with the objective of determining possible targets for further study of reticulocyte maturation. First, a broad comparison of protein expression changes between reticulocytes and erythrocytes was performed. In accordance with previous literature, where a loss of protein content is expected in maturation,20,21 both native and cultured reticulocytes have a higher abundance of most proteins, as shown in Figure 1B. However, protein clusters where expression is higher in the erythrocyte can also be observed and were found to encompass both contaminating material (serum proteins and keratins) and functional proteins (e.g. carbonic anhydrase 1/2/3). Only six proteins were significantly increased (average log2 fold-change >1) in the erythrocyte population, and all corresponded to contamination except for carbonic anhydrase 1. This observation reflects the fact that residual protein synthesis still takes place in the reticulocyte by means of the ribosomes and RNA present in the cell, and carbonic anhydrase 1 in particular is known to be synthesized during maturation.22 Some inter-sample variation was observed between the same cell types, likely due to variable purity in the CD71 isolation protocol (Online Supplementary Figure S1). Furthermore, the heterogeneous nature of reticulocyte populations undergoing maturation creates inherent difficulties for comparative proteomics. Nonetheless, native and in vitro-derived (cultured) reticulocytes were found to exhibit a very similar global protein expression profile (Figure 1C). A summary

Figure 2. Non-muscle myosin IIA components localize proximally to autophagic vesicles. (A) Erythrocytes (top) and cultured reticulocytes (bottom) were co-labeled for phospho-non-muscle myosin IIA (S1943) (green) and non-muscle myosin IIA (NMIIA) (red) and subjected to confocal imaging. Cells were fixed, labeled and cytospun as described in the Methods section. Scale bars: 10 mm. (B) Untreated reticulocytes (top) and reticulocytes treated with lambda phosphatase (bottom) for a dephosphorylation assay were labeled for phospho-NMIIA (S1943) (green). Cells were fixed, treated with lambda phosphatase, labeled and cytospun as described in the Methods section. Scale bars: 10 mm. (C) Western blot of the phospho-NMIIA signal before and after the dephosphorylation assay with lambda phosphatase. Protein 4.2 was used as a loading control. (D) Cultured reticulocytes were co-labeled with phospho-NMIIA (S1943) (green) and LC3B (red), and imaged by confocal microscopy. A zoomed-in section is shown to facilitate identification of the structures. Three-dimensional reconstruction was done using arivis Vision4D software, maximizing contrast for clarity. Scale bar in initial image: 10 mm. Scale bar in zoomed-in image: 5 mm. (E) Fixed and permeabilized cultured reticulocytes were duallabeled for phosphorylated myosin light chain (red) and phosphatidylserine or trypsin sensitive glycophorin A epitope R10 as indicated (green) and imaged using confocal microscopy. For R10 staining, reticulocytes were previously treated with trypsin. Images are shown in two-dimensional phase overlay, two-dimensional fluorescence, three-dimensional reconstructions and three-dimensional reconstructions with zoom highlighted. Scale bars: 5 mm.

2000

haematologica | 2018; 103(12)

NMII participates in reticulocyte maturation

Figure 3. In vitro circulation of cultured reticulocytes recapitulates aspects of reticulocyte maturation. (A) Diagram of the circulation system used to study the impact of shear stress on reticulocyte maturation. Cells are subjected to a continuous difference in pressure with the use of a peristaltic pump, and are circulated overnight at 37ºC in 5% CO2. A matched control is kept in culture overnight, without circulation. (B) Cross-sectional area profile of red blood cells, reticulocytes kept in culture overnight without circulation (Ctrl), and reticulocytes circulated overnight (Circ). The profile is plotted as the average of the proportion of cells within each area “bin” (corresponding to 5 mm2). Error bars correspond to the standard deviation of each average. Data were obtained using an Automated Rheoscope and Cell Analyzer (ARCA). (C) ARCA contour plots demonstrating the probability distribution of erythrocytes, uncirculated reticulocytes and circulated reticulocytes with cross-sectional area plotted against deformability index. The probability density functions for the data were generated through kernel-density estimation of three pooled samples per group. (D) Deformability index (length/width ratio) profile of red blood cells, uncirculated reticulocytes and circulated reticulocytes. The profile is plotted as the average of the proportion of cells within each deformability index “bin” (corresponding to 0.1 ratio units). Error bars correspond to the standard deviation of each average. Data were obtained using the ARCA. (E) Heatmap visualization of the tandem mass tag (TMT) proteome dataset generated from log2 fold-change values of expression in circulated reticulocytes with uncirculated reticulocytes as a baseline. Individual samples were visualized as separate rows. Blue denotes lower expression, and red denotes higher expression. Proteins are displayed horizontally by order of average log2 fold-change (n. of proteins analyzed: 2447). (F) Subset of Figure 3D detailing the 40 proteins with the lowest log2 fold-change, or, alternatively, the 40 proteins in which abundance is reduced to the greatest degree in circulated reticulocytes. Mitochondrial proteins are indicated with red arrows. (G) Subset of Figure 3D detailing erythroid proteins of interest. Proteins are displayed vertically by order of average log2 fold-change.

haematologica | 2018; 103(12)

2001

P.L. Moura et al. A

Figure 4. Blebbistatin treatment abrogates circulation-induced differences in reticulocyte cross-sectional area. (A) Automated Rheoscope and Cell Analyzer (ARCA) contour plots demonstrating the probability distribution of erythrocytes, uncirculated reticulocytes and circulated reticulocytes with cross-sectional area plotted against deformability index. Reticulocytes were untreated, treated with blebbistatin(-) or treated with blebbistatin(+) as displayed in each column. Cell types are plotted pairwise for ease of visualization. The probability density functions for the data were generated through kernel-density estimation of three pooled samples per group. (B) Comparison of the area difference to red blood cells between untreated, blebbistatin(-)-treated and blebbistatin(+)-treated cells that were left in culture (gray) or were circulated overnight (white). *Indicates a P-value of under 0.05, with n.s.s. (not statistically significant) indicating P-values above 0.05. P-values for relevant comparisons are shown underneath the graph. All comparisons were made with a paired two-tailed t-test between arrays of area difference to red blood cells. Data are represented as mean Âą standard deviation (n=3).

of significant differences in abundance is shown in Figure 1D. The proteins with the greatest difference in abundance between cultured and native reticulocytes were processed using the STRING database to create proteinprotein interaction networks (Online Supplementary Figure S2), which ease data visualization. It was observed that protein families containing a large number of proteins skewed the resulting networks. Thus, ribosomal proteins, initiation factors and tRNA synthetases were removed from the dataset and the network was re-analyzed using the Ingenuity Pathway Analysis software. The top scoring networks are summarized in Online Supplementary Table S2 and the top four networks are shown in Online Supplementary Figure S3. One of the most striking features observed in cultured reticulocytes compared to native reticulocytes was a global increased abundance of metabolic enzymes. Particularly, enzymes related to glycolysis and processing of ornithine were significantly upregulated (Online Supplementary Figure S4), in accordance with previous reports.23,24 We hypothesize that the nutrient abundance present in culture media may cause the observed metabolic changes, rather than an inherent difference between native and cultured reticulocytes. Interestingly, the Ingenuity Pathway Analysis software also showed underlying differences in the abundance of various cytoskeletal proteins between native and cultured reticulocytes. Since the process of reticulocyte maturation requires extensive remodeling of the cytoskeleton,5 the Gene Ontology category of â&#x20AC;&#x2DC;Actin Cytoskeleton and Remodelingâ&#x20AC;&#x2122; was further scrutinized (Figure 1E). We confirmed significant differences in the abundance of several proteins within this category, most notably non-muscle 2002

myosin IIA and IIB (NMIIA/MYH9 and NMIIB/MYH10) and talin-1. Loss of both myosin7 and talin-121 has been previously associated with reticulocyte maturation, which leads us to hypothesize that cultured reticulocytes may represent an intermediate maturational stage compared to native reticulocytes, with additional stimuli required to complete the maturation process. Post-translational modification of proteins via phosphorylation provides an additional level of regulation beyond that achieved through differences in protein expression and is of particular importance within the transcriptionally deficient red blood cell. Phosphorylation of membrane and cytoskeletal proteins is already known to influence red blood cell membrane properties and protein interactions7,25-27 and increased phosphorylation of the cytoskeletally-associated junctional protein complex component protein 4.1R has previously been reported in murine reticulocytes compared to erythrocytes.7 In order to assess the protein phosphorylation landscape of human reticulocytes compared to erythrocytes, qualitative phosphoproteomics data for the same samples were acquired and are summarized in the diagram shown in Figure 1F, with the full dataset in Online Supplementary Table S3. The table in Figure 1G summarizes high-confidence data (false discovery rate = 1%, proteins with one single peptide across all samples disregarded), in which it can be observed that many proteins are phosphorylated exclusively in reticulocytes, consistent with their higher kinase activity.28

Non-muscle myosin IIA localizes proximally to vesicle compartments in the reticulocyte As the most highly upregulated cytoskeletal proteins in the cultured compared to native reticulocytes, myosins haematologica | 2018; 103(12)

NMII participates in reticulocyte maturation

were selected for further investigation. Coincidentally, both NMIIA and IIB were consistently phosphorylated in native and/or cultured reticulocyte samples, but not in erythrocytes. Interestingly, the occurrence of phosphorylation at the S1943 site on NMIIA has previously been implicated in cargo binding in natural killer cells.29 Since reticulocyte maturation involves vesicle extrusion,4 we examined whether phospho-NMIIA could be the driving force behind this activity. Using a well-characterized commercially available phospho-residue specific antibody we co-labeled fixed and permeabilized reticulocytes for phospho-NMIIA together with its non-phosphorylated form. Non-phosphorylated NMIIA localize at the plasma membrane (Figure 2A, Online Supplementary Figure S5), as anticipated based upon its known integration with the spectrin-actin cytoskeleton.30 However, phosphorylated NMIIA localized to punctae that were observed to be exclusive to reticulocytes. The distinct localization of phospho-NMIIA in reticulocytes was particularly striking, and we therefore decided to validate the signal of the phospho-antibody through a dephosphorylation assay with lambda phosphatase. After lambda phosphatase treatment, the phospho-NMIIA signal was largely abrogated in the reticulocyte (Figure 2B), and this result was confirmed by western blotting (Figure 2C). Unexpectedly, not all phosphoNMIIA-positive structures were found to co-label with non-phosphorylated NMIIA. We speculate that this may reflect differing binding accessibility of the antibodies used against the phosphorylated and non-phosphorylated forms of the protein. To further investigate the localization of phosphoNMIIA and to identify the adjacent vesicular compartment, reticulocytes were co-labeled with an antibody specific to the lipidated form of LC3B, an established marker of autophagic membranes.31 Figure 2D shows fluorescent labeling of phospho-NMIIA immediately proximal to the autophagic vesicles. Phosphorylated myosin light chain (S20, MLC), a marker for increased myosin activity,32 was then detected with additional vesicle markers as shown in Figure 2E. Using protease treatment methodology to detect membrane proteins that have been internalized from the plasma membrane,33 phosphorylated MLC was found adjacent to glycophorin A-positive vesicles inside the cell, emerging from the cell and in trypsin-treated reticulocytes. These combined data suggest that NMIIA may have a role in the movement of vesicles in the reticulocyte.

final system has a very simple design and can be constructed with readily available parts, as shown in the diagram (Figure 3A). Deformability and cross-sectional area of unstimulated reticulocytes and erythrocytes were measured using an Automated Rheoscope and Cell Analyzer (ARCA)19 and the measurements were used as a basis to assess alterations in these parameters in response to circulation. Figure 3B demonstrates a decrease in cross-sectional area to a range more akin to that observed for erythrocytes following overnight circulation of reticulocytes, providing important evidence that circulatory shear stress ex vivo is able to induce alterations in the morphology of reticulocytes associated with maturation. Figure 3C shows that the erythrocyte and reticulocyte populations are visually distinct when the cross-sectional area and deformability measurements are combined, with the distribution of circulated reticulocytes approaching the distribution of erythrocytes. Figure 3D shows no reduction in the deformability index profile, confirming that the cells maintain their functional viability and do not form microspherocytes. Matched control and circulated reticulocytes were submitted for TMT-based quantitative proteomics (Online Supplementary Table S4), which showed that the global outlook is of reduced protein content after circulation (Figure 3E). However, not all proteins decreased in the same manner. As expected, the amount of CD71 decreased following circulation, albeit incompletely and to a variable degree between replicates. Figure 3F shows that the most greatly decreased proteins after circulation include a significant number of mitochondrial and ribosomal proteins. Erythroid-specific proteins of the dataset are illustrated in Figure 3G, showing that the decrease is also consistent with previous literature on the protein changes inherent to reticulocyte maturation in mice7 (e.g. lower decreases of glycophorin A, band 4.1, 4.2 and spectrins compared to adducin). Of note, a large number of serum-derived proteins were shown to exhibit increased abundance within the samples prepared from circulated compared to uncirculated reticulocytes. Since the cells were incubated in identical media, it is likely that circulation results in increased adhesion of these proteins to the extracellular membrane of the reticulocytes. Although this is likely inconsequential, we cannot exclude a role for serum protein binding in facilitating maturation, therefore, a detailed list of these proteins is provided in Online Supplementary Figure S6.

In vitro circulation of cultured reticulocytes simulates the maturation process

Non-muscle myosin IIA inhibition significantly affects the response of reticulocytes to shear stress

The most striking difference between in vitro cell culturing systems and the conditions to which maturing reticulocytes are exposed in the body is the absence of shear stress caused by circulation in the bloodstream. Therefore, to investigate the effects of this influencing factor, we attempted to construct a system that could induce shear stress conditions similar to those of in vivo circulation. Microcirculation systems have previously been used to study the effects of circulation in other cell types,34,35 however, our objective was to build a relatively inexpensive and scalable system. Thus, rather than use lithography for the creation of microcapillary-scale tubing (as this would limit the volumes used), we decided to use wider and more flexible gas-permeable tubing. The

Having established a system that can be used to characterize the response of cultured reticulocytes to shear stress, we next tested the impact of blocking NMIIAmediated activity with the use of blebbistatin, a selective and high-affinity inhibitor of NMIIA and NMIIB which preferentially binds to the ATPase intermediate and slows down phosphate release.36 Blebbistatin is particularly useful for experiments of this nature because of the existence of both an active and an inactive enantiomer, blebbistatin(-) and blebbistatin(+), respectively, which allows for the use of a matched control. It has been reported that NMIIB levels are undetectable through western blotting of red blood cells, with NMIIA being the predominant isoform in human erythrocytes.37 Therefore, although bleb-

haematologica | 2018; 103(12)

2003

P.L. Moura et al.

Figure 5. Blebbistatin treatment abrogates circulation-induced differences in reticulocyte vesicle content. (A) Permeabilized trypsin-treated reticulocytes were probed with R10 (pseudo-colored green) and protein disulfide isomerase (PDI), calreticulin, giantin, LAMP-1, Mitotracker or LC3B (pseudo-colored red) and imaged using confocal microscopy. Scale bars: 5 mm. (B) Example of the machine detection algorithm used for quantitative imaging of cells containing mitochondria (Online Supplementary Methods). Images were obtained using confocal imaging. The top row shows the individual channels corresponding to Calcein Blue (pseudo-colored red) and Mitotracker (pseudo-colored green), and an overlay of the channels. In the bottom row machine detection of the Calcein Blue signal is shown through yellow borders surrounding detected cells, machine detection of the Mitotracker signal is shown with red borders surrounding detected mitochondria, and cells containing mitochondria are shown with yellow borders in the overlay column. Scale bars: 10 mm. (C) Comparison of the reduction in the proportion of cells with mitochondria after circulation using automated counting, in both untreated reticulocytes and blebbistatin(-)-treated reticulocytes. Data are represented as mean ± standard deviation (n=3).

bistatin is unable to distinguish between NMII isoforms, its effect is far more likely to have an impact on NMIIA activity. Thus, we analyzed reticulocyte deformability and circulation-mediated area loss after overnight circulation concurrent with the use of blebbistatin enantiomers. Figure 4A summarizes the multiple experiments performed. Use of either enantiomer did not affect reticulocyte deformability in a significant manner; however, the reduction in cross-sectional area resulting from circulation was diminished to non-significant levels in reticulocytes treated with blebbistatin(-). Interestingly, Smith et al. previously reported an increase in deformability of blebbistatin-treated erythrocytes as assessed through use of a microfluidic device37 in contrast to our ARCA-based observations on reticulocytes. This may reflect differences in the specific parameters assessed by each method as well as variances in the intrinsic cytoskeletal properties of the two cell types. Paired comparison graphs of the cross-sectional area and deformability of untreated, blebbistatin(-)-treated and blebbistatin(+)-treated reticulocytes are shown in Online Supplementary Figure S7. Statistical analysis of the data with pairwise comparisons of relevant pairs is shown in Figure 4B. After having determined that NMIIA inhibition leads to a decrease in the capacity of the reticulocyte to respond to shear stress, the underlying mechanism causing this difference was investigated. As the quantitative proteomics comparison between circulated and uncirculated reticulocytes showed a significant loss of mitochondrial proteins following circulation, we decided to characterize the impact of circulation on the mitochondrial content of the reticulocyte. First, in order to confirm that mitochondrial content in the reticulocyte is associated with vesicular localization (as has been previously described4), permeabilized, trypsin-treated 2004

reticulocytes were co-labeled with R10 (an antibody to glycophorin A with trypsin-sensitive epitope) and markers for mitochondria (Mitotracker), endoplasmic reticulum proteins (calreticulin and PDI), Golgi (giantin), lysosomes (LAMP-1) and autophagosomes (LC3B). Vesicles with contents positive for all of these markers were observed (Figure 5A). Cells stained with PDI antibody were manually quantified through the use of Vision4D software before and after circulation for the presence of PDI-positive vesicles, with a significant decrease (3-fold, P<0.05) in the percentage of vesicle-containing cells being observed after circulation (Online Supplementary Figure S8). To further explore the possibility of increased vesicle clearance upon circulation in a high-throughput quantitative manner, a high-resolution live-imaging protocol was developed using Vision4D software, which facilitates algorithm-based segmentation in large planes created from individual image stitching.38 For this, reticulocytes were co-labeled with Calcein Blue AM™ and the live cell imaging-compatible mitochondrial marker Mitotracker Deep Red FM™, both cell-permeable dyes. Calcein Blue was used to avoid problems related to edge detection in bright-field images, as internal edges are typically displayed in reticulocytes due to their irregular structure. The Calcein Blue-labeled cell was then used as a binary image mask for the detection of internal mitochondria. Figure 5B shows the algorithm’s method with a representative immunofluorescence image. Despite the presence of a fluorescent background in both channels used, our method is able to successfully identify cells in the image, mitochondria in the cells, and recognize cells that contain mitochondria. Automated image processing in this way also avoids user-created bias. Having developed this protocol, circulated reticulocytes haematologica | 2018; 103(12)

NMII participates in reticulocyte maturation

were left untreated or were treated with blebbistatin(-). In all cases, a reduction in the percentage of cells with mitochondria upon circulation was identified, which was significantly affected by treatment with blebbistatin(-), as shown in Figure 5C. Thus, we have demonstrated a link between mitochondrial loss and NMII activity. In concert with the observation of mitochondrial co-localization with vesicle markers, proximity of NMIIA to those same vesicle markers and the impact of NMIIA activity in the loss of cross-sectional area in the circulating reticulocyte, we have demonstrated a functional link between NMIIA activity and vesicle clearance in reticulocyte maturation.

Discussion The study of the process of human reticulocyte maturation has been massively enhanced by the ability to generate large numbers of late reticulocytes by laboratory culture methods.3 To date, much of the published literature has focused on comparison of in vitro-derived reticulocytes to circulating mature erythrocytes.20,21,39 Here we report the first quantitative protein abundance and qualitative phosphoproteomic datasets comparing human donor-matched adult native reticulocytes, erythrocytes and in vitro culturederived reticulocytes. These data highlight the broad proteomic equivalence of cultured reticulocytes to their natively derived counterparts and provide a basis for more detailed exploration of maturation processes and mechanisms that occur in vivo and in vitro. In order to explore the contribution of circulatory shear stress to the maturation of reticulocytes and dissect this mechanical element from that of cell-cell interactions and other features of the in vivo circulatory system, a simple circulation system was developed that is easily adaptable and was shown to have a significant impact on reducing cross-sectional area, vesicle/mitochondrial content and general protein content of the circulated in vitro-derived reticulocytes in a way that recapitulates their natural progression through maturation. These ex vivo data highlight the influence that shear stress is able to exert on biological processes fundamental to reticulocyte maturation. We demonstrate that while the mechanical process of circulation results in a consistent reduction of cell volume and protein abundance, circulated reticulocytes maintain partial expression of the transferrin receptor, CD71. Loss of CD71 has been described in the literature as occurring via exosome release, a process which is independent from autophagosome release.8 Thus, our data provide further evidence for these pathways being uncoupled in maturation. Importantly, these data demonstrate that shear stress is necessary but not sufficient to generate a fully mature erythrocyte and suggest that cell-cell interactions or other influencing aspects present during in vivo circulation are required to facilitate the complete maturation process. In addition to loss of cell volume, loss of organellar content is one of the distinguishing hallmarks of reticulocyte maturation in vivo, and here also shown to be stimulated by ex vivo circulation. Organelles such as mitochondria are cleared from the reticulocyte in a process partially dependent on autophagy.40 The presence of autophagic vacuoles in human reticulocytes was described by Kent et al. back in 1966.41 Subsequent studies demonstrated that the maturation of late circulating R2 reticulocytes involves the haematologica | 2018; 103(12)

generation of endocytic vesicles which fuse with autophagosomes to create large autophagic vesicles,4 corresponding to the vacuoles described by Kent et al.41 However, the process leading to transport and extrusion of these vesicles is currently undefined. Proteomic profiling of native and in vitro-derived reticulocytes identified NMIIA as being among the most differentially expressed proteins in these cells compared to mature erythrocytes. NMIIA interacts with actin to contribute to various cellular processes, such as cell migration,42 adhesion43 and cytokinesis.44 NMIIA has also been implicated in autophagosome maturation and lysosome fusion through association with autophagy-related receptors.45 In erythroid cells, the NMIIB isoform has been described as having an essential role in enucleation of the differentiating erythroblast.46 However, there is currently no defined role for NMIIA in erythroid cells other than as a component of the cytoskeleton.37 We show that reticulocytes exhibit a NMIIA phosphorylation that has been previously associated with vesicle transport29 and is undetectable in the mature red blood cell, both by mass spectrometric analysis and by immunofluorescence. Phosphorylation of the S1943 site has been associated with filament destabilization,47 which could be necessary for the regulation of NMIIA assembly dynamics.48 Phosphorylated NMIIA localizes proximally to LC3B, a known autophagic vesicle marker which was observed to colocalize with other known erythroid autophagic vesicle markers. Moreover, the phosphorylated myosin light chain was found in a similar localization. The observed concurrent localization of the phosphorylated NMIIA heavy chain and the active light chain in proximity to reticulocyte vesicles led us to speculate that NMIIA is responsible for autophagic vesicle movement in the maturing reticulocyte. We explored this hypothesis through pharmacological inhibition of NMII activity by blebbistatin, a selective and potent inhibitor, and show that while inhibition of NMII activity does not affect reticulocyte viability or capacity to deform, it leads to a significant decrease in the cellâ&#x20AC;&#x2122;s ability to respond to shear stress by loss of cell volume and autophagosome-mediated mitochondrial clearance. Finally, it is notable that shear stress-induced calcium ion (Ca2+) influx is a well-described phenomenon in erythrocytes49,50 as well as other cell types.51,52 NMIIA activity is regulated by phosphorylation of its light chain, which in turn is regulated by Ca2+ influx and interaction with calmodulin.32 It is therefore attractive to speculate that shear stress-mediated induction of Ca2+ influx may indirectly modulate NMIIA activity and thereby influence vesicle transport in the reticulocyte. However, further work is required to confirm this hypothesis. In conclusion, our results have uncovered a previously undescribed mechanism of shear stress response in the human reticulocyte which is dependent on NMII activity for vesicle clearance and cell volume reduction. Acknowledgments We would like to thank Dr. Kate Heesom and Dr. Marieangela Wilson, as well as the proteomics facility of the University of Bristol, for proteomics sample processing and data acquisition. We also thank Dr. Emile van den Akker and Sanquin (Amsterdam, the Netherlands) for providing the Sanquin reticulocyte stabilizing reagent, Matthias Rust (arivis, Rostock, Germany) for the training provided with Vision4D, Dr. 2005

P.L. Moura et al.

Jiandi Wan (Rochester Institute of Technology, New York, USA), for helpful discussions on the setup of an ex vivo circulation system and the Watson-Marlow Fluid Technology Group for permission to reproduce an image of their pump systems in Figure 3A. Funding This work was funded by the European Union (F.A. H2020MSCA-ITN-2015, â&#x20AC;&#x153;RELEVANCEâ&#x20AC;?, Grant agreement N. 675117), NHS Blood and Transplant (NHSBT) R&D grants

References 1. Bessman JD. Reticulocytes. In: Walker HK, Hall WD, Hurst JW, eds. Clinical Methods: The History, Physical, and Laboratory Examinations. Boston, 1990. 2. Gifford SC, Derganc J, Shevkoplyas SS, Yoshida T, Bitensky MW. A detailed study of time-dependent changes in human red blood cells: from reticulocyte maturation to erythrocyte senescence. Br J Haematol. 2006;135(3):395-404. 3. Giarratana MC, Rouard H, Dumont A, et al. Proof of principle for transfusion of in vitro-generated red blood cells. Blood. 2011;118(19):5071-5079. 4. Griffiths RE, Kupzig S, Cogan N, et al. Maturing reticulocytes internalize plasma membrane in glycophorin A-containing vesicles that fuse with autophagosomes before exocytosis. Blood. 2012;119(26): 6296-6306. 5. Chasis JA, Prenant M, Leung A, Mohandas N. Membrane assembly and remodeling during reticulocyte maturation. Blood. 1989;74(3):1112-1120. 6. Malleret B, Xu F, Mohandas N, et al. Significant biochemical, biophysical and metabolic diversity in circulating human cord blood reticulocytes. PLoS One. 2013;8(10):e76062. 7. Liu J, Guo X, Mohandas N, Chasis JA, An X. Membrane remodeling during reticulocyte maturation. Blood. 2010;115(10):20212027. 8. Blanc L, Vidal M. Reticulocyte membrane remodeling: contribution of the exosome pathway. Curr Opin Hematol. 2010;17(3): 177-183. 9. Koury MJ, Koury ST, Kopsombut P, Bondurant MC. In vitro maturation of nascent reticulocytes to erythrocytes. Blood. 2005;105(5):2168-2174. 10. Johnstone RM, Adam M, Hammond JR, Orr L, Turbide C. Vesicle formation during reticulocyte maturation. Association of plasma membrane activities with released vesicles (exosomes). J Biol Chem. 1987;262 (19):9412-9420. 11. Lee E, Choi HS, Hwang JH, Hoh JK, Cho YH, Baek EJ. The RNA in reticulocytes is not just debris: it is necessary for the final stages of erythrocyte formation. Blood Cells Mol Dis. 2014;53(1-2):1-10. 12. Blanc L, Liu J, Vidal M, Chasis JA, An X, Mohandas N. The water channel aquaporin-1 partitions into exosomes during reticulocyte maturation: implication for the regulation of cell volume. Blood. 2009;114(18): 3928-3934.

2006

(WP15-04 and WP15-05) and a National Institute for Health Research Blood and Transplant Research Unit (NIHR BTRU) in Red Blood Cell Products at the University of Bristol in partnership with NHSBT (NIHR-BTRU-2015-10032). We acknowledge the Wolfson Bioimaging Facility of the University of Bristol for use of their confocal systems, as well as the MRC for establishing the Facility and the BBSRC Alert 13 capital grant (BB/L014181/1) for funding the acquisition of the Leica SP8. The views expressed are those of the authors and not necessarily of the NHS, the NIHR or the Department of Health.

13. Stolz A, Ernst A, Dikic I. Cargo recognition and trafficking in selective autophagy. Nat Cell Biol. 2014;16(6):495-501. 14. Holm TM, Braun A, Trigatti BL, et al. Failure of red blood cell maturation in mice with defects in the high-density lipoprotein receptor SR-BI. Blood. 2002;99(5):18171824. 15. Zhang J, Ney PA. Role of BNIP3 and NIX in cell death, autophagy, and mitophagy. Cell Death Differ. 2009;16(7):939-946. 16. King JS, Veltman DM, Insall RH. The induction of autophagy by mechanical stress. Autophagy. 2011;7(12):1490-1499. 17. Holroyde CP, Gardner FH. Acquisition of autophagic vacuoles by human erythrocytes. Physiological role of the spleen. Blood. 1970;36(5):566-575. 18. Kupzig S, Parsons SF, Curnow E, Anstee DJ, Blair A. Superior survival of ex vivo cultured human reticulocytes following transfusion into mice. Haematologica. 2017;102(3):476-483. 19. Dobbe JG, Streekstra GJ, Hardeman MR, Ince C, Grimbergen CA. Measurement of the distribution of red blood cell deformability using an automated rheoscope. Cytometry. 2002;50(6):313-325. 20. Prenni JE, Vidal M, Olver CS. Preliminary characterization of the murine membrane reticulocyte proteome. Blood Cells Mol Dis. 2012;49(2):74-82. 21. Chu TTT, Sinha A, Malleret B, et al. Quantitative mass spectrometry of human reticulocytes reveal proteome-wide modifications during maturation. Br J Haematol. 2018;180(1):118-133. 22. Meyers NL, Brewer GJ, Tashian RE. Enzymatic synthesis of carbonic anhydrases by human reticulocytes. Biochim Biophys Acta. 1969;195(1):176-185. 23. Srivastava A, Evans KJ, Sexton AE, Schofield L, Creek DJ. Metabolomics-based elucidation of active metabolic pathways in erythrocytes and HSC-derived reticulocytes. J Proteome Res. 2017;16(4):14921505. 24. Darghouth D, Giarratana MC, Oliveira L, et al. Bio-engineered and native red blood cells from cord blood exhibit the same metabolomic profile. Haematologica. 2016;101(6):e220-222. 25. Manno S, Takakuwa Y, Mohandas N. Modulation of erythrocyte membrane mechanical function by protein 4.1 phosphorylation. J Biol Chem. 2005;280(9): 7581-7587. 26. Gauthier E, Guo X, Mohandas N, An X. Phosphorylation-dependent perturbations of the 4.1R-associated multiprotein complex of the erythrocyte membrane.

Biochemistry. 2011;50(21):4561-4567. 27. Koshino I, Mohandas N, Takakuwa Y. Identification of a novel role for dematin in regulating red cell membrane function by modulating spectrin-actin interaction. J Biol Chem. 2012;287(42):35244-35250. 28. Fairbanks G, Palek J, Dino JE, Liu PA. Protein kinases and membrane protein phosphorylation in normal and abnormal human erythrocytes: variation related to mean cell age. Blood. 1983;61(5):850-857. 29. Sanborn KB, Mace EM, Rak GD, et al. Phosphorylation of the myosin IIA tailpiece regulates single myosin IIA molecule association with lytic granules to promote NK-cell cytotoxicity. Blood. 2011;118(22): 5862-5871. 30. Basu A, Harper S, Pesciotta EN, Speicher KD, Chakrabarti A, Speicher DW. Proteome analysis of the triton-insoluble erythrocyte membrane skeleton. J Proteomics. 2015;128:298-305. 31. Hansen TE, Johansen T. Following autophagy step by step. BMC Biol. 2011;9:39. 32. Mizuno Y, Isotani E, Huang J, Ding H, Stull JT, Kamm KE. Myosin light chain kinase activation and calcium sensitization in smooth muscle in vivo. Am J Physiol Cell Physiol. 2008;295(2):C358-364. 33. Mankelow TJ, Griffiths RE, Trompeter S, et al. Autophagic vesicles on mature human reticulocytes explain phosphatidylserinepositive red cells in sickle cell disease. Blood. 2015;126(15):1831-1834. 34. Wan J, Fan R, Emery T, et al. In vitro microfluidic circulatory system for circulating cancer cells. Protoc Exch. 2016;2016. 35. Chen Y, Chan HN, Michael SA, et al. A microfluidic circulatory system integrated with capillary-assisted pressure sensors. Lab Chip. 2017;17(4):653-662. 36. Kovacs M, Toth J, Hetenyi C, MalnasiCsizmadia A, Sellers JR. Mechanism of blebbistatin inhibition of myosin II. J Biol Chem. 2004;279(34):35557-35563. 37. Smith AS, Nowak RB, Zhou S, et al. Myosin IIA interacts with the spectrinactin membrane skeleton to control red blood cell membrane curvature and deformability. Proc Natl Acad Sci USA. 2018;115(19):E4377-E4385. 38. Gutzeit E, Scheel C, Dolereit T, Rust M. Contour based split and merge segmentation and pre-classification of zooplankton in very large images. Proceedings of the 2014 9th International Conference on Computer Vision Theory and Applications (Visapp), Vol 1. 2014;417-424. 39. Wilson MC, Trakarnsanga K, Heesom KJ, et al. Comparison of the proteome of adult

haematologica | 2018; 103(12)

NMII participates in reticulocyte maturation

40.

41. 42.

43.

44.

and cord erythroid cells, and changes in the proteome following reticulocyte maturation. Mol Cell Proteomics. 2016;15(6):19381946. Zhang J, Ney PA. Reticulocyte mitophagy: monitoring mitochondrial clearance in a mammalian model. Autophagy. 2010;6(3):405-408. Kent G, Minick OT, Volini FI, Orfei E. Autophagic vacuoles in human red cells. Am J Pathol. 1966;48(5):831-857. Vicente-Manzanares M, Ma X, Adelstein RS, Horwitz AR. Non-muscle myosin II takes centre stage in cell adhesion and migration. Nat Rev Mol Cell Biol. 2009;10(11):778-790. Ebrahim S, Fujita T, Millis BA, et al. NMII forms a contractile transcellular sarcomeric network to regulate apical cell junctions and tissue geometry. Curr Biol. 2013;23 (8):731-736. Newell-Litwa KA, Horwitz R, Lamers ML.

haematologica | 2018; 103(12)

45.

46.

47.

48.

49.

Non-muscle myosin II in disease: mechanisms and therapeutic opportunities. Dis Model Mech. 2015;8(12):1495-1515. Tang HW, Wang YB, Wang SL, Wu MH, Lin SY, Chen GC. Atg1-mediated myosin II activation regulates autophagosome formation during starvation-induced autophagy. EMBO J. 2011;30(4):636-651. Ubukawa K, Guo YM, Takahashi M, et al. Enucleation of human erythroblasts involves non-muscle myosin IIB. Blood. 2012;119(4):1036-1044. Dulyaninova NG, Malashkevich VN, Almo SC, Bresnick AR. Regulation of myosin-IIA assembly and Mts1 binding by heavy chain phosphorylation. Biochemistry. 2005;44 (18):6867-6876. Betapudi V, Gokulrangan G, Chance MR, Egelhoff TT. A proteomic study of myosin II motor proteins during tumor cell migration. J Mol Biol. 2011;407(5):673-686. Brain MC, Pihl C, Robertson L, Brown CB.

50.

51.

52.

53.

Evidence for a mechanosensitive calcium influx into red cells. Blood Cells Mol Dis. 2004;32(3):349-352. Danielczok JG, Terriac E, Hertz L, et al. Red blood cell passage of small capillaries is associated with transient Ca(2+)-mediated adaptations. Front Physiol. 2017;8:979. Ando J, Komatsuda T, Kamiya A. Cytoplasmic calcium response to fluid shear stress in cultured vascular endothelial cells. In Vitro Cell Dev Biol. 1988;24(9):871-877. Ikeda Y, Handa M, Kamata T, et al. Transmembrane calcium influx associated with von Willebrand factor binding to GP Ib in the initiation of shear-induced platelet aggregation. Thromb Haemost. 1993;69(5): 496-502. Hornbeck PV, Zhang B, Murray B, Kornhauser JM, Latham V, Skrzypek E. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 2015;43(Database issue):D512-520.

2007

ARTICLE

Red Cell Biology & its Disorders

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2008-2015

The phenotypic spectrum of germline YARS2 variants: from isolated sideroblastic anemia to mitochondrial myopathy, lactic acidosis and sideroblastic anemia 2

Lisa G. Riley,1,2,* Matthew M. Heeney,3,4,* Joëlle Rudinger-Thirion,5 Magali Frugier,5 Dean R. Campagna,6 Ronghao Zhou,3 Gregory A. Hale,7 Lee M. Hilliard,8 Joel A. Kaplan,9 Janet L. Kwiatkowski,10,11 Colin A. Sieff,3,4 David P. Steensma,12,13 Alexander J. Rennings,14 Annet Simons,15 Nicolaas Schaap,16 Richard J. Roodenburg,17 Tjitske Kleefstra,15 Leonor Arenillas,18 Josep Fita-Torró,19 Rasha Ahmed,20 Miguel Abboud,20 Elie Bechara,21 Roula Farah,21 Rienk Y. J. Tamminga,22 Sylvia S. Bottomley,23 Mayka Sanchez,19,24,25 Gerwin Huls,26 Dorine W. Swinkels,27 John Christodoulou1,2,28,29,# and Mark D. Fleming3,6,13,# *LGR and MMH contributed equally to this work. #JC and MDF contributed equally to this work as co-senior authors.

Genetic Metabolic Disorders Research Unit, Kids Research Institute, Children’s Hospital at Westmead, Sydney, Australia; 2Discipline of Child & Adolescent Health, Sydney Medical School, University of Sydney, Australia; 3Dana Farber-Boston Children’s Center for Cancer and Blood Disorders, Boston, MA, USA; 4Department of Pediatrics, Harvard Medical School, Boston, MA, USA; 5Architecture et Réactivité de l’ARN, Université de Strasbourg, CNRS, IBMC, Strasbourg, France; 6Department of Pathology, Boston Children's Hospital, Boston, MA, USA; 7Johns Hopkins All Children's Hospital, St. Petersburg, FL, USA; 8Division of Pediatric Hematology Oncology, University of Alabama at Birmingham, AL, USA; 9Levine Children's Hospital, Charlotte, NC, USA; 10The Children’s Hospital of Philadelphia, Division of Hematology, Philadelphia, PA, USA; 11University of Pennsylvania School of Medicine, Philadelphia, PA, USA; 12Adult Leukemia Program, Dana-Farber Cancer Institute, Boston, MA, USA; 13Harvard Medical School, Boston, MA USA; 14Department of Internal Medicine, Radboud University Medical Centre, Nijmegen, the Netherlands; 15Department of Human Genetics, Radboud University Medical Centre, Nijmegen, the Netherlands; 16Department of Hematology, Radboud University Medical Centre, Nijmegen, the Netherlands; 17Radboud Center for Mitochondrial Medicine, Translational Metabolic Laboratory, Department of Pediatrics, Radboud University Medical Centre, Nijmegen, the Netherlands; 18Laboratorio Citología Hematológica, Servicio Patología, GRETNHE, IMIM Hospital del Mar Research Institute, Hospital del Mar, Barcelona, Spain; 19Iron metabolism: regulation and disease group, Josep Carreras Leukaemia Research Institute (IJC), Campus ICO-Germans Trias i Pujol, Campus Can Ruti, Carretera de Can Ruti, Cami de les Escoles, Badalona, Spain; 20Department of Pediatrics and Adolescents, American University of Beirut Medical Center, Beirut, Lebanon; 21Department of Pediatrics, Saint George Hospital University Medical Center, Beirut, Lebanon; 22Beatrix Children’s Hospital, Department of Pediatric Hematology, University Medical Center Groningen, University of Groningen, the Netherlands; 23 Department of Medicine, University of Oklahoma College of Medicine, Oklahoma City, OK, USA; 24Programme of Predictive and Personalized Medicine of Cancer, Germans Trias i Pujol Research Institute (PMPPC-IGTP), Badalona, Spain; 25BloodGenetics, S.L., Esplugues de Llobregat, Barcelona, Spain; 26Department of Hematology, University Medical Center Groningen, the Netherlands; 27Department of Laboratory Medicine, Translational Metabolic Laboratory, Radboud University Medical Centre, Nijmegen, the Netherlands; 28Neurodevelopmental Genomics Research Group, Murdoch Childrens Research Institute, Melbourne, Australia and 29Department of Paediatrics, Melbourne Medical School, University of Melbourne, Australia 1

Correspondence: john.christodoulou@mcri.edu.au

Received: October 17, 2017. Accepted: July 12, 2018. Pre-published: July 19, 2018. doi:10.3324/haematol.2017.182659 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2008 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2008

ABSTRACT

ARS2 variants have previously been described in patients with myopathy, lactic acidosis and sideroblastic anemia 2 (MLASA2). YARS2 encodes the mitochondrial tyrosyl-tRNA synthetase, which is responsible for conjugating tyrosine to its cognate mt-tRNA for mitochondrial protein synthesis. Here we describe 14 individuals from 11 families presenting with sideroblastic anemia and YARS2 variants that we identified using a sideroblastic anemia gene panel or exome sequencing. The phenotype of these patients ranged from MLASA to isolated congenital sideroblastic anemia. As in previous cases, inter- and intrafamilial phenotypic variability was observed, however, this report haematologica | 2018; 103(12)

YARS2 congenital sideroblastic anemia

includes the first cases with isolated sideroblastic anemia and patients with biallelic YARS2 variants that have no clinically ascertainable phenotype. We identified ten novel YARS2 variants and three previously reported variants. In vitro amino-acylation assays of five novel missense variants showed that three had less effect on the catalytic activity of YARS2 than the most commonly reported variant, p.(Phe52Leu), associated with MLASA2, which may explain the milder phenotypes in patients with these variants. However, the other two missense variants had a more severe effect on YARS2 catalytic efficiency. Several patients carried the common YARS2 c.572 G>T, p.(Gly191Val) variant (minor allele frequency = 0.1259) in trans with a rare deleterious YARS2 variant. We have previously shown that the p.(Gly191Val) variant reduces YARS2 catalytic activity. Consequently, we suggest that biallelic YARS2 variants, including severe loss-of-function alleles in trans of the common p.(Gly191Val) variant, should be considered as a cause of isolated congenital sideroblastic anemia, as well as the MLASA syndromic phenotype.

Introduction Sideroblastic anemia is defined by the presence of bone marrow ringed sideroblasts, which are erythroblasts containing pathological intramitochondrial iron deposits.1 Congenital sideroblastic anemias (CSAs) are caused by a growing list of genetic variants that affect mitochondrial pathways, including heme synthesis, iron-sulfur cluster biogenesis, mitochondrial protein synthesis, and oxidative phosphorylation.2,3 Variants in YARS2 have been associated with myopathy, lactic acidosis, and sideroblastic anemia 2 (MLASA2; OMIM #613561),4-8 and recently cases of YARS2-related myopathy in the absence of sideroblastic anemia have been reported.9 YARS2 encodes the mitochondrial tyrosyl-tRNA synthetase, YARS2, which is responsible for the ATP-dependent conjugation of tyrosine to its cognate tRNA, required to support mitochondrial protein synthesis.10 YARS2 catalyses this reaction in a two-step process. In the first step, tyrosine and ATP bind to the catalytic domain to form the tyrosyl-adenylate intermediate. In the second step, cognate tRNATyr binds the synthetase and the tyrosyl moiety is transferred to the tRNA CCA-end. The resulting tyrosyl-tRNATyr will be delivered to the ribosome. The most commonly reported YARS2 variant, p.(Phe52Leu), prevalent in patients of Lebanese Christian descent, has been shown to reduce YARS2 amino-acylation catalytic efficiency by approximately 9-fold, and leads to a reduction in mitochondrial protein synthesis in patients with MLASA2.4 Here we report YARS2 variants, some of which were associated with milder effects on amino-acylation, in patients with isolated CSA, or CSA with mild myopathy and lactic acidosis. In addition, we describe two pairs of genotypically identical siblings with divergent, affected and unaffected, clinical phenotypes. Importantly, some patients carry a common YARS2 c.572 G>T, p.(Gly191Val), that we and others have previously shown has a mild effect on amino-acylation activity,5,11 and suggest that these milder alleles may be the basis of the reduced penetrance and expressivity.

Methods

Hospital, USA, the Radboud University Medical Center, the Netherlands, and the Hospital Germans Trias i Pujol, Badalona, Spain. In each case, CSA was ascertained by complete blood counts (CBCs), and peripheral blood or bone marrow morphology. Detailed clinical histories are provided in the Online Supplementary Appendix.

Variant detection Targeted sequencing of nuclear encoded CSA genes,12 and the mitochondrial genome as well as mitochondrial DNA deletion analysis was performed on the probands of families 1-3 and 5-9. Genomic DNA was isolated from peripheral blood or skin fibroblasts, using the Puregene DNA Purification Kit (Qiagen, Valencia, CA, USA). DNA templates for sequencing were amplified from genomic DNA by PCR, enzymatically cleaned, bidirectionally sequenced using fluorescent dye termination sequencing chemistry, and analyzed with the Sequencher 5.3 DNA sequence assembly software (Gene Codes, Ann Arbor, MI, USA), as previously described.12 Exome sequencing for Patient 4 was performed on genomic DNA isolated from whole blood. The experimental workflow was performed at BGI Europe (Bejing Genome Institute Europe, Copenhagen, Denmark) using an Illumina Hiseq (Illumina, CA, USA) platform. Variants in genes previously associated with Mendelian diseases (OMIM), including CSAs, were analyzed bioinformatically. Patient 10 DNA was analyzed using a targeted gene panel for congenital and acquired sideroblastic anemias, including ABCB7, ALAS2, GLRX5, PUS1, SF3B1, SLC19A2, SLC25A38, STEAP3, TRNT1 and YARS2. The library was constructed using the Custom HaloPlexâ&#x201E;˘ Target Enrichment System (Agilent, Santa Clara, CA, USA) and sequenced on a MiSeq platform (Illumina, San Diego, CA, USA). Data were analyzed with SureCall software (Agilent, Santa Clara, CA, USA). Patient 11 DNA was analyzed using a targeted gene panel for sideroblastic anemia (ABCB7, ALAS2, GLRX5, HSCB, HSPA9, PUS1, SLC25A38, STEAP3, YARS2) and ion semiconductor sequencing as developed by Ion Torrent systems.13 In silico predictions of variant pathogenicity were performed using the Alamut Visual suite of genetic analysis software (Interactive Biosoftware, Rouen, France), and linking externally to the PolyPhen2 and SIFT analytical tools.14,15 Minor allele frequencies are reported as in gnomAD (gnomad.broadinstitute.org) current as of September 2017.16

Clinical data The patients and their immediate family members were referred to MMH, MDF, NS or LA for clinical consultation. Written informed consent was obtained from participants in the study, as approved by the Institutional Review Boards of Boston Childrenâ&#x20AC;&#x2122;s haematologica | 2018; 103(12)

Amino-acylation assays Recombinant wild-type and the p.(Leu61Val), p.(Met195Ile), p.(Ser203Ile), p.(Tyr236Cys) and p.(Gly244Ala) YARS2 variants were expressed in E. coli, purified to homogeneity and assayed for 2009

L.G. Riley et al. tyrosylation activity as previously described.10 Apparent kinetic parameters were determined from Lineweaver-Burk plots in the presence of 4.8 to 6.5 nM YARS2 and 0.28 to 1.12 mM native E. coli tRNATyr (Sigma, St. Louis, MO, USA). Experimental errors on kcat and Km varied at most by 20%. Numerical values are averages of at least two independent experiments.

Results Phenotypic spectrum Eleven probands with CSA were identified with potentially pathogenic YARS2 variants by targeted gene

sequencing panels or exome sequencing (Table 1A and 1B). The majority of these families were derived from a group of more than 200 probands with CSA referred to SSB, MDF and MMH, in which approximately 4% of cases were attributed to YARS2 variants. YARS2 variants have previously been identified in patients with myopathy, lactic acidosis and sideroblastic anemia 2 (MLASA2);4 however, some patients in this study did not have overt clinical features of MLASA2 other than CSA, and several individuals with biallelic variants had no phenotype whatsoever. In two families, the proband had moderate sideroblastic anemia (P8a and P9a), while a sibling with the same YARS2 genotype was not anemic and was otherwise

Table 1A. Clinical data.

Participant ID

YARS2 variant 1 (NM_001040436.2) YARS2 variant 2

c.156C>G p.(Phe52Leu) c.156C>G p.(Phe52Leu) 1988 Female Lebanese/ American No 14 years Severe, transfusion dependent from 27 years

Year of birth Gender Ethnicity Consanguinity Age at presentation Sideroblastic anemia

Hemoglobin, g/dL MCV, fL Abs Retic, M/mL Retic, % WBC x109/L ANC x109/L Platelets x109/L RS, % of BM erythroblasts Transferrin saturation, % Ferritin, ng/mL Chelation (year started) Lactic acidosis Myopathy Other clinical features

Vital status

9.9 84.5 0.101 3.1 2.41 780 294 10 60 (2002) 34.4 (2002) No

P2a

P2b

c.156C>G c.156C>G c.156C>G p.(Phe52Leu) p.(Phe52Leu) p.(Phe52Leu) c.156C>G c.156C>G c.156C>G p.(Phe52Leu) p.(Phe52Leu) p.(Phe52Leu) 2007 2009 2007 Male Female Male Lebanese Lebanese Lebanese Yes 6 years Severe, transfusion dependent from 6 years 10.5 111 0.035 1 4.35 2960 195 ND 97 (2016) 825 (2016) Yes (2016) ND

Severe 9.1 mmol/L Severe None Sinus Atrial tachycardia, septal pericardial effusion, defect neutropenia thrombocytopenia, primary ovarian failure Deceased at 28 y Alive

c.181C>G p.(Leu61Val) c.181C>G p.(Leu61Val) 1986 Male Caucasian/ Dutch Yes No Yes 4 years 9 years 19 years Mild Moderate Severe, transfusion dependent intermittently from 20 years 11.5 9.5 6.6 102 92.4 81 0.13 0.132 ND 2.9 2 1.1 6.6 5.2 5.4 3480 2012 ND 305 216 374 ND 30 56 80 (2016) 45 90 (2014) 296 (2016) 61 683 (2014) Yes No Yes (2017) (2012) ND Severe Exercise 9.5 mmol/L induced only None Mild None None Diarrhea, Successful hepatosplenomegaly stem cell transplant at 28 years

Alive

c.585G>A c.572G>T p.(Met195Ile) p.(Gly191Val) c.1165_1166insG c.590_625del p.(Leu389Cysfs*6) p.(Thr197_Leu208del) 2001 1998 Female Female Caucasian/ African American American No No 2 years 20 months Severe, Severe, transfusion transfusion dependent dependent intermittently from 20 months from 2 years 3 2 82.6 101 0.0175 0.016 1.7 2.3 3.63 6.2 617 861-2070 163 324 ND >15 53 91 (2000) 93 256 (2000) Yes Yes (2011) (2004) Premortem Mild only Mild None Mild Thrombocytopenia, cardiomyopathy intermittent neutropenia

Deceased at 12 y

Alive

MCV: mean corpuscular volume; retic: reticulocytes; WBC: white blood cell count; ANC: absolute neutrophil count; RS: ringed sideroblasts; BM: bone marrow; ND: not determined; y: years.

2010

haematologica | 2018; 103(12)

YARS2 congenital sideroblastic anemia

asymptomatic (P8b and P9b) (Table 1B). In a third family (P2a and P2b) (Table 1A), the proband was identified with a severe, new onset anemia at six years of age, and, subsequent to her brotherâ&#x20AC;&#x2122;s diagnosis, the younger sibling was found to be anemic. Four of the probands presented within the first two years of life (P5, P6, P7, P9a), and 4 presented in adolescence (P1, P4, P8a, P11). Two patients have died (P1, P5), both from multi-organ failure, one of these following two unsuccessful hematopoietic stem cell transplantations (HSCTs). One patient (P4) has undergone successful HSCT. The 11 probands all had moderate to severe normocytic to macrocytic anemia. In nine probands, the presence of ringed sideroblasts, ranging from 10% to over 50% of bone marrow erythroblasts, was documented on bone marrow aspiration; marrows were not examined in 3

other patients and 2 clinically unaffected siblings (Table 1A and B). Eight patients required transfusion; however, one patient spontaneously became transfusion independent at 16 months of age (P7), and 3 patients had periods of hematologic remission (P4, P5, P9a), transiently becoming RBC transfusion independent. In addition to anemia, 3 probands had variable neutropenia and/or thrombocytopenia (P1, P6, P8a). Four patients were treated with pyridoxine with no improvement in their anemia (P4, P5, P6, P11). Two patients had severe lactic acidosis (P1, P3), but the remaining cases in which it was studied had mild or no lactic acidosis (Table 1A and B). Two patients had elevated blood lactate upon light exercise (P4, P8a); those with mild lactic acidosis also tended to have mild myopathy, although one patient with no reported lactic acidosis had

Table 1B. Clinical data.

Participant ID

P8a

P8b

P9a

P9b

P10

P11

YARS2 variant 1 (NM_001040436.2) YARS2 variant 2

c.[572G>T;731G>C] c.572G>T c.572G>T c.98C>A c.98C>A c.608G>T c.933C>G p.(Gly191Val); (Gly244Ala)] p.(Gly191Val) p.(Gly191Val) p.(Ser33*) p.(Ser33*) p.(Ser203Ile) p.(Asp311Glu) c.933C>G c.1360_1361insG c.1360_1361insG c.707A>G c.707A>G c.1104-1G>A c.933C>G p.(Asp311Glu) p.(Ile454Serfs*10) p.(Ile454Serfs*10) p.(Tyr236Cys) p.(Tyr236Cys) p.? p.(Asp311Glu) Year of birth 1999 1963 1965 2010 2010 1992 2003 Gender Female Female Female Male Male Female Male Ethnicity Caucasian/ Caucasian/ Caucasian/ Caucasian/ Caucasian/ Caucasian/ Caucasian / American American American American American Spanish Dutch Consanguinity No No No No No Unknown No Age at presentation 3 months 18 years 49 years 3 months 3 months 23 years 13 years (Asymptomatic) (Asymptomatic) Sideroblastic Severe, transfusion Moderate None Severe, None Moderate Severe anemia dependent until transfused transfusion 16 months intermittently from dependent from 3 months 13 years Hemoglobin, g/dL 5.8 9.9 13.9 2.4 12.8 9.6 6.6 MCV, fL 94.6 111.9 82 113.8 94.1 86 95 Abs Retic, M/mL 0.037 0.059 0.106 0.015 0.053 0.088 0.018 Retic, % 1.8 2.3 2.1 2.4 1.3 2.38 0.8 WBC x109/L 8.01 6 6.8 10.1 9.8 7.65 4.9 ANC x109/L 1201 3600 4340 1919 3180 4280 1700 Platelets x109/L 337 149 182 537 414 243 257 RS, % of BM erythroblasts 47 40 ND >50 ND 32 81 Transferrin saturation, % ND 66.7 (2015) Unknown 45 (2016) 51 (2015) 79.4 (2015) 62 (2016) Ferritin, ng/mL ND 387 (2015) Unknown 225 (2016) 42 (2015) 295 (2015) 180 (2016) Chelation No No No No No No Yes (2017) (Year started) Lactic acidosis None Exercise induced only None None ND None Mild Myopathy Moderate Mild None None None None Mild Other clinical Intermittent Dependent None Facial Facial Asthenia None features diarrhea and edema, leukopenia, dysmorphism dysmorphism abdominal pain thrombocytopenia, atypical pulmonary carcinoid tumor (age 53) Vital status Alive Alive Alive Alive Alive Alive Alive MCV: mean corpuscular volume; retic: reticulocytes; WBC: white blood cell count; ANC: absolute neutrophil count; RS: ringed sideroblasts; BM: bone marrow; ND: not determined.

haematologica | 2018; 103(12)

2011

L.G. Riley et al. A

Figure 1. Representation of mutated YARS2 proteins. (A) Schematic view of YARS2 domains: MTS: mitochondrial targeting sequence; ACB: anticodon binding domain; S4-Like: S4 ribosomal protein-like domain. Amongst all the variants identified, only those tested in this study are shown in cyan. Note that the recombinant YARS2 used in the amino-acylation assays is deprived of the MTS. (B) Model of YARS2 p.(Thr197-Leu208del), built with I-TASSER.28 The structural domains from (A) are shown with the same color code. The locations of the variants, which have the weakest effects on amino-acylation [p.(Leu61Val), p.(Met195Ile), p.(Tyr236Cys)] are shown in cyan. (C) Crystal structure of YARS2 catalytic domain19 with the tyrosyl-adenylate analog (TyrAMS, magenta) bound to the active site. The locations of variants p.(Ser203Ile) and p.(Gly244Ala), characterized by the strongest effects on amino-acylation, are indicated in cyan.

Table 2. In silico predictions of pathogenicity for YARS2 missense variants.

YARS2 variant

SIFT prediction

SIFT score

PolyPhen2 prediction

PolyPhen2 score

gnomAD frequency (%)

p.(Leu61Val) p.(Met195Ile) p.(Ser203Ile) p.(Tyr236Cys) p.(Gly244Ala)

Deleterious Tolerated Deleterious Tolerated Deleterious

0.03 0.17 0.02 0.09 0.00

Benign Possibly damaging Probably damaging Probably damaging Probably damaging

0.001 0.827 0.989 1.000 0.995

0.0016* 0 0 0.0008* 0.0047*

*No homozygotes reported.

moderate myopathy (P7). Patient 1 (P1) with severe lactic acidosis and myopathy had combined respiratory chain deficiency in skeletal muscle, and the muscle biopsy showed histopathological features typical of a mitochondrial myopathy, including ragged red fibers on trichrome stain and “parking lot” inclusions and whorled arrays of mitochondrial cristae by transmission electron microscopy (data not shown). In one family, the proband (P9a) and his clinically unaffected, but genotypically identical sibling (P9b), had distinctive “triangular” faces, unlike their parents or genotypically normal sibling, which has not previously been reported in association with YARS2 variants, but has been described in mitochondrial myopathy with lactic acidosis and sideroblastic anemia 1 (MLASA1; OMIM #600462) due to pseudouridine synthase 1 (PUS1) variants.17 2012

YARS2 variants in patients with congenital sideroblastic anemia We identified three previously described YARS2 variants and ten novel variants in patients with CSA: the Lebanese Christian founder variant, p.(Phe52Leu),4 was in the homozygous state in 4 patients; the p.(Asp311Glu) variant8 homozygous in one patient; and a novel variant, p.(Leu61Val) homozygous in one patient. The remaining six families had compound heterozygous variants including four novel missense variants: p.(Met195Ile), p.(Ser203Ile), p.(Tyr236Cys), p.(Gly244Ala); a novel nonsense variant p.(Ser33*); three novel indels, p.(Thr197_Leu208del), p.(Leu389Cysfs*6), p.(Ile454Serfs*10); one novel splicing variant, c.1104-1G>A; and two previously reported missense variants, p.(Gly191Val) and p.(Asp311Glu).5,8 No patient had two indel or splicing variants. haematologica | 2018; 103(12)

YARS2 congenital sideroblastic anemia

The five novel missense variants all lie in the catalytic domain of YARS2 (Figure 1A) and are rare in the gnomAD database (gnomad.broadinstitute.org) (Table 2). In silico predictions of pathogenicity for p.(Leu61Val), p.(Met195Ile) and p.(Tyr236Cys) vary between the SIFT and PolyPhen2 prediction programs while p.(Ser203Ile) and p.(Gly244Ala) are predicted to be damaging to the YARS2 protein by both algorithms (Table 2 and Figure 1B). Conservation among species for each missense variant is shown in Online Supplementary Figure S1. The nonsense variant, the splicing variant and three novel indels are likely to be deleterious. The splicing variant c.1104-1G>A alters a canonical position in the 3ʹ splice acceptor site of intron 3 and it is predicted to result in skipping of exon 4. The YARS2 c.98C>A, p.(Ser33*) nonsense variant and the c.1165_1166insG, p.(Leu389Cysfs*6) frameshift variant both lie greater than 55 nucleotides upstream of the last exon-exon junction and are most likely targeted for nonsense mediated decay.18 The p.(Thr197_Leu208) in frame deletion results in loss of 12 residues in α-helical regions of the catalytic domain, and more precisely of cluster 1, which is important for tRNA 1B). The acceptor end recognition19 (Figure c.1360_1361insG, p.(Ile454Serfs*10) variant lies in the last exon of YARS2 and is not predicted to be targeted for nonsense mediated decay.18 This variant would cause a frameshift at position 454 in the S4-like domain, which is found in all prokaryotic and organellar tyrosyl-tRNA synthetases, and is thought to stabilize the interaction between the tRNA and YARS2.19,20

Amino-acylation activity of YARS2 missense variants Amino-acylation assays are commonly used to evaluate the effect of variants on aminoacyl-tRNA synthetase activity, with reduced activity being a strong predictor of pathogenicity.21 Consequently, the effect of the five missense variants, p.(Leu61Val), p.(Met195Ile), p.(Ser203Ile), p.(Tyr236Cys) and p.(Gly244Ala) on amino-acylation activity was measured by the incorporation of [14C]-tyrosine into an E. coli tRNATyr substrate and compared to wildtype YARS2 activity. In vitro studies of the YARS2 variants revealed that amino-acylation efficiency was mildly reduced for p.(Leu61Val) and, p.(Met195Ile), while p.(Tyr236Cys) was not affected as compared to the wildtype enzyme (Table 3). YARS2 p.(Ser203Ile) and p.(Gly244Ala) demonstrated a 17-fold loss in catalytic efficiency. The reduced activity of YARS2 p.(Ser203Ile) is a consequence of an increased Km, indicating that its affinity for tRNATyr was reduced. On the other hand, the YARS2 p.(Gly244Ala) is characterized by a 13-fold lower kcat suggesting that the variant hinders efficient transfer of the tyrosyl moiety from the active site to the tRNA.

Discussion Here we expand the clinical spectrum associated with YARS2 variants and describe patients with milder phenotypes who do not display all the features of MLASA2. Rather, most of the patients we describe presented principally with a normo- or macro-cytic CSA; they are mostly non-syndromic and unlike the most common forms of non-syndromic sideroblastic anemia (e.g. ALAS2 or SLC25A38 deficiency), the anemia is not microcytic. Nevertheless, in addition to ringed sideroblasts, some of haematologica | 2018; 103(12)

Table 3. Kinetic parameters for tyrosylation of E. coli tRNATyr by YARS2 wild-type and novel missense variant recombinant proteins.

YARS2 variant

Km (mM)

WT p.(Leu61Val) p.(Met195Ile) p.(Ser203Ile) p.(Tyr236Cys) p.(Gly244Ala)

0.75 1.45 1.90 18.60 0.70 1.00

kcat kcat /Km Loss of (min-1) (efficiency) efficiency* (fold change) 20.0 9.1 25.5 28.6 16.5 1.5

26.7 6.3 13.4 1.5 23.6 1.5

1 4.2 2.0 17.3 0.9 17.8

*Loss of efficiency is calculated relative to wild-type (WT) YARS2.

these patients had vacuolization of marrow precursors and/or other cytopenias that are often seen in the syndromic sideroblastic anemias (e.g. Pearson syndrome), which may be a diagnostic clue. Patients 1 and 3 had all the typical features of MLASA2, whereas Patients 2a, and 2b, who share homozygosity for the YARS2 Lebanese founder allele, p.(Phe52Leu), had only anemia. Patient 1 also had other features not typically associated with MLASA2, including neutropenia, thrombocytopenia, pericardial effusion, and premature ovarian failure. Neutropenia and pericardial effusion have each been reported in one other patient homozygous for the p.(Phe52Leu) variant.5,22 Two other patients in the current series with other genotypes also had mild or intermittent neutropenia. Premature ovarian failure is associated with variants in several mitochondrial aminoacyl-tRNA synthetase-encoding genes including HARS2, LARS2 and AARS2,23-25 and thus may be a feature common to mitochondrial protein synthesis defects. There are now 10 reported individuals homozygous for the YARS2 p.(Phe52Leu) variant5,22 and all have been symptomatic, supporting complete penetrance of this allele. However, the great range of phenotypic severity strongly suggests the presence of other genetic and environmental influences that can modify the effects of YARS2 deficiency. Patient 4 presented in late adolescence with sideroblastic anemia without myopathy and has a homozygous p.(Leu61Val) variant that diminished the amino-acylation catalytic efficiency 4-fold. Leu61 is located in a region of the catalytic domain specific to mitochondrial YARSs that was proposed to contact the tRNATyr acceptor helix (Figure 1B).19 In this case, HSCT appeared to be an effective treatment, restoring the patient’s hemoglobin levels to normal. Patient 5 presented in infancy with CSA and was transfusion dependent other than a remission occurring between three and six years of age; she had no myopathy until her post-HSCT terminal illness. This patient had a YARS2 c.1165_1166insG variant predicted to result in a null allele, and a novel p.(Met195Ile) variant which lies within cluster 1, in a region involved in recognition of the tyrosine accepting arm of tRNATyr (Figure 1B).19 Some YARS proteins (e.g. yeast) have an isoleucine (Ile) at this position, suggesting that it might be a milder allele. Indeed, in vitro this mutant had little effect on YARS2 catalytic efficiency. Patient 10 is a compound heterozygote for a splicing mutation (c.1104-1G>A) predicted to cause skipping of exon 4, and a missense variant p.(Ser203Ile), also located 2013

L.G. Riley et al.

in cluster 1. YARS2 (p.Ser203Ile) led to a reduced affinity for tRNATyr, resulting in a 17-fold loss in catalytic efficiency (Figure 1C). Patient 10 has no lactic acidosis or myopathy, and presented with isolated normocytic anemia and asthenia, and has not required transfusion. Patient 7 presented with anemia in infancy requiring two transfusions within the first 16 months of life and then became transfusion independent. She has moderate myopathy and no lactic acidosis and a compound heterozygous genotype: a missense variant, p.(Gly244Ala), occurring in cis with p.(Gly191Val) and in trans with the p.(Asp311Glu) variant. Gly244 is a critical residue for tyrosyl-adenylate binding.19 YARS2 p.(Gly244Ala) only affected the kcat indicating that, as predicted, this variant hinders binding of the tyrosyl-adenylate in the active site (Figure 1C). YARS2 Asp311 is involved in the recognition of anticodon residue G34 of tRNATyr.19 The p.(Asp311Glu) variant is respiratory deficient in a yeast model, and patients homozygous for this allele also have transfusion-dependent sideroblastic anemia in the first year of life; however, in contrast to patient 7, they have lactic acidosis but no myopathy.8 Further phenotypic variability for the p.(Asp311Glu) variant was observed in Patient 11 who was homozygous for p.(Asp311Glu), with transfusiondependent MLASA2. In two families in this study (Families 6 and 8), affected patients have the common p.(Gly191Val) allele (MAF = 0.1259) in trans of a predicted null allele. Importantly, all of the unaffected carriers of predicted null alleles in these and other families, where probands had the ancestral p.Gly191 variant in trans, were asymptomatic (data not shown). Patient 6 presented in infancy with CSA requiring transfusions every three weeks. She has mild lactic acidosis, no myopathy and intermittent neutropenia. She has a c.590_645del variant resulting in a 12 amino acid deletion in the catalytic domain (Figure 1A), which would almost certainly lead to a completely dysfunctional protein, in trans with p.(Gly191Val). Individuals 8a and 8b also carry p.(Gly191Val) in trans with a predicted null or severe loss-of-function allele, c.1360_1360insG, p.(Ile454Serfs*10). This variant truncates the S4-like domain which is thought to stabilize the interaction with tRNATyr, and the deletion of the YARS2 S4-like domain leads to a 100-fold reduced amino-acylation activity in vitro.20 Patient 8a had sideroblastic anemia and edema. Lactate was elevated only on exertion and the patient did not have myopathy. Her sister (P8b) is asymptomatic. Patient 8a also had a somatic mutation in SF3B1 p.(Lys700Glu) that is strongly associated with myelodysplastic syndromes with ringed sideroblasts.26 Based on the childhood presentation of her anemia and exercise intolerance that was exacerbated significantly decades later, and the fact that a mutation in SF3B1 would be exceptional in a patient under 30 years of age, we infer the YARS2 mutations to be the primary cause of her anemia with the SF3B1 mutation occurring as a secondary somatic event, which exacerbated her anemia, bringing her to clinical attention. In addition to the reduced activity in vitro,5 support of the notion that YARS2 p.(Gly191Val) contributes to the disease phenotype in these patients comes from the observation that this variant is a disease modifier in Leber Hereditary Optic Neuropathy (LHON); the three common LHON

2014

mitochondrial DNA mutations have incomplete penetrance. However, all patients who carry both the LHON m.11778G>A mtDNA disease-associated variant in combination with a homozygous YARS2 p.(Gly191Val) genotype were symptomatic.11 Patients 9a and 9b carried the YARS2 c.98C>A, p.(Ser33*) nonsense variant, which would result in a null allele, and the p.(Tyr236Cys) variant (Figure 1A and B) that did not alter amino-acylation activity in vitro. In addition, in silico analysis using Alamut did not predict that this variant would lead to alteration of an exonic splicing enhancer site. Patient 9a presented in infancy with sideroblastic anemia that has come and gone throughout his life. He has no lactic acidosis or myopathy. He and his unaffected brother have some dysmorphic features, which have not previously been reported in association with YARS2 variants, but are typical of MLASA1 patients with pseudouridine synthase 1 (PUS1) mutations.17,27 His genotypically concordant fraternal twin (P9b) has only mild anemia and similar facial dysmorphology, once again highlighting the potential for decreased penetrance and/or expressivity of the disorder. Interestingly, some YARS2 patients with myopathy, but no sideroblastic anemia, have recently been reported by Sommerville et al.9 They report siblings with a homozygous YARS2 p.(Leu392Ser) variant who had MLASA2, while another individual homozygous for the same variant had myopathy without sideroblastic anemia or lactic acidosis. To summarize, the inter- and intra-familial phenotypic variability, intermittent transfusion dependence of some YARS2 cases, and the association of a common variant with disease, suggest that all MLASA2 phenotypes may be susceptible to subtle changes in YARS2 function, which may in turn be influenced by genetic and/or environmental modifiers. This study shows that YARS2 variants can result in CSA in the absence of clinically significant myopathy or lactic acidosis. Thus, we recommend that YARS2 variants be considered as a cause of isolated sideroblastic anemia as well as MLASA2 or mitochondrial myopathy. Funding This research was supported by Australian NHMRC grant 1026891 to JC, NIH DK087992 to MDF, and grants SAF2015-70412-R from the Spanish Secretary of Research, Development and Innovation (MINECO), DJCLS R14/04 from Deutsche José Carreras Leukämie Stiftung, 2014 SGR225 (GRE) from Generalitat de Catalunya and economical support from Fundació Internacional Josep Carreras and from Obra Social “la Caixa” Spain to MS. Acknowledgments We thank Katinka Redert for her help in data collection. We thank Beatriz Cadenas from Josep Carreras Leukaemia Research Institute (IJC) and Whole Genix, S.L. for excellent technical and bioinformatic assistance, and Dr. Carme Pedro and Dr. Sara Montesdeoca from IMIM Hospital del Mar for medical assistance for P10. We also gratefully acknowledge donations to JC by the Crane and Perkins families as well as the participation of the research subjects. The research conducted at the Murdoch Children’s Research Institute was supported by the Victorian Government's Operational Infrastructure Support Program.

haematologica | 2018; 103(12)

YARS2 congenital sideroblastic anemia

References 1. Cartwright GE, Deiss A. Sideroblasts, siderocytes, and sideroblastic anemia. N Engl J Med. 1975;292(4):185-193. 2. Bottomley SS, Fleming MD. Sideroblastic anemia: diagnosis and management. Hematol Oncol Clin North Am. 2014;28(4):653-700. 3. Fleming MD. Congenital sideroblastic anemias: iron and heme lost in mitochondrial translation. Hematology Am Soc Hematol Educ Program. 2011;2011:525-531. 4. Riley LG, Cooper S, Hickey P, et al. Mutation of the mitochondrial tyrosyltRNA synthetase gene, YARS2, causes myopathy, lactic acidosis, and sideroblastic anemia--MLASA syndrome. Am J Hum Genet. 2010;87(1):52-59. 5. Riley LG, Menezes MJ, Rudinger-Thirion J, et al. Phenotypic variability and identification of novel YARS2 mutations in YARS2 mitochondrial myopathy, lactic acidosis and sideroblastic anaemia. Orphanet J Rare Dis. 2013;8:193. 6. Sasarman F, Nishimura T, Thiffault I, Shoubridge EA. A novel mutation in YARS2 causes myopathy with lactic acidosis and sideroblastic anemia. Hum Mut. 2012;33(8):1201-1206. 7. Nakajima J, Eminoglu TF, Vatansever G, et al. A novel homozygous YARS2 mutation causes severe myopathy, lactic acidosis, and sideroblastic anemia 2. J Hum Genet. 2014;59(4):229-232. 8. Ardissone A, Lamantea E, Quartararo J, et al. A Novel Homozygous YARS2 Mutation in Two Italian Siblings and a Review of Literature. JIMD Rep. 2015; 20:95-101. 9. Sommerville EW, Ng YS, Alston CL, et al. Clinical Features, Molecular Heterogeneity, and Prognostic Implications in YARS2Related Mitochondrial Myopathy. JAMA Neurol. 2017;74(6):686-694. 10. Bonnefond L, Fender A, Rudinger-Thirion J, GiegĂŠ R, Florentz C, Sissler M. Toward the full set of human mitochondrial aminoacyl-

haematologica | 2018; 103(12)

11.

12.

13. 14.

15.

16.

17.

18.

19.

tRNA synthetases: characterization of AspRS and TyrRS. Biochemistry. 2005;44(12):4805-4816. Jiang P, Xiaofen J, Peng Y, et al. The exome sequencing identified the mutation in YARS2 encoding the mitochondrial tyrosyltRNA synthetase as a nuclear modifier for the phenotypic manifestation of Leber's hereditary optic neuropathy-associated DNA mutation. Hum Mol Genet. 2016;25(3):584-596. Bergmann AK, Campagna DR, McLoughlin EM, et al. Systematic molecular genetic analysis of congenital sideroblastic anemia: evidence for genetic heterogeneity and identification of novel mutations. Pediatr Blood Cancer. 2010;54(2):273-278. Rusk N. Torrents of sequence. Nat Methods. 2011;8:44. Adzhubei I, Jordan DM, Sunyaev SR. Predicting functional effect of human missense mutations using PolyPhen-2. Curr Protoc Hum Genet. 2013;Chapter 7:Unit7. 20. Sim NL, Kumar P, Hu J, Henikoff S, Schneider G, Ng PC. SIFT web server: predicting effects of amino acid substitutions on proteins. Nucleic Acids Res. 2012;40(Web Server issue):W452-457. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016;536(7616):285-291. Zeharia A, Fischel-Ghodsian N, Casas K, et al. Mitochondrial myopathy, sideroblastic anemia, and lactic acidosis: an autosomal recessive syndrome in Persian Jews caused by a mutation in the PUS1 gene. J Child Neurol. 2005;20(5):449-452. Popp MW, Maquat LE. Leveraging rules of nonsense-mediated mRNA decay for genome engineering and personalized medicine. Cell. 2016;165(6):1319-1322. Bonnefond L, Frugier M, TouzĂŠ E, et al. Crystal structure of human mitochondrial tyrosyl-tRNA synthetase reveals common and idiosyncratic features. Structure.

2007;15(11):1505-1516. 20. Paukstelis P, Chari N, Lambowitz A, Hoffman D. NMR structure of the C-terminal domain of a tyrosyl-tRNA synthetase that functions in group I intron splicing. Biochemistry. 2011;50(18):38163826. 21. Oprescu SN, Griffin LB, Beg AA, Antonellis A. Predicting the pathogenicity of aminoacyl-tRNA synthetase mutations. Methods. 2017;113:139-151. 22. Shahni R, Wedatilake Y, Cleary MA, Lindley KJ, Sibson KR, Rahman S. A distinct mitochondrial myopathy, lactic acidosis and sideroblastic anemia (MLASA) phenotype associates with YARS2 mutations. Am J Med Genet Part A. 2013;161a(9):2334-2338. 23. Dallabona C, Diodato D, Kevelam SH, et al. Novel (ovario) leukodystrophy related to AARS2 mutations. Neurology. 2014; 82(23):2063-2071. 24. Pierce SB, Chisholm KM, Lynch ED, et al. Mutations in mitochondrial histidyl tRNA synthetase HARS2 cause ovarian dysgenesis and sensorineural hearing loss of Perrault syndrome. Proc Natl Acad Sci USA. 2011;108(16):6543-6548. 25. Solda G, Caccia S, Robusto M, et al. First independent replication of the involvement of LARS2 in Perrault syndrome by wholeexome sequencing of an Italian family. J Hum Genet. 2016;61(4):295-300. 26. Yoshida K, Sanada M, Shiraishi Y, et al. Frequent pathway mutations of splicing machinery in myelodysplasia. Nature. 2011;478(7367):64-69. 27. Bykhovskaya Y, Casas K, Mengesha E, Inbal A, Fischel-Ghodsian N. Missense mutation in pseudouridine synthase 1 (PUS1) causes mitochondrial myopathy and sideroblastic anemia (MLASA). Am J Hum Genet. 2004;74(6):1303-1308. 28. Yang J, Yan R, Roy A, Xu D, Poisson J, Zhang Y. The I-TASSER Suite: Protein structure and function prediction. Nat Methods. 2015;12(1):7-8.

2015

ARTICLE

Chronic Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2016-2025

BCR-ABL1 mediated miR-150 downregulation through MYC contributed to myeloid differentiation block and drug resistance in chronic myeloid leukemia

Klara Srutova,1 Nikola Curik,1,2 Pavel Burda,1,2 Filipp Savvulidi,2 Giovannino Silvestri,3 Rossana Trotta,4 Hana Klamova,1,5 Pavla Pecherkova,1 Zofie Sovova,1 Jitka Koblihova,1 Tomas Stopka,6 Danilo Perrotti3 and Katerina Machova Polakova1,5 SK and CN contributed equally to this work.

1 Institute of Hematology and Blood Transfusion, Prague, Czech Republic; 2Institute of Pathological Physiology, First Medical Faculty, Charles University, Prague, Czech Republic; 3Department of Medicine, Marlene and Stewart Greenebaum Comprehensive Cancer Center, University of Maryland Baltimore, MD, USA; 4Department of Microbiology and Immunology, Marlene and Stewart Greenebaum Comprehensive Cancer Center, University of Maryland Baltimore, MD, USA; 5Institute of Clinical and Experimental Hematology, First Medical Faculty, Charles University, Prague, Czech Republic and 6BIOCEV, First Medical Faculty, Charles University, Vestec, Czech Republic

ABSTRACT

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2016

he fusion oncoprotein BCR-ABL1 exhibits aberrant tyrosine kinase activity and it has been proposed that it deregulates signaling networks involving both transcription factors and non-coding microRNAs that result in chronic myeloid leukemia (CML). Previously, microRNA expression profiling showed deregulated expression of miR150 and miR-155 in CML. In this study, we placed these findings into the broader context of the MYC/miR-150/MYB/miR-155/PU.1 oncogenic network. We propose that up-regulated MYC and miR-155 in CD34+ leukemic stem and progenitor cells, in concert with BCR-ABL1, impair the molecular mechanisms of myeloid differentiation associated with low miR-150 and PU.1 levels. We revealed that MYC directly occupied the -11.7 kb and -0.35 kb regulatory regions in the MIR150 gene. MYC occupancy was markedly increased through BCR-ABL1 activity, causing inhibition of MIR150 gene expression in CML CD34+ and CD34â&#x20AC;&#x201C; cells. Furthermore, we found an association between reduced miR-150 levels in CML blast cells and their resistance to tyrosine kinase inhibitors (TKIs). Although TKIs successfully disrupted BCR-ABL1 kinase activity in proliferating CML cells, this treatment did not efficiently target quiescent leukemic stem cells. The study presents new evidence regarding the MYC/miR-150/MYB/miR-155/PU.1 leukemic network established by aberrant BCR-ABL1 activity. The key connecting nodes of this network may serve as potential druggable targets to overcome resistance of CML stem and progenitor cells.

ÂŠ2018 Ferrata Storti Foundation

Introduction

Correspondence: katerina.machova@uhkt.cz

Received: March 9, 2018. Accepted: July 19, 2018. Pre-published: July 26, 2018. doi:10.3324/haematol.2018.193086

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.

2016

Chronic myeloid leukemia (CML) is a malignant myeloproliferative disease originating from hematopoietic stem cells. The hallmark of CML is the presence of the BCR-ABL1 fusion gene due to the reciprocal translocation t(9;22)(q34;11). The constitutively active tyrosine kinase activity of the chimeric BCR-ABL1 protein causes deregulation and reprogramming of downstream signaling pathways, and drives the oncogenic process by altering cell proliferation, differentiation and survival. An understanding of CML pathogenesis consequently allowed a rational therapeutic strategy targeting BCR-ABL1 oncoprotein using tyrosine kinase inhibitors (TKIs) to be developed. The introduction of TKIs represented a breakthrough in CML therapy and achieved a large improvement in patient prognosis and outcome, and TKIs became the gold standard for first-line treatment.1 haematologica | 2018; 103(12)

MYC inhibits miR-150 expression in CML

Despite the high efficacy of TKIs, 20-30% of CML patients develop resistance during the chronic phase (CP). The frequency of TKI resistance significantly increases as the disease transforms from the CP to fatal blast crisis, which is initially a BCR-ABL1-dependent process;2 however, an established network further transforms the condition to BCR-ABL1 independence, resulting in a switch to a more aggressive acute leukemia-like disease.3 Although TKI treatment can successfully ablate the tumor cell population, it does not permanently cure CML because quiescent CML stem cells (LSCs) are often insensitive to TKIs.4,5 CML LSCs survive and are able to re-initiate the disease after the discontinuation of TKI treatment in some patients.6 The dysregulated epigenetic mechanisms previously described in CML involve microRNAs. We and others have shown that miR-150 levels are significantly reduced in CML.7-10 miR-150 is an inhibitor of the oncogenic transcription factor MYB, which regulates hematopoiesis at the early progenitor levels,11 while its inappropriate levels during later stages block cell differentiation.12,13 In a mouse model of CML blast crisis, c-MYB was shown to be required for BCR-ABL1-dependent leukemogenesis.14 We previously showed that miR-150 and MYB levels are inversely related, and these levels reciprocally respond to TKI treatment.10 CML in blast crisis shares certain features of acute leukemia. MYB is an upstream factor of acute myeloid leukemia (AML) aggressiveness that positively regulates miR-155. miR-155 inhibits the tumor suppressor and pro-differentiation factor PU.1.15,16 MYB expression is directly activated by the oncogenic transcription factor MYC in murine virus-induced myeloid leukemia tumor cells.17 MYC and its partner MAX directly bind the BCR promoter and up-regulate BCR-ABL1 expression.18 The functional connections among miR-150, MYC and BCR-ABL1 and the mechanism of the MYB/miR-155/PU.1 network, which is involved in acute leukemogenesis and affects its aggressiveness, led us to evaluate their relationship in CML and TKI resistance.

the University of Maryland, USA, and used for miRNA sequencing (Figure 2). The PBMC samples from CML patients in CP at the time of diagnosis (n=3) and additional PBMC samples of healthy donors (n=3) were handled according to the Ethics Committee of the UHKT. Samples were separated into CD34+ and CD34– cells, and used in the study of MYC binding to MIR150 regulatory regions by chromatin immunoprecipitation.

Leukemic cell lines The BCR-ABL1-positive CML cell lines K562, MEG-01 and KCL-22 and the BCR-ABL1-negative AML cell lines HL-60 and KG-1 were obtained from a publicly accessible biological resource center (Leibniz Institute - Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH/DSMZ, Braunschweig, Germany). The cell lines were handled and cultivated in appropriate medium according to the recommendations of the supplier. The K562R and KCL-22R cell lines resistant to imatinib were established by gradually exposing naive parental cells to increasing concentrations of imatinib in the medium (See Online Supplementary Appendix for details). The leukemic cell lines were used to measure gene expression and for functional experiments.

Statistical analysis Statistical analyses were performed using Student t-test in MS Excel (Microsoft Corporation, Redmond, WA, USA); *P<0.05, **P<0.01 and ***P<0.001. All graphs were generated using MATLAB version R2015b and GraphPad (GraphPad, La Jolla, CA, USA). Correlation coefficients calculated by non-parametric Spearman tests were used to determine the positive and negative correlations. Cloud-based MeV4 software (Multiple Experiment Viewer; http://mev.tm4.org/)20 was used to visualize the correlation data. The combinatory effect of miR-150 overexpression and imatinib treatment on MYB expression was assessed using the Highest Single Agent approach (CIB-HSA).21 Details regarding patients and primary cell samples, cell sorting and separation, cell lines, transient transfections, siRNAs design, nucleic acid isolation, RT-qPCR assays, protein isolation and immunoblotting, miRNA RNA sequencing, cell cycle and viability analyses are described in detail in the Online Supplementary Appendix.

Methods Patients’ samples Chronic myeloid leukemia patients were diagnosed and treated at the Institute of Hematology and Blood Transfusion in Prague (UHKT), Czech Republic, and the Marlene and Stuart Greenebaum Comprehensive Cancer Center at the University of Maryland, USA. The bone marrow (BM) samples (n=46) from the CML patients in CP (n=41) and peripheral blood mononuclear cells (PBMCs) samples (n=10) from healthy volunteers were obtained with written informed consent according to the principles of the Declaration of Helsinki and approval by the UHKT Ethics Committee. The samples were collected at time of diagnosis (n=28) and at time of TKI resistance (n=18). The therapeutic response was scored according to the European LeukemiaNet recommendations.19 The response to first-line treatment was assessed after 12 months of therapy (Online Supplementary Table S1). Patients' BM samples were used for FACS sorting, and subsequently to evaluate gene expression (Figure 1A-F and Online Supplementary Figure S1A and B) and the correlations among them (Online Supplementary Figure S1C). Additional BM samples (n=6) from CML patients in CP (n=3) or blast crisis (n=3), and the BM samples from healthy donors (n=3), were obtained and handled according to the Ethics Committee of haematologica | 2018; 103(12)

Results A putative BCR-ABL1/MYC/miR-150/MYB/miR-155 regulatory pathway is activated in chronic myeloid leukemia To outline relationships among the studied molecules, gene expression levels were evaluated in primary cells sorted according to CD34 expression (see Online Supplementary Appendix) from the BM of CML patients at the time of diagnosis (n=28) and at the time of resistance to TKIs (n=18) (Figure 1). We observed significantly lower miR-150 levels and increased MYC expression in CD34+ CML cells (P<0.0001 for miR-150 and MYC) and CD34– CML cells (P=0.0001 for miR-150; P=0.0013 for MYC) at the time of diagnosis compared with those in the CD34+ and CD34– cells of healthy donors (n=10), respectively. MYB expression was significantly down-regulated in CD34+ CML cells (P<0.0001) and significantly up-regulated in CD34– CML cells (P<0.0001). miR-155 levels were significantly up-regulated in the CD34+ CML cells at the time of diagnosis (P=0.0002), while the expression of the pro-differentiation transcription factor PU.1, which is the 2017

K. Srtutova et al. validated target of miR-155 in B cells,15 was significantly down-regulated (P<0.0001) (Figure 1). These data highlight the notable similarity between the expression profiles of CML-CP CD34+ and CD34– cells at the time of diagnosis and TKI resistance, respectively. To obtain detailed information regarding the expression of studied molecules in CML cells, CD34+ and CD34– leukemic cells were further sorted into five distinct subpopulations according to CD38 expression: CD34+CD38–; CD34+CD38+; CD34–CD38+; CD34–CD38+low; and CD34–CD38–. A marked gradual increase in miR-150 levels and a decrease in its target MYB11 were found both among the sorted subpopulations at the time of diagnosis and at the time of TKI resistance (Online Supplementary Figure S1A). The increased levels of miR-150 and decreased levels of MYB coincided with a higher degree of cell differentiation (Online Supplementary Figure S1B). Negative correlations between the levels of miR-150 and the levels of oncogenes (BCR-ABL1, MYC and MYB) were observed across all five sorted leukemic BM cell subpopulations (Online Supplementary Figure S1C).

To investigate the potential roles of miR-150 and miR155 in different disease phases, we also measured miR150 and miR-155 expression in the CD34+CD38– and CD34+CD38+ subpopulations from BM samples of CML patients in CP (n=3) and in blast crisis (n=3). The levels of miR-150 and miR-155 in the corresponding cell populations in the CP were not significantly different from those in blast crisis (Figure 2).

BCR-ABL1 deregulated miR-150 and MYB in CML cells The expression of the studied molecules was further evaluated in CML (K562, KCL-22 and MEG-01) and AML (KG-1 and HL-60) cell lines to address whether in vitro models could be used to study functional relationships between the molecules (Online Supplementary Figure S2). In accordance with data from primary CML samples, miR150 levels were significantly decreased in CD34– CML cell lines compared with the levels in the pool of CD34– cells from healthy donors (n=10) (P<0.001), while the miR-150 levels were 5-fold lower in the CML than those in the BCR-ABL1-negative AML cells (P<0.01), implying that

Figure 1. Expression of genes involved in a putative oncogenic pathway in primary chronic myeloid leukemia (CML) and healthy cells. The expression of MYC (A), miR-150 (B), MYB (C), miR-155 (D), PU.1 (E) and BCR-ABL1 (F) in CD34+ (left) and CD34– (right) subpopulations from CML-chronic phase (CP) patients at diagnosis (n=28) or at the time of resistance to tyrosine kinase inhibitors (TKIs) (n=18) compared to that in the CD34+ or CD34– subpopulations of healthy donors (n=10). Gene expression in each specific sample is indicated by dots; the boxes represent box-and-whisker plots. Red: bone marrow (BM) CML-CP samples; black: healthy peripheral blood mononuclear cells (PBMC); blue lines: median expression values. Significant differences (P-values) in the gene expression between CML and healthy CD34+ and CD34– subpopulations are illustrated in the tables below each box-and-whisker plot. Unpaired two-tailed Student t-test was used to determine P-values.

2018

haematologica | 2018; 103(12)

MYC inhibits miR-150 expression in CML

miR-150 expression is likely regulated by BCR-ABL1 activity (Online Supplementary Figure S2). We tested this hypothesis by overexpressing miR-150 (Online Supplementary Figure S3) and inhibiting BCR-ABL1 in CML cells. Imatinib (the most frequently used TKI in CML treatment) and/or BCR-ABL1-specific siRNA were used to determine the effect of the BCR-ABL1 activity/expression inhibition. First, miR-150 overexpression significantly decreased MYB mRNA and protein expression levels in both the BCR-ABL1-positive (K562) and BCR-ABL1-negative (HL60) cells. MYB protein levels were decreased in KCL-22 cells (Figure 3A-C); the expression of c-MYB (75 KDa) was inhibited in both CML cell lines, while the expression of the spliced 37-KDa MYB variant was affected in HL-60 cells (the 75-KDa MYB variant was not expressed in these cells). MYB expression was down-regulated after the inhibition of BCR-ABL1 activity and expression in the CML cells (Figure 3B-D). Furthermore, simultaneously restoring miR-150 and inhibiting BCR-ABL1 had a positive combination effect on the inhibition of MYB mRNA levels (Highest Single Agent approach value: CI_HSA 0.9654) in

the K562 cells and protein levels in the K562 and KCL-22 cells (Figure 3B and C). There were no significant differences between the effects of imatinib alone or its combination with a miR-150 mimic with regard to the inhibition of MYB expression. Interestingly, miR-155 levels were significantly increased in both K562 and KCL-22 cells after the suppression of BCR-ABL1 activity by imatinib (Figure 3E). The inhibition of BCR-ABL1 activity also significantly decreased PU.1 in K562 cells but significantly increased PU.1 in KCL-22 cells (Figure 3F). This finding is of interest, as K562 cells represent leukemic erythroblasts, whose proper differentiation requires low PU.1 levels, while KCL-22 cells represent leukemic myeloblasts, whose differentiation requires high PU.1 levels. Next, we found that suppressing BCR-ABL1 activity by imatinib significantly decreased MYC mRNA expression in K562 and KCL-22 cells, while imatinib had no impact on MYC expression in the BCR-ABL1-negative HL-60 cells (Figure 3G). We investigated the possible inhibitory role of BCR-ABL1 and its putative downstream target MYC in miR-150 expression and maturation. The levels of the miR150 precursors and mature miR-150 consistently increased

P=0.16

P=0.22

P=0.14

P=0.45

Figure 2. Comparison of miR-150 and miR-155 expression in leukemic stem/progenitor cells (LSPCs) in blast crisis and chronic phase (CP) chronic myeloid leukemia (CML). Pie charts illustrate the proportions of read count median values for miR-150 (left column) and miR-155 (right column) expression obtained by miRNA RNA sequencing. The read counts that overlapped with miR-150 and miR-155 after alignment with human genome GRCh38 were calculated using miRDeep quantifier module and normalized to counts per million (CPM). Primary bone marrow (BM) cell populations used for the measurement and comparison are indicated for each pair of charts. P-values in the miR expression between the compared cell populations are illustrated below each chart. Paired two-tailed Student t-test was used for determining of P-values. BC: blast crisis CML (n=3); CP: chronic phase CML (n=3).

haematologica | 2018; 103(12)

2019

K. Srtutova et al.

upon the BCR-ABL1 inhibition induced by imatinib in the K562 cells. Thus, BCR-ABL1 inhibits miR-150 transcription without affecting miR-150 maturation (Figure 3H). The inhibition of BCR-ABL1 using imatinib significantly increased the G0/G1 phase and decreased the S phase in K562 cells, which was accompanied by an increase in apoptosis (Online Supplementary Figure S4). Sensitivity to imatinib differed significantly between cell lines, as 50% of the K562 cells but only 5% of the KCL-22 cells underwent apoptosis 96 hours (h) after imatinib treatment

2020

(Online Supplementary Figure S4B and D). This difference in imatinib sensitivity had been previously discovered during the development of the resistant cell lines (see Online Supplementary Appendix) and is consistent with the ability of KCL-22 cells to continuously grow in the presence of 1 mM imatinib and relapse early due to the acquisition of resistant BCR-ABL1 mutations or to BCR-ABL1-independent mechanisms.22 Restoration of miR-150 had no impact on the cell cycle or viability in the K562 cells within 96 h post transfection (Online Supplementary Figure S4).

Figure 3. The effects of miR-150 overexpression and BCR-ABL1 silencing on the expression of genes in oncogenic pathways in leukemic cells. (A) MYB transcript and MYB protein expression in HL-60 cells 48 hours (h) after the miR-150 transfection. The numbers shown at the top indicate the MYB protein expression relative to that in the untreated samples (β-actin normalized). Expression of MYB transcripts (48 h) and c-MYB protein (96 h) in (B) KCL-22 and (C) K562 after miR-150 transfection and/or BCR-ABL1 activity inhibition with 1 mM imatinib. (D) MYB expression in K562 cells 48 h miR-150 transfection and/or the inhibition of BCR-ABL1 expression by siRNA. (E) miR-155 levels in KCL-22 and K562 cells 96 h after 30 nM miR-150 transfection and/or BCR-ABL1 activity inhibition with 1 mM imatinib. (F) PU.1 expression in KCL-22 and K562 cells 96 h after miR-150 transfection and/or BCR-ABL1 activity inhibition with 1 mM imatinib. (G) MYC expression in KCL22, K562 and HL-60 cells after exposure to 1 mM imatinib. (H) Pri-miR150, pre-miR150 and miR-150 levels in K562 cells upon BCR-ABL1 inhibition. Generally, cells were transfected with 30 nM miR-150 mimic or 50 nM siRNA BCR-ABL1. The expression data represent the expression fold change (FC; 2-ΔΔCt) in relation to untreated, control (Ctrl) sample normalized to 1. Unpaired two-tailed Student t-test was used to determine P-values. *P<0.05, **P<0.01, and ***P<0.001. Error bars represent standard deviations.

haematologica | 2018; 103(12)

MYC inhibits miR-150 expression in CML

MYC repressed expression of miR-150 in CML cells Based on the inverse trend of MYC expression and miR150 levels (Online Supplementary Figure S1C), we hypothesized that MYC is directly involved in the repression of MIR150 gene expression in CML. Publicly available ChIPSeq data from the UCSC Genome Browser (http://genome.ucsc.edu/index.html) were used to predict the protein binding regions surrounding the MIR150 gene in K562 cells. We identified loci with a high probability of binding to MIR150 gene regulatory factors such as MYC and its partner MAX (Online Supplementary Figure S5). We investigated whether MYC binds these regions in naive and imatinib-treated K562, KCL-22 and HL-60 cells using chromatin immunoprecipitation (ChIP). We observed the MYC occupancy at the loci -11.7 kb and -0.35 kb upstream of the MIR150 gene TSS in untreated K562 and at the loci -11.7 kb in untreated KCL-22 cells, but MYC was not detected at these and other tested loci in the HL-60 cells. The 48-h treatment with imatinib significantly decreased the MYC occupancy in K562 cells and, to a lesser degree, in KCL-22 cells (Figure 4A and B), but the treatment did not impact MYC non-occupancy in HL-60 cells (Figure 4C). To determine the effect of profound MYC suppression on miR-150 expression, K562 and KCL-22 cells were treated with bromodomain inhibitor JQ1.23 JQ1 treatment significantly, and to a similar extent, decreased MYC mRNA expression in K562 and KCL-22 cells (Online Supplementary Figure S6A), without affecting BCR-ABL1 expression (Online Supplementary Figure S6B), and increased miR-150 levels in KCL-22 cells but not in K562 cells (Online Supplementary Figure S6C). Next, we investigated MYC binding to MIR150 gene regulatory regions in sorted CD34+ and CD34– CML primary cells compared with that in healthy donors. We found marked MYC binding to the -11.7 kb and -0.35 kb loci in CD34+ cells isolated from CML-CP patients (n=3) at the time of diagnosis. In contrast, MYC occupancy at these loci was weak in CD34+ cells from healthy donors (n=3) (Figure 5A). Decreased MYC occupancy to MIR150 gene regulatory regions was detected in the leukemic CD34– subpopulation, while MYC binding in the tested loci was undetectable in CD34– cells of healthy donors (Figure 5B). These data highlight the role of MYC in regulating the MIR150 gene in CML. To investigate a potential regulatory loop between the miR-150 target MYB and different molecules, we investigated the effect of silencing MYB expression using siRNA in K562 and KCL-22 cells. No significant effects on BCRABL1, MYC, miR-150, miR-155 and PU.1 levels were observed following near complete MYB inhibition (Online Supplementary Figure S7), suggesting that MYB is not involved in the direct regulation of these molecules in CML.

counterparts. PU.1 levels were significantly decreased in KCL-22R cells (P<0.001) and significantly increased in K562R cells (P<0.01) compared with the levels in their sensitive counterparts (Figure 6). These data suggest that further PU.1 deregulation may be associated with the blockade of erythroid (K562) and myeloid (KCL-22) cell differentiation, respectively, in resistant CML.

Model of BCR-ABL1/MYC/miR-150/MYB/miR155/PU.1 regulatory links in leukemic primary cells We assessed the BCR-ABL1/MYC/miR-150/MYB/miR155/PU.1 network of CD34+ and CD34– CML-CP cells at

Resistant chronic myeloid leukemia cells further down-regulate miR-150 To address whether the studied leukemic network is involved in TKI resistance in CML, we analyzed gene expression in resistant CML cells using two established distinct models of resistance (see Online Supplementary Methods). BCR-ABL1 was strongly over-expressed in all CML cell cultures. Similar BCR-ABL1 transcript levels were observed between the resistant and sensitive parental cell lines (Figure 6). miR-150 and miR-155 levels were significantly lower (P<0.001) in both resistant CML cell lines compared with the levels in their respective imatinib-sensitive haematologica | 2018; 103(12)

Figure 4. miR-150 regulation by MYC in leukemic blasts. MYC occupancy at putative regulatory loci of the MIR150 gene in (A) K562, (B) KCL-22 and (C) HL60 cells with or without imatinib treatment. Columns represent the fold change (FC) in the % of DNA input obtained in the control (gray) or imatinib-treated (dark) cells compared with the non-specific IgG precipitation and equalized to 1 (dashed line). The promoter region of the miR-15a/16-1 cluster was used as the positive control of MYC DNA binding. Unpaired two-tailed Student t-test was used to determine P-values. *P<0.05, **P<0.01. Error bars represent standard deviations.

2021

K. Srtutova et al.

the time of diagnosis (Figure 7). The proposed scheme combines the data obtained from expression analyses, the data obtained from functional experiments and chromatin immunoprecipitation assays performed in cell cultures and primary cells to present a simplified model of the regulatory interactions in CML hematopoiesis. Based on our data and that from the literature, we propose a model illustrating relatively lower levels of MYC and higher levels of miR-150 in normal CD34+ and CD34– cells (Figure 7A and B) compared with those observed in corresponding leukemic cells. As the binding of MYC to the regulatory regions of MIR150 is weak in normal CD34+ cells (i.e. slightly above background) (Figure 5), the model assumes no inhibition of miR-150 expression by MYC in CD34+ cells to ensure the physiological levels of miR-150 required for the early stages of hematopoiesis. MYC transcription has previously been shown to be repressed by PU.1.24 MYB levels are high in the normal CD34+ cells, suggesting that MYB is required during the early stages of hematopoiesis;11,12 however, MYB levels are subsequently sharply reduced by abundant miR-150 and, supposedly, transcriptional repression by PU.125 in CD34– cells (Figure 7B). The miR-155 levels were lower in healthy CD34+ cells (compared with those in the corresponding leukemic subpopulation), allowing for the proper expression of its targets MYB and PU.1 in healthy hematopoietic stem and progenitor cells. Indeed, PU.1 expression in the normal cells was further increased in the more mature subpopulation, which is consistent with its requirement for proper myeloid lineage development and differentiation.26 In summary, the model proposes the existence of few 'core' interactions between the studied molecules ensuring the strict regulation of MYC and MYB expression during normal hematopoietic cell differentiation. In contrast, the suggested model of interactions among the studied molecules in CD34+ and the more differentiated CD34– CML cells (Figure 7C and D) is characterized by the establishment of several aberrant connections caused by BCR-ABL1 activity, resulting in the consistently increased expression of MYC and MYC occupancy in the

MIR150 regulatory regions in CML. As a consequence, the expression levels of miR-150 and PU.1 were lower than those observed in the corresponding cell populations from healthy controls. Due to the repressed expression of miR150, we would expect MYB overexpression in CML CD34+ cells compared with the expression in normal CD34+ cells, which was not demonstrated. Decreased levels of MYB may be explained by the increased levels of miR-155 targeting MYB in leukemic CD34+ cells.27 Moreover, miR-155 overexpression in CML CD34+ cells suggests the presence of inhibitory effect on the expression of its target PU.1.15 The significantly higher expression of MYC in concert with the BCR-ABL1 activity resulted in downregulation of miR-150 and, consequently, in higher MYB expression levels in CD34– cells compared with the levels in cells from healthy donors. No marked differences in miR-155 levels were observed between the leukemic and healthy cells in the CD34– subpopulation, and the role of miR-155 is uncertain in these more differentiated cells. The trend for increased PU.1 levels and decreased MYB and BCR-ABL1 levels was found in leukemic CD34– compared with CD34+ cells, corresponding to the fact that the myeloid lineage development and differentiation in CML-CP is not fully arrested (Online Supplementary Figure S1A and B).

Discussion This study provides new evidence to show that BCRABL1 establishes the MYC/miR-150/MYB/miR-155/PU.1 leukemic pathway in CML pathogenesis. First, we determined the expression levels of these molecules in sorted cell populations from CML-CP patients. Consistent with previous reports7,8 and our prior data,10 we observed that miR-150 levels in primary CML samples were lower than those in the corresponding cells of healthy donors. Next, we investigated the mechanism of miR-150 suppression in CML. Previously, the repression of miR-150 expression by MYC had been demonstrated in an immor-

Figure 5. miR-150 regulation by MYC in chronic myeloid leukemia (CML) primary cells. MYC occupancy at putative regulatory loci of the MIR150 gene in separated (A) CD34+ and (B) CD34– cell subpopulations isolated from healthy donors [control (Ctrl) n=3] and CML-chronic phase (CP) patients at the time of diagnosis (CML; n=3), respectively. Separated cells from 2 healthy donors were pooled due to very low yields. The columns represent the fold change (FC) in the % of DNA input obtained in the leukemic or healthy cells compared with the non-specific IgG precipitation and equalized to 1 (dashed line). Unpaired two-tailed Student t-test was used to determine P-values. *P<0.05, **P<0.01. Error bars represent standard deviations.

2022

haematologica | 2018; 103(12)

MYC inhibits miR-150 expression in CML

talized human B-cell model; however, the tested mechanism of direct MYC binding to the regulatory loci of the MIR150 gene was not confirmed.28 In the present study, we identified previously untested loci -11.7 kb and -0.35 kb upstream of the MIR150 TSS that were occupied by MYC in CD34+ and CD34– CML-CP primary cells. The occupancy of MYC at both loci was found in K562 and, in the case of the -11.7 kb locus, in KCL-22 cells but not in BCR-ABL1-negative HL-60 cells. Inhibiting BCR-ABL1 activity with imatinib decreased MYC expression and depleted MYC occupancy at these specific loci in the K562

and KCL-22 cells. In contrast, imatinib treatment had no impact on either MYC expression or its occupancy on the studied MIR150 loci in the HL-60 cells. A pronounced decrease in MYC expression was observed without an impact on BCR-ABL1 activity using JQ1, which significantly increased miR-150 levels in KCL-22 but not K562 cells. This finding may be explained by the assumption that a profound inhibition of both MYC and BCR-ABL1 is required to induce miR-150 expression in K562 cells. Moreover, we tested dual MYC regulatory mechanisms of MIR150 expression, including direct transcriptional acti-

Figure 6. Comparison of the oncogenic pathway expression profiles in imatinib-sensitive (naive) K562, KCL-22 and imatinib-resistant K562R, KCL-22R chronic myeloid leukemia (CML) cell lines. BCR-ABL1, MYB, MYC, PU.1, miR-150 and miR-155 transcript levels in (A) KCL-22 and KCL-22R and in (B) K562 and K562R. Gene expression data for resistant cells represent the fold change (FC; 2-ΔΔCt) in relation to corresponding gene expression for naive cells normalized to 1. Unpaired twotailed Student t-test was used to determine P-values. *P<0.05, ** P<0.01, and ***P<0.001. Error bars represent standard deviations.

haematologica | 2018; 103(12)

Figure 7. The suggested model of relationships among BCR-ABL1, MYC, MYB, PU.1, miR-150 and miR155 in CD34+ and CD34– cells from chronic myeloid leukemia (CML) patients and healthy donors. The schemes represent the expression levels and mutual relationships among the molecules in CD34+ (A) and CD34– (B) cells from healthy donors compared with the profiles of CD34+ (C) or CD34– (D) cells, respectively, from CML patients at diagnosis. The circle sizes denote the expression levels of specific molecules. The black lines represent the active stimulatory (arrowheads) or inhibitory (flat-heads) relationships described in the manuscript. The red line emphasizes the novelty mechanism of miR-150 repression by MYC in CML CD34+ cells. By the dashed lines we outlined regulations that we assumed based on published observations.15,17,18,24,25,27

2023

K. Srtutova et al.

vation by MYC and the post-transcriptional inhibition of miR-150 maturation by MYC-driven Lin28 described using in vitro and ex vivo AML models with MLL rearrangements.29 However, our data did not reveal the repression of an miR150 maturation blockade following the BCR-ABL1 inhibition in CML cells. The specific regulatory mechanisms of miR-150 expression were identified in CML, as we identified MYC-binding loci neighboring the MIR150 TSS in K562 and KCL-22 cells, while there was no MYC binding in AML HL-60 cells. Recently, MYC was reported to exert opposite effects on the expression of a specific gene depending on the local epigenetic pattern;30 this would be interesting to investigate for the MIR150 gene given the presence of a CpG island near the -11.7 kb locus. Together our data suggest that BCR-ABL1 inhibits miR150 expression in CML cells via the transcriptional activation of MYC and its simultaneous recruitment to a specific -11.7 and -0.35 kb loci of the MIR150 gene, where MYC binds and acts as a direct repressor of miR-150 transcription (Figure 7). This conclusion is reliable since MYC has recently been found to act with p53 as a key CML regulator to maintain CML LSC survival.31 Furthermore, our data indicated that low miR-150 levels drive MYB expression in CD34– CML cells and hinder cell differentiation,9 which is consistent with other reports32,33 and our finding regarding low PU.1 expression in the primary CML–CP CD34+ populations and CD34– CML blast cell lines. Our data showed a significant downregulation of PU.1 following imatinib treatment in K562 cells (a model of erythroid lineage), while the opposite effect on PU.1 expression was observed following imatinib treatment in KCL-22 cells (a model of myeloid lineage), likely reflecting the distinct nature of these cell lines. We hypothesize that BCR-ABL1 inhibition generally relieves blocked cell differentiation by manipulating PU.1 levels because either low or high levels of PU.1 are required for the terminal differentiation of K562 (erythroblasts) and KCL-22 (myeloblasts) cells, respectively. Conversely, both PU.1 upregulation in the K562R imatinib-resistant cells and PU.1 downregulation in the KCL-22R imatinib-resistant cells, together with reduced miR-150 levels, may impose a differentiation block, similar to that described in a murine erythroleukemia (MEL) cell model.34 In addition, we found that miR-155 levels in the CD34+ CML-CP cells from patients at diagnosis and with TKI

References 1. Chereda B, Melo JV. Natural course and biology of CML. Ann Hematol. 2015;94(2S):107-121. 2. Pellicano F, Scott MT, Helgason GV, et al. The antiproliferative activity of kinase inhibitors in chronic myeloid leukemia cells is mediated by FOXO transcription factors. Stem Cells. 2014;32(9):2324-2337. 3. Wagle M, Eiring AM, Wongchenko M, et al. A role for FOXO1 in BCR-ABL1-independent tyrosine kinase inhibitor resistance in chronic myeloid leukemia. Leukemia. 2016;30(7):1493-1501. 4. Graham SM, Jorgensen HG, Allan E, et al. Primitive, quiescent, Philadelphia-positive stem cells from patients with chronic myeloid leukemia are insensitive to STI571

2024

resistance were higher than those in healthy CD34+ cells. miR-155 was previously identified to be an oncogenic miRNA that is up-regulated in a variety of malignancies.3537 Notably, the sustained miR-155 overexpression in the CML was associated with the induction of myeloid disorder in mice.38 However, the inhibition of BCR-ABL1 activity by imatinib increased miR-155 levels in the KCL-22 and K562 cells, which is consistent with a previous report showing miR-155 upregulation in leukocytes from CML patients following prolonged imatinib treatment.39 The functionally distinct, dose-dependent effects of miR-155 expression have been recently described in AML, highlighting the importance of the cell context and fine regulation in assessing its role in leukemogenesis.40 In summary, we propose a leukemic network model including a novel mechanism of BCR-ABL1-dependent recruitment of MYC oncoprotein to bind and inhibit MIR150 gene expression in CML cells. Compared with healthy cells, the CD34– leukemic cells showed that downregulation of miR-150 expression by MYC resulted in significantly higher MYB levels. miR-150 expression is reduced at the time of TKI resistance and further diminished in resistant CML cell lines, emphasizing the increased aggressiveness of the disease as well as the links between TKI resistance and disrupted cell differentiation. The key connecting nodes of the described leukemic network established by aberrant activity of BCR-ABL1 may serve as potential druggable targets, as was recently shown for the transcription factor MYC31 to overcome the resistance of CML LSCs. Funding This study was supported by the Ministry of Education Youth and Sports grant project MSMT LH15104, Charles University grant GAUK 178215, GACR 18-18407S, Czech Ministry of Health project no. 00023736 for the conceptual development of a research institute, Charles University Research Centers grant UNCE/MED/016 and program Progres Q26 and NCI-NIH grant CA163800. Acknowledgments We thank Prof. Jianjun Chen, Section of Hematology/ Oncology, Department of Medicine, University of Chicago, Chicago, IL, USA for providing the primer sequences for pri-miR-150 and pre-miR-150 transcript quantification.

in vitro. Blood. 2002;99(1):319-325. 5. Bhatia R, Holtz M, Niu N, et al. Persistence of malignant hematopoietic progenitors in chronic myelogenous leukemia patients in complete cytogenetic remission following imatinib mesylate treatment. Blood. 2003;101(12):4701-4707. 6. Corbin AS, Agarwal A, Loriaux M, Cortes J, Deininger MW, Druker BJ. Human chronic myeloid leukemia stem cells are insensitive to imatinib despite inhibition of BCRABL activity. J Clin Invest. 2011;121(1):396409. 7. Agirre X, Jimenez-Velasco A, San JoseEneriz E, et al. Down-regulation of hsamiR-10a in chronic myeloid leukemia CD34+ cells increases USF2-mediated cell growth. Mol Cancer Res. 2008;6(12):18301840. 8. Flamant S, Ritchie W, Guilhot J, et al.

10.

11.

12.

Micro-RNA response to imatinib mesylate in patients with chronic myeloid leukemia. Haematologica. 2010;95(8):1325-1333. Morris VA, Zhang A, Yang T, et al. MicroRNA-150 expression induces myeloid differentiation of human acute leukemia cells and normal hematopoietic progenitors. PLoS One. 2013;8(9):e75815. Machová Poláková K, Lopotová T, Klamová H, et al. Expression patterns of microRNAs associated with CML phases and their disease related targets. Mol Cancer. 2011;10:41. Lin YC, Kuo MW, Yu J, et al. c-Myb is an evolutionary conserved miR-150 target and miR-150/c-Myb interaction is important for embryonic development. Mol Biol Evol. 2008;25(10):2189-2198. Xiao C, Calado DP, Galler G, et al. MiR-150 controls B cell differentiation by targeting

haematologica | 2018; 103(12)

MYC inhibits miR-150 expression in CML

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

the transcription factor c-Myb. Cell. 2007; 131(1):146-159. Gonda TJ, Metcalf D. Expression of myb, myc and fos proto-oncogenes during the differentiation of a murine myeloid leukaemia. Nature. 1984;310(5974):249251. Lidonnici MR, Corradini F, Waldron T, Bender TP, Calabretta B. Requirement of cMyb for p210(BCR/ABL)-dependent transformation of hematopoietic progenitors and leukemogenesis. Blood. 2008; 111(9):4771-4779. Vigorito E, Perks KL, Abreu-Goodger C, et al. microRNA-155 regulates the generation of immunoglobulin class-switched plasma cells. Immunity. 2007;27(6):847-859. Gerloff D, Grundler R, Wurm AA, et al. NFkappaB/STAT5/miR-155 network targets PU.1 in FLT3-ITD-driven acute myeloid leukemia. Leukemia. 2015;29(3):535-547. Wolff L, Schmidt M, Koller R, et al. Three genes with different functions in transformation are regulated by c-Myb in myeloid cells. Blood Cells Mol Dis. 2001;27(2):483488. Sharma N, Magistroni V, Piazza R, et al. BCR/ABL1 and BCR are under the transcriptional control of the MYC oncogene. Mol Cancer. 2015;14:132. Baccarani M, Deininger MW, Rosti G, et al. European LeukemiaNet recommendations for the management of chronic myeloid leukemia: 2013. Blood. 2013;122(6):872884. Saeed AI, Sharov V, White J, et al. TM4: a free, open-source system for microarray data management and analysis. Biotechniques. 2003;34(2):374-378. Lehรกr J, Zimmermann GR, Krueger AS, et al. Chemical combination effects predict connectivity in biological systems. Mol Syst Biol. 2007;3:80. Yuan H, Wang Z, Gao C, et al. BCR-ABL

haematologica | 2018; 103(12)

23.

24.

25.

26.

27.

28.

29.

30.

31.

gene expression is required for its mutations in a novel KCL-22 cell culture model for acquired resistance of chronic myelogenous leukemia. J Biol Chem. 2010; 285(7):5085-5096. Delmore JE, Issa GC, Lemieux ME, et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell. 2011;146(6):904-917. Kihara-Negishi F, Yamamoto H, Suzuki M, et al. In vivo complex formation of PU.1 with HDAC1 associated with PU.1-mediated transcriptional repression. Oncogene. 2001;20(42):6039-6047. Bellon T, Perrotti D, Calabretta B. Granulocytic differentiation of normal hematopoietic precursor cells induced by transcription factor PU.1 correlates with negative regulation of the c-myb promoter. Blood. 1997;90(5):1828-1839. Dahl R, Simon MC. The importance of PU.1 concentration in hematopoietic lineage commitment and maturation. Blood Cells Mol Dis. 2003;31(2):229-233. Hornick NI, Doron B, Abdelhamed S, et al. AML suppresses hematopoiesis by releasing exosomes that contain microRNAs targeting c-MYB. Sci Signal. 2016;9(444):ra88. Chang TC, Yu D, Lee YS, et al. Widespread microRNA repression by Myc contributes to tumorigenesis. Nat Genet. 2008; 40(1):43-50. Jiang X, Huang H, Li Z, et al. Blockade of miR-150 maturation by MLLfusion/MYC/LIN-28 is required for MLLassociated leukemia. Cancer Cell. 2012;22(4):524-535. Adams CM, Hiebert SW, Eischen CM. Myc Induces miRNA-Mediated Apoptosis in Response to HDAC Inhibition in Hematologic Malignancies. Cancer Res. 2016;76(3):736-748. Abraham SA, Hopcroft LE, Carrick E, et al. Dual targeting of p53 and c-MYC selective-

32.

33.

34.

35.

36.

37.

38.

39.

40.

ly eliminates leukaemic stem cells. Nature. 2016;534(7607):341-346. Albajar M, Gutierrez P, Richard C, et al. PU.1 expression is restored upon treatment of chronic myeloid leukemia patients. Cancer Lett. 2008;270(2):328-336. Yang H, Liang H, Yan JS, Tao R, Hao SG, Ma LY. Down-regulation of hematopoiesis master regulator PU.1 via aberrant methylation in chronic myeloid leukemia. Int J Hematol. 2012;96(1):65-73. Stopka T, Amanatullah DF, Papetti M, Skoultchi AI. PU.1 inhibits the erythroid program by binding to GATA-1 on DNA and creating a repressive chromatin structure. EMBO J. 2005;24(21):3712-3723. Kluiver J, Poppema S, de Jong D, et al. BIC and miR-155 are highly expressed in Hodgkin, primary mediastinal and diffuse large B cell lymphomas. J Pathol. 2005; 207(2):243-249. Calin GA, Croce CM. Chronic lymphocytic leukemia: interplay between noncoding RNAs and protein-coding genes. Blood. 2009;114(23):4761-4770. Volinia S, Calin GA, Liu CG, et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proc Natl Acad Sci USA. 2006;103(7):2257-2261. O'Connell RM, Rao DS, Chaudhuri AA, et al. Sustained expression of microRNA-155 in hematopoietic stem cells causes a myeloproliferative disorder. J Exp Med. 2008; 205(3):585-594. Rokah OH, Granot G, Ovcharenko A, et al. Downregulation of miR-31, miR-155, and miR-564 in chronic myeloid leukemia cells. PLoS One. 2012;7(4):e35501. Narayan N, Morenos L, Phipson B, et al. Functionally distinct roles for different miR-155 expression levels through contrasting effects on gene expression, in acute myeloid leukaemia. Leukemia. 2017; 31(4):808-820.

2025

ARTICLE

Chronic Myeloid Leukemia

Ferrata Storti Foundation

BCR-ABL1 genomic DNA PCR response kinetics during first-line imatinib treatment of chronic myeloid leukemia Ilaria S. Pagani,1,2 Phuong Dang,1 Ivar O. Kommers,3 Jarrad M. Goyne,1 Mario Nicola,4 Verity A. Saunders,1 Jodi Braley,4 Deborah L. White,1,2,5,6,7 David T. Yeung,1,2,8,9 Susan Branford,2,4,6,10 Timothy P. Hughes,1,2,8,9,10 and David M. Ross1,2,8,9,10,11

Haematologica 2018 Volume 103(12):2026-2032

Cancer Theme, South Australian Health & Medical Research Institute, Adelaide, Australia; 2School of Medicine, Faculty of Health Sciences, University of Adelaide, Australia; 3VU University Medical Center, Amsterdam, the Netherlands; 4Genetic and Molecular Pathology, SA Pathology, Adelaide, Australia; 5School of Biological Sciences, Faculty of Sciences, University of Adelaide, Australia; 6School of Paediatrics, Faculty of Health Sciences, University of Adelaide, Australia; 7Health Sciences UniSA, Adelaide, Australia; 8Australasian Leukaemia and Lymphoma Group, Melbourne, Australia; 9 Department of Haematology, Royal Adelaide Hospital and SA Pathology, Australia; 10 Centre for Cancer Biology, School of Pharmacy and Medical Science, University of South Australia, Adelaide, Australia and 11Flinders University and Medical Centre, Adelaide, Australia 1

ABSTRACT

Correspondence: David.Ross@sa.gov.au

Received: January 29, 2018. Accepted: July 4, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.189787 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2026 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2026

ccurate quantification of minimal residual disease (MRD) during treatment of chronic myeloid leukemia (CML) guides clinical decisions. The conventional MRD method, RQ-PCR for BCR-ABL1 mRNA, reflects a composite of the number of circulating leukemic cells and the BCR-ABL1 transcripts per cell. BCR-ABL1 genomic DNA only reflects leukemic cell number. We used both methods in parallel to determine the relative contribution of the leukemic cell number to molecular response. BCR-ABL1 DNA PCR and RQ-PCR were monitored up to 24 months in 516 paired samples from 59 newly-diagnosed patients treated with first-line imatinib in the TIDEL-II study. In the first three months of treatment, BCR-ABL1 mRNA values declined more rapidly than DNA. By six months, the two measures aligned closely. The expression of BCR-ABL1 mRNA was normalized to cell number to generate an expression ratio. The expression of e13a2 BCR-ABL1 was lower than that of e14a2 transcripts at multiple time points during treatment. BCR-ABL1 DNA was quantifiable in 48% of samples with undetectable BCR-ABL1 mRNA, resulting in MRD being quantifiable for an additional 5-18 months (median 12 months). These parallel studies show for the first time that the rapid decline in BCR-ABL1 mRNA over the first three months of treatment is due to a reduction in both cell number and transcript level per cell, whereas beyond three months, falling levels of BCR-ABL1 mRNA are proportional to the depletion of leukemic cells.

Introduction Real-time reverse transcriptase quantitative PCR (RQ-PCR) for BCR-ABL1 mRNA is widely used for the routine monitoring of chronic myeloid leukemia (CML) patients receiving tyrosine kinase inhibitor (TKI) therapy. The achievement of molecularly-defined therapeutic targets during TKI treatment is associated with superior progression-free and overall survival.1 The BCR-ABL1 mRNA level is a composite measurement that reflects both the proportion of leukemic cells in the sample, and the expression of BCR-ABL1 relative to its control gene. Pre-analytical factors, such as the rate of degradation of the target mRNA, and methodological factors, such as the efficiency of reverse transcription or the choice of control gene, may have a significant influence on the final result of RQ-PCR.2,3 Substantial effort has been invested to minimize variation due to such factors through the development of an International Scale (IS) for BCR-ABL1.4 haematologica | 2018; 103(12)

BCR-ABL1 DNA monitoring of CML

An alternative approach to overcome the variability in RQ-PCR is to measure BCR-ABL1 genomic DNA, since the overwhelming majority of chronic phase CML patients will have a single copy of BCR-ABL1 and two copies of an autosomal control gene in each leukemic cell. In the past, this approach was not practical due to the complexity of sequencing individual genomic breakpoints. Almost all CML patients express one or both of the two common BCR-ABL1 mRNA transcripts (e13a2, e14a2), whereas the genomic fusion sequences involve introns that are spliced out from the mRNA, and are essentially unique to each individual patient.5 Advances in sequencing technology have made it relatively simple to detect BCR-ABL1 genomic breakpoints, and several methods have been published.6,7 It should be emphasized that DNA PCR and RQ-PCR are not expected to yield identical results. This is perhaps best exemplified by the comparison of RQ-PCR with metaphase karyotyping in CML, which shows that a partial cytogenetic response [≤35% Philadelphia-positive (Ph+) cells] is roughly equivalent to BCR-ABL1IS ≤10%.8 Whereas both techniques are clinically useful, measures of the size of the CML clone the end point of each assay is qualitatively different. BCR-ABL1 DNA PCR is analogous to fluorescence in situ hybridization, in that both methods measure the simple proportion of cells in a sample that carry the Philadelphia rearrangement. We used quantitative BCR-ABL1 DNA techniques, QPCR and digital PCR (dPCR), to monitor a cohort of patients in the Australasian Leukaemia and Lymphoma Group (ALLG) CML9 study (TIDEL-II).9 These results were compared with routine RQ-PCR monitoring. Since the number of copies of BCR-ABL1 DNA is directly related to the number of leukemic cells in a sample, we used DNA and mRNA-based methods in order to determine the relative contribution of cell number and expression changes to molecular response in CML. Secondly, where there were differences between RQ-PCR and DNA PCR, we explored whether these differences might provide additional predictive information concerning treatment response.

Methods

SA Pathology, Adelaide, Australia, using the BCR control gene.11 The results were reported as BCR-ABL1/BCR% applying an IS conversion factor (Online Supplementary Appendix).4 Chromosome banding analysis was routinely performed at diagnosis in the respective local laboratories. Samples with fewer than 10 metaphases were excluded from this analysis. All samples were collected with informed consent in accordance with the Institutional Ethics-approved protocols and with reference to the Declaration of Helsinki.

Breakpoint detection The BCR-ABL1 genomic DNA breakpoint was determined, as previously described, in blood samples collected at diagnosis using long range PCR with a single forward primer in BCR and multiple reverse primers in ABL1 to amplify the breakpoint (Online Supplementary Appendix).12,13

Quantification of BCR-ABL1 DNA Genomic DNA was extracted from peripheral blood leukocytes. The amount of amplifiable DNA in each sample was measured using the GUSB control gene. The earlier assays were performed using real-time Q-PCR with standard curves for both BCR-ABL1 (patient’s diagnostic DNA assigned a value of 100%) and GUSB (plasmid) diluted in non-human DNA. Later assays used digital PCR (dPCR) for both BCR-ABL1 and GUSB with the aim of improving precision. Results were reported as BCR-ABL1/GUSB% (corrected for the two copies of GUSB per cell) normalized against the individual patient’s diagnostic sample. Further details are provided in the Online Supplementary Appendix and Online Supplementary Figures S1-S3.

Statistical analysis Statistical analysis was performed using the GraphPad Prism 7 statistical software (GraphPad Prism Inc., La Jolla, CA, USA). Agreement between assays was assessed using the method of Bland and Altman.14 Correlation between non-parametric values was assessed using Spearman rank coefficient. Differences between BCR-ABL1 DNA and mRNA measurements were compared using a Mann-Whitney test. The cumulative incidence of MMR and MR4.5 was calculated using the Fine and Gray regression method in R. Any event leading to the permanent discontinuation of imatinib/nilotinib (including treatment failure, intolerance, and death) was treated as a competing risk. P<0.05 was considered statistically significant.

Patients’ characteristics and samples Fifty-nine newly diagnosed chronic phase CML patients from the TIDEL-II clinical trial9 were included in our study. Details of these patients and of the samples analyzed are presented in the Online Supplementary Appendix and Online Supplementary Tables S1 and S2. The overall clinical characteristics and treatment responses of the selected cohort were not significantly different from those of the overall study population. The subset of patients included here were selected in three categories: undetectable MRD (UMRD) achieved within the first 2 years (n=26); treatment failure (n=9); and 24 additional patients not falling into either of the first two categories. Treatment failure was defined following the European LeukemiaNet (ELN) criteria as loss of complete hematologic response, loss of complete cytogenetic response, loss of major molecular response (MMR; BCR-ABL1IS ≤0.1%), kinase domain mutations, or progression to accelerated phase/blast crisis).10 Peripheral blood samples for molecular analysis were collected prior to commencing TKI treatment (baseline); at one, two, and three months; and every three months thereafter up to 24 months. RQ-PCR was performed centrally in the diagnostic laboratory of haematologica | 2018; 103(12)

Results Comparison between DNA and mRNA before treatment Since DNA Q-PCR quantifies BCR-ABL1 relative to the diagnostic DNA, we considered only the absolute dPCR values at diagnosis (n=29) and compared these values with the corresponding mRNA levels and the percentage of Ph+ bone marrow metaphase cells. The median value of BCR-ABL1 DNA prior to TKI treatment was 100% by karyotyping (range, 85-100%) and 84% (range, 45-164%) by dPCR. The corresponding median BCR-ABL1IS value was 70%, with values ranging from 3.7% to 425% (Figure 1A). Two of the 3 patients (#1 and #3) with low BCRABL1IS mRNA had stored peripheral blood cells available for interphase fluorescence in situ hybridization, which showed excellent agreement with the BCR-ABL1 DNA values obtained by dPCR (Table 1). Two of these 3 patients experienced treatment failure (blast crisis/secondary resistance with a kinase domain mutation) and the third patient had ELN warning features at baseline (high 2027

I.S. Pagani et al. A

Figure 1. DNA and mRNA prior to treatment. (A) Proportion of leukemic cells and BCR-ABL1 expression before treatment assessed by conventional cytogenetic analysis (green), DNA dPCR (red) and RQ-PCR (blue). Three patients with discrepant DNA and mRNA values are highlighted (red square). (B-D) Molecular response of patients (pts) with BCR-ABL1IS <10% despite DNA values close to 100%. Absolute DNA dPCR values are represented at diagnosis. TKI: tyrosine kinase inhibitor; Ph+: Philadelphia positive; Und: undetectable.

Sokal score and additional clonal chromosomal abnormalities in Ph+ cells) and failed to reach MMR by 12 months and MR4.5 by 24 months. All 3 of these patients expressed e13a2 BCR-ABL1 transcripts (one expressed both e13a2 and e14a2) and all had unusually low white blood cell counts at study entry (<20x109/L) (Online Supplementary Table S3). The post-treatment molecular responses of these 3 patients are shown in Figure 1B-D.

Table 1. BCR-ABL1 values in patients with low mRNA values relative to DNA values.

Patient ID #1 #2 #3

mRNA, IS%

DNA dPCR, %

iFISH, %

6.96 5.10 3.70

88.0 81.2 101.8

99.4 N/A 87.8

IS: International Scale; dPCR: digital PCR; iFISH: interphase fluorescence in situ hybridization; N/A: not available.

Agreement of Q-PCR and dPCR for BCR-ABL1 DNA Forty-six serial samples from 9 patients on TKI treatment were quantified by both Q-PCR and dPCR for BCRABL1 DNA. The results were highly correlated (r=0.94, P<0.0001). Agreement between the two methods was further assessed using a Bland-Altman plot (Online Supplementary Figure S4). The mean bias was -0.11-log with the 95% limits of agreement ranging from -1.02-log to 0.80-log (Âą8.1 fold) indicating that there was no systematic difference between the results obtained by the two DNA PCR methods after diagnosis. In subsequent analyses of BCR-ABL1 DNA during treatment the two sets of data were combined.

Faster reduction in BCR-ABL1 mRNA than DNA early in treatment In our cohort of 59 patients, the quantified BCR-ABL1 mRNA and DNA results (undetectable values excluded) were highly correlated across the range of values during TKI treatment (r=0.88; P<0.0001) (Figure 2A). However, during the first three months of therapy BCR-ABL1 DNA values were significantly higher than mRNA, whereas 2028

from six months onwards there was good agreement between methods (Figure 2B). The median reduction in BCR-ABL1IS from baseline to three months was 2.05-log versus 1.75-log for BCR-ABL1 DNA (Online Supplementary Figure S5). This bias was independent of the BCR-ABL1 DNA quantification method (seen with both dPCR and QPCR; see Online Supplementary Figure S6).

Early molecular response assessment by BCR-ABL1 mRNA and DNA

A reduction in BCR-ABL1IS to â&#x2030;¤10% at three months [early molecular response (EMR)] has emerged as an early treatment milestone that is strongly associated with later achievement of optimal response and progression-free survival.15,16 The predictive effect of EMR was confirmed in the overall TIDEL-II study population of 210 patients.9 In this smaller subgroup, no patient with BCR-ABL1IS levels >10% at three months went on to achieve MMR or MR4.5. We tested the predictive value of BCR-ABL1 levels by both mRNA and DNA at the 3-month landmark using haematologica | 2018; 103(12)

BCR-ABL1 DNA monitoring of CML Figure 2. Comparison between mRNA and DNA quantification of BCR-ABL1. (A) Positive values from DNA (red) and mRNA (blue) were compared during treatment until 24 months. The quantifiable mRNA and DNA values were highly correlated, but at very low levels 42 samples were positive by DNA PCR only (red square). (B) The mRNA and DNA values (number, median and interquartile range) are shown for individual time points up to 24 months. Note that the apparent increase in RQ-PCR after 18 months is due to the exclusion of RQ-PCR samples in which there was undetectable BCR-ABL1. *P<0.05; ***P<0.0001; Und.: undetectable.

the established BCR-ABL1 transcript IS cut-offs of 10% and 1%. Both mRNA and DNA levels were predictive of later MMR and MR4.5, and the BCR-ABL1 DNA level did not improve the predictive value of conventional RQ-PCR (Online Supplementary Figure S7). The optimal BCR-ABL1 DNA cut-off for prediction of later molecular response could not be determined in this study due to the small number of patients and the potential bias due to the selection of patients on the basis of response.

Transcript type and molecular response It has previously been reported that the BCR-ABL1 transcript type may influence treatment outcomes (reviewed by Marum and Branford17). Consequently, we compared molecular responses in patients having only e13a2 transcripts (n=32) or only e14a2 transcripts (n=17). There was no significant difference between BCR-ABL1IS levels according to transcript type at any individual time point (Figure 3A). However, BCR-ABL1 DNA was significantly higher in e13a2 patients at multiple time points during treatment (Figure 3B). The median BCR-ABL1 expression ratio (mRNA%:DNA%) was 0.5 for e13a2 versus 1.09 for e14a2 (P=0.0005) (Figure 3C). This analysis was repeated using BCR-ABL1 DNA values from dPCR and Q-PCR separately and a similar pattern was observed (Online Supplementary Figure S8). haematologica | 2018; 103(12)

Sensitivity of RQ-PCR and DNA PCR The median limit of detection achieved by RQ-PCR was MR4.6 (range, 3.2-5.1 log) in comparison with MR5.2 (range, 4.6-5.7 log) for DNA PCR. BCR-ABL1 DNA was detected in 42 of 86 samples with undetectable mRNA (49%) with a median value of 0.002% (range, 0.00020.07%). Two samples were mRNA-positive, DNA-negative with BCR-ABL1IS values of 0.003 and 0.02% (Figure 2A). The remaining 44 samples had undetectable BCR-ABL1 by both methods. The higher degree of sensitivity using BCR-ABL1 DNA led to MRD being quantifiable for an additional 5-18 months (median 12 months) of follow up. Samples collected after 24 months were not analyzed, so in some patients the duration of quantifiable MRD may have been longer than this estimate.

Discussion BCR-ABL1 molecular monitoring by RQ-PCR is relied upon to ensure that TKI-treated patients are on track to achieve an optimal response, to define the end points of clinical trials, and to determine criteria for a safe trial of cessation of TKI therapy after having sustained a deep molecular response.18-20 Molecular responses defined by RQ-PCR have been shown to be robust indicators of clin2029

I.S. Pagani et al.

ical outcome, yet the biology of BCR-ABL1 molecular response is relatively complex. Key to this complexity is the composite nature of the response: a reduction in the ratio of BCR-ABL1 mRNA to a control gene could be due to a reduction in the proportion of CML cells in the sample, a reduction in the expression of BCR-ABL1, an increase in the expression of the control gene, or even a

change in the relative stability of these mRNA transcripts. Since the number of copies of genomic BCR-ABL1 is directly proportional to the number of leukemic cells, we reasoned that measuring both BCR-ABL1 DNA and mRNA would lead to a better understanding of the main determinants of variation in molecular response. During the first three months of treatment, the

Months since starting TKI treatment

C P=0.0005

Figure 3. BCR-ABL1 transcript type and molecular response. (A) Comparison of the BCR-ABL1IS values during the first two years of tyrosine kinase inhibitor (TKI) treatment (e13a2 shown in green and e14a2 shown in black). (B) Comparison of e13a2 and e14a2 BCR-ABL1 DNA values in the same patients. Diagnostic values were assigned a value of 100%. Note that at later time points the proportion of e14a2 patients with undetectable BCR-ABL1 DNA was higher than for e13a2, which may result in an underestimation of the difference between the two transcript types. (C) Box and whiskers plot comparing BCR-ABL1 expression ratio (mRNA:DNA) for e13a2 and e14a2 transcripts. *P<0.05; ***P<0.001.

2030

haematologica | 2018; 103(12)

BCR-ABL1 DNA monitoring of CML

BCR-ABL1 DNA values were significantly higher than the corresponding BCR-ABL1IS values. After three months, the reduction in BCR-ABL1IS levels (2.05-log) was primarily due to depletion of CML cells (1.75-log), with only a small contribution from expression changes (0.3-log reduction; 2-fold decrease). A proportionally greater decline in expression than in cell number is likely due to the early depletion of higher expressing cells. From six months of treatment onwards, there was generally excellent agreement between the level of MRD measured by BCR-ABL1 DNA and by RQ-PCR, indicating that the decline in BCR-ABL1IS is closely paralleled by declining numbers of BCR-ABL1-positive cells, as has been predicted using mathematical models based on RQ-PCR data.21,22 Several studies dating back at least two decades have reported inferior treatment responses among CML patients with e13a2 BCR-ABL1 transcripts.17,23 Differences in molecularly-defined end points might simply reflect differing amplification efficiency in the BCR-ABL1 assay, especially in those systems that use a common forward primer in BCR e13, resulting in a 76 bp difference in amplicon length between the two transcripts.24 In BCR-ABL1 DNA PCR, every patient-specific assay will have differing properties due to varying amplicon size and nucleotide composition. These differences are determined by factors related to the precise intronic location of the breakpoints, and therefore independent of the transcript type (Online Supplementary Figure S3). We took advantage of this to compare the relative expression of e13a2 and e14a2 BCR-ABL1 transcripts by normalizing the BCR-ABL1IS value against BCR-ABL1 DNA. Patients expressing both e13a2 and e14a2 transcripts were excluded from this analysis; in those cases the genomic BCR breakpoint is after exon 14, so e14a2 is the dominant transcript with a fraction of e13a2 expressed due to alternative splicing.25 Despite the small number of patients with each transcript type, we were able to show a significant difference in expression per cell at multiple time points during treatment. These findings require independent confirmation in a larger cohort. When treatment decisions are made according to molecular landmark responses, this may lead to incorrect classification of some e13a2 patients as optimal responders, and could contribute to adverse outcomes. Intriguingly, we identified 3 e13a2 patients who at diagnosis had discordant low BCR-ABL1 mRNA values (<10%) despite having close to 100% BCR-ABL1I-positive cells by DNA PCR and metaphase karyotyping. All 3 of these patients experienced treatment failure or warning by ELN criteria. The significance of this finding is unclear, given the small number of patients in this subgroup. It is, however, consistent with the experimental observation that imatinib sensitivity was reduced in BCR-ABL1-trans-

References 1. Hughes TP, Kaeda J, Branford S, et al. Frequency of major molecular responses to imatinib or interferon alfa plus cytarabine in newly diagnosed chronic myeloid leukemia. N Engl J Med. 2003; 349(15):1423-1432. 2. van der Velden VH, Boeckx N, Gonzalez M, et al. Differential stability of control

haematologica | 2018; 103(12)

duced murine cells selected for low BCR-ABL1 expression.26 The median limit of detection of BCR-ABL1 DNA was MR5.2 versus MR4.6 for conventional RQ-PCR. This improvement in sensitivity led to around half of the samples with undetectable BCR-ABL1 mRNA having measurable MRD and extended the period of time in which there was detectable BCR-ABL1 by around a year. The median limit of detection for dPCR was MR5.2. These results are similar to those obtained by Alikian et al., who used dPCR for both BCR-ABL1 DNA and mRNA, and found higher sensitivity with the DNA-based assay.6 Whilst the comparison of PCR methods was not the aim of this study, we found that dPCR was more precise, especially at diagnosis, but the FluidigmÂŽ system has the disadvantage that more than 80% of the input DNA is lost in the dead space of the microfluidic circuit. More sensitive mRNA-based methods have also been developed,6,17,27 and comparisons using different technology may yield different results. Nevertheless, genomic DNA-based methods have the advantage of greater specificity since cross-contamination between samples from different patients cannot cause false-positive results when the assays are patient-specific. A previous study showed that some patients with undetectable BCR-ABL1 mRNA by RQ-PCR had MRD detected by dPCR, and that these patients had a lower probability of successful treatmentfree remission (TFR) after imatinib discontinuation.27 In the ENESTfreedom study of discontinuation of first-line nilotinib, patients with MR4.5 on every measurement for 12 months prior to stopping nilotinib were more likely to maintain TFR at 12 months than patients with one or more results above the MR4.0 threshold.28 More sensitive PCR methods might, therefore, have clinical utility as a means of refining estimates of the probability of TFR. Miminal residual disease measurement by genomic DNA PCR provides insights into the kinetics of molecular response that are not provided by conventional RQ-PCR. The strong correlation between RQ-PCR and DNA-based PCR in follow-up samples beyond three months indicates that the major determinant of RQ-PCR values is the number of circulating leukemic cells, rather than variable expression of BCR-ABL1. Acknowledgments The authors wish to thank the patients, investigators, study coordinators, and ALLG staff who provided the samples and clinical data for this study, and acknowledge the ALLG as the sponsor of the TIDEL-II clinical trial. Novartis Pharmaceuticals provided a research grant for this project and additional funding was obtained from the South Australian Health Services Charitable Gifts Board.

gene and fusion gene transcripts over time may hamper accurate quantification of minimal residual disease--a study within the Europe Against Cancer Program. Leukemia. 2004;18(4):884-886. 3. Hughes T, Deininger M, Hochhaus A, et al. Monitoring CML patients responding to treatment with tyrosine kinase inhibitors: review and recommendations for harmonizing current methodology for detecting BCR-ABL transcripts and kinase domain

mutations and for expressing results. Blood. 2006;108(1):28-37. 4. Branford S, Fletcher L, Cross NC, et al. Desirable performance characteristics for BCR-ABL measurement on an international reporting scale to allow consistent interpretation of individual patient response and comparison of response rates between clinical trials. Blood. 2008;112(8):33303338. 5. Ross DM, O'Hely M, Bartley PA, et al.

2031

I.S. Pagani et al.

10.

11.

12.

2032

Distribution of genomic breakpoints in chronic myeloid leukemia: analysis of 308 patients. Leukemia. 2013;27(10):2105-2107. Alikian M, Ellery P, Forbes M, et al. NextGeneration Sequencing-Assisted DNABased Digital PCR for a Personalized Approach to the Detection and Quantification of Residual Disease in Chronic Myeloid Leukemia Patients. J Mol Diagn. 2016;18(2):176-189. Linhartova J, Hovorkova L, Soverini S, et al. Characterization of 46 patient-specific BCRABL1 fusions and detection of SNPs upstream and downstream the breakpoints in chronic myeloid leukemia using next generation sequencing. Mol Cancer. 2015; 14:89. Ross DM, Branford S, Moore S, Hughes TP. Limited clinical value of regular bone marrow cytogenetic analysis in imatinib-treated chronic phase CML patients monitored by RQ-PCR for BCR-ABL. Leukemia. 2006;20(4):664-670. Yeung DT, Osborn MP, White DL, et al. TIDEL-II: first-line use of imatinib in CML with early switch to nilotinib for failure to achieve time-dependent molecular targets. Blood. 2015;125(6):915-923. Baccarani M, Deininger MW, Rosti G, et al. European LeukemiaNet recommendations for the management of chronic myeloid leukemia: 2013. Blood. 2013;122(6):872884. Branford S, Hughes TP, Rudzki Z. Monitoring chronic myeloid leukaemia therapy by real-time quantitative PCR in blood is a reliable alternative to bone marrow cytogenetics. Br J Haematol. 1999; 107(3):587-599. Score J, Calasanz MJ, Ottman O, et al. Analysis of genomic breakpoints in p190 and p210 BCR-ABL indicate distinct mechanisms of formation. Leukemia. 2010;

24(10):1742-1750. 13. Ross DM, Branford S, Seymour JF, et al. Patients with chronic myeloid leukemia who maintain a complete molecular response after stopping imatinib treatment have evidence of persistent leukemia by DNA PCR. Leukemia. 2010;24(10):17191724. 14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307-310. 15. Hanfstein B, Muller MC, Hehlmann R, et al. Early molecular and cytogenetic response is predictive for long-term progression-free and overall survival in chronic myeloid leukemia (CML). Leukemia. 2012;26(9):2096-2102. 16. Hughes TP, Saglio G, Kantarjian HM, et al. Early molecular response predicts outcomes in patients with chronic myeloid leukemia in chronic phase treated with frontline nilotinib or imatinib. Blood. 2014;123(9):1353-1360. 17. Marum JE, Branford S. Current developments in molecular monitoring in chronic myeloid leukemia. Ther Adv Hematol. 2016;7(5):237-251. 18. Hughes TP, Ross DM. Moving treatmentfree remission into mainstream clinical practice in CML. Blood. 2016;128(1):17-23. 19. Mahon FX, Etienne G. Deep molecular response in chronic myeloid leukemia: the new goal of therapy? Clin Cancer Res. 2014;20(2):310-322. 20. Saussele S, Richter J, Hochhaus A, Mahon FX. The concept of treatment-free remission in chronic myeloid leukemia. Leukemia. 2016;30(8):1638-1647. 21. Michor F, Hughes TP, Iwasa Y, et al. Dynamics of chronic myeloid leukaemia. Nature. 2005;435(7046):1267-1270.

22. Roeder I, Horn M, Glauche I, Hochhaus A, Mueller MC, Loeffler M. Dynamic modeling of imatinib-treated chronic myeloid leukemia: functional insights and clinical implications. Nat Med. 2006;12(10):11811184. 23. Jain P, Kantarjian H, Patel KP, et al. Impact of BCR-ABL transcript type on outcome in patients with chronic-phase CML treated with tyrosine kinase inhibitors. Blood. 2016;127(10):1269-1275. 24. Gabert J, Beillard E, van der Velden VH, et al. Standardization and quality control studies of 'real-time' quantitative reverse transcriptase polymerase chain reaction of fusion gene transcripts for residual disease detection in leukemia - a Europe Against Cancer program. Leukemia. 2003; 17(12):2318-2357. 25. Branford S, Hughes TP, Rudzki Z. Dual transcription of b2a2 and b3a2 BCR-ABL transcripts in chronic myeloid leukaemia is confined to patients with a linked polymorphism within the BCR gene. Br J Haematol. 2002;117(4):875-877. 26. Kumari A, Brendel C, Hochhaus A, Neubauer A, Burchert A. Low BCR-ABL expression levels in hematopoietic precursor cells enable persistence of chronic myeloid leukemia under imatinib. Blood. 2012;119(2):530-539. 27. Mori S, Vagge E, le Coutre P, et al. Age and dPCR can predict relapse in CML patients who discontinued imatinib: the ISAV study. Am J Hematol. 2015;90(10):910-914. 28. Ross DM, Masszi T, Gomez Casares MT, et al. Durable treatment-free remission in patients with chronic myeloid leukemia in chronic phase following frontline nilotinib: 96-week update of the ENESTfreedom study. J Cancer Res Clin Oncol. 2018; 144(5):945-954.

haematologica | 2018; 103(12)

ARTICLE

Acute Myeloid Leukemia

Arsenic trioxide is required in the treatment of newly diagnosed acute promyelocytic leukemia. Analysis of a randomized trial (APL 2006) by the French Belgian Swiss APL group Lionel Adès,1 Xavier Thomas,2 Agnes Guerci Bresler,3 Emmanuel Raffoux,1 Olivier Spertini,4 Norbert Vey,5 Tony Marchand,6 Christian Récher,7 Arnaud Pigneux,8 Stephane Girault,9 Eric Deconinck,10 Claude Gardin,11 Olivier Tournilhac,12 Jean Francois Lambert,13 Patrice Chevallier,14 Stephane de Botton,15 Julie Lejeune,1 Hervé Dombret,1 Sylvie Chevret1 and Pierre Fenaux1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2033-2039

Hopital Saint Louis, Université Paris Diderot, France; 2Centre Hopitalo-Universitaire Lyon, France; 3Centre Hopitalo-Universitaire Nancy, France; 4Service d'Hématologie CHUV Lausanne, Switzerland; 5Institut Paoli Calmette, Marseille, France; 6Rennes University Hospital, France; 7Toulouse University Hospital, France; 8Bordeaux University Hospital, France; 9Limoges University Hospital, France; 10Besancon University Hospital, France; 11Bobigny University Hospital, France; 12Clermont-Ferrand University Hospital, France; 13Centre Hopitalo-Universitaire Vaudois, Switzerland; 14Nantes University Hospital, France and 15Institut Gustave Roussy, Villejuif, France 1

ABSTRACT

n standard-risk acute promyelocytic leukemia, recent results have shown that all-trans retinoic acid plus arsenic trioxide combinations are at least as effective as classical all-trans retinoic acid plus anthracycline-based chemotherapy while being less myelosuppressive. However, the role of frontline arsenic trioxide is less clear in higher-risk acute promyelocytic leukemia, and access to arsenic remains limited for frontline treatment of standard-risk acute promyelocytic leukemia in many countries. In this randomized trial, we compared arsenic, all-trans retinoic acid and the “classical” cytarabine for consolidation treatment (after all-trans retinoic acid and chemotherapy induction treatment) in standard-risk acute promyelocytic leukemia, and evaluated the addition of arsenic during consolidation in higher-risk disease. Patients with newly diagnosed acute promyelocytic leukemia with a white blood cell count <10x109/L, after an induction treatment consisting of all-trans retinoic acid plus idarubicin and cytarabine, received consolidation chemotherapy with idarubicin and cytarabine, arsenic or all-trans retinoic acid. Patients with a white blood cell count >10x109/L received consolidation chemotherapy with or without arsenic. Overall, 795 patients with acute promyelocytic leukemia were enrolled in this trial. Among those with standard-risk acute promyelocytic leukemia (n=581), the 5-year event-free survival rates from randomization were 88.7%, 95.7% and 85.4% in the cytarabine, arsenic and all-trans retinoic acid consolidation groups, respectively (P=0.0067), and the 5-year cumulative incidences of relapse were was 5.5%, 0% and 8.2%. (P=0.001). Among those with higher-risk acute promyelocytic leukemia (n=214), the 5-year event-free survival rates were 85.5% and 92.1% (P=0.38) in the chemotherapy and chemotherapy plus arsenic groups, respectively, and the corresponding 5-year cumulative incidences of relapse were 4.6% and 3.5% (P=0.99). Given the prolonged myelosuppression that occurred in the chemotherapy plus arsenic arm, a protocol amendment excluded cytarabine during consolidation cycles in the chemotherapy plus arsenic group, resulting in no increase in relapse. Our results therefore advocate systematic introduction of arsenic in the first-line treatment of acute promyelocytic leukemia, but probably not concomitantly with intensive chemotherapy, a situation in which we found myelosuppression to be significant. (ClinicalTrials.gov Identifier: NCT00378365) haematologica | 2018; 103(12)

Correspondence: pierre.fenaux@aphp.fr

Received: May 28, 2018. Accepted: July 17, 2018. Pre-published: July 19, 2018. doi:10.3324/haematol.2018.198614 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2033 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2033

L. Adès et al.

Introduction Acute promyelocytic leukemia (APL) is a specific subtype of acute myeloid leukemia (AML) characterized by its morphology, the presence of t(15;17), and marked sensitivity to the differentiating effect of all-trans retinoic acid (ATRA) and the pro-apoptotic effect of arsenic trioxide (ATO).1 The combination of ATRA and anthracycline-based chemotherapy has been the mainstay of the treatment of newly diagnosed APL over the last two decades.2–4 Published results have shown that cytarabine (cytosine arabinoside, AraC) could be omitted from chemotherapy in standard-risk APL [i.e., with a baseline white blood cell count (WBC) <10x109/L] but appeared to be useful in high-risk APL (with a WBC >10x109/L), possibly at high doses, to reduce the incidence of relapse.5 A beneficial role for prolonged maintenance treatment with continuous low-dose chemotherapy (6-mercaptopurine and methotrexate) and intermittent ATRA was also suggested, especially in high-risk APL, following in particular randomized results from our group,5,6 and from a recent meta-analysis7 of several trials. With regards to anthracyclines, at least one study suggested that idarubicin gave better results than daunorubicin,8 while non-randomized studies suggested a potential benefit of adding ATRA during consolidation cycles, at least if AraC was omitted.2,4 Recently, however, ATO has been demonstrated to have pronounced efficacy in newly diagnosed APL. In particular, it was shown in two large randomized trials that the combination of ATO and ATRA without chemotherapy was at least equal and, with longer term follow-up, even superior to ATRA plus chemotherapy combinations in standard-risk APL.9,10,11 In high-risk APL, ATO plus ATRA combinations, with very limited added chemotherapy, also appear very promising,10,12 and are currently being compared with the conventional ATRA chemotherapy approach in randomized trials. When the APL 2006 trial was launched, ATO was mainly considered as an adjunct to ATRA chemotherapy combinations in the first-line treatment of APL, aimed at reducing the relapse rate (especially in high-risk APL) and/or diminishing the amount of chemotherapy administered (especially in standard-risk APL). Based on the results of the APL 2006 trial, reported here, we evaluated the role of ATO in the treatment of standardand high-risk APL, in addition to the “classical” ATRA plus chemotherapy backbone regimens.

Patients with a baseline WBC <10x109/L who achieved a complete remission were randomized for consolidation between three groups given treatment containing AraC, ATO or ATRA. The AraC group (standard group) received a first consolidation course with idarubicin 12 mg/m2/day for 3 days and AraC 200 mg/m2/day for 7 days, a second consolidation course with idarubicin 9 mg/m2/day for 3 days and AraC 1 g/m2/12 h for 4 days, and maintenance therapy for 2 years with intermittent ATRA 15 days/3 months and continuous treatment with 6 mercaptopurine (90 mg/m2/day orally) and methotrexate (15 mg/m2/week orally). The ATO and ATRA groups received the same treatment as the AraC group, but AraC was replaced by, respectively, ATO 0.15 mg/kg/day on days 1 to 25 and ATRA 45 mg/m2/day on days 1 to 15 for both consolidation courses. The rationale for the ATRA consolidation treatment was based on results of a Spanish PETHEMA group trial, suggesting that AraC could be omitted from chemotherapy consolidation cycles in standard-risk APL, and that there could be a benefit from adding ATRA to consolidation cycles. The use of prolonged maintenance treatment was based on our previous results in a randomized phase III trial supporting the interest of this approach in reducing relapses after a conventional ATRA chemotherapy regimen. Patients with a baseline WBC >10x109/L were randomized to consolidation with either chemotherapy or chemotherapy combined with ATO. The chemotherapy group received a first consolidation course with idarubicin 12 mg/m2/day for 3 days and AraC 200 mg/m2/day for 7 days, a second consolidation course with idarubicin 9 mg/m2/day for 3 days and AraC 1 g/m2/12 h for 4 days, and 2-year maintenance therapy with intermittent ATRA and continuous 6-mercaptopurine plus methotrexate. The chemotherapy plus ATO group received the same treatment except that ATO 0.15 mg/kg/day was added from day 1 to day 25 during both consolidation courses. After a first interim analysis in September 2010 on data from 81 patients, AraC was deleted from consolidation cycles of the chemotherapy plus ATO group. Treatment of coagulopathy during the induction phase was based on platelet support to maintain the platelet count at a level greater than 50x109/L until the disappearance of the coagulopathy. The use of heparin, tranexamic acid, fresh-frozen plasma, and fibrinogen transfusions was optional, according to each center's policy. Prophylaxis and treatment of ATRA syndrome consisted of dexamethasone 10 mg/12 h given intravenously for at least 3 days if the WBC was above 10x109/L (before or during treatment with ATRA) or at the earliest sign of the ATRA syndrome (dyspnea, lung infiltrates, pleural effusion, unexplained renal failure). In the absence of rapid improvement of symptoms (within 24 h), ATRA was transiently stopped until clinical control was obtained.

Methods Statistical methods Patients Between 2006 and 2013, patients from French, Belgian and Swiss centers with documented (by cytogenetics and or molecular biology), newly diagnosed APL who were aged 70 years or less were eligible for inclusion in the APL 2006 trial, after giring informed consent. The trial was approved by local ethical committees (ClinicalTrials.gov Identifier: NCT00378365). Eligibility criteria in this trial were a morphological diagnosis of APL based on French-American-British criteria and no contraindication to intensive chemotherapy. No minimal performance status was required and patients with therapy-related APL could be enrolled. Induction treatment consisted of ATRA 45 mg/m2/day until complete remission with idarubicin 12 mg/m2/day for 3 days and AraC 200 mg/m2/day for 7 days starting on day 3. 2034

The primary endpoint was event-free survival from the time of achieving complete remission. Relapse, survival, side effects of the treatment and duration of hospitalization were secondary endpoints. Analyses were performed on a modified intent-to-treat principle, excluding only diagnostic errors and withdrawals of consent. Censored endpoints were estimated by the nonparametric KaplanMeier method13 and then compared between randomized groups by the log-rank test. In estimating relapses, we took into account competing risks, i.e., deaths in first complete remission, using cumulative incidence curves and then compared results using the Gray test, whereas a cause-specific Cox model was used to estimate cause-specific hazard ratios.14 The type I error was fixed at the 5% level. All tests were two-tailed. Statistical analyses were haematologica | 2018; 103(12)

Role of arsenic trioxide in APL

performed using SAS 9.1 (SAS Inc, Cary, NC, USA) and R software packages. Here we present the results based on all patients included in the trial and data collected before June, 2017.

Results Eight-hundred and seven patients were included in the trial. The diagnosis of APL could be confirmed in 795 of the patients who had t(15;17) and/or a PML-RAR rearrangement. The remaining 12 patients were excluded as diagnostic errors. The further analyses only dealt with the 795 patients with a confirmed diagnosis of APL who gave their consent to participation in the study and comprised 581 patients with standard-risk APL and 214 with high-risk APL.

Standard-risk acute promyelocytic leukemia Of the 581 patients with standard-risk APL; 570 (98.1%) achieved a complete remission; the others died early. Fortythree patients were not randomized for consolidation treatment, including the 11 patients who did not achieve a complete remission, 15 due to adverse events, 12 due to the patientsâ&#x20AC;&#x2122; decision and five for other reasons (Figure 1). Five-hundred and thirty-eight patients were randomized to different consolidation treatment (178, 180 and 180 in the AraC, ATO and ATRA arms, respectively). Pre-treatment characteristics were well-balanced between the three consolidation groups (Table 1).

Overall, 8, 0, and 14 patients relapsed (P=0.001) and 11, 5, and 11 patients (P=0.28) died in complete remission in the AraC, ATO and ATRA consolidation groups, respectively. Causes of death in complete remission were sepsis (n=2), hemorrhage (n=16), AML/myelodysplastic syndrome (MDS) (n=2), other (n=7). Overall, five patients developed AML/MDS including two, one and two patients treated in the AraC, ATO and ATRA arms, respectively. Five-year event-free survival rates from randomization were 88.7%, 95.7% and 85.4% in the AraC, ATO and ATRA consolidation groups, respectively (P=0.0067). The 5-year cumulative incidences of relapse were 5.5%, 0% and 8.2% (P=0.001) and the 5-year overall survival rates were 93.6%, 95.7% and 91.9% (P=0.349) in the AraC, ATO and ATRA consolidation groups, respectively (Figure 2). The median times to an absolute neutrophil count >1x109/L after the first consolidation course were 23.5, 22.8 and 18 days in the AraC, ATO and ATRA groups, respectively (P<0.0001). Similarly, the times to an absolute neutrophil count >1x109/L after the second consolidation course were 23.3, 18.2 and 13.8 days (P<0.0001). The median durations of hospitalization after the first and the second consolidation courses were 31.5, 32.2, and 19.5 days (P<0.0001) and 28.2, 29.9, and 16.5 days in the AraC, ATO and ATRA group, respectively (P<0.0001).

High-risk acute promyelocytic leukemia Of the 214 patients with high-risk APL, 205 (95.7%) achieved a complete remission, seven (3.2%) died early (1 from bleeding, 3 from thrombosis, 1 from sepsis and 2 from

Figure 1. Consort diagrams. (A) Patients with a white blood cell count <109/L; (B) Patients with a white blood cell count >109/L. APL: acute promyelocytic leukemia; ARAC: cytarabine; ATO: arsenic trioxide; ATRA: all-trans retinoic acid.

haematologica | 2018; 103(12)

2035

L. AdĂ¨s et al.

other causes) and two (0.9%) had resistant leukemia. Seventeen patients were not randomized to consolidation treatment, including the nine patients who did not achieve a complete remission, three due to adverse events and five consequent to the patientsâ&#x20AC;&#x2122; decision. One hundred and ninety-seven patients were randomized to consolidation therapy, 99 in the chemotherapy group and 98 in the chemotherapy plus ATO groups. Pretreatment characteristics were well balanced between the two groups (Table 2). With a median follow-up of 52 months, eight patients (4 in the chemotherapy group versus

4 in the chemotherapy plus ATO group) had relapsed leading to 5-year cumulative incidence rates of 4.6% [95% confidence interval (95% CI: 1.5; 10.6) and 3.5% (95% CI: 0.9; 9.2), P=0.99] and 13 patients had died in complete remission including nine in the chemotherapy arm and four in the chemotherapy plus arm (P=0.98). One patient, randomized to the chemotherapy plus ATO arm, developed AML/MDS. The 5-year overall survival rates were 90% and 93% in the chemotherapy and chemotherapy plus ATO groups, respectively (P=0.62), while the corresponding 5-

Table 2. Baseline characteristics of the patients aged less than 70 years with high-risk acute promyelocytic leukemia. Table 1. Baseline characteristics of the patients aged <70 years with standard-risk acute promyelocytic leukemia.

Median [Q1-Q3]

Median [Q1-Q3] Age (years) WBC (x109/L) Platelets(x109/L) Fibrinogen (g/L) %M3v Previous cancer

AraC Arm N=178

ATO Arm N=180

ATRA Arm N=180

Age (years)

45.4 [32.45; 55.95] 1.3 [0.8; 2.3] 44.5 [25; 71] 1.8 [1.2; 2.7] 3% 12%

49.4 [38.85; 57.8] 1.4 [0.95; 2.615] 46 [20.75; 77] 1.7 [1.1; 2.65] 8% 12%

50.5 [38.4; 60.8] 1.51 [0.8; 3.45] 42 [20; 68] 1.8 [1.2; 2.6] 5% 9%

WBC (x109/L)

QI: first quartile; Q3: third quartile; AraC: cytarabine; ATO: arsenic trioxide; ATRA: alltrans retinoic acid; WBC: white blood cells; %M3v: microgranular variant (%).

Platelets (x109/L) Fibrinogen (g/L) %M3v % Previous cancer

CT Arm n=99

CT + ATO Arm n=98

39.2 [29.6 ; 54.1] 23.7 [14.7; 37.3] 26.5 [13.3; 47.8] 1.3 [1.1; 1.7] 34% 8%

45.0 [33.85; 58.9] 19.7 [13; 34.2] 32 [18; 51] 1.3 [0.9; 1.8] 28% 11%

QI: first quartile; Q3: third quartile; CT: chemotherapy; ATO: arsenic trioxide; WBC: white blood cells; %M3v: microgranular variant (%).

Figure 2. Event-free survival, cumulative incidence of relapse and overall survival in patients with standardrisk acute promyelocytic leukemia. ARAC: cytarabine; ATO: arsenic trioxide; ATRA: all-trans retinoic acid; CR: complete remission.

2036

haematologica | 2018; 103(12)

Role of arsenic trioxide in APL

year event-free rates were 85.5% and 92.1% (P=0.38) (Figure 3). Excluding AraC (after the protocol amendment) from the consolidation cycles in the chemotherapy plus ATO group did not increase the 5-year cumulative incidence of relapse (4.6% in the chemotherapy arm, 5.3% in the chemotherapy plus ATO with AraC arm and 2.7% in the chemotherapy plus ATO without AraC arm, P=0.61). On the other hand, excluding AraC from consolidation cycles in the chemotherapy plus ATO arm significanty reduced myelosuppression: the median times to an absolute neutrophil count >1x109/L after the second consolidation course were 22, 25 and 18 days in, respectively, the chemotherapy arm, the chemotherapy plus ATO with AraC arm, and the chemotherapy plus ATO without AraC arm (P<0.001), while the median times to a platelet count >50x109/L were 24, 26 and 18 days (P<0.001). Similarly, the median durations of hospitalization after the first and the second consolidation courses were 29 days, 34 days, and 33 days (P<0.0001) and 28 days, 32 days and 31 days (P=0.0005), respectively.

6-mercaptopurine, methotrexate and intermittent ATRA, which may also contribute to reducing the relapse rate.7 This reference treatment proved effective, as the incidence of relapse after 5 years was only 5.5%. Substituting ATRA for AraC did not significantly increase the relapse rate (8.2% at 5 years, compared to 5.5% in the AraC group) , but the replacement significantly reduced the time to recovery from neutropenia after the first and second consolidation cycles, and the duration of hospitalization during those two consolidation cycles. It did not reduce the incidence of deaths in complete remission, but among the six, six and five deaths in complete remission occurring in the three consolidation groups, only two in each arm were

Discussion The main results of this study are that, in standard-risk APL, addition of ATO to a “classical” ATRA chemotherapy regimen further reduces the incidence of relapse and that, in high-risk APL, AraC (including high-dose AraC) can be replaced by ATO without increasing the relapse risk and with more limited myelosuppression, thus potentially reducing the risk of death in complete remission. A first finding was the very high complete remission rate obtained in the APL 2006 trial, both in standard-risk and high-risk APL (98.1% and 95.7%, respectively), even though patients could be included up to the age of 70. Recent reports have suggested that, even in the ATRA era, early death rates could be as high as 15% to 20% in “reallife” APL patients.15–19 On the other hand, we previously published that, during the 2006 to 2011 period, 75% of the patients in the 17 French largest centers participating in the APL 2006 trial could be included in the trial, while 25% could not, mainly based on age, major comorbidities or direct admission to an intensive care unit.15 The overall complete remission rate was 91.4% and the overall rate of early death was 8.6%. All studies suggest that, if APL is suspected and before the diagnosis is confirmed, the immediate institution of ATRA treatment can reduce the risk of early death. Intensive platelet support during induction treatment can probably also contribute to reducing the risk of early death, particularly in patients with high-risk APL. In the APL 2006 trial, in standard-risk APL patients aged less than 70 years of age, our aim was to show that by substituting ATO or ATRA for AraC during consolidation cycles, we would not increase the relapse rate, but would reduce myelosuppression, thereby potentially reducing the incidence of deaths in complete remission, which was 5% in our previous experience with AraC-containing consolidation cycles (at a conventional dose for the first consolidation cycle, and intermediate dose for the second). The ATRA and chemotherapy regimen chosen appeared to be an “optimal” regimen, using in particular high cumulative doses of anthracyclines, idarubicin rather than daunorubicin (as the latter may lead to more relapses8), AraC during consolidation and prolonged maintenance treatment with haematologica | 2018; 103(12)

Figure 3. Overall survival, cumulative incidence of relapse and event-free survival in patients with high-risk acute promyelocytic leukemia. Chemo: chemotherapy; ATO: arsenic trioxide; CR: complete remission.

2037

L. Adès et al.

due to myelosuppression (the remaining being due to intercurrent disease or secondary AML/MDS). However, the main result in this standard-risk APL group was that no relapses were seen in the ATO arm, and that this relapse rate was significantly lower than in the AraC and ATRA consolidation arms. These results suggest that adding ATO to an already highly effective ATRA chemotherapy regimen may further improve the regimen’s anti-leukemic effect, and that ATO may not be dispensable in the treatment of standard-risk APL. On the other hand, substituting ATO for AraC did not reduce the duration of neutropenia after consolidation cycles, and neutropenia was longer with ATO and idarubicin than with ATRA and idarubicin consolidation cycles. This finding suggests that ATO, a non-myelosuppressive drug when used alone or combined with ATRA, may worsen myelosuppression when used concomitantly with chemotherapy. The duration of hospitalization was, however, shorter after ATO and idarubicin than after AraC and idarubicin consolidation cycles. The incidence of deaths in complete remission was not reduced in the ATO group, but only two deaths in complete remission were attributable to myelosuppression in the three consolidation arms. Finally, the incidence of secondary AML/MDS was similar in the three treatment arms, and similar to that reported in APL patients treated with ATRA chemotherapy regimens, i.e., between 1% and 2%.20–22 By contrast, in the follow up of the two main clinical trials that used ATRA-ATO regimens without chemotherapy in newly diagnosed APL, no case of secondary AML/MDS has been reported so far (Lo Coco and Russell, personal communications). Thus, in standard-risk APL, and in spite of very high complete remission and very low relapse rates obtained with ATRA chemotherapy combinations, our results confirm that the rates can be further improved by using ATO during the consolidation regimen. ATO in this situation did indeed reduce the relapse risk in standard-risk APL, confirming results of two recent, large studies.10,11 Long-term results of one of them, the Italian German study, show in particular that an ATRA-ATO regimen is not just equivalent but superior to ATRA chemotherapy regimens in terms of relapse rate and overall survival. Thus, ATRA-ATO (chemotherapy free) regimens are becoming reference treatments for standard-risk APL. With regards to high-risk APL, only limited studies of ATO-ATRA regimens without chemotherapy have been published, and in those studies patients often also received

References 1. Sanz MA, Grimwade D, Tallman MS, et al. Management of acute promyelocytic leukemia: recommendations from an expert panel on behalf of the European LeukemiaNet. Blood. 2009;113(9):1875– 1891. 2. Lo-Coco F, Avvisati G, Vignetti M, et al. Front-line treatment of acute promyelocytic leukemia with AIDA induction followed by risk-adapted consolidation for adults younger than 61 years: results of the AIDA2000 trial of the GIMEMA group. Blood.

2038

myelosuppressive drugs, mainly gentuzumab.10,12 In the British study, this approach was found to give results equivalent to those of an ATRA chemotherapy regimen, but the overall number of patients included in the randomized study was only 56.10 A US intergroup study showed that addition of ATO to a classical ATRA chemotherapy regimen significantly reduced the relapse rate. The ATRA chemotherapy regimen was, however based on daunorubicin instead of idarubicin (with a total scheduled dose of 500 mg/m2), which may have contributed to higher relapse rates. In the present study, among the patients with high-risk APL there was a very high complete remission rate (97.4%) and, contrary to the US intergroup study, a very low relapse rate (2.5%) was seen in the chemotherapy consolidation arm (without ATO), confirming our previous results.23 The fact that substituting ATO for AraC was not associated with an increased incidence of relapse (5.3% versus 2.7%), but with a reduced incidence of deaths in complete remission (from 7.8% to 0%) was, therefore, an important finding. This substitution also lead to less myelosuppression and less hospitalization for consolidation cycles. By contrast, the chemotherapy plus ATO consolidation therapy, combining AraC and ATO, used during the first part of the trial, did not further reduce the relapse rate (which was, it should be noted, already very low in the conventional AraC arm) but was associated with increased myelosuppression and a 5% rate of deaths in complete remission. This finding supports the fact that ATO worsens myelosuppression when used concomitantly with chemotherapy, as in the standard-risk group. Our results therefore support the addition of ATO during consolidation cycles, in high-risk APL, at least in order to reduce the amount of chemotherapy administered and, therefore, the rate of deaths in complete remission (as in our study) but also the relapse rate (according to other studies, including the US intergroup study). While ATO-ATRA regimens without chemotherapy can now probably be substituted for ATRA chemotherapy regimens in standard-risk APL, ongoing clinical trials will show to what extent chemotherapy can also be reduced or even avoided in highrisk APL. Funding This study was supported by the programme Hospitalier de Recherche Clinique and the Association pour la Recherche sur le Cancer (ARC).

2010;116(17):3171–3179. 3. Adès L, Chevret S, Raffoux E, et al. Longterm follow-up of European APL 2000 trial, evaluating the role of cytarabine combined with ATRA and daunorubicin in the treatment of nonelderly APL patients. Am J Hematol. 2013;88(7):556–559. 4. Sanz MA, Montesinos P, Rayón C, et al. Risk-adapted treatment of acute promyelocytic leukemia based on all-trans retinoic acid and anthracycline with addition of cytarabine in consolidation therapy for highrisk patients: further improvements in treatment outcome. Blood. 2010;115(25): 5137– 5146.

5. Fenaux P, Chastang C, Chevret S, et al. A randomized comparison of all transretinoic acid (ATRA) followed by chemotherapy and ATRA plus chemotherapy and the role of maintenance therapy in newly diagnosed acute promyelocytic leukemia. The European APL group. Blood. 1999;94(4): 1192–1200. 6. Adès L, Guerci A, Raffoux E, et al. Very longterm outcome of acute promyelocytic leukemia after treatment with all-trans retinoic acid and chemotherapy: the European APL group experience. Blood. 2010;115(9):1690–1696. 7. Muchtar E, Vidal L, Ram R, Gafter-Gvili A,

haematologica | 2018; 103(12)

Role of arsenic trioxide in APL

10.

11.

12.

Shpilberg O, Raanani P. The role of maintenance therapy in acute promyelocytic leukemia in the first complete remission. Cochrane Database Syst Rev. 2013;(3): CD009594. Adès L, Sanz MA, Chevret S, et al. Treatment of newly diagnosed acute promyelocytic leukemia (APL): a comparison of French-Belgian-Swiss and PETHEMA results. Blood. 2008;111(3):1078–1084. Lo-Coco F, Avvisati G, Vignetti M, et al. Retinoic acid and arsenic trioxide for acute promyelocytic leukemia. N Engl J Med. 2013;369(2):111–121. Burnett AK, Russell NH, Hills RK, et al. Arsenic trioxide and all-trans retinoic acid treatment for acute promyelocytic leukaemia in all risk groups (AML17): results of a randomised, controlled, phase 3 trial. Lancet Oncol. 2015;16(13):1295–1305. Platzbecker U, Avvisati G, Cicconi L, et al. Improved outcomes with retinoic acid and arsenic trioxide compared with retinoic acid and chemotherapy in non-high-risk acute promyelocytic leukemia: final results of the randomized Italian-German APL0406 Trial. J Clin Oncol. 2017;35(6):605–612. Abaza Y, Kantarjian H, Garcia-Manero G, et al. Long-term outcome of acute promyelo-

haematologica | 2018; 103(12)

13. 14.

15.

16.

17.

18.

cytic leukemia treated with all-trans-retinoic acid, arsenic trioxide, and gemtuzumab. Blood. 2017;129(10):1275–1283. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53457–53481. Fine J, Gray R. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446)496– 509. Rahmé R, Thomas X, Recher C, et al. Early death in acute promyelocytic leukemia (APL) in French centers: a multicenter study in 399 patients. Leukemia. 2014;28(12): 2422–2424. Lehmann S, Ravn A, Carlsson L, et al. Continuing high early death rate in acute promyelocytic leukemia: a populationbased report from the Swedish Adult Acute Leukemia Registry. Leukemia. 2011;25(7): 1128–1134. McClellan JS, Kohrt HE, Coutre S, et al. Treatment advances have not improved the early death rate in acute promyelocytic leukemia. Haematologica. 2012;97(1):133– 136. Altman JK, Rademaker A, Cull E, et al. Administration of ATRA to newly diagnosed patients with acute promyelocytic

19.

20.

21.

22.

23.

leukemia is delayed contributing to early hemorrhagic death. Leuk Res. 2013;37(9): 1004–1009. Paulson K, Serebrin A, Lambert P, et al. Acute promyelocytic leukaemia is characterized by stable incidence and improved survival that is restricted to patients managed in leukaemia referral centres: a pan-Canadian epidemiological study. Br J Haematol. 2014;166(5):660–666. Batzios C, Hayes LA, He SZ, et al. Secondary clonal cytogenetic abnormalities following successful treatment of acute promyelocytic leukemia. Am J Hematol. 2009;84(11):715–719. Lobe I, Rigal-Huguet F, Vekhoff A, et al. Myelodysplastic syndrome after acute promyelocytic leukemia: the European APL group experience. Leukemia. 2003;17(8): 1600–1604. Andersen MK, Pedersen-Bjergaard J. Therapy-related MDS and AML in acute promyelocytic leukemia. Blood. 2002;100 (5):1928–1929. Kelaidi C, Chevret S, De Botton S, et al. Improved outcome of acute promyelocytic leukemia with high WBC counts over the last 15 years: the European APL Group experience. J Clin Oncol. 2009;27(16):2668–2676.

2039

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2040-2048

Physician uncertainty aversion impacts medical decision making for older patients with acute myeloid leukemia: results of a national survey

Pierre Bories,1,2 Sébastien Lamy,3,4 Célestine Simand,5 Sarah Bertoli,2 Cyrille Delpierre,3 Sandra Malak,6 Luc Fornecker,5 Stéphane Moreau,7 Christian Récher2 and Antoine Nebout8 Regional Cancer Network Onco-Occitanie, Toulouse University Institute of CancerOncopole; 2Department of Hematology, Toulouse University Institute of CancerOncopole; 3INSERM Unit 1027, Faculty of Medicine, Toulouse; 4Department of Clinical Pharmacology, Toulouse University Hospital; 5Department of Hematology, Strasbourg University Hospital; 6Department of Hematology, Rene Huguenin Hospital, Curie Institute, Saint-Cloud; 7Department of Hematology, Limoges University Hospital and 8 INRA, UR 1303 ALISS, Ivry-sur-Seine, France 1

ABSTRACT

Correspondence: pierre.bories@onco-occitanie.fr

Received: March 2, 2018. Accepted: July 12, 2018. Pre-published: July 13, 2018. doi:10.3324/haematol.2018.192468 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2040 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2040

lderly patients with acute myeloid leukemia can be treated with intensive chemotherapy, low-intensity therapy such as low-dose aracytine or hypomethylating agents, or best supportive care. The choice between these treatments is a function of many patient-related and disease-related factors. We investigated how physicians’ behavioral characteristics affect medical decision-making between intensive and non-intensive therapy in this setting. A nationwide cross-sectional online survey of hematologists collected data on medical decision-making for 6 clinical vignettes involving older acute myeloid leukemia patients that were representative of routine practice. Questionnaires elicited physicians’ demographic and occupational characteristics along with their individual behavioral characteristics according to a decision theory framework. From the pattern of responses to the vignettes, a Kmeans clustering algorithm was used to distinguish those who were likely to prescribe more intensive therapy and those who were likely to prescribe less intensive or no therapy. Multivariate analyses were used to identify physician’s characteristics predictive of medical decision-making. We obtained 230 assessable answers, which represented an adjusted response rate of 45.4%. A multivariate model (n=210) revealed that physicians averse to uncertainty recommend significantly more intensive chemotherapy: Odds Ratio (OR) [95% Confidence Interval (CI)]: 1.15 [1.01;1.30]; P=0.039. Male physicians who do not conform to the expected utility model (assumed as economically irrational) recommend more intensive chemotherapy [OR (95% CI) = 3.45 (1.34; 8.85); P=0.01]. Patient volume per physician also correlated with therapy intensity [OR (95% CI)=0.98 (0.96; 0.99); P=0.032]. The physicians’ medical decisionmaking was not affected by their age, years of experience, or hospital facility. The significant association between medical decision and individual behavioral characteristics of the physician identifies a novel nonbiological factor that may affect acute myeloid leukemia patients’ outcomes and explain variations in clinical practice. It should also encourage the use of validated predictive models and the description of novel biomarkers to best select patients for intensive chemotherapy or low-intensity therapy. haematologica | 2018; 103(12)

Dealing with treatment uncertainty in elderly AML patients

Introduction Outside clinical trials, therapy options offered to elderly acute myeloid leukemia (AML) patients are limited.1,2 They can be summarized as intensive chemotherapy (IC), low-intensity therapy (LIT) or best supportive care (BSC) depending on patient-specific3 and AML-related4 prognostic factors. Although scoring systems have been proposed to rationalize the medical decision-making (MDM) between intensive and non-intensive approaches,5,6 large variations in clinical practice remain,7 which underlines the paucity of evidence supporting medical decisions. International guidelines define available intensive or lowintensity options, but in most cases they give the physician the responsibility of determining which option should be recommended for a particular patient. The AZA-AML-0018 and DACO 169 phase III studies failed to demonstrate the superiority of hypomethylating agents (azacitidine and decitabine, respectively) over conventional chemotherapy for patients over 65 years of age with non-proliferative AML, which increases the uncertainty regarding the optimum strategy for any individual patient. In the AZA-AML-001 trial, only 18% of patients were allocated to AZA versus IC as compared to 82% to AZA versus low-dose aracytine (LDAC) or BSC, suggesting that physicians’ decisions were already biased toward LIT. In addition, when physician-investigators in the UK National Cancer Research Institute’s AML-14 trial10,11 were offered the possibility of an optional randomization between intensive and non-intensive therapy, they preferred to bypass this randomization and allocate their patients directly into the intensive or non-intensive arms. Multivariate comparison of the characteristics of the patients treated intensively or non-intensively in this AML trial revealed that the physician was a strong determinant of the choice, which clearly demonstrates a physician effect in this setting. Previous studies have investigated the impact of physician’s professional characteristics on their decision-making for hematologic malignancies, particularly in the setting of allogeneic stem cell transplantation,12,13 but these studies mostly focused on age, specialty, and hospital facility. Much less attention has been given to an individual physician’s non-professional characteristics. Uncertainty is a crucial, multifaceted component of the therapeutic decision for older patients with AML.14 Intensive chemotherapy offers the greatest chance of complete remission (CR) but is associated with a significant risk of early death (ED), while hypomethylating agents yield a lower chance of CR but lower risk of ED. Thus, for physicians treating an older AML patient, uncertainty is a pre-condition for the decision itself, which underscores the need to investigate how physicians deal with it.15,16 In behavioral economics and decision sciences, attitudes towards risk and uncertainty are crucial psychological traits that may explain medical choices and practices.17,18 They aim to describe individual decisions in situations where choices have uncertain consequences. Risk- or uncertainty-averse individuals prefer a safer option (with greater chances of a smaller gain) than risk- or uncertainty-seeking individuals who will choose a riskier option (with lower chances of a larger gain). The difference between risk and uncertainty is that, for an uncertain option, the probability of success (or gain) is unknown. In economics, the gold standard of rationality is the expected haematologica | 2018; 103(12)

utility model19 (EU) although much experimental evidence in behavioral sciences shows departures from EU.20 The Allais paradoxes21,22 are decision tasks used to classify individuals as conforming to EU or not (non-EU). The most popular non-EU model is prospect theory.23 Investigating the association between practice variations in AML therapy and physician’s behavioral characteristics (such as risk or uncertainty aversion) and types (EU vs. non–EU) may help define new determinants of these variations and to propose corrective measures to improve the quality of care.24,25 We hypothesized that individual physicians’ attitudes towards risk and uncertainty have an impact on their decision-making process for elderly patients with AML.

Methods Survey design We conducted a national cross-sectional online survey of French hematologists to evaluate the impact of demographic, occupational and behavioral characteristics on medical decision-making for selected clinical cases of older patients with AML presented as clinical vignettes. As compared with other tools such as chart abstraction or standardized patients, clinical vignettes have been validated as a simple case-mix adjusted method for measuring quality of care and practice variations.26 All the hematologistoncologists practising in France who provide direct patient care for adults with AML were eligible. A first draft of the questionnaire was developed and subsequently modified after pilot testing with 20 hematologists. Overall, the survey contained 27 questions and took 10-15 minutes to complete. The questionnaire is available in Online Supplementary Appendix Section I.

Survey instrument Physician’s demographic and occupational characteristics included age, gender, medical specialty, subspecialty, hospital facility, hierarchical position, year of graduation, patient volume (number of AML patients aged 60 years or older each physician treated annually), and self-evaluation of expertise in the field of AML. Four hypothetical AML patients aged 60 years or older were selected as representative of clinical practice and were summarized by 3 local specialists (PB, SB and CR) as Vignettes #1 to #4 (Table 1). Each of these cases highlighted distinct and difficult representative situations regarding their age, comorbidity, family environment or AML biology. Vignettes #5 and #6 were similar to Vignette #4 but included a unique variation related to age (increased from 63 to 73-years old in Vignette #5) or white blood cell (WBC) count (increased from 2.5 to 40 x109/L in Vignette #6). For each of these 6 vignettes, the close-ended treatment options were: 1) intensive chemotherapy; 2) low-intensity therapy; or 3) best supportive care. To measure physicians’ attitudes towards risk and uncertainty, we used four different elicitation methods (Figure 1) that have been validated in representative national surveys.27 The first two measures are certainty equivalent elicitation and the third one consisted of two binary lottery choices. These tasks involve risky choices with financial consequences. The fourth method is a Likert scale that measures willingness to take risks in four different domains (Online Supplementary Appendix Section II).

Survey implementation The Ethics committee of the French Society of Hematology approved the study and provided an incentive email accompanying the online survey invitation. Physicians identified from the French 2041

P. Bories et al. Table 1. Clinical Vignettes of older AML patients derived from real life activity. General instructions • Six clinical cases of AML patients derived from real life activity are presented. • You are not alone to decide but we are asking you to state which treatment option would you recommend for each of these patients among: 1. Intensive chemotherapy 2. Low-intensity therapy (hypomethylating agent or low-dose cytarabine) 3. Best supportive care • These patients have announced they would accept medical treatment decision • You do not have any clinical trial to offer them. • You have unlimited possibilities of hospitalization as inpatient or outpatient Vignette#1: A 72-year old female, with no comorbidity. Normal cardiac function. Untreated low-risk myelodysplastic syndrome for 3 years (refractory anemia, IPSS 0.5). CBC: WBC 1x109/L incl. neutrophil count 0.3x109/L and PB count 5%, Hb 11g/dL, platelets 120x109/L. BMA: FAB1 AML with 40% marrow blast infiltrate, and adverse karyotype (monosomy 7). Vignette#2: A 75-year old male, coronary artery disease with anterior interventricular artery stenting in 2010, controlled ischemic cardiopathy with medication (LVEF 52%), ECOG 2, recent weight loss 4 kg. CBC: WBC count 75x109/L, PB blast count 40%, Hb 10 g/dL, platelets 50x109/L. BMA: FAB2 AML (marrow blast infiltrate 60%) with normal karyotype. Vignette#3: A 77-year old female, with an 8-year history of hypertension controlled with angiotensin-converting-enzyme inhibitor, a recent echocardiogram showed LVEF of 55%. She is natural carer of her husband affected by Alzheimer’s disease. CBC: WBC 18x109/L incl. 25% peripheral blast, Hb 10g/dL, platelets 80x109/L, BMA: FAB4 AML with favorable karyotype (inv16). Vignette#4: A 63-year old male, with a 5-year history of asymptomatic Parkinson disease and recently diagnosed with an asymptomatic carotid artery stenosis (90%). CBC: WBC 2x109/L incl. 5% PB blast count, Hb 8g/dL, platelets 35x109/L. BMA: FAB2 AML (marrow blast infiltrate 30%, tri-lineage dysplasia) with complex Karyotype incl. inv3, -5q, -7. Vignette#5: Patient from the Vignette#4 but 73-year old. Vignette#6: Patient from the Vignette#4 but with WBC count 40x109/L incl. PB blast count 25%. IPSS: International Prognosis Scoring System; CBC: complete blood count; WBC: white blood cell count; Hb: hemoglobin; FAB: French-American-British classification system; BMA: bone marrow aspirate; PB peripheral blast; LVEF: left ventricular ejection fraction; ECOG: Eastern Cooperative Oncology Group Performance Status.

Society of Hematology mailing list received a unique link to enter the survey or opt out. After duplicate names were removed, the panel of potentially eligible subjects contained 1337 physicians, including 220 residents with an email address. On November 30th 2015 we emailed the survey link; non-responders received three subsequent reminders every eight weeks. Before entering the survey, physicians were informed they would not be compensated for their participation. Consent was implied based on reading the survey goals and participating. Assessable respondents included those who answered the 6 vignettes. Data were collected from November 30th 2015 to June 6th 2016, and analyzed at the Toulouse University Cancer Institute and Toulouse Faculty of Medicine.

vignettes. The results of this score and its association with the Kmeans clustering are given in Online Supplementary Table S1. We first tested the physicians’ demographic, occupational and behavioral characteristics associated with belonging to the IC group in bivariate analyses at the threshold of 0.2. These variables were then included in a multivariate model systematically adjusted for age and gender. From this step, the variables to keep in the final parsimonious model were determined using a stepwise backward selection based on log-likelihood tests between nested models. All analyses were made using STATA release 14 (Stata Corp LP, College Station, TX, USA).

Statistical analysis

Results

We described physicians’ characteristics using counts and frequency for qualitative data, and mean and standard deviation for quantitative data. To assess the clinician pattern of decision-making for the 6 clinical vignettes, we used K-means clustering to define clinician groups with homogeneous patterns of responses to the clinical cases.28 The aim of this method was to define clusters of subjects by maximizing the between-cluster differences in the subjects’ medical choices and by minimizing the within-cluster differences in subjects’ medical choices. This allowed us to define two clusters: clinicians more likely to choose intensive chemotherapy (IC), i.e. the “intensive treatment group” (IC group), and those who were more likely to choose less intensive therapy, i.e. the “nonintensive treatment group” (Non-IC group). Only the K-means analyses are presented in this paper. A 6-18 point MDM-score was also calculated for each physician, assigning 1 point for IC, 2 points for LIT, and 3 points for BSC from the responses in the 6 2042

Of the 1337 invitations sent out to complete the questionnaire, 1295 were eligible and 230 completed questionnaires were received (response rate: 17.20%). According to the American Association of Public Opinion Reporting standard definitions, and by taking the characteristics of the mailing list and the interrogated population into account, the adjusted response rate was 45.4% (see Online Supplementary Table S2). No differences were found according to gender or by geographical area between respondents and non-respondents (see Online Supplementary Table S3).

Respondents’ demographic and occupational characteristics The median age of the respondent cohort was 42 years [standard deviation (SD)±11.2], 123 were male (54%), 160 haematologica | 2018; 103(12)

Dealing with treatment uncertainty in elderly AML patients

Figure 1. Behavioral tasks. (A) Physician’s individual risk aversion evaluation. The closer the scroll bar is to 500 euros, the more risk-seeking the behavior; the lower the score bar, the greater the aversion to risk. E.g. if the scroll bar is at 200 euros, the person prefers a 50% chance of winning 500 euros to a 100% chance of winning 190 euros. If the scroll bar is at 300 Euros, the person prefers a 50% chance of winning 500 euros to a 100% chance of winning 290 euros. The latter is riskier since you are giving up more certain money (290 vs. 190) for a chance to win the same amount (500 euros). (B) Physician’s individual uncertainty aversion evaluation. The same line of reasoning applies to the uncertainty aversion evaluation except that for option A, the probability of gain is unknown. The closer the scroll bar is to 500 euros the more uncertainty-seeking the behavior; the lower the scroll bar, the greater the aversion to uncertainty. (C) Classic binary choices from Kahneman and Tversky. Choice patterns AC and BD conform to the expected utility theory. Choice patterns AD and BC do not conform to expected utility theory (for further details see Online Supplementary Appendix, Section 2). (D) Self-evaluation of the willingness to take risk in four different domains.

were attending physicians or professors (70%), 166 worked in an academic center (72%), 197 were specialized in hematology (86%), and the mean patient volume per physician was 20.7 (SD±17.1).

Medical decision-making among clinical vignettes The physicians’ decisions about the 6 clinical vignettes assessing front-line therapy for older AML patient are summarized in Figure 2. The most controversial case was Vignette #4 for which 50.8% of respondents recommendhaematologica | 2018; 103(12)

ed IC versus 49.2% opting for a non-intensive approach. Increasing the age of this patient in Vignette #5 resulted in a marked decrease (from 50.8% to 6.9%) in the proportion of respondents choosing IC. Alternatively, increasing the WBC in Vignette #6 increased the proportion of physicians who recommended an IC from 50.8% to 64.7%. These practice variations induced by modifying classical AML prognosis factors were expected, and they thus provide internal quality control of non-random responses to the online survey (internal coherence criteria). 2043

P. Bories et al.

Aversion towards risk and uncertainty The mean certainty equivalent under risk was 277 euros (SD±130). This was significantly different from the expected value of the lottery (250 euros), revealing global risk-seeking among our sample (unpaired t-test, P=0.03).

Under uncertainty conditions, the mean certainty equivalent was 241 euros (SD±136). This was significantly lower than under risk (paired t-test, P<0.001) and reveals an overall ambiguity aversion among our sample. More specifically, of the 212 respondents, 110 are ambiguity averse, 52

Table 2. Panel Description (n=230), and Bivariate Comparison of the Physicians from IC and non-IC Groups.

Overall sample n Demographical characteristics Gender Men Women Age (n=230) (mean± SD) Occupational characteristics Clinical speciality Other than hematology Hematology Workplace Non-academic centers Academic centers Region North East West South-west South Méditerranée Rhone-Alpes/Auvergne Ile de France Status Few or no decision-making role Decision-making role Occupational experience (n=210) (means ± SD) Activity in AML pts aged > 60 / year (n=210)(means ± SD) Behavioral characteristics Attitude towards uncertainty (n=213) (means ± SD) Attitude towards risks (n=212) (means ± SD) Expected utility yes (n=101) no (n=109) General attitude towards the risk (means ± SD) regarding their personal life (n=221) regarding their money (n=220) regarding their patients’ health (n=220) regarding their own health (n=219) Responses to the vignettes MDM score (mean ± SD)

Non-IC group

Global P-value of the bivariate analysis of the characteristics related with belonging to IC groupa

123 53 107 47 42.0 ±11.2

31 44 39 56 42.3 ±11.1

92 68

57 43 41.8 ±11.3

64 166

28 72

18 52

26 74

46 114

29 71

0.636

64 166

28 72

18 52

26 74

46 114

29 71

0.636

34 38 30 45 27 24 32

15 17 13 20 12 10 14

6 8 11 13 9 9 14

9 11 16 19 13 13 20

28 30 19 32 18 15 18

18 19 12 20 11 9 11

0.227

70 30 160 70 16.9±10.9

23 47

33 67

0.597

17.3±10.8

47 29 113 71 16.8±11

0.772

20.7±17.1

24.5±18.5

19.1±16.2

0.034

242±136

276±149

228±127

0.031b

277±130

296±150

269±120

0.131b

101 109

48 52

IC group

36 27

57 43

65 82

44 56

0.064 0.792

0.086

5.1±2.1 4.1±2.4 5.3±2.1 5.7±2.1

4.8±2.1 4.1±2.6 5.2±2.4 5.5±2.2

5.2±2.1 4.2±2.3 5.3±2.0 5.8±2.0

0.194 0.922 0.630 0.243

10.2±1.6

11.7±1.4

9.6±1.3

<0.001

SD: Standard Deviation; pts: patients; MDM: medical decision-making. Studenta or Wilcoxonb tests for continuous variables; χ2 test for categorial variables.

2044

haematologica | 2018; 103(12)

Dealing with treatment uncertainty in elderly AML patients

are ambiguity neutral, and 50 are ambiguity seeking. Regarding the Allais paradox, the AC, BD, AD and BC choice patterns were found in 90 (42.9%), 11 (5.2%), 102 (48.6%), and 7 (3.3%) assessable subjects, respectively, which represents 101 EU respondents (48.1%) and 109 non-EU respondents (51.9%). Mean self-reported willingness to take risks is 4.1 (SD±2.4) in the financial domain, 5.1 (SD±2.1) in their personal life, 5.3 (SD±2.1) for a patient’s health, and 5.7 (SD±2.1) for their own health.

higher patient volume increases the probability of being in the non-IC group [OR (95%CI): 0.98 (0.96;0.99); P=0.032]. We found an interaction between gender and the Allais paradox resulting in a statistically significant increase in the probability of being pro-IC among men who do not conform to the Expected Utility model [OR (95%CI): 3.45 (1.34;8.85); P=0.01], but such an effect was not found among women.

K-means clustering identifies two populations of physicians

Discussion

From the pattern of responses to the 6 clinical vignettes, the K-means clustering (see Online Supplementary Appendix Section III for details) allowed us to separate two groups of physicians: one group of clinicians with lower MDMscores, i.e. more likely to choose intensive therapy (IC group), and another group of clinicians with higher MDM-scores, i.e. more likely to choose non-intensive therapy (Non-IC group).

In this cross-sectional national survey, we evaluated the impact of physicians’ behavioral characteristics on their medical decision-making in older patients with AML. We hypothesized that physicians’ behavioral traits such as risk and uncertainty aversion or rationality could be correlated with their choice between intensive and less-intensive therapy. To our knowledge, this is the first evidence that physicians belonging to the uncertainty-tolerant group recommend IC significantly less often than uncertainty-averse physicians, and that male physicians considered as “economically irrational” prescribe more IC. Several non-biological factors (NBF) are known to be associated with a patient’s health-related outcomes such as socio-economic status (SES), area of residence or marital status.29,30 In the spectrum of NBFs affecting cancer patient outcomes, physician’s characteristics have been described as therapy determinants in the setting of allogeneic stem cell transplantation for hematologic malignancies12 and solid tumors.31 In our study, neither age, hierarchical status or years of experience influenced the tendency of physicians to belong to the IC or Non-IC group, while individual uncertainty aversion was a strong determinant of practice variations in multivariate analysis. Volume-outcome relationships at treatment facility level32,33 and at physician level34 are well described with NBF affecting the outcome of patients with cancer. It is worthy of note that our study gets behind the volumeoutcome relationship in AML, while connecting physician’s patient volume with medical decision-making and more precisely with therapy intensity. Verma et al.35 stated that physicians should learn about the individual behavioral mechanisms underpinning choices under uncertainty. Our findings go one step fur-

IC-physicians harbor specific behavioral characteristics compared with non-IC-physicians A bivariate comparison of the characteristics of physicians in the IC and Non-IC groups is summarized in Table 2. We detected significantly more aversion toward uncertainty within the IC cluster (mean certainty equivalents of 228 euros for the Non-IC group vs. 276 euros for the IC group; P=0.031) (Table 2). For the Allais paradox, we detected a trend toward significance (P=0.086) with more EU subjects in the Non-IC group (57%) than in the IC group (44%). Among the demographical and occupational characteristics, only the patient volume was associated with these clusters, with a mean number of older AML patients treated annually of 19.1 in the IC group versus 24.5 in the non-IC group (P=0.03). Although we found more male and risk-averse physicians in the IC group, these differences were not significant (P=0.06 and P=0.131, respectively). The logistical regression of the IC versus non-IC groups, on individual characteristics that were significant with a P-value <0.20 in the bivariate analysis, is presented in Table 3. This analysis confirmed that aversion towards uncertainty increases the probability of belonging to the IC group [OR (95%CI): 1.17 (1.01;1.37); P=0.043] and that

Figure 2. Medical decision-making among the 6 clinical vignettes. Proportion of physicians choosing intensive chemotherapy, low-intensity therapy or best supportive care for each of the 6 clinical vignettes.

haematologica | 2018; 103(12)

2045

P. Bories et al. Table 3. Characteristics associated with belonging to the intensive care (IC) group. Results from the multivariate logistic regression among the 210 clinicians for whom complete data of variable selected based on the bivariate analysis.* Age (per additional year) Aversion towards risks Aversion towards uncertainty (for each 50 euro decrease) General attitude towards the risk regarding personal life Activity in AML pts. ≥60y/year Gender among rational Men Women Expected utility among men Yes No Interaction term = difference between irrational effect among women and men

[95% Confidence Interval]

1.00 1.00 1.17 1.10 0.98

[0.97 ; 1.03] [0.99 ; 1.01] [1.01 ; 1.37] [0.95 ; 1.29] [0.96 ; 0.99]

0.757 0.875 0.043 0.208 0.032

ref. 0.93

[0.39 ; 2.20]

0.865

ref. 3.45 0.253

[1.34 ; 8.85] [0.07 ; 0.91]

0.01 0.035

OR: Odds Ratio; pts: patients; AML: acute myeloid leukemia; ref: reference value. *Results of the parsimonious model constructed from a backward-stepwise-procedure (see the Methods section for details) which was initially additionally adjusted for clinicians’ general attitude regarding their own health and aversion towards risks.

ther and show that a behavioral characteristic such as uncertainty aversion is directly correlated with the clinician’s therapeutic choice. We evaluated physician behavioral characteristics with tools validated in behavioral economics. Although such tasks may be incentivized (paid for real) in experimental economics, we decided to use hypothetical incentives following Kahneman and Tversky (1979) who claimed that: “the method of hypothetical choices emerges as the simplest procedure by which a large number of theoretical questions can be investigated. The use of the method relies on the assumption that people often know how they would behave in actual situations of choice, and on the further assumption that the subjects have no special reason to disguise their true preferences”. Concerns have been raised about the correlation between a clinician’s medical behavior and their uncertainty aversion as measured by economical tools in a nondomain-specific manner.36 Our study confirms, as previously reported,17,18 that economic behavior and its underlying psychological traits can predict medical behavior. Our cohort was globally risk-seeking and ambiguityaverse; 109 (51.9%) of the physicians did not conform with the EU model, which is in line with evidence in behavioral economics.37,38 Mean self-reported willingness to take risks was consistent with previous results but higher for the patient’s health domain.39 This finding may be explained by the toxicity related to the intensive therapy that physicians are used to prescribe. We detected an interaction between physician’s gender and the Allais paradox, with an impact of departures from the EU model on decision-making, in male physicians. Gender effect for risk and uncertainty attitudes is a well-established stylized fact in behavioral economics,40,41 albeit the impact of the interaction between gender and Allais paradox on MDM has, to our knowledge, never been documented empirically. Although our findings provide novel insights into the clinical debate pitting intensive versus low-intensity approaches for older patients with AML, we acknowledge that our study has limitations. The respondent panel was 2046

representative of the surveyed French hematologist population in terms of gender, hierarchical status and geographical area. Respondents more often belonged to academic centers than surveyed physicians, which can be explained by the French healthcare system’s organization for AML patients usually being oriented towards academic centers. Physicians were asked to recommend how to treat an AML patient in an experimental framework. We deliberately proposed clinical situations where patients had announced they would accept medical decisions, and patient choices did not appear in the vignettes. Since informed decision-making has emerged as the new normative standard for health care,42 concerns about the increase in complexity provided by this mode of decision making have been raised.43 To encompass this increase in complexity, and presumably of uncertainty, physicians were asked to state which therapy they would ideally recommend, irrespective of the patient’s choice. We also evaluated individual clinician’s choices, whereas multidisciplinary team (MDT) decision-making is currently the standard of care in cancer. Even though a treatment plan devised by an MDT may differ from that of a single physician, it is noteworthy that the MDT constitutes an area of exchange between healthcare professionals where they may clearly state which treatment they consider to be appropriate in any clinical situation. Another limitation is the construction of the clinical vignettes. We focused the proposed treatments on intensive, low-intensity therapy and best supportive care, and did not propose any investigational drug or therapeutic strategy through clinical trial enrollment. We anticipated this would have swayed the physicians’ answers in favor of trial participation. Even though clinical trial enrollment remains an ideal scenario for all AML patients, real-life data provided by the Netherland’s registry show that only a small number of patients over 60-years of age could benefit from such innovative strategy.44 We did not provide any potentially druggable molecular markers such as FLT3-ITD, NPM1 or IDH1/2 in the clinical scenarios, because, to date, a large proportion of French centers do not have access to these haematologica | 2018; 103(12)

Dealing with treatment uncertainty in elderly AML patients

markers during front-line therapeutic decision-making. The increasing use of next generation sequencing (NGS) technologies in AML will soon allow the identification of molecular markers in almost all patients, which will have various consequences. For patients with an actionable molecular marker,45 or with accurate genomics-based outcome prediction,46 NGS technologies will presumably reduce treatment uncertainty. Alternatively, for patients with a non-actionable marker or markers with unknown prognostic significance, NGS will likely add another level of uncertainty. Even though the MDM process cannot be

References 1. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017; 129(4):424-447. 2. Bories P, Bertoli S, Bérard E, et al. Intensive chemotherapy, azacitidine, or supportive care in older acute myeloid leukemia patients: an analysis from a regional healthcare network. Am J Hematol. 2014;89(12):E244-252. 3. Klepin HD, Rao AV, Pardee TS. Acute myeloid leukemia and myelodysplastic syndromes in older adults. J Clin Oncol. 2014;32(24):2541-2552. 4. Tsai C-H, Hou H-A, Tang J-L, et al. Genetic alterations and their clinical implications in older patients with acute myeloid leukemia. Leukemia. 2016; 30(7):14851492. 5. Krug U, Röllig C, Koschmieder A, et al. Complete remission and early death after intensive chemotherapy in patients aged 60 years or older with acute myeloid leukaemia: a web-based application for prediction of outcomes. Lancet. 2010; 376(9757):2000-2008. 6. Klepin HD, Geiger AM, Tooze JA, et al. Geriatric assessment predicts survival for older adults receiving induction chemotherapy for acute myelogenous leukemia. Blood. 2013;121(21):4287-4294. 7. Loberiza FR, Cannon AC, Cannon AJ, Bierman PJ. Insights on practice variations in the management of lymphoma and leukemia. Leuk Lymphoma. 2014; 55(11):2449-2456. 8. Dombret H, Seymour JF, Butrym A, et al. International phase 3 study of azacitidine vs conventional care regimens in older patients with newly diagnosed AML with >30% blasts. Blood. 2015;126(3):291-299. 9. Kantarjian HM, Thomas XG, Dmoszynska A, et al. Multicenter, randomized, openlabel, phase III trial of decitabine versus patient choice, with physician advice, of either supportive care or low-dose cytarabine for the treatment of older patients with newly diagnosed acute myeloid leukemia. J Clin Oncol. 2012;30(21):26702677. 10. Burnett AK, Milligan D, Goldstone A, et al. The impact of dose escalation and resistance modulation in older patients with acute myeloid leukaemia and high risk myelodysplastic syndrome: the results of the LRF AML14 trial. Br J Haematol. 2009;145(3):318-332.

haematologica | 2018; 103(12)

restricted to a computational process, novel methods such as decision-making tools supported by knowledge banks of matched genomic-clinical data47 are warranted. They will help physicians absorb large amounts of complex information and likely act as moderators of uncertainty. Pending the validation of such tools in daily practice, our study (which found a strong physician-effect on treatment decisions) should encourage the use of validated prognostic scores to rationalize the decision-making process in this setting.48 It should also encourage further exploration of the role of physicians’ attitudes in decision-making.

11. Burnett AK, Milligan D, Prentice AG, et al. A comparison of low-dose cytarabine and hydroxyurea with or without all-trans retinoic acid for acute myeloid leukemia and high-risk myelodysplastic syndrome in patients not considered fit for intensive treatment. Cancer. 2007;109(6):1114-1124. 12. Lee SJ, Joffe S, Artz AS, et al. Individual physician practice variation in hematopoietic cell transplantation. J Clin Oncol. 2008; 26(13):2162-2170. 13. Lee SJ, Astigarraga CC, Eapen M, et al. Variation in supportive care practices in hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2008; 14(11):1231-1238. 14. Dhawale T, Steuten LM, Deeg HJ. Uncertainty of Physicians and Patients in Medical Decision Making. Biol Blood Marrow Transplant. 2017;23(6):865-869. 15. Hunter DJ. Uncertainty in the Era of Precision Medicine. N Engl J Med. 2016; 375(8):711-713. 16. Simpkin AL, Schwartzstein RM. Tolerating uncertainty - the next medical revolution? N Engl J Med. 2016; 375(18):1713-1715. 17. Michel-Lepage A, Ventelou B, Nebout A, Verger P, Pulcini C. Cross-sectional survey: risk-averse French GPs use more rapidantigen diagnostic tests in tonsillitis in children. BMJ Open. 2013;3(10):e003540. 18. Massin S, Ventelou B, Nebout A, Verger P, Pulcini C. Cross-sectional survey: riskaverse French general practitioners are more favorable toward influenza vaccination. Vaccine. 2015;33(5):610-614. 19. von Neumann J, Morgenstern O. Theory of Games and Economic Behavior. 2nd Ed. Princeton University Press; 1947. 20. Starmer C. Developments in Non-expected Utility Theory: The Hunt for a Descriptive Theory of Choice under Risk. J Econ Lit. 2000;38(2):332-382. 21. Kahneman D, Tversky A. Prospect Theory: An Analysis of Decision under Risk. Econometrica. 1979;47(2):263-291. 22. Allais M. Le Comportement de l’Homme Rationnel devant le Risque: Critique des Postulats et Axiomes de l’Ecole Americaine. Econometrica. 1953;21(4):503-546. 23. Wakker PP. Prospect Theory: For Risk and Ambiguity. 1st ed. Cambridge University Press; 2010. 24. Lipitz-Snyderman A, Sima CS, Atoria CL, et al. Physician-Driven Variation in Nonrecommended Services Among Older Adults Diagnosed With Cancer. JAMA Intern Med. 2016;176(10):1541-1548. 25. Sikkens JJ, van Agtmael MA, Peters EJG, et

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

al. Behavioral Approach to Appropriate Antimicrobial Prescribing in Hospitals: The Dutch Unique Method for Antimicrobial Stewardship (DUMAS) Participatory Intervention Study. JAMA Intern Med. 2017;177(8):1130-1138. Peabody JW, Luck J, Glassman P, Dresselhaus TR, Lee M. Comparison of vignettes, standardized patients, and chart abstraction: a prospective validation study of 3 methods for measuring quality. JAMA. 2000;283(13):1715-1722. Dohmen T, Falk A, Huffman D, Sunde U, Schupp J, Wagner GG. Individual risk attitudes: Measurement, determinants, and behavioral consequences. J Eur Econ Assoc. 2011;9(3):522-550. Selim SZ, Ismail MA. K-means-type algorithms: a generalized convergence theorem and characterization of local optimality. IEEE Trans Pattern Anal Mach Intell. 1984; 6(1):81-87. Kristinsson SY, Derolf AR, Edgren G, Dickman PW, Björkholm M. Socioeconomic differences in patient survival are increasing for acute myeloid leukemia and multiple myeloma in sweden. J Clin Oncol. 2009;27(12):2073–2080. Borate UM, Mineishi S, Costa LJ. Nonbiological factors affecting survival in younger patients with acute myeloid leukemia. Cancer. 2015; 121(21):38773884. Luo R, Giordano SH, Zhang DD, Freeman J, Goodwin JS. The role of the surgeon in whether patients with lymph node-positive colon cancer see a medical oncologist. Cancer. 2007;109(5):975-982. Giri S, Pathak R, Aryal MR, Karmacharya P, Bhatt VR, Martin MG. Impact of hospital volume on outcomes of patients undergoing chemotherapy for acute myeloid leukemia: a matched cohort study. Blood. 2015;125(21):3359-3360. Go RS, Bartley AC, Crowson CS, et al. Association Between Treatment Facility Volume and Mortality of Patients With Multiple Myeloma. J Clin Oncol. 2017; 35(6):598-604. Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol. 2000;18(11):2327-2340. Verma AA, Razak F, Detsky AS. Understanding choice: why physicians should learn prospect theory. JAMA. 2014; 311(6):571-572. Han PKJ, Reeve BB, Moser RP, Klein WMP. Aversion to ambiguity regarding medical

2047

P. Bories et al.

37.

38.

39.

40.

2048

tests and treatments: measurement, prevalence, and relationship to sociodemographic factors. J Health Commun. 2009;14(6): 556-572. Nebout A, Dubois D. When Allais meets Ulysses: Dynamic axioms and the common ratio effect. J Risk Uncertain. 2014;48(1):19-49. Trautmann ST, van de Kuilen G. Ambiguity Attitudes. In: : G. Keren and G. Wu, editors. The Wiley Blackwell Handbook of Judgment and Decision Making. John Wiley & Sons, Ltd. 2015. p.89-116. Nebout A, Cavillon M, Ventelou B. Comparing GPsâ&#x20AC;&#x2122; risk attitudes for their own health and for their patientsâ&#x20AC;&#x2122; : a troubling discrepancy? BMC Health Serv Res. 2018;18(1):283. Eckel CC and Grossman PJ. Men, Women and Risk Aversion: Experimental Evidence.

41. 42.

43.

44.

In: Plott, C. and Smith, V., Editors. Handbook of Experimental Economics Results. Elsevier. 2008. p. 1061-1073. Croson R, Gneezy U. Gender Differences in Preferences. J Econ Lit. 2009; 47(2):448474. Gattellari M, Voigt KJ, Butow PN, Tattersall MHN. When the treatment goal is not cure: are cancer patients equipped to make informed decisions? J Clin Oncol. 2002; 20(2):503-513. Han PKJ, Klein WMP, Lehman T, Killam B, Massett H, Freedman AN. Communication of uncertainty regarding individualized cancer risk estimates: effects and influential factors. Med Decis Mak Int J Soc Med Decis Mak. 2011;31(2):354-366. Dinmohamed AG, Visser O, van Norden Y, et al. Treatment, trial participation and survival in adult acute myeloid leukemia: a population-based study in the

45.

46.

47.

48.

Netherlands, 1989-2012. Leukemia. 2016; 30(1):24-31. Medeiros BC, Fathi AT, DiNardo CD, Pollyea DA, Chan SM, Swords R. Isocitrate dehydrogenase mutations in myeloid malignancies. Leukemia. 2017;31(2):272-281. Welch JS, Petti AA, Miller CA, et al. TP53 and Decitabine in Acute Myeloid Leukemia and Myelodysplastic Syndromes. N Engl J Med. 2016;375(21): 2023-2036. Gerstung M, Papaemmanuil E, Martincorena I, et al. Precision oncology for acute myeloid leukemia using a knowledge bank approach. Nat Genet. 2017; 49(3):332-340. Sorror ML, Storer BE, Fathi AT, et al. Development and Validation of a Novel Acute Myeloid Leukemia-Composite Model to Estimate Risks of Mortality. JAMA Oncol. 2017;3(12):1675-1682.

haematologica | 2018; 103(12)

ARTICLE

Non-Hodgkin Lymphoma

Bromodomain and extra-terminal domain inhibition modulates the expression of pathologically relevant microRNAs in diffuse large B-cell lymphoma

Afua A. Mensah,1* Luciano Cascione,1,2,3* Eugenio Gaudio,1 Chiara Tarantelli,1 Riccardo Bomben,4 Elena Bernasconi,1 Domenico Zito,5,6 Andrea Lampis,5,6 Jens C. Hahne,5,6 Andrea Rinaldi,1 Anastasios Stathis,3 Emanuele Zucca,3 Ivo Kwee,1,2,7 Valter Gattei,4 Nicola Valeri,5,6 Maria E Riveiro8 and Francesco Bertoni1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2049-2058

Università della Svizzera italiana (USI), Institute of Oncology Research (IOR), Bellinzona, Switzerland; 2Swiss Institute of Bioinformatics (SIB), Lausanne, Switzerland; 3Oncology Institute of Southern Switzerland, Bellinzona, Switzerland; 4Clinical and Experimental Onco-Hematology Unit, Centro di Riferimento Oncologico, Aviano, Italy; 5The Institute of Cancer Research, London, UK; 6The Royal Marsden NHS Foundation Trust, London and Surrey, UK; 7Dalle Molle Institute for Artificial Intelligence (IDSIA), Manno, Switzerland and 8Oncology Therapeutic Development, Clichy, France. 1

*These authors contributed equally to this work.

ABSTRACT

berrant changes in microRNA expression contribute to lymphomagenesis. Bromodomain and extra-terminal domain inhibitors such as OTX015 (MK-8628, birabresib) have demonstrated preclinical and clinical activity in hematologic tumors. MicroRNA profiling of diffuse large B-cell lymphoma cells treated with OTX015 revealed changes in the expression levels of a limited number of microRNAs, including miR-92a-1-5p, miR-21-3p, miR-155-5p and miR96-5p. Analysis of publicly available chromatin immunoprecipitation sequencing data of diffuse large B-cell lymphoma cells treated with bromodomain and extra-terminal domain (BET) inhibitors showed that the BET family member BRD4 bound to the upstream regulatory regions of multiple microRNA genes and that this binding decreased following BET inhibition. Alignment of our microRNA profiling data with the BRD4 chromatin immunoprecipitation sequencing data revealed that microRNAs downregulated by OTX015 also exhibited reduced BRD4 binding in their promoter regions following treatment with another bromodomain and extra-terminal domain inhibitor, JQ1, indicating that BRD4 contributes directly to microRNA expression in lymphoma. Treatment with bromodomain and extra-terminal domain inhibitors also decreased the expression of the arginine methyltransferase PRMT5, which plays a crucial role in B-cell transformation and negatively modulates the transcription of miR-96-5p. The data presented here indicate that in addition to previously observed effects on the expression of coding genes, bromodomain and extra-terminal domain inhibitors also modulate the expression of microRNAs involved in lymphomagenesis.

Introduction The important role of non-coding elements of the genome, specifically microRNAs (miRNAs), in mediating cellular transformation was first demonstrated in chronic lymphocytic leukemia.1 Since then, numerous miRNAs have been shown to function as tumor suppressors or oncogenes in both hematologic and solid tumors.2-8 miRNAs are short sequences of 19 – 25 nucleotides that function as part of an RNA-induced silencing complex (RISC).9 In humans, they function primarily by destabilizing messenger RNA (mRNA) and inhibiting the translation of mRNA into protein. This is achieved through binding of the 5’ seed region of a haematologica | 2018; 103(12)

Correspondence: frbertoni@mac.com

Received: February 20, 2018. Accepted: July 31, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.191684 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2049 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2049

A.A. Mensah et al.

miRNA to its recognition sequence in the 3’ untranslated region of its target gene.10 A single miRNA can recognize multiple target genes and, conversely, different miRNAs can target an individual gene.11 Thus, in the context of cancer, miRNAs can intricately and markedly influence individual driver genes and entire signaling pathways crucial to the survival of cancer cells. Furthermore, a number of miRNAs have been shown to participate in a feedback loop with the protein product of their target gene.11 Diffuse large B-cell lymphoma (DLBCL) is an aggressive lymphoma that accounts for approximately 35-40% of all lymphoma cases.12 DLBCL frequently harbors mutations in chromatin-modifying enzymes indicating that perturbation of epigenetic regulation is an important trigger for B-cell transformation.13,14 A class of epigenetic drugs that has recently shown promising results in pre-clinical and clinical settings, and particularly in DLBCL, inhibits members of the bromodomain and extra-terminal domain (BET) protein family.15-25 In mammals, the BET family comprises four proteins, BRD2, BRD3, BRD4 and BRDT, which all share two highly conserved N-terminal bromodomains (BRD) and a C-terminal extra-terminal (ET) domain. BET proteins specifically bind to acetylated lysine residues via their dual BRD motifs, acting as epigenetic readers of acetyl-lysine marks. They therefore constitute an important component of the write-read-erase model via which epigenetic information is interpreted by cells.17 BET inhibitors act by preventing the interaction of BRD4 with acetylated histones.26 Here we show direct and indirect regulation of miRNA expression in DLBCL by a BET inhibitor.

Methods Cell lines and molecules Established human cell lines derived from DLBCL were cultured according to recommended conditions. Two germinal-center B-cell type DLBCL (GCB-DLBCL) cell lines, DOHH-2 and OCI-LY-1, were cultured in Roswell Park Memorial Institute medium and Iscove's Modified Dulbecco's Medium, respectively. The activated B-cell–like DLBCL (ABC-DLBCL) cell lines SUDHL-2 and HBL-1 were cultured in Roswell Park Memorial Institute medium. Cell lines were obtained as previously described,27 and their identity was authenticated by short tandem repeat DNA profiling (IDEXX BioResearch, Ludwigsburg, Germany). All media were supplemented with fetal bovine serum (10%; DOHH-2 and OCI-LY-1 or 20%; SU-DHL-2 and HBL-1), penicillin-streptomycin-neomycin (5,000 units penicillin, 5 mg streptomycin and 10 mg neomycin/mL, Sigma) and L-glutamine (1%). OTX015 (MK-8628, birabresib) was provided by Oncoethix (Lausanne, Switzerland).

In vivo xenograft model The xenograft model used here has been described elsewhere.15 Total RNA, previously extracted from these tumors, was used to analyze OTX015-mediated modulation of miRNA expression in vivo.

Western blotting analysis Protein extractions, sodium dodecylsulfate polyacrylamide gel electrophoresis and immunoblotting were performed as previously described.15 The antibodies used were anti-PRMT5 (A1520; NeoBiolab), anti-GAPDH (9131; Cell Signaling) and anti-BRD4 (A301-985A; Bethyl). 2050

One-step quantitative reverse transcription - polymerase chain reaction Total RNA was extracted from cells treated with dimethyl sulfoxide (DMSO) or OTX015 using TRIzol (Thermo Scientific, Lausanne, Switzerland). One-step quantitative reverse transcription - polymerase chain reaction (qRT-PCR) was performed as previously described15 using 20 ng of RNA for each reaction. Forward and reverse primers used for quantification of PRMT5 mRNA were, respectively, 5’-TCTCATGGTTTCCCATCCTC-3’ and 5’ACACAGATGGTTTGGCCTTC-3’. Quantification of GAPDH expression served as an endogenous control. GAPDH primer sequences were, 5’-CGACCACTTTGTCAAGCTCA-3’ (forward) and 5’-CCCTGTTGCTGTAGCCAAAT-3’ (reverse). Expression of GAPDH was verified to be stable between the analyzed groups.

MicroRNA expression profiling Total RNA was extracted as previously described.15 miRNA expression profiling was performed on RNA from DLBCL cell lines treated with DMSO or OTX015 using the Agilent Human microRNA microarray v. 3 or Nanostring nCounter Human V3A miRNA Expression Assay Kits. Profiling was done on RNA extracted from untreated lymphoma cell lines27,28 using the Nanostring nCounter Human V2. All samples were processed as previously described.29,30 Profiling data are available at the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo) database under the GEO project number GSE99208.

MicroRNA quantification with TaqMan microRNA assays Profiling results for selected miRNAs were validated using the following TaqMan MicroRNA Assays (Applied Biosystems): hsamiR-96-5p, assay ID: 000186; hsa-miR-92a-1-5p, assay ID: 002137, hsa-miR-21-3p, assay ID: 002438; hsa-miR-155-5p, assay ID: 002623; RNU6B, assay ID: 001093. Reverse transcription and quantitative PCR were performed using the TaqMan MicroRNA Reverse Transcription Kit and the TaqMan Universal PCR Master Mix according to the manufacturer’s instructions. Briefly, for each sample, 10 ng of total RNA was used for reverse transcription and 1.33 mL of the reverse transcription product was used in triplicate wells for the quantitative PCR (qPCR). All qPCR reactions were performed on an Applied Biosystems StepOnePlus System. Amplification of RNU6B served as a normalizing control for RNA quantity. Data were analyzed using the ΔΔCt method to obtain relative quantities. Expression of RNU6B was verified as stable between the analyzed groups.

Data mining miRNA expression data obtained from each profiling platform were analyzed independently. For Agilent arrays, the hybridization signal values for the multiple probes were obtained using the Agilent Feature Extraction Software 10.7.3 (Agilent Technologies). For the Nanostring nCounter, raw expression data were log-transformed and normalized by the quantile method after application of manufacturer-supplied correction factors. For both platforms, differentially expressed miRNAs were defined using R/Bioconductor with the linear model for microarray data analysis (limma) with a contrast matrix for the comparisons of interest on the datasets filtered to exclude features below the detection threshold (defined for each sample by a cut-off corresponding to twice the standard deviation of negative control probes plus the means) in at least half of the samples. The transcripts bearing an absolute log-fold change greater than 0.2 and a P-value less than 0.05 at any experimental time point were defined as differentially expressed. Overlapping among lists was performed using the VENNY on-line tool.31 Experimentally validated transcript targets haematologica | 2018; 103(12)

BET inhibition modulates miRNAs in DLBCL

of the miRNAs were obtained using the MicroRNA Target Filter in Ingenuity Pathway Analysis (Qiagen). Functional annotation of the targets was performed with the Gene Set Enrichment Analysis tool for overlap analysis using the hallmarks and the c2.cp of the Molecular Signatures Database 5.232 and hypergeometric P-values after correction for multiple hypothesis testing according to Benjamini and Hochberg. Publicly available chromatin immunoprecipitation (ChIP) sequencing datasets obtained by ChIP followed by high-throughput DNA sequencing were downloaded and re-analyzed. They comprised datasets obtained in the ABC-DLBCL cell line HBL-1 (SRP043524)23 and in the GCB-DLBCL cell line OCI-LY-1 (SRP022129),22 both treated with the BET inhibitor JQ1 or DMSO alone. Sequence reads obtained from ChIP fragments were aligned to the human reference genome hg19 using Bowtie, allowing up to one mismatch per fragment length. Redundant reads were removed and only reads uniquely mapping to the reference genome were used for further analysis. The peaks that were genomic regions enriched by ChIP, relative to the background reads, were detected using HOMER (v2.6), a suite of tools for Motif Discovery and next-generation sequencing analysis, with a default option (false discovery rate = 0.001 and Poisson P-value cut-off = 1e-04). Differential peaks were defined as having at least a four-fold difference in enrichment within a 200 bp region between the two conditions (DMSO versus JQ1) and a Poisson enrichment P-value less than 1e-04. All discovered putative peaks were ranked by their normalized tag counts (number of tags found at the peak, normalized to 10x106 total mapped tags) and annotated with the annotatePeaks.pl subroutine. We defined miRNA promoters using FANTOM533 and the precursors of microRNAs downloaded from miRBase (v20)34 to annotate the BRD4 ChIP sequencing datasets. We defined enriched regions located within 5 kb regions of predicted promoters and pre-miRNAs as candidate BRD4 binding sites. For global ChIP sequencing visualization, we used ngs.plot (https://code.google.com/p/ngsplot/) for inspection of both average and ‘laid out’ coverages as curves or heatmaps.

Chromatin immunoprecipitation Cells (SU-DHL-2 and HBL-1) were cross-linked with 1% formaldehyde. Crosslinking was quenched with 125 mM glycine. Cells were washed with ice-cold phosphate-buffered saline containing 1 x HALT protease inhibitor (Thermo Scientific, Lausanne, Switzerland) and resuspended in sodium dodecylsulfate lysis buffer (ChIP Assay Kit, Millipore, Schaffhausen, Switzerland) before sonication using the Bioruptor Plus. For each immunoprecipitation reaction, chromatin from 1x106 cells was incubated overnight with anti-PRMT5 (A1520; NeoBiolab), anti-BRD4 (A301-985A; Bethyl) or 3 mg of the negative control antibody, antiIgG (Millipore). Immune complexes were collected by incubation with 20 mL magnetic protein G beads at 4°C for 1.5 h. Protein Gbound complexes were sequentially washed with Low Salt Wash Buffer, High Salt Wash Buffer, LiCl Wash Buffer and twice with TE Buffer (ChIP Assay Kit, Millipore). Protein/DNA complexes were eluted using 1% sodium dodecylsulfate and 0.1 M NaHCO3. Following reversal of crosslinks (65°C overnight), samples were treated with RNAse A and then Proteinase K. DNA samples were purified using the QIAquick PCR purification kit (Qiagen, Hombrechtikon, Switzerland). Chromatin samples to which no antibody had been added were processed in parallel as input references. For qPCR analysis of ChIP samples, triplicate wells containing 1 mL of purified ChIP DNA plus PCR master mix were prepared. Reactions were performed on a StepOnePlus Real-Time PCR system (Applied Biosystems). Standard curves were constructed using sonicated and purified chromatin. ChIP-qPCR was performed using primers specific for the upstream regulatory haematologica | 2018; 103(12)

regions of PRMT5 and miR-96-5p. Primer sequences were as follows: PRMT5 forward; 5’-AGCGCGAGGAGAAAGATG-3’, PRMT5 reverse; 5’-CTATTTCGGGGACGCAATTC-3’, miR-96 forward; 5’-AGCTGGGAGACCTTGCTTC-3’, miR-96 reverse; 5’-TCACCCCTCCTAACCCAAAT-3’.

Results BET inhibition modulates the expression of a subset of microRNAs We have previously shown that the BET inhibitor OTX015 modulates the expression of multiple coding transcripts in DLBCL cells.15 Here we assessed the effect of OTX015 on global miRNA expression. GCB-DLBCL OCILY-1 and ABC-DLBCL HBL-1 cells were treated with 500 nM OTX015 for 4 and 24 h. Total RNA isolated from vehicle- and OTX015-treated cells was interrogated with the Nanostring nCounter. Fourteen miRNAs were modulated (5 downregulated, 9 upregulated) by the BET inhibitor in OCI-LY-1 cells and 11 (5 downregulated, 6 upregulated) in HBL-1 cells (Table 1). Additionally, we used the Agilent Human miRNA microarray v.3. platform to perform miRNA profiling on two more DLBCL cell lines, DOHH-2 (GCB-DLBCL) and SU-DHL-2 (ABC-DLBCL), the same cell lines we had previously used for mRNA profiling of OTX015-treated cells.15 In this case, seven miRNAs (3 downregulated, 4 upregulated) were affected by OTX015 in the GCBDLBCL, and five (2 downregulated, 3 upregulated) in the ABC-DLBCL cell line (Table 1). A few miRNAs were affected in more than one cell line, although we could not determine clear subtype-specific differences in miRNA modulation since only one GCBand one ABC-DLBCL cell line were interrogated on each profiling platform. The oncogenic miR-92a-1-5p,35 belonging to the miR-17-92 cluster, was downregulated in three of four cell lines (2 ABC-DLBCL, 1 GCB-DLBCL). miR204-5p, involved in BRAF resistance in melanoma,36 was downregulated and miR-487b-3p, expressed at lower levels in DLBCL versus follicular lymphoma,37 was upregulated in both cell lines analyzed with the Nanostring platform. The tumor suppressor miR-96-5p38,39 was upregulated in HBL-1 and DOHH-2 cells. Besides these, among the miRNAs modulated by the BET inhibitor there were others known to be involved in lymphomagenesis. The oncogenic miRNAs hsa-miR-21-3p40-44 and miR-15545,46 were downregulated, while, besides miR-96-5p, another miRNA with a tumor suppressor function, miR-16-5p,47 was also upregulated by the BET inhibitor. qRT-PCR was used to validate the expression of two lymphoma oncomiRNAs modulated by BET inhibition: miR-155-5p and miR-92a-1-5p (Online Supplementary Figure S1). The latter also appeared significantly downregulated after in vivo treatment of SU-DHL-2 xenografts (Online Supplementary Figure S2).

microRNAs modulated by BET inhibition control important pathways in diffuse large B-cell lymphoma Functional annotation analysis identified the p53 pathway, apoptosis, MYC-targets, cell cycle regulation, B-cell receptor signaling, interleukin-6 signaling, the STAT3 pathway, PI3K and nuclear factor-kB signaling among the biological processes significantly associated with the miRNAs that exhibited expression changes in HBL-1 and 2051

A.A. Mensah et al.

OCI-LY-1 DLBCL cells treated with OTX015 (Online Supplementary Table S1). The same pathways were predicted to be affected based on the modulated miRNAs in DOHH-2 and SU-DHL-2 cells treated with OTX015 (Online Supplementary Table S1). These signaling pathways and processes were similar to those we previously observed when analyzing the gene expression profiles of OTX015-treated DLBCL cells.15

BRD4 binds to the upstream regulatory regions of multiple miRNAs To further study the role of BET proteins in miRNA regulation, we took advantage of two publicly available ChIP-sequencing datasets obtained in the ABC-DLBCL cell line HBL-1 (SRP043524)23 and in the GCB-DLBCL cell line OCI-LY-1 (SRP022129),22 both treated with the BET inhibitor JQ1 or DMSO alone. Analysis of these datasets revealed that half of the regions bound by BRD4 were in intronic and intergenic regions where miRNAs are often located.48 We detected 794 miRNAs with at least one BRD4-binding event within their regulatory regions in ABC-DLBCL HBL-1 cells and 757 in the GCB-DLBCL OCI-LY-1 cell line (Online Supplementary Table S2). To determine whether BRD4 binding was associated with the expression of miRNA genes we profiled miRNA expression levels in a panel of 35 lymphoma cell lines using the Nanostring nCounter (Online Supplementary Table S3). Comparison of miRNA expression levels and BRD4 binding sites demonstrated that BRD4 peaks were more prevalent in the proximity of expressed miRNAs than non-expressed miRNAs (P<0.001) and were positively correlated with miRNA expression levels (Figure 1A). When we compared BRD4 binding in the presence or absence of the BET inhibitor JQ1, we identified 707 miRNAs with decreased BET bromodomain binding after exposure to the BET inhibitor in ABC-DLBCL HBL-1 cells and 348 in GCB-DLBCL OCI-LY-1 cells (Online Supplementary Table S2). Downregulation of miR-92a-1-5p and miR-155p expression following BET inhibitor-mediated reduction of BRD4 binding was also confirmed by qRTPCR analysis (Figure 1B).

BET inhibition mediates upregulation of miR-96-5p by downregulating PRMT5 expression The observed upregulation of the tumor suppressor miR-96-5p after exposure of DLBCL cell lines to OTX015 could not be explained by a direct BRD4-mediated effect of the BET inhibitor on the miRNA promoter. Thus, the miRNA profiling results were further validated by qRTPCR in two GCB-DLBCL cell lines (DOHH-2, OCI-LY-1) and two ABC-DLBCL cell lines (SU-DHL-2, HBL-1) treated with OTX015 for 4, 24 and 48 h. For DOHH-2, SUDHL-2 and HBL-1 cells, there was a time-dependent upregulation of miR-96-5p. For OCI-LY-1 cells, miR-96-5p was similarly upregulated at all three time points (Figure 2A). In lymphomas, miR-96-5p expression is regulated as part of a negative feedback loop with the protein arginine methyltransferase, PRMT5.39 Overexpression of PRMT5 mediates transcriptional repression of this miRNA via symmetric dimethylation of histones H3 and H4 in the promoter of miR-96-5p. Conversely, binding of miR-96-5p to the 3â&#x20AC;&#x2122; untranslated region of PRMT5 inhibits its translation.39,49 We hypothesized that the upregulation of miR-965p observed in OTX015-treated DLBCL cells could be due 2052

Table 1. miRNAs modulated by the BET inhibitor OTX015 in four DLBCL cell lines.

microRNA_ID hsa-miR-639 hsa-miR-204-5p ^ hsa-miR-6511a-5p hsa-miR-106a-5p+hsa-miR-17-5p hsa-miR-3613-5p hsa-miR-1254 hsa-miR-760 hsa-miR-221-5p hsa-miR-498 hsa-miR-487b-3p ^ hsa-miR-16-5p hsa-miR-182-5p hsa-miR-3136-5p hsa-miR-3605-3p

microRNA_ID hsa-miR-204-5p ^ hsa-miR-133a-5p hsa-miR-155-5p hsa-miR-580-3p hsa-miR-92a-1-5p ^ hsa-miR-487b-3p ^ hsa-miR-652-5p hsa-miR-191-5p hsa-miR-96-5p ^ hsa-miR-433-3p hsa-miR-582-3p

microRNA_ID hsa-miR-196a-3p hsa-miR-21-3p hsa-miR-92a-1-5p ^ hsa-miR-630 hsa-miR-935 hsa-miR-1181 hsa-miR-96-5p ^

microRNA_ID hsa-miR-92a-1-5p ^ hsa-miR-29b-1-5p hsa-miR-765 hsa-miR-1246 hsa-miR-345-5p

OCI-LY-1 (GCB-DLBCL) 4 hours 24 hours log2FC P value log2FC P value -1.38 -0.46 -0.43 -0.33 0.30 0.96 1.07 1.46 1.84 -0.02 0.07 0.24 0.43 0.48

0.005 n.s. n.s. n.s. n.s. 0.031 0.012 0.004 0.013 n.s. n.s. n.s. n.s. n.s.

-0.29 -1.61 -1.38 -0.95 -3.30 0.73 0.84 -0.07 -0.48 1.58 1.22 0.95 1.51 1.47

n.s. 0.048 0.034 0.035 <0.001 n.s. n.s. n.s. n.s. 0.016 0.012 0.025 0.013 0.014

HBL-1 (ABC-DLBCL) 4 hours 24 hours log2FC P value log2FC P value -1.59 -1.13 -0.67 -0.22 -0.45 1.59 2.08 0.16 0.41 0.49 0.72

0.011 0.020 n.s. n.s. n.s. 0.007 <0.001 n.s. n.s. n.s. 0.059

-0.58 0.26 -1.11 -1.11 -1.93 -0.32 0.56 1.17 0.99 1.75 1.88

n.s. n.s. 0.017 0.017 0.009 n.s. n.s. 0.007 0.023 0.014 0.001

DOHH2 (GCB-DLBCL) 4 hours 8 hours log2FC P value log2FC P value -0.44 -0.37 -0.32 0.63 0.36 0.07 0.01

0.039 0.045 n.s. 0.036 0.049 n.s. n.s.

0.29 -0.48 -0.52 -0.33 1.45 0.61 0.39

n.s. n.s. 0.0177 n.s. n.s. 0.0086 0.0415

SU-DHL-2 (ABC-DLBCL) 4 hours 8 hours log2FC P value log2FC P value -0.78 -0.25 0.39 0.23 0.13

0.004 n.s. 0.102 n.s. n.s.

-2.01 -0.46 0.48 0.64 0.58

0.004 0.043 0.011 0.025 0.015

^:modulated in more than one cell line; n.s., not significant. GCB: gerinal center B-cell; ABC: activated B-cell; DLBCL: diffuse B-cell lymphoma; FC: fold change.

haematologica | 2018; 103(12)

BET inhibition modulates miRNAs in DLBCL

to a perturbation of its downregulation by PRMT5. To assess this, we performed qRT-PCR analysis of PRMT5 in DLBCL cells treated with OTX015 for 4, 24 and 48 h. PRMT5 mRNA was markedly downregulated at 4 and 24 h. At 48 h PRMT5 levels were similar in DMSO- and OTX015-treated cells for all four cell lines (Figure 2B). At the protein level (Figure 2C), moderate downregulation of PRMT5 was evident at 24 h in DOHH-2 cells treated with OTX015. In SU-DHL-2 cells, PRMT5 was moderately downregulated at 24 h and was negligible at 48 h. These results indicate that OTX015 could downregulate the levels of expression of both PRMT5 RNA and protein in DLBCL cells. Hence upregulation of miR-96-5p following OTX015 treatment was associated with downregulation of PRMT5 protein, particularly in SU-DHL-2 cells.

BRD4 binds to the 5’ regulatory region of PRMT5 in diffuse large B-cell lymphoma cells and treatment with a BET inhibitor reduces BRD4 binding As OTX015 mediates transcriptional repression by displacing BRD4 from chromatin,15,21 we hypothesized that PRMT5 was transcriptionally regulated by BRD4. To assess this, we re-analyzed the two public ChIP sequencing datasets of DLBCL cells treated with the BET inhibitor JQ1,22,23 which has a similar mechanism of action to OTX015 and exhibits an overlapping profile of targeted genes.15,22 This revealed that BRD4 bound to the 5’ region of PRMT5 and that this binding was reduced following treatment with the BET inhibitor (Figure 3A). In agreement with the public ChIP sequencing data of JQ1-treated DLBCL cells, when we performed ChIP-

Figure 1. BRD4 binds to the regulatory regions of microRNAs. (A) The genomic regions within ± 1 kb of miRNA promoters that are bound by BRD4 obtained using ngs.plot. Lines represent the average expression profiles of “expressed” (red line) and “not expressed” (green line) miRNA. (B) Analysis of publicly available chromatin immunoprecipitation sequencing data of diffuse large B-cell lymphoma (DLBCL) cells showed that BET inhibitor treatment reduces BRD4 binding at the 5’ regulatory regions of the miR-17-92 cluster, which contains miR-92a-1-5p, and the miR-155 host gene (left panel). An activated B-cell (ABC)-DLBCL cell line (SU-DHL-2) and a germinal center B-cell (GCB)-DLBCL cell line (DOHH-2) were treated with OTX015 for 4 and 24 h before TaqMan quantitative reverse transcription polymerase chain reaction analysis of miR-92a-1-5p and miR155-5p expression (right panel). miR-1555p expression is only shown for SU-DHL-2 as it is an ABC-DLBCL specific oncomiR. Expression of RNU6B was used for normalization. For each timepoint, the mean foldchange relative to the dimethyl sulfoxide (DMSO) control is shown. Charts show the mean of at least three independent experiments. *P<0.05; **P<0.01. Error bars denote the standard error.

haematologica | 2018; 103(12)

2053

A.A. Mensah et al.

qPCR analysis of SU-DHL-2 cells treated with OTX015, we observed decreased binding of BRD4 to the 5â&#x20AC;&#x2122; region of PRMT5 (Figure 3B). The decrease in PRMT5 expression following OTX015 treatment was, therefore, likely due to a reduced association of BRD4 to the 5â&#x20AC;&#x2122; regulatory region of PRMT5. Additionally, DNAse hypersensitivity sites and H3K27 acetylation, both marks of active transcription, were enriched at the BRD4 binding site in

PRMT5 (data not shown). As PRMT5 and miR-96-5p regulate each other in a negative feedback loop, we tested for binding of PRMT5 to the promoter of miR-96-5p. OTX015 treatment led to reduced recruitment of PRMT5 to the miR-96-5p promoter (Figure 3C) indicating that upregulation of miR-96-5p in BET inhibitor-treated cells was mediated through downregulation of PRMT5 (Figure 3D).

2054

Figure 2. OTX015 modulates microRNA-96-5p expression in diffuse large B-cell lymphoma models. (A) OTX015 upregulates miR-96-5p in a time-dependent manner. Two germinal center B-cell (GCB)-diffuse large B-cell lymphoma (DLBCL) cell lines (DOHH-2, OCI-LY-1) and two activated B-cell (ABC)-DLBCL cell lines (SU-DHL2, HBL-1) were treated with dimethyl sulfoxide (DMSO) or 500 nM OTX015 for 4, 24, and 48 h. Expression of miR-96-5p was determined by TaqMan quantitative reverse transcription polymerase chain reaction (qRT-PCR). Expression of RNU6B was used for normalization. For each timepoint, the mean fold-change relative to the DMSO control is shown. (B) OTX015 treatment of DLBCL cells downregulates PRMT5. Two GCBDLBCL (DOHH-2, OCI-LY-1) and two ABC-DLBCL (SU-DHL-2, HBL-1) cell lines were treated with DMSO or 500 nM OTX015 for 4, 24, and 48 h. Expression of PRMT5 was determined by qRT-PCR. GAPDH expression was used for normalization. For each timepoint, the mean fold-change relative to the DMSO control is shown. (C) OTX015 reduces PRMT5 protein levels in DOHH-2 and SU-DHL-2 cells treated with DMSO or 500 nM OTX015. GAPDH was used as a loading control. PRMT5 signals were quantified using ImageJ (http://rsbweb.nih.gov/ij/) and normalized to GAPDH signals. Representative images of two independent Western blot analyses are shown. The graphs show the mean of three independent experiments. **P<0.01. Error bars denote the standard error.

haematologica | 2018; 103(12)

BET inhibition modulates miRNAs in DLBCL

haematologica | 2018; 103(12)

Figure 3. OTX015 reduces binding of BRD4 to PRMT5 and diminishes recruitment of PRMT5 to the microRNA-96-5p promoter. (A) Analysis of publicly available chromatin immunoprecipitation (ChIP) sequencing data of diffuse large B-cell lymphoma (DLBCL) cells treated with the BET inhibitor JQ1 showed that BRD4 binds to the 5â&#x20AC;&#x2122; regulatory region of PRMT5 and that BET inhibitor treatment reduces BRD4 recruitment to PRMT5. (B,C) ChIP was performed for DLBCL cells treated with dimethyl sulfoxide (DMSO) or 500 nM OTX015 for 48 h. Anti-BRD4, anti-PRMT5 and anti-IgG (negative control) antibodies were used for immunoprecipitations. (B) Chromatin pulled down with antiBRD4 and anti-IgG in DMSO- and OTX015-treated SU-DHL-2 cells was amplified with primers specific for the 5â&#x20AC;&#x2122; regulatory region of PRMT5 identified by the analysis in (A). (C) Chromatin pulled down with anti-PRMT5 and anti-IgG antibodies in DMSO- and OTX015-treated HBL-1 cells was amplified with primers specific for the promoter of miR-96-5p. Amplification of the same immunoprecipitated chromatin samples was performed with primers specific for the chromosome 4 human alpha satellite sequence as an additional negative control (representative results from one of two biological replicates are shown). The graphs show the mean fold-difference between DMSO- and OTX015-treated cells after normalization to input and IgG background subtraction. ChIP-quantitative polymerase chain reaction experiments were repeated twice in triplicate. (D) Proposed model for the upregulation of miR-96-5p expression following treatment of DLBCL cells with OTX015. Upper panel; previous work by others has shown that PRMT5 is overexpressed in lymphoma cells in which it mediates transcriptional repression of miR-96 and that overexpression of miR-96 negatively regulates PRMT5 translation.39,49 Lower panel; here we showed that BRD4 binds to the upstream regulatory region of PRMT5. Treatment of DLBCL cells with a BET inhibitor reduced BRD4 occupancy at the PRMT5 locus and also reduced the expression of PRMT5 mRNA and protein. Additionally, the BET inhibitor diminished the occupancy of PRMT5 at the miR96 promoter and increased miR-96 expression. **P<0.01. Error bars denote the standard error.

2055

A.A. Mensah et al.

Discussion The present study shows that the BET inhibitor OTX015 modulates the expression of miRNAs in DLBCL cells. The regulation may occur directly, due to the binding of BRD4 to the regulatory regions of specific miRNAs, or indirectly as demonstrated for miR-96-5p, a miRNA with important functions in the proliferation and survival of B-cell malignancies.49 The ability of OTX015 to alter miRNA expression demonstrates that the effects of BET inhibition on the transcriptome extend beyond coding genes to comprise also non-coding regions of the genome. miRNAs function by regulating the expression of genes at the transcript level, where they can mediate mRNA degradation, reduce mRNA stability or prevent translation of mRNA into protein. By analyzing publicly available ChIP sequencing data22,23 in combination with our miRNA profiling data of baseline and BET inhibitor-treated lymphoma cells, we determined that a subset of miRNAs were bound by BRD4 and that this binding decreased after BET inhibitor treatment. For a number of these miRNAs reduced binding of BRD4 after BET inhibition was associated with reduced expression. Our finding that BRD4 directly binds to the regulatory regions of miRNA genes to regulate their expression in lymphomas complements the recent report describing that components of the miRNA processing machinery, namely DGCR8 and Drosha, are localized to super-enhancers of miRNAs in a tissue-specific manner and that association of these proteins with superenhancers is reduced following treatment with the BET inhibitor JQ1.50 Indeed, these observations indicate that the targeted effects of BET inhibition on specific genes in different cellular contexts also likely comprises non-coding transcripts that are specifically modulated in different transformed cell types. With respect to this, we observed that the promoters of two established lymphoma oncomiRNAs, miR-155-5p and miR-92a-1-5p, were bound by BRD4. When lymphoma cells were treated with a BET inhibitor, BRD4 was diminished at these sites and this was associated with downregulation of miRNA expression. miR-155-5p is upregulated in activated B cells and ABC-DLBCL,45,46,51 often because of amplification of its locus.45 Its high expression is associated with poor outcome and resistance to R-CHOP therapy and its knockdown compromises the viability of ABC-DLBCL cells.43,45 Of interest, BET inhibition decreased miR-155-5p expression only in the two ABC-DLBCL cell lines, while the GCB-DLBCL cell line showed upregulation of miR-1555p suggesting that different mechanisms may regulate the transcription of this miRNA gene in different DLBCL subtypes. miR-92a-1-5p is one of the six members of the miR-1792 cluster located on chromosome 13q31.3 and is overexpressed in different lymphoma subtypes including DLBCL.35 It is a transcriptional target of the MYC oncoprotein,35 which is itself rapidly and robustly downregulated by treatment with a BET inhibitor.15 Downregulation of miR-92a-1-5p was already very pronounced after 4 h of BET inhibitor treatment. Additionally, miR-92a-1-5p was the only miRNA commonly identified by the two different platforms that we utilized for miRNA profiling. The miR-17-92 cluster is involved in activation of the PI3K/AKT/mTOR pathway, lymphoma pathogenesis and chemoresistance.52,53 The 2056

miR-17-92 cluster can acquire super-enhancers during neoplastic transformation50,54 and BRD4 exhibits a preference for binding at super-enhancers.22 This provides further support for our observation of direct regulation of the miR-17-92 cluster by BRD4. miR-21-3p was also downregulated by BET inhibitor treatment. This miRNA is overexpressed in B-cell lymphomas.40-42 It inhibits translation of the tumor suppressor PTEN43 and knockdown of miR-21 increases the sensitivity of lymphoma cells to CHOP treatment.44 When we performed functional annotation analysis of the OTX015-modulated miRNAs, we identified the same signaling pathways and processes that we had previously identified from gene expression profiles of OTX015-treated DLBCL cells,15 indicating that changes in miRNA expression likely contribute to modulating some of the transcripts and pathways that have been previously identified in cells treated with a BET inhibitor.15,22,23 In lymphoma cells, which overexpress PRMT5, the negative feedback loop comprising miR-96-5p and PRMT5 is usually poised in favor of PRMT5.39 PRMT5 catalyzes the symmetric methylation of arginine residues on histones H3 and H4, giving rise to the repressive epigenetic marks H3R8me2S and H4R3me2S. Inhibition of PRMT5 expression, either pharmacologically or by the use of antisense oligonucleotides, severely compromises lymphoma cell viability and induces apoptosis.39,49 In the DLBCL cells we treated with OTX015, PRMT5 protein was negligible after 48 h of treatment in the ABC-DLBCL cell line SU-DHL-2, which we previously showed undergoes pronounced apoptosis in response to OTX015.15 DOHH-2 cells, in which we did not observe apoptosis following OTX015 treatment,15 exhibited moderate PRMT5 downregulation. The less marked downregulation of PRMT5 protein in DOHH-2 cells treated with OTX015 was in contrast to its pronounced downregulation at the transcript level. It was nevertheless associated with a pronounced increase in miR-96-5p levels indicating that, for this cell line, factors other than PRMT5 downregulation may contribute to releasing miR-96-5p from transcriptional repression. Inhibition of PRMT5 has been shown to release miR-96-5p from transcriptional repression in lymphoma cells.39 We observed that BET inhibitor-mediated inhibition of PRMT5 was associated with decreased binding of BRD4 to the 5â&#x20AC;&#x2122; regulatory region of PRMT5 and that this resulted in reduced occupancy of PRMT5 at the miR-96-5p promoter. The disruption of the negative feedback loop comprising PRMT5 and miR-96-5p by OTX015 shows how the anti-tumor effects of BET inhibition are further amplified by modulation of secondary targets, which might themselves also contribute to further suppress direct targets of BET inhibitors. With respect to this, overexpression of miR96-5p downregulates phosphorylated STAT3 (p-STAT3) without affecting levels of total STAT3 in T-cell anaplastic large-cell lymphoma cells.38 We have previously shown that OTX015 treatment decreases p-STAT3 in ABCDLBCL cells.15 It is therefore possible that in addition to the direct effects of OTX015 on the expression of genes involved in JAK/STAT signaling,15 the overexpression of miR-96-5p contributes to maintaining p-STAT3 repressed. Our study provides the first evidence of BET inhibitormediated modulation of miRNAs in lymphomas. This modulation can occur by inhibiting the interaction of haematologica | 2018; 103(12)

BET inhibition modulates miRNAs in DLBCL

BRD4 with genes whose products regulate miRNA expression, or through the direct inhibition of BRD4 at miRNA regulatory regions, or, as recently suggested, by interfering with the processing of pri-miRNA to pre-miRNAs.50 Unlike coding transcripts, miRNAs are highly stable in blood and as such, levels of circulating miRNAs have been used for diagnosis and screening in a number of diseases. In lymphomas, the overexpression of specific miRNAs in plasma and serum samples has been shown to be an accurate biomarker for diagnosis, prognosis and response to therapy.45

References 1. 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 USA. 2002;99(24):1552415529. 2. Johnson SM, Grosshans H, Shingara J, et al. RAS is regulated by the let-7 microRNA family. Cell. 2005;120(5):635-647. 3. Ambs S, Prueitt RL, Yi M, et al. Genomic profiling of microRNA and messenger RNA reveals deregulated microRNA expression in prostate cancer. Cancer Res. 2008;68(15): 6162-6170. 4. Yanaihara N, Caplen N, Bowman E, et al. Unique microRNA molecular profiles in lung cancer diagnosis and prognosis. Cancer Cell. 2006;9(3):189-198. 5. Pekarsky Y, Santanam U, Cimmino A, et al. Tcl1 expression in chronic lymphocytic leukemia is regulated by miR-29 and miR181. Cancer Res. 2006;66(24):11590-11593. 6. Kluiver J, Poppema S, de Jong D, et al. BIC and miR-155 are highly expressed in Hodgkin, primary mediastinal and diffuse large B cell lymphomas. J Pathol. 2005;207 (2):243-249. 7. He L, Thomson JM, Hemann MT, et al. A microRNA polycistron as a potential human oncogene. Nature. 2005;435(7043):828-833. 8. Hezaveh K, Kloetgen A, Bernhart SH, et al. Alterations of microRNA and microRNAregulated messenger RNA expression in germinal center B-cell lymphomas determined by integrative sequencing analysis. Haematologica. 2016;101(11):1380-1389. 9. Gregory RI, Chendrimada TP, Cooch N, Shiekhattar R. Human RISC couples microRNA biogenesis and posttranscriptional gene silencing. Cell. 2005;123(4):631640. 10. Lai EC. Micro RNAs are complementary to 3' UTR sequence motifs that mediate negative post-transcriptional regulation. Nat Genet. 2002;30(4):363-364. 11. Bracken CP, Scott HS, Goodall GJ. A network-biology perspective of microRNA function and dysfunction in cancer. Nat Rev Genet. 2016;17(12):719-732. 12. Teras LR, DeSantis CE, Cerhan JR, Morton LM, Jemal A, Flowers CR. 2016 US lymphoid malignancy statistics by World Health Organization subtypes. CA Cancer J Clin. 2016;66(6):443-459. 13. Morin RD, Mendez-Lago M, Mungall AJ, et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature. 2011;476(7360):298-303. 14. Pasqualucci L, Dominguez-Sola D,

haematologica | 2018; 103(12)

15.

16.

17. 18.

19.

20.

21.

22.

23.

24.

25.

26.

The circulating miRNAs that have been identified as biomarkers in lymphoma are among those that we have identified as regulated by BET inhibition (miR-92, miR-21, miR-155). The assessment of circulating miRNAs could, therefore, be used as a robust and non-invasive way to monitor response to BET inhibitor treatment. In conclusion, our observations contribute to a better understanding of the targeted effects of BET inhibitors, revealing a novel aspect of the activity of this class of compounds in lymphomas.

Chiarenza A, et al. Inactivating mutations of acetyltransferase genes in B-cell lymphoma. Nature. 2011;471(7337):189-195. Boi M, Gaudio E, Bonetti P, et al. The BET bromodomain inhibitor OTX015 affects pathogenetic pathways in preclinical B-cell tumor models and synergizes with targeted drugs. Clin Cancer Res. 2015;21(7):16281638. Filippakopoulos P, Knapp S. Targeting bromodomains: epigenetic readers of lysine acetylation. Nat Rev Drug Discov. 2014;13 (5):337-356. Stathis A, Bertoni F. BET Proteins as targets for anticancer treatment. Cancer Discov. 2018;8(1):24-36. Stathis A, Zucca E, Bekradda M, et al. Clinical response of carcinomas harboring the BRD4-NUT oncoprotein to the targeted bromodomain inhibitor OTX015/MK-8628. Cancer Discov. 2016;6(5):492-500. Amorim S, Stathis A, Gleeson M, et al. Bromodomain inhibitor OTX015 in patients with lymphoma or multiple myeloma: a dose-escalation, open-label, pharmacokinetic, phase 1 study. Lancet Haematol. 2016;3(4):e196-204. Berthon C, Raffoux E, Thomas X, et al. Bromodomain inhibitor OTX015 in patients with acute leukaemia: a dose-escalation, phase 1 study. Lancet Haematol. 2016;3(4):e186-195. Henssen A, Althoff K, Odersky A, et al. Targeting MYCN-driven transcription by BET-bromodomain inhibition. Clin Cancer Res. 2016;22(10):2470-2481. Chapuy B, McKeown MR, Lin CY, et al. Discovery and characterization of superenhancer-associated dependencies in diffuse large B cell lymphoma. Cancer Cell. 2013;24(6):777-790. Ceribelli M, Kelly PN, Shaffer AL, et al. Blockade of oncogenic IkappaB kinase activity in diffuse large B-cell lymphoma by bromodomain and extraterminal domain protein inhibitors. Proc Natl Acad Sci USA. 2014;111(31):11365-11370. Riveiro ME, Astorgues-Xerri L, Vazquez R, et al. OTX015 (MK-8628), a novel BET inhibitor, exhibits antitumor activity in nonsmall cell and small cell lung cancer models harboring different oncogenic mutations. Oncotarget. 2016;7(51):84675-84687. Vazquez R, Riveiro ME, Astorgues-Xerri L, et al. The bromodomain inhibitor OTX015 (MK-8628) exerts anti-tumor activity in triple-negative breast cancer models as single agent and in combination with everolimus. Oncotarget. 2017;8(5):75987613. Noel JK, Iwata K, Ooike S, Sugahara K, Nakamura H, Daibata M. Development of

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

the BET bromodomain inhibitor OTX015. Mol Cancer Ther. 2013;12(11 Suppl): C244. Chila R, Basana A, Lupi M, et al. Combined inhibition of Chk1 and Wee1 as a new therapeutic strategy for mantle cell lymphoma. Oncotarget. 2015;6(5):3394-3408. Tarantelli C, Gaudio E, Arribas AJ, et al. PQR309 is a novel dual PI3K/mTOR inhibitor with preclinical antitumor activity in lymphomas as a single agent and in combination therapy. Clin Cancer Res. 2018;24(1):120-129. Valeri N, Braconi C, Gasparini P, et al. MicroRNA-135b promotes cancer progression by acting as a downstream effector of oncogenic pathways in colon cancer. Cancer Cell. 2014;25(4):469-483. Dal Bo M, D'Agaro T, Gobessi S, et al. The SIRT1/TP53 axis is activated upon B-cell receptor triggering via miR-132 up-regulation in chronic lymphocytic leukemia cells. Oncotarget. 2015;6(22):19102-19117. Oliveros JC. VENNY. An interactive tool for comparing lists with Venn Diagrams. 2007 [cited; Available from: http://bioinfogp.cnb.csic.es/tools/venny/index.html. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA. 2005;102(43):15545-15550. Lizio M, Harshbarger J, Shimoji H, et al. Gateways to the FANTOM5 promoter level mammalian expression atlas. Genome Biol. 2015;16:22. Kozomara A, Griffiths-Jones S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res. 2014;42(Database issue):D68-73. Dal Bo M, Bomben R, Hernandez L, Gattei V. The MYC/miR-17-92 axis in lymphoproliferative disorders: a common pathway with therapeutic potential. Oncotarget. 2015;6(23):19381-19392. Diaz-Martinez M, Benito-Jardon L, Alonso L, Koetz-Ploch L, Hernando E, Teixido J. miR-204-5p and miR-211-5p contribute to BRAF inhibitor resistance in melanoma. Cancer Res. 2018;78(4):1017-1030. Culpin RE, Proctor SJ, Angus B, Crosier S, Anderson JJ, Mainou-Fowler T. A 9 series microRNA signature differentiates between germinal centre and activated B-cell-like diffuse large B-cell lymphoma cell lines. Int J Oncol. 2010;37(2):367-376. Vishwamitra D, Li Y, Wilson D, et al. MicroRNA 96 is a post-transcriptional suppressor of anaplastic lymphoma kinase expression. Am J Pathol. 2012;180(5):17721780. Pal S, Baiocchi RA, Byrd JC, Grever MR, Jacob ST, Sif S. Low levels of miR-92b/96

2057

A.A. Mensah et al.

40.

41.

42.

43.

44.

2058

induce PRMT5 translation and H3R8/H4R3 methylation in mantle cell lymphoma. EMBO J. 2007;26(15):3558-3569. Arribas AJ, Gomez-Abad C, Sanchez-Beato M, et al. Splenic marginal zone lymphoma: comprehensive analysis of gene expression and miRNA profiling. Mod Pathol. 2013;26(7):889-901. Baraniskin A, Kuhnhenn J, Schlegel U, et al. Identification of microRNAs in the cerebrospinal fluid as marker for primary diffuse large B-cell lymphoma of the central nervous system. Blood. 2011;117(11):3140-3146. Junker F, Chabloz A, Koch U, Radtke F. Dicer1 imparts essential survival cues in Notch-driven T-ALL via miR-21-mediated tumor suppressor Pdcd4 repression. Blood. 2015;126(8):993-1004. Go H, Jang JY, Kim PJ, et al. MicroRNA-21 plays an oncogenic role by targeting FOXO1 and activating the PI3K/AKT pathway in diffuse large B-cell lymphoma. Oncotarget. 2015;6(17):15035-15049. Bai H, Wei J, Deng C, Yang X, Wang C, Xu

45.

46.

47.

48. 49.

R. MicroRNA-21 regulates the sensitivity of diffuse large B-cell lymphoma cells to the CHOP chemotherapy regimen. Int J Hematol. 2013;97(2):223-231. Iqbal J, Shen Y, Huang X, et al. Global microRNA expression profiling uncovers molecular markers for classification and prognosis in aggressive B-cell lymphoma. Blood. 2015;125(7):1137-1145. Lawrie CH, Soneji S, Marafioti T, et al. MicroRNA expression distinguishes between germinal center B cell-like and activated B cell-like subtypes of diffuse large B cell lymphoma. Int J Cancer. 2007;121(5):1156-1161. Cimmino A, Calin GA, Fabbri M, et al. miR15 and miR-16 induce apoptosis by targeting BCL2. Proc Natl Acad Sci USA. 2005;102(39):13944-13949. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004;116(2):281-297. Alinari L, Mahasenan KV, Yan F, et al. Selective inhibition of protein arginine

50.

51.

52.

53.

54.

methyltransferase 5 blocks initiation and maintenance of B-cell transformation. Blood. 2015;125(16):2530-2543. Suzuki HI, Young RA, Sharp PA. Superenhancer-mediated RNA processing revealed by integrative MicroRNA network analysis. Cell. 2017;168(6):1000-1014. e1015. Ahmadvand M, Eskandari M, Pashaiefar H, et al. Over expression of circulating miR155 predicts prognosis in diffuse large B-cell lymphoma. Leuk Res. 2018;70:45-48. Rao E, Jiang C, Ji M, et al. The miRNA-1792 cluster mediates chemoresistance and enhances tumor growth in mantle cell lymphoma via PI3K/AKT pathway activation. Leukemia. 2012;26(5):1064-1072. Benhamou D, Labi V, Novak R, et al. A cMyc/miR17-92/Pten axis controls PI3K-mediated positive and negative selection in B cell development and reconstitutes CD19 deficiency. Cell Rep. 2016;16(2):419-431. Hnisz D, Abraham BJ, Lee TI, et al. Superenhancers in the control of cell identity and disease. Cell. 2013;155(4):934-947.

haematologica | 2018; 103(12)

ARTICLE

Non-Hodgkin Lymphoma

Cyclin-dependent kinase 9 as a potential specific molecular target in NK-cell leukemia/lymphoma

Shiori Kinoshita,1 Takashi Ishida,1,2 Asahi Ito,1 Tomoko Narita,1 Ayako Masaki,1,3 Susumu Suzuki,4 Takashi Yoshida,1 Masaki Ri,1 Shigeru Kusumoto,1 Hirokazu Komatsu,1 Norio Shimizu,5 Hiroshi Inagaki,3 Taruho Kuroda,6 Arne Scholz,7 Ryuzo Ueda,4 Takaomi Sanda8 and Shinsuke Iida1

Department of Hematology and Oncology, Nagoya City University Graduate School of Medical Sciences, Japan; 2Division of Hematology and Oncology, Department of Internal Medicine, School of Medicine, Iwate Medical University, Japan; 3Department of Pathology and Molecular Diagnostics, Nagoya City University Graduate School of Medical Sciences, Japan; 4Department of Tumor Immunology, Aichi Medical University School of Medicine, Japan; 5Department of Virology, Division of Medical Science, Medical Research Institute, Tokyo Medical and Dental University, Japan; 6Bayer Yakuhin, Ltd., Osaka, Japan; 7Bayer AG Pharmaceuticals Division, Berlin, Germany; 8Cancer Science Institute of Singapore, National University of Singapore 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2059-2068

ABSTRACT

AY 1143572 is a highly selective inhibitor of cyclin-dependent kinase 9/positive transcription elongation factor b. It has entered phase I clinical studies. Here, we have assessed the utility of BAY 1143572 for treating natural killer (NK) cell leukemias/lymphomas that have a poor prognosis, namely extranodal NK/T-cell lymphoma, nasal type and aggressive NK-cell leukemia, in a preclinical mouse model in vivo as well as in tissue culture models in vitro. Seven NK-cell leukemia/lymphoma lines and primary aggressive NK-cell leukemia cells from two individual patients were treated with BAY 1143572 in vitro. Primary tumor cells from an aggressive NK-cell leukemia patient were used to establish a xenogeneic murine model for testing BAY 1143572 therapy. Cyclin-dependent kinase 9 inhibition by BAY 1143572 resulted in prevention of phosphorylation at the serine 2 site of the C-terminal domain of RNA polymerase II. This resulted in lower c-Myc and Mcl-1 levels in the cell lines, causing growth inhibition and apoptosis. In aggressive NK-cell leukemia primary tumor cells, exposure to BAY 1143572 in vitro resulted in decreased Mcl-1 protein levels resulting from inhibition of RNA polymerase II C-terminal domain phosphorylation at the serine 2 site. Orally administering BAY 1143572 once per day to aggressive NK-cell leukemia-bearing mice resulted in lower tumor cell infiltration into the bone marrow, liver, and spleen, with less export to the periphery relative to control mice. The treated mice also had a survival advantage over the untreated controls. The specific small molecule targeting agent BAY1143572 has potential for treating NK-cell leukemia/lymphoma.

Correspondence: itakashi@iwate-med.ac.jp

Received: February 15, 2018. Accepted: July 30, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.191395 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2059 ÂŠ2018 Ferrata Storti Foundation

Introduction Extranodal natural killer (NK)/T-cell lymphoma (ENKTL), nasal type and aggressive NK-cell leukemia (ANKL), are both representative NK-cell leukemia/lymphoma in which Epstein-Barr virus (EBV) is considered to play a critical role.1,2 ANKL is a systemic neoplastic proliferation of NK cells that has an aggressive clinical course, and a seriously poor prognosis, with a median survival of < 2 months.25 There can be overlap with ENKTL, nasal type, showing systemic organ involvement; thus, it is unclear whether ANKL is the leukemic counterpart of ENKTL, nasal type.1,2 A regimen not containing anthracyclines, SMILE (dexamethasone, methotrexate, ifosfamide, L-asparaginase, and etoposide) has brought some haematologica | 2018; 103(12)

2059

S. Kinoshita et al.

improvement in the treatment of these systemic NK-cell leukemia/lymphoma.6,7 However, the prognosis of these neoplasms is still unsatisfactory,8,9 and the development of novel therapeutic agents remains an urgent issue. Nevertheless, to the best of our knowledge, until now there have been very few preclinical studies on the development of novel antitumor agents targeting NK-cell leukemia/lymphoma. We have been focusing on cyclin-dependent kinase 9 (CDK9) as a potential molecular target for NK-cell leukemia/lymphoma. CDK9 is a serine (Ser)/threonine kinase, and constitutes a subunit of the positive transcription elongation factor b (P-TEFb) complex. This plays a vital role in regulating gene transcription elongation via phosphorylation of the C-terminal domain (CTD) of RNA polymerase II (RNAPII).10-12 Accumulating reports indicate that CDK9 kinase activity is crucial during the evolution and/or maintenance of many types of human malignancy.10-17 CDK9 is also known to have an important role for Epstein–Barr Nuclear Antigen 2 (EBNA2)-dependent transcriptional activation and immortalization of EBV-infected cells.18-20 Taken together, these findings suggest that CDK9 could represent a new molecular target for treating systemic NK-cell neoplasms, such as ENKTL, nasal type with systemic organ involvement, as well as ANKL. Here, we begin to test this hypothesis by investigating the therapeutic potential of BAY 1143572 (Bayer AG Pharmaceuticals Division, Berlin, Germany), which is a new, highly selective inhibitor of CDK9/P-TEFb.21

Animals All in vivo experiments of NOD/Shi-scid, IL-2Rγnull (NOG) mice were performed as previously described.17

Establishment of the primary ANKL cell-bearing mouse model Patient A’s PBMC, consisting of almost 90% CD56-positive atypical lymphoid tumor cells, were injected intraperitoneally (i.p.) into naïve NOG mice (1 x 107/mouse). Three to 4 weeks after i.p. injection, the NOG mice became weaker and exhibited clinical features of cachexia. The tumor cells were recovered and i.p. inoculated into other naïve NOG mice, and after three to four weeks, they displayed features almost identical to those of the donor mice. This procedure of transfer from mouse to mouse was repeated successfully until at least the fifth passage.

Primary ANKL cell-bearing mice treated with BAY 1143572

Methods

Leukemic cells from ANKL patient A, which could be serially transplanted into NOG mice, were i.p. injected into 10 naïve NOG mice (1x107/mouse). The animals were randomly divided into two groups seven days after ANKL cell inoculation, and were treated with oral application of 12.5 mg/kg BAY 1143572 or vehicle, for 15 days (7–21 days after tumor inoculations). Therapeutic efficacy was then evaluated 22 days after tumor inoculation. In another experiment, ANKL cells from the mice suspended were also inoculated i.p. into another 12 naive NOG mice (0.8x107/mouse). These animals were randomly divided into two groups and were treated by oral application of 12.5 mg/kg BAY1143572 or vehicle for 15 days (7–21 days after tumor inoculation). The therapeutic efficacy of BAY 1143572 was evaluated by survival times.

NK-cell leukemia/lymphoma lines

Flow cytometry analysis of cells inoculated into mice

SNT-8, SNK-1, SNT-16, NK-92 and KAI-3 are EBV-positive, but MTA and KHYG-1 are EBV-negative lines.22-26 NK-92 was purchased from ATCC (Manassas, VA), and MTA, KAI-3 and KHYG1 were purchased from the Japanese Collection of Research Bioresources Cell Bank (Osaka, Japan).

The following mAbs were used for flow cytometry: BD MultitestTM CD3/CD16+CD56/CD45/CD19 (No. 342416, BD Biosciences), and stained cells were analyzed as previously described.17

Statistical analysis Primary tumor cells from patients with ANKL and cells from control subjects Primary tumor cells were isolated using anti-human CD56 microbeads (Miltenyi Biotec, Bergisch Gladbach) from peripheral blood mononuclear cells (PBMC) of two patients (patient A and B, Online Supplementary Figure S1). Five healthy volunteers participated as control subjects, and their CD56-positive cells were isolated from their PBMC in the same manner. All donors provided written informed consent before blood sampling according to the Declaration of Helsinki, and the present study was approved by the institutional ethics committees of Nagoya City University Graduate School of Medical Sciences.

Cell proliferation and apoptosis assays Cell proliferation and apoptosis were assessed as previously described.17,27

Western blotting Antibodies to RNAPII (N-20), c-Myc, Mcl-1, actin, phosphorylated RNAPII (phospho-RNAPII) (Ser2 of the CTD) and phosphoRNAPII (Ser5 of the CTD) were as previously described.17,28

Histological analysis Hematoxylin and eosin (HE) staining was performed on formalin-fixed, paraffin-embedded sections, as previously described.17,29 2060

All statistical analyses were performed using SPSS Statistics 17.0 software (SPSS Inc., Chicago, IL), as previously described.17

Results In vitro inhibitory effect of BAY 1143572 on the proliferation of NK-cell leukemia/lymphoma lines BAY 1143572 was found to inhibit NK-cell leukemia/lymphoma cell line proliferation in a dosedependent manner (Figure 1A). IC50 values for BAY1143572 after 72 hours of incubation for SNT-8, SNK-1, SNT-16, NK-92, MTA, KAI-3, and KHYG-1 were 0.35, 0.22, 0.24, 0.63, 0.20, 0.57, and 0.50 mM, respectively.

BAY 1143572 induces apoptosis in NK-cell leukemia/lymphoma lines BAY 1143572 was found to induce apoptosis of the tested NK-cell leukemia/lymphoma lines in a dosedependent manner. Nearly half of the MTA and KAI-3 cells, 70% of SNT-16 and KHYG-1 cells, 80% of SNK-1 and NK-92 cells, and 100% of SNT-8 cells underwent apoptosis on treatment with 1.0 mM BAY 1143572 for 72 hours (Figure 1B). haematologica | 2018; 103(12)

CDK9 as a potential molecular target in NK cell leukemia/lymphoma

Effect of BAY 1143572 on CDK9 activity in NK-cell leukemia/lymphoma lines

BAY 1143572 inhibits growth of primary ANKL cells isolated from patients

BAY 1143572 had very little effect on the total amount of RNAPII protein in any of the cell lines tested but inhibited phosphorylation of RNAPII CTD at the Ser2 site in a dose-dependent manner. In contrast, phosphorylation of RNAPII at the Ser5 site was not affected by this agent in any of the lines. BAY 1143572 treatment did decrease cMyc and Mcl-1 protein levels in a dose-dependent manner in all of the lines (Figure 2).

IC50 values for BAY 1143572 after 24 hours of incubation with CD56-positive cells isolated from five healthy volunteers were 0.65, 0.56, and 0.27â&#x20AC;&#x201C;1.34 mM, (mean, median, and range, respectively). Cell viability curves for BAY 1143572 showed an initial decrease at concentrations up to around 1.0 uM, almost reaching a plateau at around 100.0 uM. Thus, some of the CD56-positive cells isolated from healthy volunteers remained viable even after expo-

Figure 1. BAY 1143572 inhibits proliferation and induces apoptosis in natural killer (NK) cell leukemia/lymphoma lines. (A) Viability of NK-cell leukemia/lymphoma lines on exposure to different concentrations of BAY 1143572 for 72 hours. The IC50 value is shown for each line. IC50 was defined as the concentration of an inhibitor that reduced cell survival to 50% of the untreated control value, with the highest viability (no inhibitor) defined as 100%, and the lowest viability defined as 0%. Each graph shows one representative result of three independent experiments. (B) NK-cell leukemia/lymphoma lines were treated with different concentrations of BAY 1143572 for 72 hours followed by assessing apoptosis via Annexin V and propidium iodide (PI; nuclear) staining. BAY 1143572 concentrations are indicated above the panels, and the percentage of cells in each quadrant is given. Each graph shows one representative result of three independent experiments.

haematologica | 2018; 103(12)

2061

S. Kinoshita et al.

sure to 100.0 uM BAY 1143572 for 24 hours (Figure 3A). On the other hand, IC50 values for BAY 1143572 after 24 hours of incubation with CD56-positive tumor cells obtained from two different patients with ANKL were 0.16 uM and 0.29 uM (Figure 3B). In contrast to the CD56-positive cells from healthy volunteers, most of the tumor cells were dead after exposure to 3.0 uM BAY 1143572 for 24 hours. Bay 1143572 inhibited phosphorylation of RNAPII CTD at the Ser2 site and reduced the expression of Mcl-1 proteins in the primary tumor cells (Figure 3C).

Phosphorylation status of RNAPII at the Ser2 site in tumor lesions of ENKTL, nasal type RNAPII in ENKTL, nasal type, tumor cells were highly phosphorylated at Ser2 sites. This was also the case in non-tumorous lymphocytes from reactive lymph nodes. Indeed, the degree of phosphorylation of RNAPII at the Ser2 site was identical in both the tumor cells and nontumor lymphocytes. Three representative cases from each group are shown in Online Supplementary Figure S2.

Macroscopic and microscopic findings in primary ANKL cell-bearing mice with or without BAY 1143572 therapy The appearance of mice treated with vehicle only and those treated with BAY 1143572, 22 days after primary ANKL cell inoculation, is shown in Figure 4A, upper and lower panels, respectively. Hepatosplenomegaly was observed in all of the untreated mice, but not in those treated with BAY 1143572 (liver and spleen are demarcated by narrow yellow lines). Immunohistological analyses revealed that the livers of the control mice were infiltrated by atypical lymphoid cells, and the normal architecture was destroyed (Figure 4B, top panels). ISH showed that these atypical cells were positive for EBER (Figure 4B, second panels from the top). On the other hand, the liver architecture of the treated mice was almost intact (Figure 4B, third panels from the top), and EBER-positive cells

were rare (Figure 4B, bottom panels). The same immunohistological analyses also revealed that the spleens of the untreated mice were infiltrated by atypical lymphoid cells, and the normal architecture was completely destroyed (Figure 4C, top panels). ISH also showed that these atypical cells were positive for EBER (Figure 4C, second panels from the top). On the other hand, the splenic architecture of the treated mice, like the liver, was essentially intact (Figure 4C, third panels from the top), and EBER-positive cells were rarely observed (Figure 4C, bottom panels).

BAY 1143572 treatment reduces primary ANKL cells in the blood of mice Twenty-two days after inoculation of primary ANKL cells, the percentage of human tumor cells (identified as CD45- and CD16/56-positive but CD19-negative) in the whole blood of control NOG mouse No.1 was 17.1% (i.e., 17.3% [CD45-positive lymphocyte population] x 98.7% [CD16/56-positive, but CD19-negative cells] = 17.1%) (Figure 5A, the two upper left panels). In control NOG mice Nos. 2, 3, 4, and 5, and in BAY1143572-treated NOG mice Nos. 1, 2, 3, 4, and 5, the percentages of ANKL cells in whole blood, calculated in the same manner, were 30.9, 45.0, 11.1, and 38.4%; and 0.6, 0.1, 0.4, 0.4, and 0.4%, respectively. Thus, BAY 1143572 treatment resulted in significantly decreased percentages of ANKL cells in the blood of these xenogeneic primary tumor-bearing mice (P=0.009; Figure 5A).

BAY 1143572 treatment reduces primary ANKL cells in the bone marrow Twenty-two days after inoculation, the percentage of ANKL tumor cells in the bone marrow of control NOG mouse No.1 was 18.3% (i.e., 26.5% [CD45-positive lymphocyte population] x 69.2% [CD16/56-positive, CD19negative] = 18.3%) (Figure 8, the two upper left panels). In control NOG mice Nos. 2, 3, 4, and 5, and in BAY 1143572-treated NOG mice Nos. 1, 2, 3, 4, and 5, the percentages of ANKL cells in bone marrow, calculated in the

Figure 2. BAY 1143572 affects CDK9 activity in NK-cell leukemia/lymphoma lines. NK-92, MTA, KAI-3, and KHYG-1 cells were treated with the indicated concentrations of BAY 1143572 for 5 hours, followed by Western blotting probing RNA polymerase II (RNAPII), phospho-RNAPII (the serine-2 residue of the C-terminal domain [CTD] [Ser2]), phospho-RNAPII (Ser5), c-Myc, and Mcl-1. Actin was the loading control.

2062

haematologica | 2018; 103(12)

CDK9 as a potential molecular target in NK cell leukemia/lymphoma

same manner, were 17.0, 26.0, 26.0, and 16.8%; and 0.2, 0.3, 0.3, 0.2, and 0.2%, respectively. Thus, BAY1143572 treatment resulted in lowered percentages of ANKL cells infiltrating into the bone marrow (P=0.009; Figure 5B).

BAY 1143572 treatment reduces primary ANKL cells in the liver The percentage of leukemia cells infiltrating into the liver of control NOG mouse No.1 was 32.4% (i.e., 36.9% [CD45-positive lymphocyte population] x 87.8% [CD16/56-positive, CD3-negative] = 32.4%) (Figure 5C, the two upper left panels), 22 days after tumor inoculation. In control NOG mice Nos. 2, 3, 4, and 5, and in BAY 1143572 treated NOG mice Nos. 1, 2, 3, 4, and 5, these figures were 27.6, 29.7, 15.8, and 21.4%; and 0.3, 0.6, 1.8, 0.5, and 0.1%, respectively. Thus, the percentage of primary ANKL cells present in the liver was also decreased by BAY1143572 treatment in these animals (P=0.009; Figure 5C).

left two panels) whereas in control NOG mice Nos. 2, 3, 4, and 5, and in BAY1143572-treated NOG mice Nos.1, 2, 3, 4, and 5, these percentages were 17.6, 21.0, 21.1, and 19.3%; and 0.3, 0.5, 1.0, 0.9, and 2.5%, respectively. Thus, BAY1143572 treatment also decreases primary ANKL cell infiltration into the spleens of these animals (P=0.009; Figure 5D).

BAY1143572 prolongs the survival of mice inoculated with primary ANKL cells Two of the 6 BAY 1143572-treated mice survived up to 34 days after inoculation of the primary ANKL tumor cells, whereas none of the vehicle-treated control mice were alive at that time (P=0.020; Figure 6). No toxicity attributable to BAY 1143572 was observed over this time period in any of the mice. Thus, we conclude that BAY 1143572-treatment of mice inoculated with primary ANKL cells significantly extended their survival relative to the untreated controls.

BAY 1143572 treatment reduces primary ANKL cells in the spleen

Discussion

Using the same experimental protocol as described above, 22 days after inoculation of primary ANKL tumor cells, the percentage of leukemia cells in the spleen of control NOG mouse No.1 was 11.7% (Figure 5D, the upper

NK-cell leukemia/lymphoma have a very poor prognosis necessitating improved treatment regimens for these patients.1-9 Here, we begin to assess whether a selective

C Figure 3. BAY 1143572 inhibits proliferation and affects CDK9 activity in primary ANKL cells. (A) Viability of human CD56-positive cells in control donor PBMC (n=5) assessed in the presence of recombinant human IL-2 (rIL-2, Miltenyi Biotec) at a final concentration of 100 IU/mL together with different concentrations of BAY 1143572 for 24 hours. The IC50 value is indicated in each graph. (B) Viability of primary aggressive NK-cell leukemia (ANKL) cells from two separate patients (patient A, left panel, and patient B, right panel) assessed in the presence of rIL-2 at 100 IU/mL together with different concentrations of BAY 1143572 for 24 hours. The IC50 value is indicated in each graph. IC50 was defined as the concentration of an inhibitor that reduced cell survival to 50% of the untreated control value, with the highest viability (no inhibitor) defined as 100%, and the lowest viability defined as 0%. (C) Primary ANKL cells from patient A were treated with the indicated concentrations of BAY 1143572 for 12 hours in the presence of 100 IU/mL rIL-2, followed by Western blotting probed with antibodies to phospho-RNAPII (Ser2), phospho-RNAPII (Ser5), and Mcl-1. Actin was the loading control.

haematologica | 2018; 103(12)

2063

S. Kinoshita et al.

2064

Figure 4. Macroscopic and microscopic findings in primary ANKL cell-bearing mice treated or not treated with BAY 1143572. (A) Macroscopic appearance of control mice treated with vehicle (control; top panels) or BAY 1143572 (bottom panels). Liver and spleen are demarcated by narrow yellow lines. (B) Photomicrographs of control mouse liver (hematoxylin and eosin [HE] staining) (top panels), and with in situ hybridization (ISH) using an Epsteinâ&#x20AC;&#x201C;Barr virus (EBV)-encoded RNA (EBER) probe (panels second from top). HE-stained liver (third panels from the top), together with ISH using an EBER Probe (bottom panels) from BAY 1143572-treated mice. Scale bars represent 100 mm. (C) HE staining of the spleen of control mice (top panels), together with ISH using an EBER Probe (Leica Microsystems Newcastle Ltd.) (second panels from the top). HE staining of the spleen (third panels from the top), and ISH using an EBER Probe (bottom panels) from mice treated with BAY 1143572. Scale bars represent 100 mm.

haematologica | 2018; 103(12)

CDK9 as a potential molecular target in NK cell leukemia/lymphoma

Figure 5. Significant therapeutic effect of BAY 1143572 in ANKL cell-bearing NOG mice. (A) Flow cytometry of mouse blood cells. Human CD45-positive cells among the mouse cells are indicated in each panel by a red square (upper panels) and are plotted to show CD16/56 and CD19 expression; ANKL cells are CD16/56-positive and CD19-negative as indicated by a red square (lower panels). The percentages of ANKL cells in whole blood of control and BAY 1142572-treated NOG mice are 28.5%, 30.9% (mean, median), and 0.4%, 0.4%, respectively. (B) Flow cytometry of mouse bone marrow cells. Human CD45-expressing cells are again indicated in each panel by red squares (upper panels) and plotted to show CD16/56 and CD19 expression; ANKL cells are CD16/56-positive and CD19-negative as indicated by red squares (lower panels). The percentages of ANKL cells in the bone marrow of control and BAY 1142572-treated NOG mice are 20.8%, 18.3% (mean, median), and 0.2%, 0.2%, respectively. (C) Flow cytometry of mouse liver cell suspensions. Human CD45-expressing cells are indicated in each panel by red squares (upper panels) and are stained for CD16/56 and CD3 expression; ANKL cells are CD16/56-positive and CD3-negative (red squares, lower panels). The percentages of ANKL cells in the livers of control and BAY 1142572-treated NOG mice are 25.4%, 27.6% (mean, median), and 0.7%, 0.5%, respectively. (D) Flow cytometry of mouse spleen cell suspensions. As above, the human CD45-expressing cells are indicated in each panel by red squares (upper panels) and ANKL cells are CD16/56-positive and CD3-negative (red squares, lower panels). The percentages of ANKL cells in the spleens of control and BAY 1142572-treated NOG mice are 18.1%, 19.3% (mean, median), and 1.0%, 0.9%, respectively. The significance of the difference (P value) in the percentages of ANKL cells between two groups is shown in each panel (A) (B) (C), and (D).

haematologica | 2018; 103(12)

2065

S. Kinoshita et al.

Figure 6. Kaplan-Meier survival curves of ANKL cell-bearing NOG mice treated with BAY 1143572 or vehicle. Mice were treated orally with 12.5 mg/kg BAY 1143572 or vehicle (n=6 for both) once daily for 15 days (7-21 days after ANKL cell inoculation). The significance of the difference in survival is shown on the graph.

CDK9/P-TEFb inhibitor, BAY 1143572, may have clinical utility in this regard. We found that BAY 1143572 possessed notable antitumor activity, not only against established NK cell leukemia/lymphoma lines, but also against primary ANKL cells in vitro and in primary ANKL cell-bearing mice in vivo. Hence, CDK9 activity is crucial for NKcell leukemia/lymphoma pathogenesis, implying that transcriptional machinery could represent an appropriate molecular target for developing new treatments for this disease. First, we aimed to investigate the mechanism of action of BAY 1143572 using NK-cell leukemia/lymphoma lines in vitro and determined that it specifically inhibited the phosphorylation of RNAPII CTD at the Ser2 but not the Ser5 site. This resulted in transcriptional repression of RNAPII and thus decreased levels of the downstream proteins c-Myc and Mcl-1. This in turn causes inhibition of growth and induction of apoptosis in NK-cell leukemia/lymphoma lines whether they originate from NK cells (SNK-1, NK-92, KAI-3, or KHY-G) or T cells (SNT-8, SNT-16, or MTA). Whether they were EBV-positive (SNT-8, SNK-1, SNT-16, NK-92, or KAI-3) or -negative (MTA or KHY-G) was also found not to affect the anticancer activity of BAY 1143572. Very similar results were obtained using primary ANKL cells freshly isolated from a patient, rather than established cell lines. BAY 1143572 markedly reduced phosphorylation of RNAPII CTD at the Ser2 site, but only weakly at the Ser5 site, which results in downregulation of the downstream protein, Mcl-1. Importantly, our study showed that NK-cell leukemia/lymphoma are more sensitive to the CDK9/PTEFb inhibitor than their normal cell counterparts. Exposure to ≥1.0 mM of BAY 1143572 for 24 hours killed almost all primary ANKL cells from two different patients, but also a fraction of CD56-positive cells from healthy 2066

volunteers. Nonetheless, after 24 hours´ exposure to 10.0 mM BAY 1143572, 20-40% of CD56-positive cells from controls were still viable. This clearly documents that BAY 1143572 is far less toxic to CD56-positive cells from healthy donors than to CD56-positive primary ANKL tumor cells. It remains unclear why NK-cell leukemia/lymphoma cells are more sensitive to the CDK9/P-TEFb inhibitor. One possibility is that CDK9 kinase and RNAPII are more highly activated in malignant than in normal cells. Supporting this, another group has reported higher CDK9 expression accompanied by greater phosphorylation of RNAPII CTD in tumor cells than in normal cells.30 Additionally, a different group reported that CDK9 expression appeared to be related to particular stages of lymphoid differentiation/activation.31 However, we did not observe any significant differences between tumor cells and non-tumor lymphocytes in the phosphorylation level of RNAPII at the Ser2 site in the present study. A possible explanation for this is that the malignant cells are more “dependent” on the transcriptional machinery involving RNAPII to sustain the expression of critical oncogenes supporting their survival and proliferation, such as Mcl-1 and c-Myc.32,33 In this context, inhibition of the transcriptional machinery would result in an acute and concurrent downregulation of these oncogenes, thereby leading to cell death. In fact, inhibition of CDK7 or BET bromodomain 4, which also block the activity of RNAPII in a similar manner to CDK9/PTEF-b inhibition, has been reported to result in potent anti-cancer activity in several malignancies.34-38 Importantly, many types of cancer and leukemia cells are more sensitive to these inhibitors than non-transformed cells.39,40 These studies suggest that expression of oncogenes may need to be maintained at a high level, and thus that cancer cells are likely to be more sensitive to RNAPII inhibition. Collectively, these earlier haematologica | 2018; 103(12)

CDK9 as a potential molecular target in NK cell leukemia/lymphoma

investigations and the present studies suggest a great advantage for the clinical use of CDK9 inhibitors. Future investigation is needed to analyze the status of regulatory elements after BAY 1143572 treatment. In the present study, we have established a mouse xenogeneic model for pre-clinical testing of a CDK9/P-TEFb inhibitor. Severe immune deficiencies in NOG mice permit engraftment of human immune cells, where they may retain very similar functions in an in vivo environment that can be manipulated experimentally.41,42 In these mice, primary EBV-positive, CD16/CD56-positive, CD3â&#x20AC;&#x201C; and CD19-negative cells from ANKL patients were found to robustly infiltrate several organs such as the spleen, liver and bone marrow. These features are very similar to those seen in the ANKL patients who donated the cells. To the best of our knowledge, this is the first report of primary ANKL cell-bearing mice in which it could be shown that ANKL cells are maintained by serial transplantations. This is a useful model because these tumor cells cannot be cultured as cell lines in vitro, implying that the in vivo microenvironment is an absolute requirement for tumor cell survival. Hence, our present in vivo ANKL model represents the in vivo human ANKL environment better than other models using established tumor cell lines. Thus, this in vivo xenogeneic primary tumor model offers a useful methodology for probing ANKL pathogenesis and provides a more relevant tool for testing novel antitumor agents. Here, we report that BAY 1143572 possesses strong antitumor activity, as demonstrated in vivo by reduced ANKL cell infiltration into blood, bone marrow, liver and spleen in this model. Moreover, this model also allowed us to demonstrate that BAY 1143572 monotherapy can prolong survival of ANKL-bearing hosts. It should also be noted that the in vivo antitumor activity of BAY 1143572 in the primary tumor cell-bearing NOG mice was actually mediated by the on-target effect of BAY 1143572; namely, the inhibition of CDK9 and subsequent inhibition of phosphorylation at the Ser2 site of the RNAPII CTD was clearly demonstrated in our previous study on ATL.17 It is also important to note that no BAY 1143572 toxicity was observed in any of the mice studied here. Taken together, these findings in the primary ANKL mice strongly suggest that targeting CDK9 in human ANKL patients

References 1. Chan JKC, Quintanilla-Martinez L, and Ferry JA. Extranodal NK/T-cell lymphoma, nasal type. In: Swerdlow SH, Campo E, Harris NL, et al., eds. WHO classification of tumours of haematopoietic and lymphoid tissues (Revised 4th Edition, pp 368-371). Lyon, France, International Agency for Research on Cancer (IARC), 2017. 2. Chan JKC, Jaffe ES, and Ko YH. Aggressive NK-cell leukaemia. In: Swerdlow SH, Campo E, Harris NL, et al., eds. WHO classification of tumours of haematopoietic and lymphoid tissues (Revised 4th Edition, pp 353-354). Lyon, France, International Agency for Research on Cancer (IARC), 2017. 3. Ishida F, Ko YH, Kim WS, et al. Aggressive natural killer cell leukemia: therapeutic potential of L-asparaginase and allogeneic hematopoietic stem cell transplantation.

haematologica | 2018; 103(12)

could be a promising therapeutic approach. To the best of our knowledge, this is the first report evaluating the efficacy of CDK9-targeted therapy in NKcell leukemia/lymphoma, including ANKL. It must be noted that several other anti-cancer agents targeting CDK9 have already been tested in the clinic, but with little benefit and they were accompanied by numerous adverse events.43-47 The latter may have been due to their insufficient selectivity for CDK9.48 In contrast, BAY 1143572 exhibits marked selectivity for CDK9/PTEFb,21 and possesses strong antitumor effects against NK-cell leukemia/lymphoma including ANKL, not only in vitro, but also in animal models in vivo. Because, in clinical practice, NK-cell leukemia/lymphoma is highly aggressive and refractory, combination strategies such as CDK9 inhibition together with PD-1 blockade should be worth considering.49,50 In conclusion, we propose that BAY 1143572 holds great promise as a potential agent to treat NK-cell leukemia/lymphoma, including ANKL patients. The mechanism of action is via CDK9/P-TEFb inhibition. Clinical evaluation of CDK9/P-TEFb selective inhibitors in patients with NK cell leukemia/lymphoma is therefore warranted. Acknowledgments We thank Ms Chiori Fukuyama for excellent technical assistance and Ms Naomi Ochiai for excellent secretarial assistance. We also thank Stuart Ince (Bayer US, LLC) for the kind contribution of obtaining BAY 1143572. Funding This work was supported by research funding from Bayer AG to Nagoya City University Graduate School of Medical Sciences. This work was also supported by grants-in-aid for scientific research (B) (No. 16H04713 to Takashi Ishida), grants-in-aid from the National Cancer Center Research and Development Fund (Nos. 29-A-3 to Takashi Ishida, and Shinsuke Iida), and grants-in-aid from the Japan Agency for Medical Research and Development (Nos. 17ck0106287h0001 to Takashi Ishida, 16cm0106301h0001 to Takashi Ishida, and 15ck0106132h0002 to Takashi Ishida, Ryuzu Ueda, and Shinsuke Iida).

Cancer Sci. 2012;103(6):1079-1083. 4. Song SY, Kim WS, Ko YH, Kim K, Lee MH, Park K. Aggressive natural killer cell leukemia: clinical features and treatment outcome. Haematologica. 2002;87(12):13431345. 5. Suzuki R, Suzumiya J, Nakamura S, et al. Aggressive natural killer-cell leukemia revisited: large granular lymphocyte leukemia of cytotoxic NK cells. Leukemia. 2004;18(4):763-770. 6. Yamaguchi M, Kwong YL, Kim WS, et al. Phase II study of SMILE chemotherapy for newly diagnosed stage IV, relapsed, or refractory extranodal natural killer (NK)/Tcell lymphoma, nasal type: the NK-Cell Tumor Study Group study. J Clin Oncol. 2011;29(33):4410-4416. 7. Kwong YL, Kim WS, Lim ST, et al. SMILE for natural killer/T-cell lymphoma: analysis of safety and efficacy from the Asia

10.

11. 12.

Lymphoma Study Group. Blood. 2012;120(15):2973-2980. Kim SJ, Yoon DH, Jaccard A, et al. A prognostic index for natural killer cell lymphoma after non-anthracycline-based treatment: a multicentre, retrospective analysis. Lancet Oncol. 2016;17(3):389-400. Jung KS, Cho SH, Kim SJ, Ko YH, Kang ES, Kim WS. L-asparaginase-based regimens followed by allogeneic hematopoietic stem cell transplantation improve outcomes in aggressive natural killer cell leukemia. J Hematol Oncol. 2016;9:41. Bywater MJ, Pearson RB, McArthur GA, Hannan RD. Dysregulation of the basal RNA polymerase transcription apparatus in cancer. Nat Rev Cancer. 2013;13(5):299-314. Morales F, Giordano A. Overview of CDK9 as a target in cancer research. Cell Cycle. 2016;15(4):519-527. Olson CM, Jiang B, Erb MA, et al.

2067

S. Kinoshita et al.

13.

14.

15.

16.

17.

18.

19.

20. 21.

22.

23.

24.

25.

2068

Pharmacological perturbation of CDK9 using selective CDK9 inhibition or degradation. Nat Chem Biol. 2018;14(2):163-170. Huang CH, Lujambio A, Zuber J, et al. CDK9-mediated transcription elongation is required for MYC addiction in hepatocellular carcinoma. Genes Dev. 2014;28(16):18001814. Gregory GP, Hogg SJ, Kats LM, et al. CDK9 inhibition by dinaciclib potently suppresses Mcl-1 to induce durable apoptotic responses in aggressive MYC-driven B-cell lymphoma in vivo. Leukemia. 2015;29(6):1437-1441. Baker A, Gregory GP, Verbrugge I, et al. The CDK9 inhibitor dinaciclib exerts potent apoptotic and antitumor effects in preclinical models of MLL-rearranged acute myeloid leukemia. Cancer Res. 2016;76(5):1158-1169. Tong Z, Chatterjee D, Deng D, et al. Antitumor effects of cyclin dependent kinase 9 inhibition in esophageal adenocarcinoma. Oncotarget. 2017;8(17):2869628710. Narita T, Ishida T, Ito A, et al. Cyclin-dependent kinase 9 is a novel specific molecular target in adult T-cell leukemia/lymphoma. Blood. 2017; 130(9):1114-1124. Bark-Jones SJ, Webb HM, West MJ. EBV EBNA 2 stimulates CDK9-dependent transcription and RNA polymerase II phosphorylation on serine 5. Oncogene. 2006; 25(12):1775-1785. Palermo RD, Webb HM, West MJ. RNA polymerase II stalling promotes nucleosome occlusion and pTEFb recruitment to drive immortalization by Epstein-Barr virus. PLoS Pathog. 2011;7(10):e1002334. Zaborowska J, Isa NF, Murphy S. P-TEFb goes viral. Inside Cell. 2016;1(2):106-116. Lucking U, Scholz A, Lienau P, et al. Identification of atuveciclib (BAY 1143572), the first highly selective, clinical PTEFb/CDK9 inhibitor for the treatment of cancer. ChemMedChem. 2017; 12(21):17761793. Zhang Y, Nagata H, Ikeuchi T, et al. Common cytological and cytogenetic features of Epstein-Barr virus (EBV)-positive natural killer (NK) cells and cell lines derived from patients with nasal T/NK-cell lymphomas, chronic active EBV infection and hydroa vacciniforme-like eruptions. Br J Haematol. 2003;121(5):805-814. Gong JH, Maki G, Klingemann HG. Characterization of a human cell line (NK92) with phenotypical and functional characteristics of activated natural killer cells. Leukemia. 1994;8(4):652-8. Emi N, Abe A, Kasai M, et al. CD4- and CD56-positive T-cell line, MTA, established from natural killer-like T-cell leukemia/lymphoma. Int J Hematol. 1999;69(3):180-185. Tsuge I, Morishima T, Morita M, Kimura H, Kuzushima K, Matsuoka H.

26.

27.

28.

29.

30.

31.

32.

33. 34. 35.

36.

37.

38.

Characterization of Epstein-Barr virus (EBV)-infected natural killer (NK) cell proliferation in patients with severe mosquito allergy; establishment of an IL-2-dependent NK-like cell line. Clin Exp Immunol. 1999;115(3):385-392. Yagita M, Huang CL, Umehara H, et al. A novel natural killer cell line (KHYG-1) from a patient with aggressive natural killer cell leukemia carrying a p53 point mutation. Leukemia. 2000;14(5):922-930. Ishida T, Iida S, Akatsuka Y, et al. The CC chemokine receptor 4 as a novel specific molecular target for immunotherapy in adult T-Cell leukemia/lymphoma. Clin Cancer Res. 2004;10(22):7529-7539. Ri M, Iida S, Ishida T, et al. Bortezomibinduced apoptosis in mature T-cell lymphoma cells partially depends on upregulation of Noxa and functional repression of Mcl-1. Cancer Sci. 2009;100(2):341-348. Ishida T, Utsunomiya A, Iida S, et al. Clinical significance of CCR4 expression in adult Tcell leukemia/lymphoma: its close association with skin involvement and unfavorable outcome. Clin Cancer Res. 2003;9(10):36253634. Zhang Y, Zhou L, Leng Y, Dai Y, Orlowski RZ, Grant S. Positive transcription elongation factor b (P-TEFb) is a therapeutic target in human multiple myeloma. Oncotarget. 2017;8(35):59476-59491. Bellan C, De Falco G, Lazzi S, et al. CDK9/CYCLIN T1 expression during normal lymphoid differentiation and malignant transformation. J Pathol. 2004;203(4):946952. Wong RWJ, Ishida T, Sanda T. Targeting general transcriptional machinery as a therapeutic strategy for adult T-cell leukemia molecules. 2018;23(5). Hnisz D, Abraham BJ, Lee TI, et al. Superenhancers in the control of cell identity and disease. Cell. 2013;155(4):934-947. Filippakopoulos P, Qi J, Picaud S, et al. Selective inhibition of BET bromodomains. Nature. 2010;468(7327):1067-1073. Delmore JE, Issa GC, Lemieux ME, et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell. 2011;146(6):904-917. Roderick JE, Tesell J, Shultz LD, et al. c-Myc inhibition prevents leukemia initiation in mice and impairs the growth of relapsed and induction failure pediatric T-ALL cells. Blood. 2014;123(7):1040-1050. Wong RWJ, Ngoc PCT, Leong WZ, et al. Enhancer profiling identifies critical cancer genes and characterizes cell identity in adult T-cell leukemia. Blood. 2017; 130(21):23262338. Hnisz D, Shrinivas K, Young RA, Chakraborty AK, Sharp PA. A phase separation model for transcriptional control. Cell. 2017;169(1):13-23.

39. Kwiatkowski N, Zhang T, Rahl PB, et al. Targeting transcription regulation in cancer with a covalent CDK7 inhibitor. Nature. 2014;511(7511):616-620. 40. Walsby E, Pratt G, Shao H, et al. A novel Cdk9 inhibitor preferentially targets tumor cells and synergizes with fludarabine. Oncotarget. 2014;5(2):375-385. 41. Mori F, Ishida T, Ito A, et al. Potent antitumor effects of bevacizumab in a microenvironment-dependent human lymphoma mouse model. Blood Cancer J. 2012; 2(4):e67. 42. Ito A, Ishida T, Utsunomiya A, et al. Defucosylated anti-CCR4 monoclonal antibody exerts potent ADCC against primary ATLL cells mediated by autologous human immune cells in NOD/Shi-scid, IL-2R gamma(null) mice in vivo. J Immunol. 2009; 183(7):4782-4791. 43. Kar JE, Garrett-Mayer E, Estey EH, et al. Randomized phase II study of two schedules of flavopiridol given as timed sequential therapy with cytosine arabinoside and mitoxantrone for adults with newly diagnosed, poor-risk acute myelogenous leukemia. Haematologica. 2012; 97(11):1736-1742. 44. Nemunaitis JJ, Small KA, Kirschmeier P, et al. A first-in-human, phase 1, dose-escalation study of dinaciclib, a novel cyclin-dependent kinase inhibitor, administered weekly in subjects with advanced malignancies. J Transl Med. 2013;11:259. 45. Le Tourneau C, Faivre S, Laurence V, et al. Phase I evaluation of seliciclib (R-roscovitine), a novel oral cyclin-dependent kinase inhibitor, in patients with advanced malignancies. Eur J Cancer 2010;46(18):32433250. 46. Tong WG, Chen R, Plunkett W, et al. Phase I and pharmacologic study of SNS-032, a potent and selective Cdk2, 7, and 9 inhibitor, in patients with advanced chronic lymphocytic leukemia and multiple myeloma. J Clin Oncol. 2010;28(18):3015-3022. 47. van der Biessen DA, Burger H, de Bruijn P, et al. Phase I study of RGB-286638, a novel, multitargeted cyclin-dependent kinase inhibitor in patients with solid tumors. Clin Cancer Res. 2014;20(18):4776-4783. 48. Asghar U, Witkiewicz AK, Turner NC, Knudsen ES. The history and future of targeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug Discov. 2015; 14(2):130-146. 49. Kwong YL, Chan TSY, D, et al. PD1 blockade with pembrolizumab is highly effective in relapsed or refractory NK/T-cell lymphoma failing l-asparaginase. Blood. 2017;129(17):2437-2442. 50. Li X, Cheng Y, Zhang M, et al. Activity of pembrolizumab in relapsed/refractory NK/T-cell lymphoma. J Hematol Oncol. 2018;11(1):15.

haematologica | 2018; 103(12)

ARTICLE

Chronic Lymphocytic Leukemia

Trisomy 12 chronic lymphocytic leukemia expresses a unique set of activated and targetable pathways

Ferrata Storti Foundation

Lynne V. Abruzzo,1* Carmen D. Herling,2 George A. Calin,3 Christopher Oakes,4 Lynn L. Barron,5 Haley E. Banks,5 Vikram Katju,6 Michael J. Keating7 and Kevin R. Coombes6*

1 Department of Pathology, The Ohio State University, Columbus, OH, USA; 2Department I for Internal Medicine and Center of Integrated Oncology, University of Cologne, Germany; 3Department of Experimental Therapeutics, The University of Texas, MD Anderson Cancer Center, Houston, TX, USA; 4Department of Internal Medicine, The Ohio State University, Columbus, OH, USA; 5Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 6Department of Biomedical Informatics, The Ohio State University, Columbus, OH, USA and 7Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

Haematologica 2018 Volume 103(12):2069-2078

*LVA and KRC contributed equally to this study

ABSTRACT

lthough trisomy 12 (+12) chronic lymphocytic leukemia (CLL) comprises about 20% of cases, relatively little is known about its pathophysiology. These cases often demonstrate atypical morphological and immunophenotypic features, high proliferative rates, unmutated immunoglobulin heavy chain variable region genes, and a high frequency of NOTCH1 mutation. Patients with +12 CLL have an intermediate prognosis, and show higher incidences of thrombocytopenia, Richter transformation, and other secondary cancers. Despite these important differences, relatively few transcriptional profiling studies have focused on identifying dysregulated pathways that characterize +12 CLL, and most have used a hierarchical cytogenetic classification in which cases with more than one recurrent abnormality are categorized according to the abnormality with the poorest prognosis. In this study, we sought to identify protein-coding genes whose expression contributes to the unique pathophysiology of +12 CLL. To exclude the likely confounding effects of multiple cytogenetic abnormalities on gene expression, our +12 patient cohort had +12 as the sole abnormality. We profiled samples obtained from 147 treatment-naĂŻve patients. We compared cases with +12 as the only cytogenetic abnormality to cases with only del(13q), del(11q), or diploid cytogenetics using independent discovery (n=97) and validation (n=50) sets. We demonstrate that CLL cases with +12 as the sole abnormality express a unique set of activated pathways compared to other cytogenetic subtypes. Among these pathways, we identify the NFAT signaling pathway and the immune checkpoint molecule, NT5E (CD73), which may represent new therapeutic targets.

Correspondence: lynne.abruzzo@osumc.edu

Received: February 5, 2018. Accepted: June 29, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.190132 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2069 ÂŠ2018 Ferrata Storti Foundation

Introduction Chromosomal abnormalities, predominantly gains and losses, are strong predictors of disease progression and survival in chronic lymphocytic leukemia (CLL). Fluorescence in situ hybridization (FISH) assays on interphase nuclei have demonstrated that approximately 80% of cases contain non-random gains or losses of chromosomal material, many with prognostic significance.1 Deletions in 13q14 (del(13q)) are most common, followed by deletions in 11q22.3-q23.1 (del(11q)), trisomy 12 (+12), and deletions in 6q21-q23 (del(6q)) and 17p13 (del(17p)).1 Del(13q), associated with a good prognosis, is the site of the microRNA genes, miR-15a/16-1, which negatively regulate BCL2 post-transcriptionally.2 Their deletion results in overexpression of the anti-apoptotic protein BCL2. In contrast, haematologica | 2018; 103(12)

2069

L.V. Abruzzo et al.

del(17p), the site of TP53, and del(11q), the site of ATM and the miR34b/c cluster, are markers of poor prognosis.1,3 Cases with +12 have an overall survival (OS) that lies in the middle range between that of cases with del(13q) and those with del(11q) or del(17p). Although +12 CLL comprises approximately 20% of cases, relatively little is known about its pathophysiology. These cases often demonstrate atypical morphological and immunophenotypic features.4 Patients with +12 CLL show a higher incidence of thrombocytopenia, Richter transformation, and other secondary cancers, their main cause of death.5 However, compared to other cytogenetically-defined CLL subtypes, few studies have attempted to identify the critical protein-coding and microRNA genes that are relevant to its pathophysiology.6-9 For many cancer types, gene dosage correlates strongly with mRNA, microRNA, and protein expression.6,9 This suggests that at least a subset of the more than 1000 protein-coding genes (and more than 40 microRNA genes) on chromosome 12 in +12 CLL are likely to show a concordant increase in expression. Conceivably, some of these proteins modulate expression of downstream targets, either on chromosome 12 or other chromosomes, resulting in aberrant gene expression. We aimed to identity protein-coding genes whose expression contributes to the unique pathophysiology of +12 CLL. We performed transcriptional profiling on CLL cases with +12 as the only cytogenetic abnormality, and compared them to cases with only del(13q), del(11q), or diploid cytogenetics. We demonstrate that CLL cases with +12 as the only cytogenetic abnormality express a unique set of activated pathways compared to other cytogenetic subtypes, several of which may represent new therapeutic targets.

cohort to the others individually and jointly. We assessed DE genes between subtypes by performing probe-by-probe ANOVA (for multiple subtypes) or t-tests (for two subtypes). We fit a betauniform-mixture (BUM) model to the set of P-values in order to find the false discovery rate (FDR). Microarray data are available at http://silicovore.com/CLL/Trisomy12. For validation, we performed transcriptional profiling using an MF-QRT-PCR assay to a subset of DE genes identified using microarrays, along with 5 endogenous control genes.13 We performed gene-by-gene ANOVA and t-tests to validate DE genes discovered from the microarrays. A discovery was considered “validated” if the unadjusted P-value was <0.05.

Survival analysis We performed time-to-event (survival) analysis using Cox proportional hazards models, and assessed significance using the logrank (score) test. To assess multivariate models, we used a forward-backward stepwise algorithm to eliminate redundant factors and optimize the Akaike Information Criterion (AIC). We performed the computations using the survival package (v.2.40-1) in R v.3.3.0, and computed median follow-up times using the reverse Kaplan-Meier estimator.14

Pathway analysis We performed pathway analysis using Ingenuity Pathway Analysis (IPA; Qiagen, Redwood City, CA, USA). The statistical significance of altered pathways was calculated using a one-sided Fisher exact test. The z-score indicates if a pathway is activated (positive) or down-regulated (negative). We also examined ratio values (the number of molecules with expression levels above or below the mean, divided by the total number of molecules in that pathway) for the canonical pathways.

Results Methods

Patients' characteristics of discovery and validation sets

Sample collection and preparation

Based on SNP genotyping, we divided cases into cytogenetic subsets defined by abnormalities that would be detected using a FISH probe panel to the common CLL cytogenetic abnormalities: del(11q), del(13q), del(17p), +12, and diploid. The proportion of cases in each subset is similar to that reported previously.7 Because patients were treatment-naïve, there was only one del(17p) case, which we excluded from subsequent analysis; a single case is insufficient to achieve statistical significance in a comparison of gene expression profiles to other cytogenetic subtypes. We focused our experiments on cases with only +12 compared to cases with only del(13q), del(11q), or diploid cytogenetics. We identified 147 CLL samples that met our inclusion criteria. One hundred and twenty-three contained a single abnormality or were diploid: 27 with +12 as the only abnormality (18%), 49 with del(13q) as the only abnormality (33%), and 47 diploid cases (32%). The discovery set was chosen to contain patients with only abnormalities who later went on to receive front-line therapy with FCR. Because there were relatively few cases with del(11q) as the only abnormality, to achieve statistical significance, we used 10 cases with del(11q) as the only abnormality for the discovery set, but included 14 cases with del(11q) and del(13q) in the validation set. Patients' characteristics are summarized in Tables 1 and 2. The discovery set contained no statistically significant

Between 2000 and 2008, we obtained peripheral blood (PB) from 250 treatment-naïve CLL patients. This study was approved by the Institutional Review Board and conducted according to principles expressed in the Declaration of Helsinki. We extracted nucleic acids from negatively-selected CLL cells (CD19+) and prepared cDNA, as described previously.10 Cases were divided into discovery and validation sets, as described in the Online Supplementary Methods.

Single nucleotide polymorphism (SNP) genotyping, IGHV and NOTCH1 mutation status, and ZAP70 protein expression We assessed genomic copy number variations (CNV) by single nucleotide polymorphism (SNP) genotyping.10 The IGHV somatic mutation status was assessed as described previously, with minor modifications. Patient and germline sequences were aligned in VBASE II. Cases with <2% mutations compared to germline were designated “unmutated”; cases with ≥2% mutations were designated “mutated”.11 NOTCH1 exon 34 mutation hotspots were assessed as described previously.12 We assessed ZAP70 protein expression by immunohistochemistry or flow cytometry.

Transcriptional profiling of protein-coding genes For discovery, we performed transcriptional profiling using gene expression microarrays (Online Supplementary Methods).13 To identify differentially expressed (DE) genes, we compared the +12 2070

haematologica | 2018; 103(12)

Trisomy 12 CLL Table 1. Patients' characteristics of the discovery set.

Age at diagnosis (years) Gender, n (%) Rai stage, n (%) WBC count, n (%) WBC count, (1x109/L) Prolymphocytes, (% in PB) β2M, n (%)L Immunophenotype, n (%) Light chain subtype, n (%) IGHV status, n (%) ZAP70 status, n (%) CD38 expression, n (%) NOTCH1 status, n (%)

Median Range Male Female 0-2 3-4 <150 G/L ≥150 G/L Median Range Median Range ≤4 mg/L >4 mg/L Atypical Typical Kappa Lambda Mutated Unmutated Positive Negative ≥30% <30% Mutated Unmutated

Sole +12 (n=15)

Sole del(13q) (n=40)

Sole del(11q) (n=10)

Diploid (n=32)

Statistic P*

57 51 – 74 12 (80%) 3 (20%) 12 (80%) 3 (20%) 12 (75%) 3 (25%) 77 23 – 364 6 1 - 22 12 (80%) 3 (10%) 7 (45%) 8 (55%) 8 (55%) 7 (45%) 5 (33%) 10 (67%) 8 (62%) 5 (38%) 5 (33%) 10 (67%) 6 (40%) 9 (60%)

55 27 – 70 30 (75%) 10 (25%) 30 (75%) 10 (25%) 33 (85%) 6 (15%) 110 9 – 319 5 0 – 19 30 (77%) 9 (23%) 5 (16%) 33 (87%) 17 (45%) 21 (55%) 24 (60%) 16 (40%) 14 (40%) 21 (60%) 3 (8%) 35 (92%) 1 (3%) 34 (97%)

57 48 – 80 9 (90%) 1 (10%) 10 (100%) 0 (0%) 10 (100%) 0 (0%) 57 30 – 135 3 0 – 18 6 (60%) 4 (40%) 3 (30%) 7 (70%) 8 (80%) 2 (20%) 0 (0%) 10 (100%) 3 (30%) 7 (79%) 4 (40%) 6 (60%) 0 (0%) 10 (100%)

52 34 – 77 24 (75%) 8 (25%) 22 (69%) 10 (31%) 26 (81%) 6 (19%) 71 21 – 372 5 0 – 13 20 (63%) 12 (37%) 6 (19%) 26 (81%) 24 (75%) 8 (25%) 7 (22%) 25 (78%) 17 (65%) 9 (35%) 9 (28%) 23 (72%) 4 (13%) 26 (87%)

F = 2.4008 P = 0.0727 χ2 = 1.1993 P = 0.7532 χ2 = 4.2644 P = 0.2343 χ2 = 2.3084 P = 0.509 F = 0.8942 P = 0.4473 F = 2.0346 P = 0.1147 χ2 =2.9538 P = 0.3988 χ2 = 7.5932 P =0.0552 χ2 = 8.633 P = 0.0346 χ2 = 18.154 P = 0.0004 χ2 =5.4413 P = 0.1422 χ2 = 8.088 P = 0.0442 χ2 = 15.076 P = 0.0017

WBC: white blood cell; PB: peripheral blood; β2M: serum β2 microglobulin; n: number. *Continuous variables were evaluated with analysis of variance (F-test); categorical variables were evaluated with a χ2 test. Values in bold are statistically significant.

differences between subtypes with respect to age at diagnosis, gender, Rai stage, white blood cell (WBC) count, prolymphocyte percentage, serum beta-2 microglobulin (β2M), or ZAP-70 expression at the time samples were obtained (Table 1). The association between +12 and atypical immunophenotype (moderate to strong expression of at least two markers including CD22, CD79b, strong surface immunoglobulin, and FMC7) approached significance (χ2 test, P=0.0552).4,15 Trisomy 12 cases were more likely to demonstrate unmutated IGHV genes (P=0.0004) and express CD38 (P=0.044) than del(13q) cases, with statistical significance. The proportion of del(11q) or diploid cases with unmutated IGHV genes and CD38 expression was similar to +12 cases. ZAP70 protein expression was not statistically different between the groups. Trisomy 12 cases contained 6 out of 15 (40%) cases with NOTCH1 PEST domain truncation mutation. Collectively, the other subtypes contained 5 out of 75 (7%) NOTCH1-mutated cases. The validation set contained no statistically significant differences between subtypes with respect to age at diagnosis, gender, Rai stage, WBC count, serum β2M, or ZAP-70 expression at the time samples were obtained (Table 2). haematologica | 2018; 103(12)

The association between +12 and atypical immunophenotype approached significance (χ2 test, P=0.06757).4,15 Trisomy 12 cases were more likely to demonstrate unmutated IGHV genes (P=0.0015), and express ZAP70 (P=0.0136) or CD38 (P=0.0275) than del(13q) cases, with statistical significance. The proportion of del(11q) or diploid cases with unmutated IGHV genes and CD38 expression was similar to +12 cases. There was no statistical difference in ZAP70 protein expression between the three groups. Corresponding to their increased frequency of poor prognostic features (i.e. unmutated IGHV status, CD38 positivity), +12 patients required treatment earlier during their disease course (median, 22.5 months) than del(13q) patients (median, 27.5 months), and at around the same timepoint as diploid patients (median, 23.1 months), but later than sole del(11q) patients (median, 9.6 months) (Figure 1A). All 97 discovery set patients and 22 out of 44 (50%; treatment information unavailable for 6 cases) validation set patients subsequently received front-line chemoimmunotherapy with fludarabine, cyclophosphamide, and rituximab (FCR). Trisomy 12 patients had longer progression-free survival (PFS) after treatment (median, >150 months) than patients with del(13q) (medi2071

L.V. Abruzzo et al.

Table 2. Patients' characteristics of the validation set.

Median Range Male Female 0-2 3-4 <150 G/L ≥150 G/L Median Range Median Range ≤4 mg/L >4 mg/L Atypical Typical Kappa Lambda Mutated Unmutated Positive Negative ≥30% <30% Mutated Unmutated

Sole +12 (n=12)

Sole del(13q) (n=9)

del(11q) (n=14)

Diploid (n=15)

Statistic P

59 39 – 82 7 (58%) 5 (42%) 11 (92%) 1 (8%) 10 (83%) 2 (17%) 73 8 – 364 7 1 – 10 9 (75%) 3 (25%) 6 (50%) 6 (50%) 8 (67%) 4 (33%) 3 (25%) 9 (75%) 4 (44%) 5 (56%) 5 (42%) 7 (58%) NA

60 38 – 70 4 (44%) 5 (56%) 7 (78%) 2 (22%) 8 (89%) 1 (11%) 75 41 – 193 1 0 – 11 9 (100%) 0 (0%) 3 (43%) 4 (57%) 4 (57%) 3 (43%) 8 (100%) 0 (0%) 0 (0%) 7 (100%) 0 (0%) 7 (100%) NA

64 40 – 81 7 (87%) 1 (13%) 8 (100%) 0 (0%) 7 (87%) 1 (13%) 82 35 – 206 1 0–5 5 (63%) 3 (37%) 0 (0%) 8 (100%) 7 (87%) 1 (13%) 1 (13%) 7 (87%) 6 (86%) 1 (14%) 6 (75%) 2 (25%) NA

52 38 – 78 7 (47%) 8 (53%) 13 (87%) 2 (13%) 12 (80%) 3 (20%) 75 30 – 209 2 0–5 9 (60%) 6 (40%) 3 (20%) 12 (80%) 11 (73%) 4 (27%) 8 (57%) 7 (43%) 7 (54%) 6 (46%) 5 (33%) 10 (67%) NA

F = 0.4872 P =0.6931 χ2 = 4.2724 P = 0.2355 χ2 = 2.2464 P = 0.5229 χ2 = 0.4172 P = 0.9367 F = 0.0388 P = 0.9896 F = 8.7786 P = 0.0001 χ2 = 5.0531 P = 0.1679 χ2 = 7.140 P = 0.06757 χ2 = 1.8725 P = 0.5993 χ2 = 15.432 P = 0.0015 χ2 = 10.681 P = 0.0136 χ2 = 9.137 P = 0.0275 NA

WBC: white blood cell count; β2M: β2 microglobulin; n: number; PB: peripheral blood. *Continuous variables were evaluated with analysis of variance (F-test); categorical variables were evaluated with a χ2 test. Values in bold are statistically significant.

an, 61.5 months), del(11q) (median, 62.5 months), or diploid cytogenetics (median, 66.2 months) (Figure 1B). With a median follow up of 146 months (95%CI: 144-157 months) from sample, and 181 months (95%CI: 168-200 months) from diagnosis, we found no statistically significant difference in overall survival (OS). Results for the cytogenetic subsets in the validation and combined datasets were similar.

Trisomy 12 cases have a unique gene expression profile Next we performed univariate probe-by-probe analysis of variance (ANOVA) and identified 1263 probes representing 1012 unique protein-coding genes, 40 ncRNAs, and 22 ESTs that were differentially expressed between at least two of the four cytogenetic subtypes (FDR=1%; unadjusted P=0.00385). Clustering samples using all 1263 probes showed that the gene expression signature of +12 cases (green) was distinct from the other subtypes (Figure 2). (See Online Supplementary Figure S1 for principal components analysis.) Similarly, most del(13q) cases formed a single cluster (blue), as did the del(11q) cases (pink). In contrast, diploid cases were found in three clusters, admixed with a subset of del(13q) cases. We identified no 2072

clinical or laboratory features to account for 4 of the diploid cases clustering with +12 cases. Because the genes selected for this analysis were based on ANOVA, and known only to be different between at least two of the cytogenetic subtypes, it is significant that the +12 cluster is clearly distinct from the other subtypes. Thus, +12 CLL has a unique gene expression profile.

Differences between +12 and del(13q), del(11q), or diploid CLL cases To identify DE probes/genes between +12 cases and del(13q), del(11q), and diploid cases, respectively, we performed univariate t-tests. Comparing +12 and del(13q) cases, we identified 1181 DE probes representing 927 unique protein-coding genes, 41 ncRNAs, and 15 ESTs (FDR=1%, unadjusted P=0.000333). Thirty-one (2.6%) of the DE probes represent genes (e.g. BCL2, EIF4B, EIF4E, PIM1) that are known or predicted targets of miR-15a/miR16-1, within the 13q minimally deleted region (MDR).16 In addition, 365 out of 1181 (31%) of the DE probes are on chromosome 12; 364 out of 365 are over-expressed in +12 compared to del(13q) CLL. Comparing +12 and del(11q) cases, we identified 736 DE probes representing 583 unique genes, 19 ncRNAs, and 7 ESTs (FDR=5%, unadhaematologica | 2018; 103(12)

Trisomy 12 CLL

justed P=0.00091). Forty-three (5.9%) of the DE probes represent genes on 11q22.3 (e.g. ATM, NPAT, DDX10, CUL5, ACAT1);17 all are expressed at higher levels in +12 and at lower levels in del(11q) cases. In addition, 208 out of 736 (28%) of the DE probes are on chromosome 12; 206 out of 208 are over-expressed in +12 compared to del(11q). Comparing +12 and diploid cases, we identified 1229 DE probes representing 964 unique genes, 49 ncRNAs, and 18 ESTs (FDR=5%, unadjusted P=0.00164); 413 out of 1229 (34%) DE probes are on chromosome 12 (Ď&#x2021;2=92.5, P<2e-16), and all 413 are over-expressed in +12 compared to diploid.

Consistent differences between +12 and other cytogenetic subtypes Because the pair-wise DE comparisons included genes whose expression characterizes other cytogenetic subtypes (e.g. BCL2 and other miR-15-a/16-1 targets when comparing +12 to del(13q) cases, or ATM and other genes in the commonly deleted region when comparing +12 to del(11q) cases), we used t-tests to identify DE genes when comparing +12 cases to the union of all other cases in the study. We found 1226 DE probes representing 953 unique genes, 40 ncRNAs, and 17 ESTs (FDR=1%, unadjusted P=0.000347); 419 out of 1226 (34%) were on chromosome 12, and 418 out of 419 were over-expressed in +12 cases. We also observed that 1194 out of 1226 (97%) of these probes were DE in the direct comparison with del(13q); 892 out of 1226 (73%) were DE in the direct comparison to diploid cases, and 526 out of 1226 (43%) were DE in the direct comparison to del(11q). The full list of DE probes is presented in Online Supplementary Table S1. Among cases with sole +12, microarray profiling data and NOTCH1 mutation status were available for 15 patients in the discovery set: n=9 wild-type (60%), n=6 mutated (40%). To identify DE genes between these subsets, we performed the following analysis. We removed low-expressing probes, and retained a probe only if its expression was >4 on the log2 scale in at least 3 out of 15 samples; 20,776 of the 47,231 probes satisfied this criterion. We performed probe-by-probe t-tests to compare expression between the NOTCH1 mutated versus wildtype samples. We found 389 DE genes with P<0.01 (FDR=45%; Online Supplementary Table S2).

Pathway analysis We performed Ingenuity Pathway Analysis (IPA) to identify differentially regulated canonical pathways that distinguish +12 from other cytogenetic subtypes individually. For del(13q), del(11q), and diploid subtypes we performed analyses using the 1181, the 736, and the 1229 DE probes described above. For each comparison, ten pathways were identified as either activated or down-regulated with the most statistical significance, based on the Ingenuity z-scores (Table 3). Complete data are listed in Online Supplementary Table S3.

Validation of potential targets To validate potential mRNA targets identified by wholegenome transcriptional profiling performed on the discovery set, we assessed expression of a subset of these genes by MF-QRT-PCR assay on an independent validation set of 50 patient samples. We used the QRT-PCR assay because it is more reproducible and has a wider dynamic range than microarray profiling.13 Of 135 genes assayed, haematologica | 2018; 103(12)

Figure 1. Kaplan Meier plots stratified by cytogenetic subtype. (A) Time to treatment, and (B) progression-free survival.

64 (47%) were fully validated for all comparisons between +12 and other pure cytogenetic subsets, and 91 (67%) were validated for at least half of the comparisons. A subset of 31 genes assayed using the MF-QRT-PCR assay are included in the network diagram in Figure 3; of these, 19 (61%) were fully validated and all 31 (100%) were validated in at last half of the comparisons of +12 with other cytogenetic groups. Complete data are listed in Online Supplementary Table S4.

A +12 specific network Using the MF-QRT-PCR data, we constructed a gene network whose expression in +12 cases differed from other subtypes. We selected all DE genes with FDR=5% (unadjusted P=0.0016) and fold change (FC) â&#x2030;Ľ2 in comparison to both del(13q) and diploid cases. Intersecting these lists yielded 109 probes that represent 92 distinct genes. Although 17 probes did not satisfy the criteria when we compared +12 cases directly to del(11q), we chose to retain them because all 109 probes satisfied the selection 2073

L.V. Abruzzo et al.

Figure 2. Results of two-way clustering according to cytogenetic subtype using the genes found to be differentially expressed. The samples include 40 del(13q) (blue), 32 diploid (brown), 10 del(11q) (pink), and trisomy 12 (+12) (green). Each column is one sample; each row contains the standardized log expression values for one gene.

criteria in the analysis that compared +12 cases to all other samples. Starting with these 92 genes, we used the network building tools in IPA. First, the “connect” operation joins any pair of genes whose interaction is supported in the literature. Then the “grow” operation adds genes from the literature (not from the initial list) that are significantly connected to genes already in the network. We performed the “connect” and “grow” operations twice; the resulting network is shown in Figure 3. Genes not connected to the main network and not on chromosome 12 were omitted from the final diagram.

Discussion Despite important differences in the clinical and pathophysiological features of +12 CLL compared to other cytogenetic subtypes, only a few transcriptional profiling studies (some on small numbers of samples) have focused on identifying dysregulated pathways that characterize +12 CLL and that may serve as therapeutic targets.7,8,18 Furthermore, these studies have classified cases using a hierarchical system, i.e. cases with more than one recurrent abnormality are categorized according to the abnormality with the poorest prognosis.1 Thus, cases with both del(13q) and +12 are classified as +12, and cases with both +12 and del(11q) are classified as del(11q). To exclude the likely confounding effects of multiple cytoge2074

netic abnormalities on gene expression, our +12 patient cohort had +12 as the only abnormality. Similar to previous studies, we found that patients with +12 and with diploid cytogenetics required treatment earlier during their disease course than patients with del(13q), but later than patients with del(11q). Following FCR chemoimmunotherapy, our +12 cohort had a longer PFS than patients with other cytogenetic subtypes, but showed no difference in OS. Thus, +12 CLL patients may require treatment earlier, but respond better to FCR, than patients with other cytogenetic abnormalities. Higher CD20 expression by +12 CLL compared to other cytogenetic subtypes may account, in part, for the high rate of response to rituximab-based therapy.19 However, despite the good response of +12 CLL patients to FCR, it is poorly tolerated by unfit patients or those over 65 years of age. Some patients develop myelosuppression and neutropenic fevers, and cannot receive a full course of therapy. Finally, a small percentage of patients treated with alkylating agents and fludarabine develop secondary myeloid malignancies.20 Thus, there is a need for less toxic, targeted therapies. As expected, we identified genes whose expression patterns are known to be associated with cytogenetic subtypes, giving us confidence in our methods. For example, we identified statistically significant differences in gene expression between +12 cases with and without NOTCH1 mutation. Although the number of cases is relhaematologica | 2018; 103(12)

Trisomy 12 CLL Table 3. Ingenuity Pathway Analysis of differentially-regulated canonical pathways.

+12 vs. del(13q) Pathway

Phospholipase C signaling

Integrin signaling

Regulation of actin-based motility by Rho Remodeling of epithelial adherens junctions Protein kinase A signaling

A A A

Role of BRCA1 in DNA damage response

RhoGDI signaling

Fcg receptor-mediated phagocytosis in macrophages and monocytes Non-small cell lung cancer signaling

Gαq signaling

+12 vs. del(11q) Pathway TNFR2 signaling HIPPO signaling CD40 signaling Death receptor signaling TWEAK signaling Sphingosine-1phosphate signaling TNFR1 signaling Signaling by Rho family GTPases Regulation of actin-based motility by Rho RhoGDI signaling

+12 vs. diploid Pathway

S D A A D D

Ceramide signaling Non-small cell lung cancer signaling NGF signaling Integrin signaling Pancreatic adenocarcinoma

S D A A A D

Huntington disease signaling

Glioma signaling 14-3-3-mediated signaling IL-8 signaling

A A

HMGB1 signaling

D A

+12 vs. all Pathway Protein kinase A signaling Integrin signaling Phospholipase C signaling Ceramide signaling Cell cycle: G1/S checkpoint regulation Role of BRCA1 in DNA damage response Insulin receptor signaling Huntington disease signaling Remodeling of epithelial adherens junctions Glioma signaling

S A A A A D

A A D

vs.: versus; S: Status of pathway compared to +12 CLL; A: activated; D: down-regulated.

atively small, our results are similar to those of previous studies.18,21-23 However, we found no differences in expression of NOTCH1 or NOTCH1 target genes (e.g. HES1, DTX1, NRARP) between NOTCH1-mutated and unmutated cases. Recently, Fabbri et al. analyzed NOTCH1 RNA expression in normal B-cell subsets and more than 100 PB CLL cases.12 They found that tonsillar naïve and memory B cells expressed NOTCH1, HES1, and MYC, while germinal center B cells were negative. Furthermore, about 50% of CLL cases that lacked NOTCH1 mutations expressed the active intracellular portion of NOTCH1. They describe a “NOTCH1 gene expression signature” that regulates critical B-cell processes, but is independent of NOTCH1 mutation. Using pathway analysis, we identified canonical pathways that are differentially regulated in +12 CLL compared to other subtypes; several converge on the BCR signaling pathway. One of the most highly activated pathways was integrin signaling. Integrins are transmembrane receptors that mediate interactions between the extracellular matrix and actin cytoskeleton. Integrins enhance adhesion, which activates signaling pathways that regulate migration, proliferation, cell survival, and other processes.24 B-cell receptor signaling is critical for CLL survival and proliferation,25 and is enhanced through interactions between CLL cells and the microenvironment.26 Signals from the BCR are transduced by downstream kinases. Therapeutic agents that target Bruton’s tyrosine kinase (BTK) and the phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit delta (PI3Kδ) are highly effective.27-29 Patients treated with kinase inhibitors experihaematologica | 2018; 103(12)

ence a rapid rise in absolute lymphocyte count (ALC) due to egress of CLL cells from lymph nodes that declines slowly over approximately eight months.27-29 The magnitude and duration of this transient redistribution lymphocytosis varies with cytogenetic subtype; del(13q) patients with IGHV-mutated genes tend to experience prolonged lymphocytosis, while +12 patients show an attenuated rise in the ALC and a more rapid reduction to baseline.29,30 Redistribution lymphocytosis is likely a consequence of inhibition of CLL migration and adhesion due to disruption of chemokine receptor and integrin signaling. For example, the BTK inhibitor ibrutinib abrogates adhesion of CLL cells to fibronectin, interfering with their ability to adhere to stromal cells.31 Recent studies indicate that +12 CLL express higher levels of several integrin proteins compared to other cytogenetic subtypes.23,32 Similarly, we found overexpression of ITGAL and ITGB2 (which encode the αL and β2 chains of LFA-1), ITGA4 (which encodes the α4 chain of VLA-4), and ITGB7. We also observed overexpression of ITGB5 and vinculin (VCL). ITGB5 encodes the β5 integrin chain, a fibronectin receptor component. Increased ITGB5 may contribute to the attenuated redistributive lymphocytosis in +12 CLL by increasing adhesion to fibronectin and interfering with the CLL egress from lymph nodes. Vinculin stabilizes integrins at the immune synapse, the interface between the B cell and the antigen-presenting cell, and is critical for activation of BCR signaling.33 Thus, increased integrin-mediated signaling may promote retention of +12 CLL cells within tissues. Alternatively, circulating +12 CLL cells may undergo more rapid apoptosis 2075

L.V. Abruzzo et al.

Figure 3. Construction of a specific trisomy 12 (+12) CLL gene expression network. Genes indicated in blue are over-expressed in +12 chronic lymphocytic leukemia compared to other cytogenetic subtypes. Genes indicated in orange are under-expressed in +12 CLL. Genes indicated in gray are not differentially expressed. Brighter colors are more statistically significant; duller colors are less statistically significant. Genes on chromosome 12 are indicated by hexagons; genes located on other chromosomes are indicated by rectangles.

due to greater reliance on microenvironmental survival signals, another factor that may contribute to their good response to FCR.30 We identified three NFAT mRNAs, NFATc1 (NFAT2), NFATc2 (NFAT1), and NFATc3 (NFAT4), that are overexpressed in +12 CLL, and involved in three activated signaling pathways: protein kinase A, phospholipase C, and integrin signaling. The NFAT (nuclear factor of activated T cells) transcription factor family contains five proteins; four are regulated by calcium and the calcineurin signaling pathway.34,35 Originally described in T cells, NFAT proteins are expressed by B cells, natural killer cells, and other cell types. They regulate genes involved in cell cycle, apoptosis, angiogenesis, and metastasis. In resting lymphocytes, inactive hyperphosphorylated NFAT proteins are confined to the cytoplasm.34,35 In B cells, BCR ligation by cognate antigen activates SYK, which phosphorylates and activates BTK.36 BTK then phosphorylates and activates PLCg2, which catalyzes the hydrolysis of inositol 1, 4, 5-trisphosphate (IP3) and diacylglycerol (DAG). IP3 mediates influx of extracellular calcium and calcium release from intracellular stores, which results in calcium/calmodulin-dependent activation of calcineurin. In the cytoplasm, calcineurin cleaves phosphate groups from inactive, hyperphosphorylated NFAT proteins, which enter the nucleus, bind to specific response elements in target gene promoters (alone or in combination with partner proteins), and activate or inhibit transcription. Thus, NFAT proteins integrate calcium signaling with other signaling pathways, including the MAPKinase, WNT, and NOTCH pathways. Dysregulated calcineurin/NFAT signaling has been reported in carcinomas and lymphoid malignancies. In large B-cell lymphomas, active NFAT interacts with NF-kB and directly regulates CD154 expression to maintain growth.37 Despite its central role in BCR signaling, there are few studies of NFAT signaling in CLL.38-41 LeRoy et al. demonstrated that BCR-NFAT signaling affects CLL clinical outcome, and suggest that BCR-NFAT intermediates 2076

serve as therapeutic targets.38 Recently, Oakes et al. compared the epigenetic programming of normal B-cell subsets with 268 CLL samples, and identified aberrant NFAT methylation in a CLL subset.42 Thus, efforts are underway to develop more specific NFAT inhibitors.43,44 Using IPA to construct a novel +12 specific network, we identified ecto-5â&#x20AC;&#x2122;-nucleotidase (NT5E, CD73) as an important element. NT5E, an immune checkpoint molecule of potential therapeutic value, is expressed in a wide variety of tissues.45-47 Immune checkpoint molecules regulate interactions between immune and tumor cells, and may stimulate or inhibit these interactions.45,46 Many cancers exploit these molecules to evade an anti-tumor immune response. Among the best described are the inhibitory molecules PD1, PD-L1, and CTLA-4. CLL is characterized by immunosuppression and an inefficient anti-tumor response that results from defects in humoral and cellular immunity, including ineffective T-cell responses and expression of exhaustion-like surface markers, such as PD-L1.48-50 Immune checkpoint inhibitors that target the PD1/PD-L1 and CTLA4 pathways are being used to treat a variety of tumors, including melanoma and prostate cancer. NT5E catalyzes the conversion of extracellular ATP to adenosine, which is critical for immune function.45,46 Among immune cells, it is expressed in macrophages, B cells, regulatory T cells, and dendritic cells. NT5E helps tumors evade the immune response by inhibiting the activation, proliferation, and homing of tumor-specific T cells, and by enhancing conversion of anti-tumor type 1 macrophages to pro-tumor type 2 macrophages. The NT5E-adenosine axis constitutes a promising new pathway in cancer immunotherapy. Targeted blockade of NT5E or adenosine receptors promotes anti-tumor immunity and enhances the activity of first-generation immune checkpoint blockers.45,46 Phase I clinical trials evaluating the efficacy of anti-NT5E or anti-A2A therapies in cancer patients are underway. However, few studies have investigated the functions of NT5E in lymphoid malignancies. In a study of CLL patients, 30% of cases expressed NT5E, haematologica | 2018; 103(12)

Trisomy 12 CLL

which was associated with aggressive disease, and CD38 and ZAP70 positivity.47 Unfortunately, this study did not assess the association between NT5E expression and cytogenetic status. Our findings suggest that targeting this pathway may be an effective therapy in patients with +12 CLL. In summary, we have demonstrated, by whole transcriptome profiling, that CLL cases with +12 as the only cytogenetic abnormality demonstrate a unique set of differentially expressed genes and pathways compared to cases with del(13q) or del(11q). Our data support the hypothesis that these differences contribute, in part, to the

References 1. Zenz T, Dohner H, Stilgenbauer S. Genetics and risk-stratified approach to therapy in chronic lymphocytic leukemia. Best Pract Res Clin Haematol. 2007; 20(3):439-453. 2. Cimmino A, Calin GA, Fabbri M, et al. miR-15 and miR-16 induce apoptosis by targeting BCL2. Proc Natl Acad Sci USA. 2005;102(39):13944-13949. 3. 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. 4. Matutes E, Oscier D, Garcia-Marco J, et al. Trisomy 12 defines a group of CLL with atypical morphology: correlation between cytogenetic, clinical and laboratory features in 544 patients. Br J Haematol. 1996; 92(2):382-388. 5. Strati P, Abruzzo LV, Wierda WG, O'Brien S, Ferrajoli A, Keating MJ. Second cancers and Richter transformation are the leading causes of death in patients with trisomy 12 chronic lymphocytic leukemia. Clin Lymphoma Myeloma Leuk. 2015; 15(7):420-427. 6. Haslinger C, Schweifer N, Stilgenbauer S, et al. Microarray gene expression profiling of B-cell chronic lymphocytic leukemia subgroups defined by genomic aberrations and VH mutation status. J Clin Oncol. 2004;22(19):3937-3949. 7. Mittal AK, Hegde GV, Aoun P, et al. Molecular basis of aggressive disease in chronic lymphocytic leukemia patients with 11q deletion and trisomy 12 chromosomal abnormalities. Int J Mol Med. 2007; 20(4):461-469. 8. Porpaczy E, Bilban M, Heinze G, et al. Gene expression signature of chronic lymphocytic leukaemia with Trisomy 12. Eur J Clin Invest. 2009;39(7):568-575. 9. Visone R, Rassenti LZ, Veronese A, et al. Karyotype-specific microRNA signature in chronic lymphocytic leukemia. Blood. 2009;114(18):3872-3879. 10. Duzkale H, Schweighofer CD, Coombes KR, et al. LDOC1 mRNA is differentially expressed in chronic lymphocytic leukemia and predicts overall survival in untreated patients. Blood. 2011;117(15):4076-4084. 11. Fais F, Ghiotto F, Hashimoto S, et al. Chronic lymphocytic leukemia B cells express restricted sets of mutated and unmutated antigen receptors. J Clin Invest. 1998;102(8):1515-1525. 12. Fabbri G, Rasi S, Rossi D, et al. Analysis of

haematologica | 2018; 103(12)

13.

14. 15.

16.

17.

18.

19.

20.

21.

22.

23.

unique pathophysiology of +12 CLL. Finally, we have identified genes and pathways, such as the checkpoint inhibitor molecule, NT5E (CD73), and the NFAT signaling pathway that may represent new therapeutic targets. Acknowledgments The authors thank the patients who donated their blood. The authors also thank Dr. David Lucas for helpful discussions and comments. Funding This work was supported in part by grants from the CLL Global Research Foundation and NIH/NCI 1 R01 CA182905-01.

the chronic lymphocytic leukemia coding genome: role of NOTCH1 mutational activation. J Exp Med. 2011;208(7):1389-1401. Abruzzo LV, Lee KY, Fuller A, et al. Validation of oligonucleotide microarray data using microfluidic low-density arrays: a new statistical method to normalize realtime RT-PCR data. Biotechniques. 2005;38(5):785-792. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials. 1996;17(4):343-346. Moreau EJ, Matutes E, A'Hern RP, et al. Improvement of the chronic lymphocytic leukemia scoring system with the monoclonal antibody SN8 (CD79b). Am J Clin Pathol. 1997;108(4):378-382. Manyam G, Ivan C, Calin GA, Coombes KR. targetHub: a programmable interface for miRNA-gene interactions. Bioinformatics. 2013;29(20):2657-2658. Stankovic T, Skowronska A. The role of ATM mutations and 11q deletions in disease progression in chronic lymphocytic leukemia. Leuk Lymphoma. 2014; 55(6):1227-1239. Del Giudice I, Rossi D, Chiaretti S, et al. NOTCH1 mutations in +12 chronic lymphocytic leukemia (CLL) confer an unfavorable prognosis, induce a distinctive transcriptional profiling and refine the intermediate prognosis of +12 CLL. Haematologica. 2012;97(3):437-441. Tam CS, Otero-Palacios J, Abruzzo LV, et al. Chronic lymphocytic leukaemia CD20 expression is dependent on the genetic subtype: a study of quantitative flow cytometry and fluorescent in-situ hybridization in 510 patients. Br J Haematol. 2008;141(1):36-40. Skarbnik AP, Faderl S. The role of combined fludarabine, cyclophosphamide and rituximab chemoimmunotherapy in chronic lymphocytic leukemia: current evidence and controversies. Ther Adv Hematol. 2017;8(3):99-105. Maura F, Mosca L, Fabris S, et al. Insulin growth factor 1 receptor expression is associated with NOTCH1 mutation, trisomy 12 and aggressive clinical course in chronic lymphocytic leukaemia. PLoS One. 2015;10(3):e0118801. Pozzo F, Bittolo T, Vendramini E, et al. NOTCH1-mutated chronic lymphocytic leukemia cells are characterized by a MYCrelated overexpression of nucleophosmin 1 and ribosome-associated components. Leukemia. 2017;31(11):2407-2415. Riches JC, O'Donovan CJ, Kingdon SJ, et al. Trisomy 12 chronic lymphocytic leukemia cells exhibit upregulation of inte-

24. 25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

grin signaling that is modulated by NOTCH1 mutations. Blood. 2014; 123(26): 4101-4110. Huttenlocher A, Horwitz AR. Integrins in cell migration. Cold Spring Harb Perspect Biol. 2011;3(9):a005074. Herishanu Y, Perez-Galan P, Liu D, et al. The lymph node microenvironment promotes B-cell receptor signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011;117(2):563-574. Burger JA. Nurture versus nature: the microenvironment in chronic lymphocytic leukemia. Hematology Am Soc Hematol Educ Program. 2011;2011:96-103. Brown JR, Byrd JC, Coutre SE, et al. Idelalisib, an inhibitor of phosphatidylinositol 3-kinase p110delta, for relapsed/refractory chronic lymphocytic leukemia. Blood. 2014;123(22):3390-3397. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):32-42. Woyach JA, Smucker K, Smith LL, et al. Prolonged lymphocytosis during ibrutinib therapy is associated with distinct molecular characteristics and does not indicate a suboptimal response to therapy. Blood. 2014;123(12):1810-1817. Thompson PA, Ferrajoli A, O'Brien S, Wierda WG, Keating MJ, Burger JA. Trisomy 12 is associated with an abbreviated redistribution lymphocytosis during treatment with the BTK inhibitor ibrutinib in patients with chronic lymphocytic leukaemia. Br J Haematol. 2015;170(1):125128. Herman SE, Mustafa RZ, Jones J, Wong DH, Farooqui M, Wiestner A. Treatment with Ibrutinib Inhibits BTK- and VLA-4Dependent Adhesion of Chronic Lymphocytic Leukemia Cells In Vivo. Clin Cancer Res. 2015;21(20):4642-4651. Zucchetto A, Caldana C, Benedetti D, et al. CD49d is overexpressed by trisomy 12 chronic lymphocytic leukemia cells: evidence for a methylation-dependent regulation mechanism. Blood. 2013;122(19): 3317-3321. Saez de Guinoa J, Barrio L, Carrasco YR. Vinculin arrests motile B cells by stabilizing integrin clustering at the immune synapse. J Immunol. 2013;191(5):2742-2751. Medyouf H, Ghysdael J. The calcineurin/NFAT signaling pathway: a novel therapeutic target in leukemia and solid tumors. Cell Cycle. 2008;7(3):297303.

2077

L.V. Abruzzo et al. 35. Mognol GP, Carneiro FR, Robbs BK, Faget DV, Viola JP. Cell cycle and apoptosis regulation by NFAT transcription factors: new roles for an old player. Cell Death Dis. 2016;7:e2199. 36. Zhong Y, Byrd JC, Dubovsky JA. The B-cell receptor pathway: a critical component of healthy and malignant immune biology. Semin Hematol. 2014;51(3):206-218. 37. Pham LV, Tamayo AT, Yoshimura LC, LinLee YC, Ford RJ. Constitutive NF-kappaB and NFAT activation in aggressive B-cell lymphomas synergistically activates the CD154 gene and maintains lymphoma cell survival. Blood. 2005;106(12):3940-3947. 38. Le Roy C, Deglesne PA, Chevallier N, et al. The degree of BCR and NFAT activation predicts clinical outcomes in chronic lymphocytic leukemia. Blood. 2012;120(2):356-365. 39. Marklin M, Heitmann JS, Fuchs AR, et al. NFAT2 is a critical regulator of the anergic phenotype in chronic lymphocytic leukaemia. Nat Commun. 2017;8(1):755. 40. Schuh K, Avots A, Tony HP, Serfling E,

2078

41.

42.

43. 44. 45.

Kneitz C. Nuclear NF-ATp is a hallmark of unstimulated B cells from B-CLL patients. Leuk Lymphoma. 1996;23(5-6):583-592. Wolf C, Garding A, Filarsky K, et al. NFATC1 activation by DNA hypomethylation in chronic lymphocytic leukemia correlates with clinical staging and can be inhibited by ibrutinib. Int J Cancer. 2018;142(2):322-333. Oakes CC, Seifert M, Assenov Y, et al. DNA methylation dynamics during B cell maturation underlie a continuum of disease phenotypes in chronic lymphocytic leukemia. Nat Genet. 2016;48(3):253-264. Mancini M, Toker A. NFAT proteins: emerging roles in cancer progression. Nat Rev Cancer. 2009;9(11):810-820. Qin JJ, Nag S, Wang W, et al. NFAT as cancer target: mission possible? Biochim Biophys Acta. 2014;1846(2):297-311. Allard D, Allard B, Gaudreau PO, Chrobak P, Stagg J. CD73-adenosine: a next-generation target in immuno-oncology. Immunotherapy. 2016;8(2):145-163.

46. Antonioli L, Yegutkin GG, Pacher P, Blandizzi C, Hasko G. Anti-CD73 in cancer immunotherapy: awakening new opportunities. Trends Cancer. 2016;2(2):95-109. 47. Serra S, Horenstein AL, Vaisitti T, et al. CD73-generated extracellular adenosine in chronic lymphocytic leukemia creates local conditions counteracting drug-induced cell death. Blood. 2011;118(23):6141-6152. 48. McClanahan F, Hanna B, Miller S, et al. PDL1 checkpoint blockade prevents immune dysfunction and leukemia development in a mouse model of chronic lymphocytic leukemia. Blood. 2015;126(2):203-211. 49. Riches JC, Davies JK, McClanahan F, et al. T cells from CLL patients exhibit features of T-cell exhaustion but retain capacity for cytokine production. Blood. 2013; 121(9):1612-1621. 50. Riches JC, Gribben JG. Understanding the immunodeficiency in chronic lymphocytic leukemia: potential clinical implications. Hematol Oncol Clin North Am. 2013; 27(2):207-235.

haematologica | 2018; 103(12)

ARTICLE

Plasma Cell Disorders

Daratumumab plus bortezomib and dexamethasone versus bortezomib and dexamethasone in relapsed or refractory multiple myeloma: updated analysis of CASTOR

Andrew Spencer,1 Suzanne Lentzsch,2 Katja Weisel,3 Hervé Avet-Loiseau,4 Tomer M. Mark,5 Ivan Spicka,6 Tamas Masszi,7 Birgitta Lauri,8 Mark-David Levin,9 Alberto Bosi,10 Vania Hungria,11 Michele Cavo,12 Je-Jung Lee,13 Ajay K. Nooka,14 Hang Quach,15 Cindy Lee,16 Wolney Barreto,17 Paolo Corradini,18 Chang-Ki Min,19 Emma C. Scott,20 Asher A. Chanan-Khan,21 Noemi Horvath,16 Marcelo Capra,22 Meral Beksac,23 Roberto Ovilla,24 Jae-Cheol Jo,25 Ho-Jin Shin,26 Pieter Sonneveld,27 David Soong,28 Tineke Casneuf,29 Christopher Chiu,28 Himal Amin,30 Ming Qi,28 Piruntha Thiyagarajah,31 A. Kate Sasser,32 Jordan M. Schecter30 and Maria-Victoria Mateos33

Malignant Haematology and Stem Cell Transplantation Service, Alfred Health-Monash University, Melbourne, Australia; 2Division of Hematology/Oncology, Columbia University, New York, NY, USA; 3Universitaetsklinikum Tuebingen der Eberhard-Karls-Universitaet, Abteilung fuer Innere Medizin II, Tübingen, Germany; 4Unite de Genomique du Myelome, CHU Rangueil, Toulouse, France; 5Department of Medicine, University of Colorado, Aurora, CO, USA; 6Clinical Department of Haematology, 1st Medical Department, Charles University in Prague, Czech Republic; 7Department of Haematology and Stem Cell Transplantation, St László Hospital, 3rd Department of Internal Medicine, Semmelweis University, Budapest, Hungary; 8Department of Hematology, Sunderbyn Hospital, Luleå, Sweden; 9Albert Schweitzer Hospital Department of Internal Medicine, Dordrecht, the Netherlands; 10Department of Hematology, Careggi Hospital and University of Florence, Italy; 11Irmandade Da Santa Casa De Misericordia De São Paulo, Brazil; 12“Seràgnoli” Institute of Hematology, Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna, Italy; 13Department of Hematology-Oncology, Chonnam National University Hwasun Hospital, Jeollanamdo, South Korea; 14Winship Cancer Institute, Emory University, Atlanta, GA, USA; 15St. Vincent's Hospital, University of Melbourne, Australia; 16Royal Adelaide Hospital, North Terrace, Australia; 17Hospital Santa Marcelina, São Paulo, Brazil; 18Fondazione IRCCS Instituto Nazionale dei Tumori, University of Milan, Italy; 19Seoul St. Mary’s Hospital, South Korea; 20Oregon Health & Science University, Portland, OR, USA; 21Mayo Clinic Florida, Jacksonville, FL, USA; 22 Instituto do Cancer-Hospital Mae de Deus, Porto Alegre, Brazil; 23Ankara University, Turkey; 24Hospital Angeles Lomas, Naucalpan de Juárez y alrededores, México; 25Ulsan University Hospital, South Korea; 26Division of Hematology-Oncology, Department of Internal Medicine, School of Medicine, Medical Research Institute, Pusan National University Hospital, Busan, South Korea; 27Erasmus Medical Center, Rotterdam, the Netherlands; 28Janssen Research & Development, LLC, Spring House, PA, USA; 29 Janssen Research & Development, Beerse, Belgium; 30Janssen Research & Development, LLC, Raritan, NJ, USA; 31Janssen Research & Development, High Wycombe, UK; 32Genmab US, Inc, Princeton, NJ, USA and 33University Hospital of Salamanca/IBSAL, Spain

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2079-2087

ABSTRACT

aratumumab, a CD38 human monoclonal antibody, demonstrated significant clinical activity in combination with bortezomib and dexamethasone versus bortezomib and dexamethasone alone in the primary analysis of CASTOR, a phase 3 study in relapsed and/or refractory multiple myeloma. A post hoc analysis based on treatment history and longer follow up is presented. After 19.4 (range: 0-27.7) months of median follow up, daratumumab plus bortezomib and dexamethasone prolonged progression-free survival (median: 16.7 versus 7.1 months; hazard ratio, 0.31; 95% confidence interval, 0.24-0.39; P<0.0001) and improved the overall response rate (83.8% versus 63.2%; P<0.0001) compared with bortezomib and dexamethasone alone. The progression-free survival benefit of daratumumab plus bortezomib and dexamethasone was most apparent in patients with 1 prior line of therapy (median: not reached versus 7.9 months; hazard ratio, 0.19; 95% conhaematologica | 2018; 103(12)

Correspondence: aspencer@netspace.net.au

Received: March 27, 2018. Accepted: August 17, 2018. Pre-published: September 20, 2018. doi:10.3324/haematol.2018.194118 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2079 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2079

A. Spencer et al.

fidence interval, 0.12-0.29; P<0.0001). Daratumumab plus bortezomib and dexamethasone was also superior to bortezomib and dexamethasone alone in subgroups based on prior treatment exposure (bortezomib, thalidomide, or lenalidomide), lenalidomide-refractory status, time since last therapy (≤12, >12, ≤6, or >6 months), or cytogenetic risk. Minimal residual disease–negative rates were >2.5-fold higher with daratumumab across subgroups. The safety profile of daratumumab plus bortezomib and dexamethasone remained consistent with longer follow up. Daratumumab plus bortezomib and dexamethasone demonstrated significant clinical activity across clinically relevant subgroups and provided the greatest benefit to patients treated at first relapse. Trial registration: clinicaltrials.gov identifier: 02136134.

Introduction As multiple myeloma (MM) progresses, a reduction in the duration and depth of response is observed with each treatment relapse, as a result of diminished sensitivity of heavily treated patients to subsequent therapies.1 Proteasome inhibitors (PIs) are widely used due to their clinical effectiveness, manageable safety profile, and combinability with other therapies.2 However, in several studies of novel PI-based regimens in relapsed and/or refractory MM (RRMM), deep clinical responses were uncommon.3-6 PI-based regimens that generate deeper responses in RRMM are an unmet need. Daratumumab, a human IgGk monoclonal antibody targeting CD38, has a direct on-tumor and immunomodulatory mechanism of action.7-12 In combination with standard of care regimens, (bortezomib and dexamethasone [Vd; CASTOR] or lenalidomide and dexamethasone [Rd; POLLUX]), daratumumab induced rapid, deep, and durable responses, reducing the risk of disease progression or death by >60%, versus Vd or Rd in relapsed patients.13,14 Based on the superior progression-free survival (PFS) benefit, daratumumab-Vd (D-Vd) and daratumumab-Rd (D-Rd) were approved in the United States and Europe for MM patients who have received ≥1 prior therapy.15,16 In addition, daratumumab plus pomalidomide and dexamethasone was approved in the United States for MM patients after 2 prior therapies including lenalidomide and a PI.15 More recently, daratumumab in combination with bortezomib, melphalan, and prednisone was approved in the United States for patients with newly diagnosed MM who are ineligible for autologous stem cell transplantation.15 At the time of the event-driven, pre-specified primary analysis (median follow up: 7.4 months) of the CASTOR study, PFS was significantly prolonged with D-Vd versus Vd (median: not reached versus 7.2 months; hazard ratio [HR], 0.39; 95% confidence interval [CI], 0.28-0.53; P<0.0001).13 This updated analysis provides an additional 12 months of follow up for efficacy and safety compared with the primary analysis, including updated PFS in the intent-to-treat population, and presents an exploratory post hoc analysis of CASTOR to identify patient subgroups that may benefit most from D-Vd.

Methods Study Design CASTOR (clinicaltrials.gov identifier: 02136134) is an ongoing multi-center, open-label, randomized, active-controlled, phase 3 study of D-Vd versus Vd in patients with RRMM who received ≥1 prior line of therapy. The study design and primary results were previously published.13 2080

Briefly, patients were randomized 1:1 to D-Vd or Vd. Randomization was balanced and stratified by International Staging System (I, II, or III) at screening (central laboratory results), number of prior lines of therapy (1 versus 2 or 3 versus >3), and prior bortezomib exposure (no versus yes). The study protocol was approved by an independent ethics committee or institutional review board at each study center, and was conducted in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice guidelines. All patients provided written informed consent.

Patients

Eligible patients had ≥1 prior line of therapy, achieved at least a partial response to ≥1 prior MM treatment, and had progressive disease per International Myeloma Working Group (IMWG) criteria17,18 on or after their last regimen. Patients refractory to bortezomib or another PI (ixazomib or carfilzomib following a protocol amendment) were ineligible.

Procedures Patients received 8 cycles of bortezomib (1.3 mg/m2 subcutaneously on Days 1, 4, 8, 11) and dexamethasone (20 mg orally on Days 1, 2, 4, 5, 8, 9, 11, 12) with or without daratumumab (16 mg/kg intravenously once weekly in Cycles 1-3, Day 1 of Cycles 4-8, then every 4 weeks until disease progression, unacceptable toxicity, or withdrawal of consent). Cycle durations were 21 days for Cycles 1 to 8 and 28 days for Cycle 9 onwards. A protocol amendment after the primary analysis allowed patients who progressed on Vd to receive daratumumab monotherapy.

Assessments and Endpoints The primary endpoint was PFS; secondary endpoints included time to disease progression, overall response rate (ORR), minimal residual disease (MRD), and safety. This exploratory, post hoc, secondary analysis examined subpopulations according to prior lines of therapy (1, 2-3, >3, or 1-3), prior treatment exposure (bortezomib, thalidomide, or lenalidomide), refractoriness to lenalidomide at the last prior line of therapy, time since last therapy (≤12, >12, ≤6, or >6 months), and cytogenetic risk assessed centrally by next-generation sequencing.19 Site investigators determined numbers of prior lines of therapy using IMWG guidelines.18 Time since last therapy was the duration between the end date of the last line of prior therapy and the randomization date. PFS, ORR, and MRD-negativity at 10–5 and 10–6 sensitivity thresholds were assessed for each subgroup. PFS based on MRD (10–5), and cytogenetic risk status was also examined. Health-related quality haematologica | 2018; 103(12)

Subgroup analyses of CASTOR

of life (HRQoL) was assessed by the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core-30 EORTC QLQ-C30 and the EuroQol 5 Dimensions Questionnaire (EQ-5D-5L) tools. The Online Supplementary Appendix provides full details of statistical analyses and MRD, cytogenetic, and HRQoL assessments.

Results Of 498 patients, 251 and 247 were randomized to D-Vd and Vd, respectively (Online Supplementary Figure S1). Patient demographics and baseline clinical characteristics were previously published and are well-balanced between groups.13 Relevant clinical characteristics, including treatment history and cytogenetic-risk status, were balanced between groups and are summarized in Table 1 and Online Supplementary Table S1. Briefly, patients in CASTOR received a median of 2 prior lines of therapy. Overall, 47.2% received 1 prior line of therapy, 28.9% received 2 prior lines, 13.9% received 3 prior lines, and 10.0% received >3 prior lines of therapy. A total of 21.1% of patients were refractory to lenalidomide at their last line of therapy. Among patients treated with D-Vd, the median duration of treatment was 13.4 months (range: 0-26.7) versus 5.2 months (range: 0.2-8.0) with Vd. Following a protocol amendment after the primary analysis, patients who progressed on Vd had the option to receive daratumumab monotherapy.13 At a median follow up of 19.4 months, all patients in both groups had discontinued or completed Vd treatment per protocol; in the D-Vd group, 41% of patients remained on daratumumab monotherapy. A total of 64 patients in the Vd group opted to receive daratumumab monotherapy following disease progression. The clinical cut-off date was January 11, 2017. At a median duration of follow up of 19.4 months (range: 027.7) months, D-Vd significantly prolonged PFS versus Vd (median: 16.7 versus 7.1 months; HR, 0.31; 95% CI, 0.240.39; P<0.0001 [Figure 1A]), with 18-month PFS rates of 48.0% and 7.9%, respectively. Among response-evaluable patients (D-Vd, n=240; Vd, n=234), ORR was significantly improved with D-Vd versus Vd (83.8% versus 63.2%; P<0.0001 [Table 2]), including higher rates of stringent complete response (CR) (8.8% versus 2.6%), CR or better (28.8% versus 9.8%; P<0.0001), and very good partial response or better (62.1% versus 29.1%; P<0.0001 [Online Supplementary Table S2]). MRD was evaluated for the ITT population at pre-specified time points using a stringent, unbiased approach with IMWG criteria of a minimum sensitivity threshold of 10–5 for next-generation sequencing evaluation.20 At this threshold, 11.6% of D-Vd–treated patients were MRDnegative versus 2.4% of Vd-treated patients (P=0.000034 [Table 2]). Consistent findings were observed at a higher sensitivity threshold of 10–6 (D-Vd: 4.8%; Vd: 0.8%; P=0.004763). Overall survival (OS) remained immature at the time of this analysis, and survival follow up will continue until 320 deaths are reported, per protocol. Subgroup analyses showed the clinical benefit of daratumumab by prolonging PFS and improving ORR and MRD negativity across all clinical populations (Table 2 and Figure 2). Patients who received D-Vd at first relapse (Dhaematologica | 2018; 103(12)

Table 1. Baseline demographics and clinical characteristics of the ITT population.

Characteristic Age (years) median (range) Median time from diagnosis (years) Number of prior lines of therapy, n (%) Median (range) 1 2-3 >3 1-3 Prior treatments, n (%) PI Bortezomib IMiD Thalidomide Lenalidomide PI and IMiD Prior ASCT, n (%) Refractory to last line of therapy, n (%) Refractory to lenalidomide at last prior line of therapy, n (%) Treatment-free interval, n (%) >12 months ≤12 months >6 months ≤6 months Cytogenetic profile, n (%)a n Standard-risk High-risk

D-Vd n=251

Vd n=247

64 (30-88)

64 (33-85)

3.9

3.7

2 (1-9) 122 (48.6) 107 (42.6) 22 (8.8) 229 (91.2)

2 (1-10) 113 (45.7) 106 (42.9) 28 (11.3) 219 (88.7)

169 (67.3) 162 (64.5) 179 (71.3) 125 (49.8) 89 (35.5) 112 (44.6) 157 (62.5) 76 (30.3) 45 (17.9)

172 (69.6) 164 (66.4) 198 (80.2) 121 (49.0) 120 (48.6) 129 (52.2) 149 (60.3) 85 (34.4) 60 (24.3)

118 (47.0) 133 (53.0) 150 (59.8) 101 (40.2)

104 (42.1) 143 (57.9) 133 (53.8) 114 (46.2)

167 123 (73.7) 44 (26.3)

186 135 (72.6) 51 (27.4)

ITT: intent-to-treat; D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone; PI: proteasome inhibitor; IMiD: immunomodulatory drug; ASCT: autologous stem cell transplantation; data are median (range) or n (%). a Cytogenetic status was determined using next-generation sequencing. High-risk cytogenetic status was defined as having ≥1 of the following abnormalities: del17p, t(4;14), or t(14;16); standard-risk cytogenetic status was defined as those who underwent cytogenetic testing and did not meet the high-risk criteria.

Vd, n=122; Vd, n=113) achieved the greatest benefit [Table 2 and Figure 2]. In this population, PFS was significantly prolonged with D-Vd versus Vd (median: not reached versus 7.9 months; HR, 0.19; 95% CI, 0.12-0.29; P<0.0001 [Figure 1B]), an 81% reduction in the risk of disease progression or death with 18-month PFS of 68.0% versus 11.5%, respectively. Among patients with 2 to 3 prior lines of therapy (D-Vd, n=107; Vd, n=106), PFS was also significantly prolonged with D-Vd versus Vd (median: 9.8 versus 6.3 months; HR, 0.51; 95% CI, 0.36-0.71; P<0.0001), with 18-month PFS of 31.2% versus 5.5%, respectively (Figure 1C). Likewise, in patients with 1 to 3 prior lines of therapy (D-Vd, n=229; Vd, n=219), D-Vd significantly prolonged PFS versus Vd (median: 18.9 versus 7.3 months; HR, 0.31; 95% CI, 0.24-0.40; P<0.0001), with 18month PFS rates of 51.2% versus 8.7%, respectively (Online Supplementary Figure S2). 2081

A. Spencer et al. A

Figure 1.PFS (A) in the ITT population and (B) in patients who received 1 prior line of therapy or (C) 2 to 3 prior lines of therapy. Kaplan-Meier estimates of PFS. in (A) the the ITT population and in patients who received (B) 1 prior line of therapy or (C) 2 to 3 prior lines of therapy. D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone; HR: hazard ratio; CI: confidence interval.

The PFS benefit of daratumumab was maintained in patients who received prior bortezomib (D-Vd, n=162; Vd, n=164; median: 12.1 versus 6.7 months; HR, 0.35; 95% CI, 0.26-0.46; P<0.0001 [Online Supplementary Figure S3]), with 18-month PFS rates of 37.9% and 1.8%, respectively. In this subgroup, D-Vd improved ORR (80.5% versus 59.5%) and increased MRD-negative rates (6.2% versus 0.6%) versus Vd (Table 2). Importantly, the PFS benefit of daratumumab was maintained in patients who received prior bortezomib in their only line of therapy (D-Vd, n=62; Vd, n=57; median: 19.6 versus 8.0 months; HR, 0.20; 95% CI, 0.12-0.35; P<0.0001 [Online Supplementary Figure S4]), with 18-month PFS rates of 58.1% and 2.1%, respectively. Patients refractory to lenalidomide at their last prior line of therapy (D-Vd, n=45; Vd, n=60) also achieved a significant PFS benefit with D-Vd versus Vd (median: 9.3 versus 4.4 months; HR, 0.36; 95% CI, 0.21-0.63; P=0.0002 [Figure 2]), with 18-month PFS rates of 33.5% versus 2.0%, respectively. In this subgroup, D-Vd improved ORR (80.5% versus 50.0%) and increased MRD negativity (8.9% versus 0%) versus Vd [Table 2]. In a pre-specified subgroup analysis of cytogenetic risk, 2082

D-Vd prolonged PFS and improved ORR versus Vd (Table 2, Figures 2 and 3A). PFS was prolonged with D-Vd versus Vd in both high-risk (median: 11.2 versus 7.2 months; HR, 0.45; 95% CI, 0.25-0.80; P=0.0053; D-Vd, n=44; Vd, n=51) and standard-risk disease (median: 19.6 versus 7.0 months; HR: 0.26; 95% CI, 0.18-0.37; P<0.0001; D-Vd, n=123; Vd, n=135 [Figures 2 and 3A]). ORRs were higher with D-Vd for both high-risk (D-Vd, n=44; Vd, n=47; 81.8% versus 61.7%; P=0.2028) and standard-risk (D-Vd, n=118; Vd, n=131; 84.7% versus 64.1%; P=0.0001) subgroups (Table 2). Higher D-Vd response rates aligned with MRD negativity. In the D-Vd group, 13.8% (17/123) of evaluable, standard-risk patients reached MRD negativity at 10–5 sensitivity versus 2.2% (3/135) in the Vd group (P=0.0003 [Table 2]). No high-risk Vd group patients (n=51) achieved MRD negativity at 10–5, unlike 13.6% (6/44) of high-risk D-Vd group patients (P=0.0018). The PFS benefit of D-Vd versus Vd was also maintained irrespective of the time since last therapy (≤12, >12, ≤6, or >6 months [Figure 2]). Regardless of treatment group, PFS was prolonged in patients who achieved MRD-negative status (median: not reached in either group [Figure 3B]). Conversely, among patients with MRD-positive status (10–5), D-Vd significanthaematologica | 2018; 103(12)

Subgroup analyses of CASTOR

(months)

months

Figure 2. PFS based on prior treatment history and cytogenetic risk (ITT population). Subgroup analysis of PFS based on prior lines of therapy, prior treatment exposure, refractoriness to lenalidomide at the last prior line of therapy, treatment-free interval, and cytogenetic risk. Patients with high-risk cytogenetics had any of t(4;14), t(14;16), or del17p cytogenetic abnormalities as determined by central next-generation sequencing. Standard-risk patients had an absence of high-risk abnormalities. PFS: progression-free survival; ITT: intent-to-treat; D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone; CI: confidence interval; NR: not reached.

ly prolonged PFS versus Vd (median: not reached versus 16.2 months; HR, 0.19; 95% CI, 0.05-0.73; P=0.0080 [Figure 3B]). The rate of MRD-negativity (10–5) continued to increase over time for patients in the overall study population who received D-Vd versus Vd (Figure 4). Within the safety population (D-Vd, n=243; Vd, n=237), longer follow up revealed a tolerability profile consistent with the primary analysis and no new emergent toxicities. Among the most common (≥15%) hematologic treatment-emergent adverse events (TEAEs) were thrombocytopenia and anemia. Among the most common (≥15%) non-hematologic TEAEs were peripheral sensory neuropathy, diarrhea, upper respiratory tract infection, and cough (Table 3). The most common (≥5%) grade 3 or 4 hematologic TEAEs included thrombocytopenia, anemia, neutropenia, and lymphopenia (Table 3). The most common (≥5%) grade 3 or 4 non-hematologic TEAEs included pneumonia, haematologica | 2018; 103(12)

hypertension, and peripheral sensory neuropathy. Discontinuations due to TEAEs remained low and balanced between groups (D-Vd: 9.5%; Vd: 9.3%). Transfusions were received by 26.3% versus 20.3% of patients (D-Vd versus Vd). With longer follow up, second primary malignancies (SPMs) occurred in 10 (4.1%) patients who received D-Vd (4 new cases following the primary analysis13 included basal and squamous cell carcinoma, Bowen disease, and prostate cancer) versus 1 (0.4%) patient who received Vd (no new cases with longer follow up). The EORTC QLQ-C30 and EQ-5D-5L tools showed that HRQoL was maintained during treatment for patients in both groups who remained on the study. Significant differences in the least squares mean changes from baseline were not observed between D-Vd and Vd at any time for the EORTC QLQ-C30 Global Health Status Scores or the EQ-5D-5L Utility Score. A significant difference was 2083

A. Spencer et al. Table 2. ORR and MRD based on prior treatment history. # of patients in group Subgroup ITTb Prior lines of therapy 1 2-3 >3 1-3 Prior therapy Bortezomib Lenalidomide Thalidomide Refractory to lenalidomide at last prior line of therapy Treatment-free interval ≤12 months >12 months ≤6 months >6 months Cytogenetic riske Highf Standard

ORR, n (%)a

# of patients in group

MRD, n (%)

D-Vd

10–5 Vd

201 (83.8) 148 (63.2) <0.0001

251

247

29 (11.6)

6 (2.4)

0.000034

12 (4.8) 2 (0.8)

0.004763

109 100 25 209

108 (90.8) 78 (78.8) 15 (68.2) 186 (85.3)

81 (74.3) 0.0014 58 (58.0) 0.0022 9 (36.0) 0.0294 139 (66.5) <0.0001

122 107 22 229

113 106 28 219

17 (13.9) 12 (11.2) N/A 29 (12.7)

3 (2.7) 3 (2.8) N/A 6 (2.7)

0.001138 0.013511 N/A <0.0001

8 (6.6) 2 (1.8) 4 (3.7) 0 (0) N/A N/A 12 (5.2) 2 (0.9)

0.059541 0.018130 N/A 0.0055

154 83 120 41

153 112 115 58

124 (80.5) 65 (78.3) 102 (85.0) 33 (80.5)

91 (59.5) <0.0001 59 (52.7) <0.0001 74 (64.3) 0.0003 29 (50.0) 0.0021

162 89 125 45

164 120 121 60

10 (6.2) 7 (7.9) 16 (12.8) 4 (8.9)

1 (0.6) 2 (1.7) 4 (3.3) 0 (0)

0.002822 0.0278 0.0049 0.008194

5 (3.1) 2 (2.2) 6 (4.8) 1 (2.2)

0 (0) 0 (0) 2 (1.7) 0 (0)

0.007830 0.0636 0.1544 0.191319

125 115 94 146

135 99 107 127

96 (76.8) 105 (91.3) 72 (76.6) 129 (88.4)

66 (48.9) <0.0001 82 (82.8) 0.0632 50 (46.7) <0.0001 98 (77.2) 0.0139

133 118 101 150

143 104 114 133

13 (9.8) 16 (13.6) 8 (7.9) 21 (14.0)

1 (0.7) 5 (4.8) 1 (0.9) 5 (3.8)

0.0002 0.0223 0.0067 0.0020

4 (3.0) 8 (6.8) 3 (3.0) 9 (6.0)

0 (0) 2 (1.9) 0 (0) 2 (1.5)

0.0151 0.0704 0.0323 0.0413

44 118

47 131

36 (81.8) 29 (61.7) 100 (84.7) 84 (64.1)

44 123

51 135

6 (13.6) 17 (13.8)

0 (0) 3 (2.2)

0.0018 0.0003

5 (11.4) 0 (0) 6 (4.9) 1 (0.7)

0.004 0.0328

D-Vd

240

234

119 99 22 218

D-Vd

0.2028 0.0001

D-Vd

10–6 Vd

ORR: overall response rate; MRD: minimal residual disease; D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone; ITT: intent-to-treat; N/A: not available. Data are n (%) based on computerized algorithm. aResponse-evaluable population. bITT population. cP-value was generated using the Cochran-Mantel-Haenszel χ2 test. dP-value was generated using the likelihood-ratio χ2 test. eBiomarker risk-evaluable population. fIncludes subjects who have either del17p, t(14;16), t(4;14), or a combination of these.

Table 3. Adverse events in the safety population.

D-Vd (n=243) Common hematologic adverse events, n (%) Thrombocytopenia Anemia Neutropenia Lymphopenia Common non-hematologic adverse events, n (%) Peripheral sensory neuropathy Diarrhea Upper respiratory tract infection Cough Fatigue Constipation Back pain Dyspnea Edema peripheral Pyrexia Insomnia Asthenia Pneumonia Hypertension

Vd (n=237)

All-grade ≥15% 145 (59.7) 69 (28.4) 46 (18.9) 32 (13.2)

Grade 3 or 4 ≥5% 111 (45.7) 37 (15.2) 33 (13.6) 24 (9.9)

All-grade ≥15% 105 (44.3) 75 (31.6) 23 (9.7) 9 (3.8)

Grade 3 or 4 ≥5% 78 (32.9) 38 (16.0) 11 (4.6) 6 (2.5)

121 (49.8) 85 (35.0) 76 (31.3) 68 (28.0) 53 (21.8) 53 (21.8) 47 (19.3) 46 (18.9) 45 (18.5) 43 (17.7) 42 (17.3) 24 (9.9) 36 (14.8) 23 (9.5)

11 (4.5) 9 (3.7) 6 (2.5) 0 (0.0) 12 (4.9) 0 (0.0) 5 (2.1) 9 (3.7) 1 (0.4) 3 (1.2) 2 (0.8) 2 (0.8) 24 (9.9) 16 (6.6)

90 (38.0) 53 (22.4) 43 (18.1) 30 (12.7) 58 (24.5) 38 (16.0) 24 (10.1) 21 (8.9) 20 (8.4) 28 (11.8) 36 (15.2) 37 (15.6) 31 (13.1) 8 (3.4)

16 (6.8) 3 (1.3) 1 (0.4) 0 (0.0) 8 (3.4) 2 (0.8) 3 (1.3) 2 (0.8) 0 (0.0) 3 (1.3) 3 (1.3) 5 (2.1) 24 (10.1) 2 (0.8)

D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone. Data are n (%). Incidences of all-grade and grade 3 or 4 adverse events occurring in ≥15% and ≥5% of patients in either treatment group are listed, respectively.

2084

haematologica | 2018; 103(12)

Subgroup analyses of CASTOR

Figure 4. Time to MRD negativity in the ITT population. MRD-negative status was evaluated over time at a sensitivity threshold of 10–5 using bone marrow aspirate samples that were prepared using Ficoll and analyzed by the clonoSEQ® assay. MRD: minimal residual disease; D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone.

Figure 3. PFS survival based on (A) cytogenetic risk and (B) MRD status. (A) Kaplan-Meier estimates of PFS among patients evaluated for cytogenetic risk. High-risk patients had any of t(4;14), t(14;16), or del17p cytogenetic abnormalities as determined by central next-generation sequencing. Standard-risk patients had an absence of high-risk abnormalities. (B) Kaplan-Meier estimates of PFS among patients in the ITT population population. MRD-negative status was evaluated at a sensitivity threshold of 10–5 using bone marrow aspirate samples that were prepared using Ficoll and analyzed by the clonoSEQ® assay. MRD: minimal residual disease; D-Vd: daratumumab plus bortezomib and dexamethasone; Vd: bortezomib and dexamethasone.

observed solely at Week 21 in favor of D-Vd for the Visual Analog Scale Score (P=0.0185). No significant differences in EORTC QLQ-C30 global health status were observed for median time to improvement (5.0 versus 5.1 months; HR, 0.99; 95% CI, 0.76-1.29; P=0.9163). Similarly, no significant differences in median time to improvement were observed for either the EQ-5D-5L Utility Score (7.7 versus 3.5 months; HR, 0.82; 95% CI, 0.62-1.08; P=0.1469) or the Visual Analog Scale Score (5.0 versus 5.0 months; HR, 1.03; 95% CI, 0.79-1.35; P=0.8072).

Discussion These data confirm that D-Vd provides significant clinical benefit to patients with RRMM. D-Vd prolonged PFS, resulting in a 69% reduction in the risk of disease progression or death versus Vd. With an additional 12 months of follow up, responses to daratumumab deepened over time haematologica | 2018; 103(12)

(≥CR: 28.8%) compared with the primary analysis (19.2%).13 Deeper responses to D-Vd were associated with significantly higher (>4 fold) MRD-negative rates at sensitivities of 10–5 and 10–6 versus Vd. We hypothesize that, as previous studies have demonstrated a correlation between MRD negativity and OS,21,22 this may translate into improved OS outcomes after longer follow up for patients treated with D-Vd. Analysis of OS is ongoing. There were consistent clinical benefits with D-Vd versus Vd across subgroups based on prior lines of therapy, treatment exposure, or refractory status. These were also observed in patients regardless of time since last therapy or cytogenetic risk, as those patient subgroups were not evaluated in the primary analysis. Importantly, the benefit of D-Vd was maintained in patients who received prior bortezomib (including as their sole prior line of therapy) and those refractory to lenalidomide at their last prior line of therapy. Bortezomib- and lenalidomide-based combinations are common MM first-line and maintenance regimens. Thus D-Vd can be considered after bortezomib (if patients are not PI-refractory) or in lenalidomide-refractory patients, which is of particular importance considering the increased lenalidomide use as maintenance therapy in newly diagnosed MM regardless of transplant eligibility.23,24 D-Vd significantly prolonged PFS versus Vd across all lines of therapy with the greatest benefit achieved in patients who received 1 prior line in comparison to those who received 2 to 3 or >3 prior lines of therapy. Response rates, including the rates of MRD-negativity, were also highest in patients who received 1 prior line of therapy. As D-Vd showed the greatest benefit at first relapse, it may represent an optimal second-line treatment for patients after frontline lenalidomide or bortezomib. 2085

A. Spencer et al.

The benefit of D-Vd was also maintained in patients regardless of cytogenetic risk, as D-Vd but not Vd induced MRD negativity in high-risk patients, suggesting that this combination may improve historically poor outcomes in this population.25-28 D-Vdâ&#x20AC;&#x201C;treated patients continued to receive daratumumab monotherapy after completing 8 cycles of Vd, reflected by the longer treatment duration (median: D-Vd, 13.4 months; Vd, 5.2 months). With longer follow up, the depth of response in the D-Vd arm, including CR rates and MRD negativity, continued to improve over time after patients entered the monotherapy phase, supporting the benefit of continued daratumumab treatment. Analyses are ongoing to quantify the therapeutic impact of maintenance therapy with single-agent daratumumab. This was the first randomized, phase 3 clinical trial of RRMM with prospective MRD evaluation. MRD-negative status was associated with prolonged PFS in both treatment groups, but D-Vd increased MRD-negative rates at all sensitivity thresholds and evaluated subgroups. Additional longitudinal MRD evaluation in CASTOR is ongoing and the potential benefit of daratumumabinduced MRD negativity is being explored in studies of newly diagnosed MM (ALCYONE clinicaltrials.gov identifier 02195479; clinicaltrials.gov identifier 02252172; clinicaltrials.gov identifier 02541383; clinicaltrials.gov identifier 03301220. These studies aim to further validate MRDnegative status as a surrogate study endpoint. Several new agents for RRMM have been approved based on robust clinical data, including carfilzomib29 and ixazomib30 (second-generation PIs), pomalidomide31,32 (a third-generation immunomodulatory drug), daratumumab13,14,33-35 and elotuzumab36 (monoclonal antibodies), and panobinostat4 (a histone deacetylase inhibitor). Approvals of many of these agents were based on superiority of PFS in phase 3 trials. These studies are beginning to report OS outcomes. In the ENDEAVOR study, carfilzomib and dexamethasone conferred an additional OS benefit of 7.6 months versus Vd.37 OS analysis in CASTOR is ongoing. Clinical trials are not usually powered to determine optimal treatment sequencing or the most effective regimen for each disease subset.38 Although meta-analyses provide useful guides for selecting treatment options, physicians need to consider many different factors to optimize individual regimens, including numbers and types of prior regimens, duration of response to prior therapy, toxicities with prior therapies, disease aggressiveness, and performance status or frailty.38,39 Based on the current findings, and the findings of other studies,40 daratumumab combined with other anti-myeloma drugs such as borte-

References 1. Kurtin SE. Relapsed or relapsed/refractory multiple myeloma. J Adv Pract Oncol. 2013;4(Suppl 1):5-14. 2. Merin NM, Kelly KR. Clinical use of proteasome inhibitors in the treatment of multiple myeloma. Pharmaceuticals (Basel). 2015;8 (1):1-20. 3. Dimopoulos MA, Moreau P, Palumbo A, et

2086

zomib or lenalidomide, may provide significant benefit in patients with early relapsed MM regardless of prior treatment exposure. It remains to be seen whether this translates to prolonged survival. The safety profile of D-Vd remained unchanged with approximately 1 year of additional follow up from the primary analysis,13 with no new unexpected TEAEs observed. Preliminary data indicated that adding a third agent to Vd did not worsen HRQoL, an evaluation that was not presented in the primary analysis. More SPMs were reported with D-Vd versus Vd (4.1% versus 0.4%); this rate is similar to the incidence of SPMs reported for patients in POLLUX (5.7% for both D-Rd and Rd; manuscript in preparation) and for RRMM patients in general (between 1%-6%).41 At clinical cut-off, all patients in the Vd group had discontinued or completed 8 treatment cycles, whereas 41% of patients receiving D-Vd remained on daratumumab treatment. Therefore, more frequent monitoring during active treatment may explain why a greater number of TEAEs (including grade 3 or 4 events) and SPMs were reported with D-Vd. After 8 cycles of D-Vd, patients were monitored every 4 weeks during daratumumab dosing, whereas patients who received Vd who did not receive daratumumab monotherapy were followed for survival via phone calls every 16 weeks following disease progression. In conclusion, the original finding of significant benefit of D-Vd over Vd was confirmed regardless of treatment history or cytogenetic risk. Importantly, this clinical benefit was achieved without any emergent safety issues or decline in HRQoL. These results provide further support for the addition of daratumumab to a standard of care regimen in RRMM, particularly at first relapse. The CASTOR study is ongoing, and the feasibility of MRD negativity as a surrogate for OS in RRMM continues to be investigated. An analysis of OS will be conducted after 320 events are observed. Funding This study was sponsored by Janssen Research & Development, LLC. The data sharing policy of Janssen Pharmaceutical Companies of Johnson & Johnson is available at https://www.janssen.com/clinical-trials/transparency. As noted on this site, requests for access to the study data can be submitted through Yale Open Data Access (YODA) Project site at http://yoda.yale.edu. Acknowledgments Medical writing and editorial support were provided by Jason Jung, PhD and Kristin Runkle, PhD of MedErgy, and were funded by Janssen Global Services, LLC.

al. Carfilzomib and dexamethasone versus bortezomib and dexamethasone for patients with relapsed or refractory multiple myeloma (ENDEAVOR): a randomised, phase 3, open-label, multicentre study. Lancet Oncol. 2016;17(1):27-38. 4. San-Miguel JF, Hungria VT, Yoon SS, et al. Panobinostat plus bortezomib and dexamethasone versus placebo plus bortezomib and dexamethasone in patients with relapsed or relapsed and refractory multiple

myeloma: a multicentre, randomised, double-blind phase 3 trial. Lancet Oncol. 2014;15(11):1195-1206. 5. San-Miguel JF, Hungria VTM, Yoon SS, et al. Final analysis of overall survival from the Phase 3 panorama 1 trial of panobinostat plus bortezomib and dexamethasone versus placebo plus bortezomib and dexamethasone in patients with relapsed or relapsed and refractory multiple myeloma. Blood. 2015;126(23):3026.

haematologica | 2018; 103(12)

Subgroup analyses of CASTOR 6. Jakubowiak A, Offidani M, Pegourie B, et al. Randomized phase 2 study of elotuzumab plus bortezomib/dexamethasone (Bd) versus Bd for relapsed/refractory multiple myeloma. Blood. 2016;127(33):2833-2840. 7. de Weers M, Tai YT, van der Veer MS, et al. Daratumumab, a novel therapeutic human CD38 monoclonal antibody, induces killing of multiple myeloma and other hematological tumors. J Immunol. 2011;186(3):18401848. 8. Lammerts van Bueren J, Jakobs D, Kaldenhoven N, et al. Direct in vitro comparison of daratumumab with surrogate analogs of CD38 antibodies MOR03087, SAR650984 and Ab79. Blood. 2014;124 (21):3474. 9. Overdijk MB, Verploegen S, Bogels M, et al. Antibody-mediated phagocytosis contributes to the anti-tumor activity of the therapeutic antibody daratumumab in lymphoma and multiple myeloma. MAbs. 2015;7(2):311-321. 10. Overdijk MB, Jansen JH, Nederend M, et al. The therapeutic CD38 monoclonal antibody daratumumab induces programmed cell death via Fcgamma receptor-mediated crosslinking. J Immunol. 2016;197(3):807-813. 11. van de Donk NWCJ, Janmaat ML, Mutis T, et al. Monoclonal antibodies targeting CD38 in hematological malignancies and beyond. Immunol Rev. 2016;270(1):95-112. 12. Krejcik J, Casneuf T, Nijhof IS, et al. Daratumumab depletes CD38+ immuneregulatory cells, promotes T-cell expansion, and skews T-cell repertoire in multiple myeloma. Blood. 2016;128(3):384-394. 13. Palumbo A, Chanan-Khan A, Weisel K, et al. Daratumumab, bortezomib, and dexamethasone for multiple myeloma. N Engl J Med. 2016;375(8):754-766. 14. Dimopoulos MA, Oriol A, Nahi H, et al. Daratumumab, lenalidomide, and dexamethasone for multiple myeloma. N Engl J Med. 2016;375(14):1319-1331. 15. DARZALEXÂŽ (daratumumab) injection, for intravenous use [package insert]. Horsham, PA: Janssen Biotech, Inc.; 2018. 16. European Medicines Agency. DARZALEX summary of product characteristics, May 2016. Available at: http:// www.ema.europa.eu/docs/en_GB/docum e n t _ l i b r a r y / E PA R _ - _ P r o d u c t _ Information/human/004077/WC500207296 .pdf. Last accessed July 2018. 17. Durie BGM, Harousseau JL, Miguel JS, et al. International uniform response criteria for multiple myeloma. Leukemia. 2006;20(9): 1467-1473. 18. Rajkumar SV, Harousseau JL, Durie B, et al. Consensus recommendations for the uniform reporting of clinical trials: report of the

haematologica | 2018; 103(12)

19.

20.

21.

22.

23. 24.

25.

26.

27.

28.

29.

International Myeloma Workshop Consensus Panel 1. Blood. 2011;117(18): 4691-4695. Chiu C, Soong D, Spicka I, et al. Next generation sequencing (NGS) methodology for determining cytogenetic risk status in the daratumumab phase 3 CASTOR and POLLUX studies in relapsed or refractory multiple myeloma (RRMM). Presented at: the 22nd Congress of the European Hematology Association (EHA); June 22-25, 2017; Madrid, Spain. Abstract S100. Kumar S, Paiva B, Anderson KC, et al. International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma. Lancet Oncol. 2016;17(8):e328e346. Martinez-Lopez J, Lahuerta JJ, Pepin F, et al. Prognostic value of deep sequencing method for minimal residual disease detection in multiple myeloma. Blood. 2014;123(20):3073-3079. Rawstron AC, Gregory WM, de Tute RM, et al. Minimal residual disease in myeloma by flow cytometry: independent prediction of survival benefit per log reduction. Blood. 2015;125(12):1932-1935. McCarthy PL, Palumbo A. Maintenance therapy for multiple myeloma. Hematol Oncol Clin North Am. 2014;28(5):839-859. 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. Dimopoulos MA, Weisel KC, Song KW, et al. Cytogenetics and long-term survival of patients with refractory or relapsed and refractory multiple myeloma treated with pomalidomide and low-dose dexamethasone. Haematologica. 2015;100(10):13271333. Avet-Loiseau H, Attal M, Moreau P, et al. Genetic abnormalities and survival in multiple myeloma: the experience of the Intergroupe Francophone du Myelome. Blood. 2007;109(8):3489-3495. Avet-Loiseau H, Hulin C, Campion L, et al. Chromosomal abnormalities are major prognostic factors in elderly patients with multiple myeloma: the intergroupe francophone du myelome experience. J Clin Oncol. 2013;31(22):2806-2809. Chng WJ, Dispenzieri A, Chim CS, et al. IMWG consensus on risk stratification in multiple myeloma. Leukemia. 2014;28(2): 269-277. Stewart AK, Rajkumar SV, Dimopoulos MA, et al. Carfilzomib, lenalidomide, and dexamethasone for relapsed multiple myeloma. N Engl J Med. 2015;372(2):142-152.

30. Moreau P, Masszi T, Grzasko N, et al. Oral ixazomib, lenalidomide, and dexamethasone for multiple myeloma. N Engl J Med. 2016;374(17):1621-1634. 31. Richardson PG, Siegel DS, Vij R, et al. Pomalidomide alone or in combination with low-dose dexamethasone in relapsed and refractory multiple myeloma: a randomized phase 2 study. Blood. 2014;123(12):18261832. 32. San Miguel J, Weisel K, Moreau P, et al. Pomalidomide plus low-dose dexamethasone versus high-dose dexamethasone alone for patients with relapsed and refractory multiple myeloma (MM-003): a randomised, open-label, phase 3 trial. Lancet Oncol. 2013;14(11):1055-1066. 33. Lokhorst HM, Plesner T, Laubach JP, et al. Targeting CD38 with daratumumab monotherapy in multiple myeloma. N Engl J Med. 2015;373(13):1207-1219. 34. Lonial S, Weiss BM, Usmani S, et al. Daratumumab monotherapy in patients with treatment-refractory multiple myeloma (SIRIUS): an open-label, randomised, phase 2 trial. Lancet. 2016; 387(10027):1551-1560. 35. Chari A, Suvannasankha A, Fay JW, et al. Daratumumab plus pomalidomide and dexamethasone in relapsed and/or refractory multiple myeloma. Blood. 2017; 130(8):974981. 36. Lonial S, Dimopoulos M, Palumbo A, et al. Elotuzumab therapy for relapsed or refractory multiple myeloma. N Engl J Med. 2015;373(7):621-631. 37. Dimopoulos MA, Goldschmidt H, Niesvizky R, et al. Carfilzomib or bortezomib in relapsed or refractory multiple myeloma (ENDEAVOR): an interim overall survival analysis of an open-label, randomised, phase 3 trial. Lancet Oncol. 2017;18:1327-1337. 38. Boudreault JS, Touzeau C, Moreau P. Triplet combinations in relapsed/refractory myeloma: update on recent phase 3 trials. Expert Rev Hematol. 2017;10(3):207-215. 39. Yee AJ, Raje NS. Sequencing of nontransplant treatments in multiple myeloma patients with active disease. Hematology Am Soc Hematol Educ Program. 2016;2016 (1):495-503. 40. van Beurden-Tan CH, Franken MG, Blommestein HM, Uyl-de Groot CA, Sonneveld P. Systematic literature review and network meta-analysis of treatment outcomes in relapsed and/or refractory multiple myeloma. J Clin Oncol. 2017;35:13131319. 41. Areethamsirikul N, Reece DE. The risk of secondary primary malignancies after therapy for multiple myeloma. Leuk Lymphoma. 2015;56(11):3012-3021.

2087

ARTICLE

Plasma Cell Disorders

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2088-2096

Daratumumab plus lenalidomide and dexamethasone versus lenalidomide and dexamethasone in relapsed or refractory multiple myeloma: updated analysis of POLLUX

Meletios A. Dimopoulos,1 Jesus San-Miguel,2 Andrew Belch,3 Darrell White,4 Lotfi Benboubker,5 Gordon Cook,6 Merav Leiba,7 James Morton,8 P. Joy Ho,9 Kihyun Kim,10 Naoki Takezako,11 Philippe Moreau,12 Jonathan L. Kaufman,13 Heather J. Sutherland,14 Marc Lalancette,15 Hila Magen,16 Shinsuke Iida,17 Jin Seok Kim,18 H. Miles Prince,19 Tara Cochrane,20 Albert Oriol,21 Nizar J. Bahlis,22 Ajai Chari,23 Lisa O’Rourke,24 Kaida Wu,24 Jordan M. Schecter,25 Tineke Casneuf,26 Christopher Chiu,24 David Soong,24 A. Kate Sasser,27 Nushmia Z. Khokhar,24 Hervé Avet-Loiseau28 and Saad Z. Usmani29

The National and Kapodistrian University of Athens, Greece; 2Clínica Universidad de Navarra-CIMA, IDISNA, CIBERONC, Pamplona, Spain; 3Department of Oncology, University of Alberta Cross Cancer Institute, Edmonton, Canada; 4QEII Health Sciences Center and Dalhousie University, Halifax, Nova Scotia, Canada; 5Service d’Hématologie et Thérapie Cellulaire, Hôpital Bretonneau, Centre Hospitalier Régional Universitaire (CHRU), Tours, France; 6St James’s Institute of Oncology, Leeds Teaching Hospitals NHS Trust and University of Leeds, UK; 7Sheba Medical Center, Tel Hashomer, Ramat Gan, Israel; 8Icon Cancer Care, South Brisbane, QLD, Australia; 9Institute of Haematology, Royal Prince Alfred Hospital, Camperdown, NSW, Australia; 10Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea; 11Department of Hematology, National Hospital Organization Disaster Medical Center of Japan, Tachikawa, Japan; 12Department of Hematology, University Hospital Hôtel-Dieu, Nantes, France; 13Winship Cancer Institute, Emory University, Atlanta, GA, USA; 14Leukemia/Bone Marrow Transplant Program, University of British Columbia, Vancouver, Canada; 15CHU de Québec Research Center, Faculty of Medicine, Laval University, Canada; 16Institute of Hematology, Davidoff Cancer Center, Beilinson Hospital, Rabin Medical Center, Petah-Tikva, Israel; 17Department of Hematology and Oncology, Nagoya City University Graduate School of Medical Sciences, Japan; 18 Yonsei University College of Medicine, Severance Hospital, Seoul, South Korea; 19 Cabrini Hospital, Epworth HealthCare and Sir Peter MacCallum Department of Oncology, University of Melbourne, VIC, Australia; 20Gold Coast University Hospital, Southport, QLD, Australia; 21Institut Català d’Oncologia i Institut Josep Carreras, Hospital Germans Trias I Pujol, Barcelona, Spain; 22University of Calgary, Arnie Charbonneau Cancer Institute, Alberta, Canada; 23Icahn School of Medicine at Mount Sinai, New York, NY, USA; 24Janssen Research & Development, LLC, Spring House, PA, USA; 25Janssen Research & Development, LLC, Raritan, NJ, USA; 26Janssen Research & Development, Beerse, Belgium; 27Genmab US, Inc, Princeton, NJ, USA; 28Unite de Genomique du Myelome, IUC-Oncopole, Toulouse, France and 29Levine Cancer Institute/Atrium Health, Charlotte, NC, USA 1

Correspondence: mdimop@med.uoa.gr

Received: March 27, 2018. Accepted: August 21, 2018. Pre-published: September 20, 2018. doi:10.3324/haematol.2018.194282 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2088 ©2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2088

ABSTRACT

n the POLLUX study, daratumumab plus lenalidomide/dexamethasone significantly reduced risk of progression/death versus lenalidomide/dexamethasone alone in relapsed/refractory multiple myeloma. We provide one additional year of follow up and include the effect on minimal residual disease and in clinically relevant subgroups. After 25.4 months of follow up, daratumumab plus lenalidomide/dexamethasone prolonged progression-free survival versus lenalidomide/dexamethasone alone (median not reached vs. 17.5 months; hazard ratio, 0.41; 95% confidence interval, 0.31-0.53; P<0.0001). The overall response rate was 92.9% versus 76.4%, and 51.2% versus 21.0% achieved a complete response or better, respectively (both P<0.0001). At the 10–5 sensitivity threshold, 26.2% versus 6.4% were minimal residual disease–negative, respectively (P<0.0001). Post hoc analyses of clinically relevant patient subgroups demonstrated that progression-free survival was significantly prolonged for daratumumab plus lenalidomide/dexamethasone versus haematologica | 2018; 103(12)

Subgroup analyses of D-Rd vs. Rd in POLLUX

lenalidomide/dexamethasone regardless of number of prior lines of therapy. Patients previously treated with lenalidomide or thalidomide and those refractory to bortezomib received similar benefits (all P<0.01). Treatment benefit with daratumumab plus lenalidomide/dexamethasone was maintained in high-risk patients (median progression-free survival 22.6 vs. 10.2 months; hazard ratio, 0.53; 95% confidence interval, 0.25-1.13; P=0.0921) and patients with treatment-free intervals of >12 and ≤12 months and >6 and ≤6 months. No new safety signals were observed. In relapsed/refractory multiple myeloma patients, daratumumab plus lenalidomide/dexamethasone continued to improve progression-free survival and deepen responses versus lenalidomide/dexamethasone. Trial Registration: clinicaltrials.gov identifier: 02076009.

Introduction Novel therapeutics approved during the last decade for relapsed or refractory multiple myeloma (RRMM), including proteasome inhibitors (PIs) and immunomodulatory drugs (IMiDs), are now standard of care treatments.1-3 However, patients eventually relapse and subpopulations such as those with high-risk cytogenetic abnormalities may not achieve the same benefits as standard-risk patients.4 Treatment choices must be based on patient factors (comorbidities, frailty, preferences), disease factors (burden, molecular risk), and treatment factors (prior therapies, refractoriness, and toxicity). Daratumumab is a human monoclonal antibody targeting CD38, a cell surface receptor highly expressed on multiple myeloma (MM) cells,5,6 that exerts its effects via direct on-tumor and immunomodulatory mechanisms of action.7-11 Daratumumab is approved in many countries as monotherapy for patients with heavily treated RRMM.12,13 In combination with standard of care regimens, including bortezomib/dexamethasone (Vd; CASTOR study)14 and lenalidomide/dexamethasone (Rd; POLLUX study),15 daratumumab significantly prolonged progression-free survival (PFS) and deepened responses in patients with RRMM. This led to regulatory approval in many countries for patients with ≥1 prior line of therapy.16 In the prespecified interim analysis of POLLUX, after a median follow up of 13.5 months, daratumumab in combination with Rd (D-Rd) resulted in a 63% reduction in the risk of disease progression or death, significantly higher overall response rates (ORRs; 92.9% vs. 76.4%), and significantly higher minimal residual disease (MRD)-negativity rates at multiple sensitivity thresholds versus Rd.15 To identify patients who may benefit more from D-Rd treatment, we conducted subgroup analyses of POLLUX after a longer follow up of 25.4 months. The efficacy of D-Rd versus Rd was compared according to the number of prior lines of therapy received, prior IMiD exposure, bortezomib refractoriness, treatment-free interval after previous treatment line, and cytogenetic risk. We also assessed the ability of daratumumab to drive deep clinical responses beyond complete responses (CRs) through assessment of MRD.

tocol was conducted in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice guidelines. The study design and primary results have been previously published.15 Briefly, eligible patients had progressive disease according to International Myeloma Working Group (IMWG) criteria18,19 during or after their last regimen and had received and responded to ≥1 line of prior therapy; lenalidomide-refractory patients were ineligible. Patients were randomized 1:1 to Rd (lenalidomide: 25 mg orally on Days 1-21 of each 28-day cycle; dexamethasone: 40 mg orally weekly) with or without daratumumab (16 mg/kg intravenously weekly for 8 weeks, every 2 weeks for 16 weeks, and then every 4 weeks) until progression.

Endpoints and Assessments The primary efficacy endpoint was PFS, and secondary efficacy endpoints included ORR, rates of very good partial response (VGPR) or better and CR or better, MRD, time to response, and overall survival. This exploratory, post hoc, secondary analysis examined patient populations according to prior lines of therapy received (1, 2-3, 1-3), prior treatment exposure (bortezomib, lenalidomide, and thalidomide), refractoriness to bortezomib, time since last therapy (>12 months, ≤12 months, >6 months, and ≤6 months prior to randomization), and cytogenetic risk. Numbers of prior lines of therapy were determined by investigators according to the IMWG consensus guidelines.19 Treatment-free interval was defined as the duration between the end date of the last line of prior therapy and randomization. Cytogenetic abnormalities were determined with CD138+ selected cells at the screening visit prior to randomization by centralized next-generation sequencing. The presence of 1 or more of t(4;14), t(14;16), or del17p (utilizing >50% deletion cutoff) defined a patient as high risk. Standard-risk cytogenetic status was assigned to those not meeting the high-risk criteria. Due to limitations with sample quality or quantity, and because cytogenetic evaluation was mandated mid-study, cytogenetic risk was not evaluated in all patients. PFS, ORR, and MRD negativity at sensitivity thresholds of 10–5 and 10–6 were assessed for each subgroup. PFS based on MRD and cytogenetic risk status was also examined. Health-related quality of life (HRQoL) was assessed using the EuroQol 5-Dimension Questionnaire (EQ-5D-5L) and the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core-30 (EORTC QLQ-C30).

Minimal Residual Disease Methods Study Design POLLUX is an ongoing randomized, open-label, multicenter, phase 3 study conducted in patients with RRMM (clinicaltrials.gov identifier: 02076009). An independent ethics committee or institutional review board at each site approved the trial. The study prohaematologica | 2018; 103(12)

MRD was assessed (blinded to treatment group) at the time of suspected CR (i.e., CR/stringent CR or VGPR with suspected interference20,21), and at 3 and 6 months after suspected CR for patients who maintained this response. The MRD-negativity rate per treatment arm was determined as the proportion of patients with negative MRD at any time point after the first dose and compared using the likelihood-ratio test. 2089

M.A. Dimopoulos et al.

Additional details on assessments for safety, MRD, cytogenetic risk, and HRQoL and statistical analyses are provided in the Appendix.

Results Of 569 enrolled patients in POLLUX, 286 were assigned to D-Rd and 283 to Rd (Appendix Figure 1). Baseline patient demographics and prior treatment history were generally balanced and have been previously reported.15 Additional baseline clinical and cytogenetic characteristics are summarized in Table 1. The median (range) duration of study treatment was 24.5 (0-32.7) months in the D-Rd group and 16.0 (0.20-32.2) months in the Rd group. At the clinical cut-off on March 7, 2017, the median (range) duration of follow up was 25.4 (0-32.7) months. Consistent with the primary analysis, D-Rd improved PFS compared with Rd (median not reached [NR] vs. 17.5 months; hazard ratio [HR], 0.41; 95% confidence interval [CI], 0.31-0.53; P<0.0001), with 24-month PFS rates of 68.0% versus 40.9%, respectively (Figure 1A). In the response-evaluable population, D-Rd (n=281) compared with Rd alone (n=276) significantly improved the ORR (92.9% vs. 76.4%, respectively; P<0.0001 [Table 2]), and rates of ≥VGPR and ≥CR (Appendix Table 1). Stringent CRs were achieved by 26.0% and 8.7% of patients receiving D-Rd and Rd, respectively (Appendix Table 1). Overall survival data remain immature, and the final analysis is planned after 330 events are observed. D-Rd significantly improved PFS in RRMM patients who received several prior lines of therapy, a population for whom more effective therapies are needed. In patients who received 1 prior line of therapy, D-Rd significantly prolonged PFS compared with Rd (D-Rd, n=149; Rd, n=146; median NR vs. 19.6 months; HR, 0.39; 95% CI, 0.26-0.58, P<0.0001 [Figure 1B, Appendix Figure 2A]), with 24-month PFS rates of 70.3% and 45.0%, respectively. A similar benefit was observed in patients who received 2 to 3 prior lines of therapy (Figure 1B, Appendix Figure 2B). ORR was also significantly improved in patients treated with 1 prior line of therapy (D-Rd, n=147; Rd, n=142; 93.2% vs. 80.3%; P=0.0003 [Table 2]) and 2 to 3 prior lines of therapy (D-Rd, n=120; Rd, n=115; 95.0% vs. 73.9%; P<0.0001 [Table 2]). It is considered preferable for MM patients who relapse to switch drug classes for subsequent therapy,22 and, in bortezomib-refractory patients, D-Rd significantly prolonged PFS compared with Rd (D-Rd, n=59; Rd, n=58; median 26.1 vs. 11.3 months; HR, 0.46; 95% CI, 0.26-0.80; P=0.0051 [Figure 1B]), with 24-month PFS rates of 60.0% and 29.7%, respectively, and improved ORR (D-Rd, n=57; Rd, n=56; 87.7% vs. 67.9%; P=0.0113 [Table 2]). Patients with high-risk cytogenetic status have had historically poor outcomes, but, regardless of cytogenetic risk status, D-Rd improved outcomes compared with Rd (Figures 1B and 2, Table 2). PFS was longer for D-Rd-treated (n=28) versus Rd-treated patients (n=37) with high cytogenetic risk (median 22.6 vs. 10.2 months; HR, 0.53; 95% CI, 0.25-1.13; P=0.0921; 24-month PFS rate, 48.1% vs. 31.7%), and significantly longer for patients with standard-risk disease (D-Rd, n=133; Rd, n=113; median NR vs. 18.5 months; HR, 0.30; 95% CI, 0.20-0.47; P<0.0001; 24-month PFS rate, 74.3% vs. 40.0% [Figures 1B and 2]). With D-Rd versus Rd, significantly higher ORRs were 2090

Table 1. Baseline demographic and clinical characteristics.

Characteristic

D-Rd (n=286)

Age, years Median (range) 65 (34-89) Median (range) time from diagnosis, years 3.48 (0.4-27.0) ECOG performance-status, n (%) 0 139 (48.6) 1 136 (47.6) 2 11 (3.8) Cytogenetic profile, n (%)a n 161 Standard risk 133 (82.6) High risk 28 (17.4) t(4;14) 16 (9.9) t(14;16) 1 (0.6) del17p 13 (8.1) Prior lines of therapy, n (%) Median (range) 1 (1-11) 1 149 (52.1) 2 to 3 123 (43.0) Time since last prior line of therapy, n (%) >12 months 140 (49.0) ≤12 months 146 (51.0) >6 months 187 (65.4) ≤6 months 99 (34.6) Prior ASCT, n (%) 180 (62.9) Prior PI, n (%) 245 (85.7) Bortezomib 241 (84.3) Prior IMiD, n (%) 158 (55.2) Lenalidomide 50 (17.5) Thalidomide 122 (42.7) Prior IMiD + PI, n (%) 125 (43.7) Refractory to bortezomib, n (%) 59 (20.6) Refractory to last line of therapy, n (%) 80 (28.0)

Rd (n=283) 65 (42-87) 3.95 (0.4-21.7) 150 (53.0) 118 (41.7) 15 (5.3) 150 113 (75.3) 37 (24.7) 21 (14.0) 3 (2.0) 13 (8.7) 1 (1-8) 146 (51.6) 118 (41.7) 149 (52.7) 134 (47.3) 188 (66.4) 95 (33.6) 180 (63.6) 242 (85.5) 238 (84.1) 156 (55.1) 50 (17.7) 125 (44.2) 125 (44.2) 58 (20.5) 76 (26.9)

D-Rd: daratumumab/lenalidomide/dexamethasone; Rd: lenalidomide/dexamethasone; ECOG: Eastern Cooperative Oncology Group; ASCT: autologous stem cell transplantation; PI: proteasome inhibitor; IMiD: immunomodulatory drug. aCentral next-generation sequencing. High-risk patients had any of t(4;14), t(14;16), or del17p. Standard-risk patients had an absence of high-risk abnormalities.

observed for both high-risk (D-Rd, n=27; Rd, n=36; 85.2% vs. 66.7%; P=0.0435) and standard-risk patients (D-Rd, n=132; Rd, n = 111; 94.7% vs. 82.0%; P=0.0004 [Table 2]). Further subgroup analyses determined that the clinical benefit of daratumumab was maintained in patients regardless of prior lines of therapy received (1-3), prior treatment exposure (thalidomide or lenalidomide), or time since last therapy (≤12, >12, ≤6, or >6 months [Figure 1B, Table 2, Appendix Figure 2C]). Although D-Rd was associated with a significant treatment benefit in subpopulations that received their last line of prior therapy either ≤12 or >12 months before randomization into the study (Figure 1B, Table 2), the proportion of patients who remained progression-free and alive at 24 months was smaller among patients who relapsed earlier in both the D-Rd (63.6% vs. 72.4%) and Rd (29.2% vs. haematologica | 2018; 103(12)

Subgroup analyses of D-Rd vs. Rd in POLLUX

B No. of progression or Median progression-free death events/total no. survival (months)

51.4%) treatment groups. However, the ORR and PFS in patients with a shorter treatment-free interval was more severely impacted in patients treated with Rd compared with D-Rd (Figure 1B, Table 2). MRD status was evaluated to determine the ability of daratumumab to generate deeper responses in patients achieving conventional CR. The IMWG established MRD-negative criteria requiring evaluation by next-generation sequencing to be conducted at a minimum sensitivity threshold of 10–5.23 At this threshold, of the ITT population, 26.2% of patients treated with D-Rd achieved MRD negativity compared with 6.4% of patients who received Rd (P<0.000001 [Table 2]). Consistent findings were also observed for D-Rd versus Rd at a sensitivity of 10–6 (12.9% vs. 2.8%; P=0.000003). D-Rd generated deeper responses and higher MRD-negativity rates than Rd in all haematologica | 2018; 103(12)

Figure 1. (A) PFS in the ITT population and (B) a forest plot summary of PFS HRs in subgroups by prior lines, prior therapies, treatment-free intervals. KaplanMeier analysis of PFS among patients in the ITT population. a Treatment-free interval was defined as the duration between the end date of the last line of prior therapy and randomization. b High-risk patients had any of t(4;14), t(14;16), or del17p as assessed by next generation c Standard-risk sequencing. patients had an absence of highrisk abnormalities. PFS: progression-free survival; ITT: intent-totreat; HR: hazard ratio; D-Rd: daratumumab/lenalidomide/dex amethasone; Rd: lenalidomide/ dexamethasone; CI: confidence interval; NR: not reached; TFI: treatment-free interval; std: standard.

subgroups evaluated (Table 2), including patients who received 1 prior line of therapy (D-Rd, n=149; Rd, n=146; 10–5: 25.5% vs. 8.2%; P=0.000053; 10–6: 10.1% vs. 4.8%; P=0.08134) and 2 to 3 prior lines of therapy (D-Rd, n=123; Rd, n=118; 10–5: 27.6% vs. 4.2%; P<0.000001; 10–6: 16.3% vs. 0.8%; P=0.000003), patients who were bortezomib refractory (D-Rd, n=59; Rd, n=58; 10–5: 20.3% vs. 6.9%; P=0.0308; 10–6: 10.2% vs. 3.4%; P=0.1410), and patients who had high-cytogenetic risk at a sensitivity threshold of 10–5 (D-Rd, n=28; Rd, n=37; 21.4% vs. 0.0%, P=0.0009) and 10–6 (D-Rd, n=28; Rd, n=37; 14.3% vs. 0.0%, P=0.0078 [Table 2]). PFS was prolonged in patients who achieved MRD negativity compared with MRD-positive patients in both treatment arms (Figure 3A-B). D-Rd also significantly prolonged PFS compared with Rd in patients with MRDpositive status at the 10–5 and 10–6 sensitivity thresholds (P<0.001 [Figure 3A-B]). At the time of the analysis, the 2091

M.A. Dimopoulos et al. Table 2. ORR and MRD disease based on prior treatment history.

ORR, n (%) a

# of patients in group

Subgroup

D-Rd

ITTb Prior lines of therapy 1 2-3 1-3 Prior therapy Bortezomib Lenalidomide Thalidomide Refractory to bortezomib Treatment-free interval ≤12 months >12 months ≤6 months >6 months Cytogenetic riske Highf Standard

281

276

261 (92.9)

211 (76.4)

147 120 267

142 115 257

137 (93.2) 114 (95.0) 251 (94.0)

237 50 119 57

232 47 123 56

143 138 98 183 27 132

MRD, n (%) 10–5 Rd

D-Rd

10–6 Rd

D-Rd

<0.0001

286

283

75 (26.2)

18 (6.4) <0.000001 37 (12.9) 8 (2.8) 0.000003

114 (80.3) 85 (73.9) 199 (77.4)

0.0003 <0.0001 <0.0001

149 123 272

146 118 264

38 (25.5) 34 (27.6) 72 (26.5)

12 (8.2) 0.000053 15 (10.1) 7 (4.8) 0.081340 5 (4.2) <0.000001 20 (16.3) 1 (0.8) 0.000003 17 (6.4) <0.0001 35 (12.9) 8 (3.0) <0.0001

218 (92.0) 42 (84.0) 109 (91.6) 50 (87.7)

175 (75.4) 32 (64.0) 87 (70.7) 38 (67.9)

<0.0001 0.0233 <0.0001 0.0113

241 50 122 59

238 50 125 58

63 (26.1) 13 (26.0) 26 (21.3) 12 (20.3)

18 (7.6) <0.000001 32 (13.3) 8 (3.4) 0.000051 2 (4.0) 0.0012 7 (14.0) 2 (4.0) 0.0729 6 (4.8) <0.0001 16 (13.1) 2 (1.6) 0.0002 4 (6.9) 0.0308 6 (10.2) 2 (3.4) 0.1410

131 145 92 184

129 (90.2) 132 (95.7) 87 (88.8) 174 (95.1)

87 (66.4) 124 (85.5) 57 (62.0) 154 (83.7)

<0.0001 0.0038 <0.0001 0.0004

146 140 99 187

134 149 95 188

34 (23.3) 41 (29.3) 21 (21.2) 54 (28.9)

7 (5.2) 11 (7.4) 6 (6.3) 12 (6.4)

<0.0001 <0.0001 0.0021 <0.0001

16 (11.0) 21 (15.0) 8 (8.1) 29 (15.5)

36 111

23 (85.2) 125 (94.7)

24 (66.7) 91 (82.0)

0.0435 0.0004

28 133

37 113

6 (21.4) 0 (0) 42 (31.6) 13 (11.5)

0.0009 0.0001

4 (14.3) 0 (0) 20 (15.0) 6 (5.3)

2 (1.5) 0.0006 6 (4.0) 0.0010 2 (2.1) 0.0514 6 (3.2) <0.0001 0.0078 0.0109

D-Rd: daratumumab/lenalidomide/dexamethasone; Rd: lenalidomide/dexamethasone; ITT: intent-to-treat. Data are based on computerized algorithm. aResponse-evaluable population. b ITT population. cP value was generated using the Cochran-Mantel-Haenszel χ2 test. dP value was generated using the likelihood-ratio χ2 test. eBiomarker risk-evaluable population. fIncludes subjects who have either del17p, t(14;16), t(4;14) or a combination of these.

majority of patients maintained MRD-negative status; patients will continue to be assessed annually. The safety profile remained unchanged from the primary analysis, with no new safety signals reported in either treatment group with longer follow up (Table 3). The most common treatment-emergent adverse events (≥15%) of any grade included neutropenia, anemia, thrombocytopenia, diarrhea, fatigue, upper respiratory tract infection, cough, constipation, muscle spasms, nasopharyngitis, and nausea (Table 3). The most common grade 3/4 treatment-emergent adverse events (≥5%) included neutropenia, febrile neutropenia, anemia, thrombocytopenia, lymphopenia, diarrhea, fatigue, and pneumonia (Table 3). The percentage of patients with adverse events leading to treatment discontinuation was similar between groups (12.0% for D-Rd, 12.8% for Rd). The most common adverse events (≥1%) leading to treatment discontinuation in D-Rd compared with Rd included pneumonia (1.4% vs. 0.7%), pulmonary embolism (0% vs. 1.1%), general physical health deterioration (1.1% vs. 0%), and renal failure (0.4% vs. 1.1%), respectively. The incidence of second primary malignancies between the D-Rd (n=283) and Rd (n=281) groups was the same (5.7% vs. 5.7%), consistent with previous observations that had no notable differences.15 The proportion of patients who received transfusions while on study drug was also similar between the D-Rd and Rd groups (24.4% vs. 25.3%). There was no decline in HRQoL measures with the addition of daratumumab to Rd. Statistically significant differences in the change from baseline were observed in favor of D-Rd at Weeks 48 and 56 with the Utility Score 2092

Figure 2. PFS by cytogenetic risk status. Cytogenetic risk was assessed via nextgeneration sequencing. High-risk patients had any of t(4;14), t(14;16), or del17p. Standard-risk patients had an absence of high-risk abnormalities. PFS: progression-free survival; D-Rd: daratumumab/lenalidomide/dexamethasone; Rd: lenalidomide/dexamethasone; HR: hazard ratio; CI: confidence interval.

and at Weeks 40, 48, and 56 with the Visual Analog Scale Score of the EQ-5D-5L questionnaire. With the EORTC QLQ-C30 Global Health Status Score, statistically significant differences in the change from baseline were observed in favor of D-Rd at Weeks 40, 48, 52, 68, 84, and 116. However, these improvements did not last beyond haematologica | 2018; 103(12)

Subgroup analyses of D-Rd vs. Rd in POLLUX

Figure 3. PFS by MRD at sensitivity threshold of (A) 10–5 and (B) 10–6. Kaplan-Meier estimates of PFS among patients in the intent-to-treat population. MRD-negative status was evaluated at sensitivity thresholds of 10–5 and 10–6 using bone marrow aspirate samples that were prepared using Ficoll and analyzed by the clonoSEQTM assay. PFS: progression-free survival; MRD: minimal residual disease; D-Rd: daratumumab/lenalidomide/dexamethasone; Rd: lenalidomide/dexamethasone.

3 consecutive assessments in either questionnaire. No significant differences for median time to improvement (6.6 months vs. 6.5 months; HR, 1.03; 95% CI, 0.81-1.30; P=0.820) were reported for EORTC QLQ-C30 Global Health Status Scores in the D-Rd and Rd groups. Similarly, no significant differences in median time to improvement were observed between treatment groups for either the EQ-5D-5L Utility Score (6.6 months vs. 10.2 months; HR, 1.23; 95% CI, 0.97-1.57; P=0.089) or Visual Analog Scale Score (6.9 months vs. 9.3 months; HR, 1.14; 95% CI, 0.891.45; P=0.283). haematologica | 2018; 103(12)

Discussion These updated analyses with nearly one further year of follow up reinforce the initial findings of deep and durable responses achieved with D-Rd in RRMM patients.15 The benefit of D-Rd over Rd was consistently maintained across patients who received 1 or 2 to 3 prior lines of therapy, and the risk of progression or death was reduced by >60%. D-Rd was also superior to Rd regardless of prior treatment exposure (i.e., lenalidomide, thalidomide, bortezomib-refractory), time since last therapy, and cyto2093

M.A. Dimopoulos et al. Table 3. Most common all grade (â&#x2030;Ľ15%) and grade 3/4 (â&#x2030;Ľ5%) treatment-emergent adverse events in the safety population

D-Rd (n=283)

Rd (n=281)

Event, n (%)

All grade

Grade 3/4

All grade

Grade 3/4

Total Hematologic Neutropenia Febrile neutropenia Anemia Thrombocytopenia Lymphopenia Nonhematologic Diarrhea Fatigue Upper respiratory tract infection Cough Constipation Muscle spasms Nasopharyngitis Nausea Insomnia Pyrexia Dyspnea Back pain Pneumonia Bronchitis Edema peripheral Vomiting Asthenia Headache

281 (99.3)

251 (88.7)

274 (97.5)

216 (76.9)

172 (60.8) 17 (6.0) 104 (36.7) 81 (28.6) 18 (6.4)

153 (54.1) 17 (6.0) 44 (15.5) 39 (13.8) 15 (5.3)

127 (45.2) 8 (2.8) 109 (38.8) 87 (31.0) 16 (5.7)

112 (39.9) 8 (2.8) 60 (21.4) 44 (15.7) 11 (3.9)

144 (50.9) 103 (36.4) 105 (37.1) 91 (32.2) 88 (31.1) 81 (28.6) 84 (29.7) 76 (26.9) 67 (23.7) 67 (23.7) 59 (20.8) 58 (20.5) 58 (20.5) 53 (18.7) 53 (18.7) 52 (18.4) 51 (18.0) 43 (15.2)

20 (7.1) 18 (6.4) 4 (1.4) 1 (0.4) 3 (1.1) 3 (1.1) 0 (0.0) 5 (1.8) 4 (1.4) 7 (2.5) 12 (4.2) 6 (2.1) 34 (12.0) 6 (2.1) 2 (0.7) 3 (1.1) 10 (3.5) 0 (0.0)

89 (31.7) 85 (30.2) 74 (26.3) 40 (14.2) 74 (26.3) 59 (21.0) 54 (19.2) 50 (17.8) 61 (21.7) 36 (12.8) 35 (12.5) 53 (18.9) 42 (14.9) 46 (16.4) 43 (15.3) 19 (6.8) 43 (15.3) 22 (7.8)

9 (3.2) 10 (3.6) 4 (1.4) 0 (0.0) 2 (0.7) 4 (1.4) 0 (0.0) 2 (0.7) 4 (1.4) 5 (1.8) 2 (0.7) 5 (1.8) 24 (8.5) 7 (2.5) 4 (1.4) 4 (1.4) 8 (2.8) 0 (0.0)

D-Rd: daratumumab/lenalidomide/dexamethasone; Rd: lenalidomide/dexamethasone.

genetic risk status. As one of the first studies to prospectively assess MRD in a phase 3 trial of RRMM, MRD-negativity rates were also significantly higher with D-Rd across all patient subgroups, including patients with high cytogenetic risk. These findings reinforce observations that highlight the depth, durability, and robustness of responses achieved with daratumumab-based regimens.24 These findings compare favorably with other studies of IMiD-containing regimens with subgroup analyses based on prior lines of therapy and/or prior treatment exposure (Appendix Table 2). In a large phase 3 study (ASPIRE) of carfilzomib-Rd (KRd) compared with Rd alone, a consistent PFS benefit was observed with KRd versus Rd among patients who received 1 and 2 to 3 prior lines of therapy.25 Prespecified PFS subgroup analyses showed a modest benefit with KRd over Rd for patients who previously received lenalidomide and patients nonresponsive to bortezomib in any previous regimen.26 The magnitude of benefit for KRd versus Rd was similar in patients with early disease relapse (â&#x2030;¤12 months from starting the first prior regimen; ORR: 79% vs. 61%).27 In a phase 3 study of elotuzumab (ELOQUENT-2), the PFS benefit in combination with Rd was maintained among patients who received 2 to 3 prior lines of therapy or had prior exposure 2094

to bortezomib but not among patients who received 1 prior line of therapy or prior lenalidomide.28,29 Ixazomib in combination with lenalidomide and dexamethasone in the phase 3 TOURMALINE-MM1 study demonstrated a modest PFS benefit versus Rd in patients who received 1 or 2 prior lines, were previously treated with an IMiD or a PI, and were refractory to their last line of therapy; interestingly, patients who received 3 prior lines of therapy demonstrated an HR of 0.37.30 D-Rd improved responses, including MRD negativity in patients with high-risk cytogenetic status, suggesting that targeting CD38 in combination with Rd may improve outcomes for this challenging-to-treat population. Moreover, despite the small number of patients with high-risk disease, D-Rd demonstrated significantly higher MRD-negativity rates compared with standard-risk patients receiving Rd. While PFS was not statistically different between D-Rd versus Rd for patients with high-risk disease, a numerical improvement in PFS was observed in high-risk patients who received D-Rd versus Rd. As the magnitude of benefit for D-Rd was lower among high-risk patients, these findings suggest that while D-Rd is able to provide improvements in efficacy, it is not able to overcome the greater risk of progression and poor outcomes associated haematologica | 2018; 103(12)

Subgroup analyses of D-Rd vs. Rd in POLLUX

with high-risk disease. Although cross-study comparisons should be performed with caution, smaller differences in PFS and ORR between standard- and high-risk patients in the CASTOR study in patients treated with daratumumab and Vd were observed.31 Daratumumab-based combinations that include both a PI and an IMiD may further improve efficacy in patients with high-risk disease. Prolonged treatment with D-Rd or Rd did not uncover new safety concerns, and the incidences of second primary malignancies were balanced between study arms. While certain adverse events, such as cough, dyspnea, and pneumonia occurred more frequently with daratumumab, the percentage of patients that discontinued treatment due to adverse events was similar between groups (12.0% for D-Rd, 12.8% for Rd). Additionally, patients in the D-Rd group were treated for longer, with a median duration of 24.5 months versus 16.0 months in the Rd group, which likely contributed to the increase in rates for certain adverse events. Finally, the addition of a third drug to the standard of care Rd regimen did not have a negative effect on HRQoL. Several limitations of the study should be noted. First, although balanced between treatment arms, only a subset of patient samples was collected for central cytogenetic testing as outlined in the Methods. Second, not all eligible patients with suspected CR or CR had available MRD data, and these patients were therefore conservatively classified as MRD positive. Based on the experiences obtained from this phase 3 study, continuous improvements in MRD testing are being implemented across ongoing and future daratumumab studies in multiple myeloma. Finally, based on the limitations of post hoc

References 1. Kumar SK, Rajkumar SV, Dispenzieri A, et al. Improved survival in multiple myeloma and the impact of novel therapies. Blood. 2008;111(5):2516-2520. 2. Kumar SK, Dispenzieri A, Lacy MQ, et al. Continued improvement in survival in multiple myeloma: changes in early mortality and outcomes in older patients. Leukemia. 2014;28(5):1122-1128. 3. Moreau P, San Miguel J, Sonneveld P, et al. Multiple myeloma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2017;28(suppl 4):iv52iv61. 4. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines): multiple myeloma. Version 4. 2018. https://www.nccn.org/professionals/physi cian_gls/pdf/myeloma.pdf 5. Lin P, Owens R, Tricot G, Wilson CS. Flow cytometric immunophenotypic analysis of 306 cases of multiple myeloma. Am J Clin Pathol. 2004;121(4):482-488. 6. Santonocito AM, Consoli U, Bagnato S, et al. Flow cytometric detection of aneuploid CD38(++) plasmacells and CD19(+) B-lymphocytes in bone marrow, peripheral blood and PBSC harvest in multiple myeloma patients. Leuk Res. 2004;28(5):469-477. 7. de Weers M, Tai YT, van der Veer MS, et al. Daratumumab, a novel therapeutic human

haematologica | 2018; 103(12)

10.

11.

12.

13. 14.

analyses, the findings presented here require further confirmation in future studies prospectively assessing the efficacy of D-Rd in these clinically relevant patient populations. In summary, these updated secondary subgroup analyses highlight the benefit of combining daratumumab with the standard of care regimen Rd in RRMM across examined subgroups. The addition of daratumumab to Rd drove deep responses, as shown by achievement of MRD negativity in many patients. These results suggest that D-Rd is a highly effective and well-tolerated regimen that may be recommended for RRMM after first relapse and beyond. Funding This study was sponsored by Janssen Research & Development, LLC. The data sharing policy of Janssen Pharmaceutical Companies of Johnson & Johnson is available at https://www.janssen.com/clinical-trials/transparency. As noted on this site, requests for access to the study data can be submitted through Yale Open Data Access (YODA) Project site at http://yoda.yale.edu. JLK consulted for BMS, Janssen, Celgene, Karyopharm, and Pharmacyclics. Acknowledgments The authors wish to thank the patients participating in this study and their families, as well as the global network of investigators, research nurses, study coordinators, and operations staff (including Sonali Trivedi, Jaime Bald, Xiang Qin, and Christopher Velas). Medical writing and editorial support were provided by Jason Jung, PhD, and Sima Patel, PhD, of MedErgy, and were funded by Janssen Global Services, LLC.

CD38 monoclonal antibody, induces killing of multiple myeloma and other hematological tumors. J Immunol. 2011;1 86(3):18401848. Lammerts van Bueren J, Jakobs D, Kaldenhoven N, et al. Direct in vitro comparison of daratumumab with surrogate analogs of CD38 antibodies MOR03087, SAR650984 and Ab79. Blood. 2014; 124(21):3474. Overdijk MB, Verploegen S, Bogels M, et al. Antibody-mediated phagocytosis contributes to the anti-tumor activity of the therapeutic antibody daratumumab in lymphoma and multiple myeloma. MAbs. 2015;7(2):311-321. Overdijk MB, Jansen JH, Nederend M, et al. The therapeutic CD38 monoclonal antibody daratumumab induces programmed cell death via Fcgamma receptor-mediated cross-linking. J Immunol. 2016; 197(3):807813. Krejcik J, Casneuf T, Nijhof IS, et al. Daratumumab depletes CD38+ immuneregulatory cells, promotes T-cell expansion, and skews T-cell repertoire in multiple myeloma. Blood. 2016;128(3):384-394. McKeage K, Lyseng-Williamson KA. Daratumumab in multiple myeloma: a guide to its use as monotherapy in the EU. Drugs Ther Perspect. 2016;32(11):463-469. McKeage K. Daratumumab: first global approval. Drugs. 2016;76(2):275-281. Palumbo A, Chanan-Khan A, Weisel K, et

15.

16. 17. 18.

19.

20.

21.

22.

al. Daratumumab, bortezomib, and dexamethasone for multiple myeloma. N Engl J Med. 2016;375(8):754-766. Dimopoulos MA, Oriol A, Nahi H, et al. Daratumumab, lenalidomide, and dexamethasone for multiple myeloma. N Engl J Med. 2016;375(14):1319-1331. Blair HA. Daratumumab: A review in relapsed and/or refractory multiple myeloma. Drugs. 2017;77(18):2013â&#x20AC;&#x201C;2024. DARZALEXÂŽ (daratumumab) injection, for intravenous use [package insert]. Horsham, PA: Janssen Biotech, Inc.; 2018. Durie BGM, Harousseau JL, Miguel JS, et al. International uniform response criteria for multiple myeloma. Leukemia. 2006; 20(9):1467-1473. Rajkumar SV, Harousseau JL, Durie B, et al. Consensus recommendations for the uniform reporting of clinical trials: report of the International Myeloma Workshop Consensus Panel 1. Blood. 2011;117(18): 4691-4695. McCudden C, Axel AE, Slaets D, et al. Monitoring multiple myeloma patients treated with daratumumab: teasing out monoclonal antibody interference. Clin Chem Lab Med. 2016;54(6):1095-1104. Durie BG, Miguel JF, Blade J, Rajkumar SV. Clarification of the definition of complete response in multiple myeloma. Leukemia. 2015;29(12):2416-2417. Sonneveld P, Broijl A. Treatment of relapsed and refractory multiple myeloma.

2095

M.A. Dimopoulos et al. Haematologica. 2016;101(4):396-406. 23. Kumar S, Paiva B, Anderson KC, et al. International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma. Lancet Oncol. 2016; 17(8):e328e346. 24. Avet-Loiseau H, Casneuf T, Chiu C, et al. Evaluation of minimal residual disease (MRD) in relapsed/refractory multiple myeloma (RRMM) patients treated with daratumumab in combination with lenalidomide plus dexamethasone or bortezomib plus dexamethasone. Blood. 2016; 128(22):246. 25. Dimopoulos A, Stewart AK, Rajkumar SV, et al. Effect of carfilzomib, lenalidomide, and dexamethasone (KRd) vs lenalidomide and dexamethasone (Rd) in patients with relapsed multiple myeloma (RMM) by line of therapy: secondary analysis from an

2096

interim analysis of the phase III study ASPIRE (NCT01080391). J Clin Oncol. 2015;33(15_suppl):8525. 26. Stewart AK, Rajkumar SV, Dimopoulos MA, et al. Carfilzomib, lenalidomide, and dexamethasone for relapsed multiple myeloma. N Engl J Med. 2015;372(2):142152. 27. Ludwig H, Dimopoulos MA, Masszi T, et al. Carfilzomib, lenalidomide, and dexamethasone (KRd) vs lenalidomide and dexamethasone (Rd) in patients with relapsed multiple myeloma (RMM) and early progression during prior therapy: Secondary analysis from the phase 3 study ASPIRE (NCT01080391). J Clin Oncol. 2016;34(15_suppl):8045. 28. Lonial S, Richardson PG, Mateos MV, et al. ELOQUENT-2 update: Phase III study of elotuzumab plus lenalidomide/dexamethasone (ELd) vs Ld in relapsed/refractory mul-

tiple myeloma (RRMM)-identifying responders by subset analysis. J Clin Oncol. 2016;34(15_suppl):8037. 29. Lonial S, Dimopoulos M, Palumbo A, et al. Elotuzumab therapy for relapsed or refractory multiple myeloma. N Engl J Med. 2015;373(7):621-631. 30. Moreau P, Masszi T, Grzasko N, et al. Oral ixazomib, lenalidomide, and dexamethasone for multiple myeloma. N Engl J Med. 2016;374(17):1621-1634. 31. Mateos MV, Estell J, Barreto W, et al. Efficacy of daratumumab, bortezomib, and dexamethasone versus bortezomib and dexamethasone in relapsed or refractory myeloma based on prior lines of therapy: updated analysis of CASTOR. Presented at the 58th Annual Meeting & Exposition of the American Society of Hematology (ASH); December 3-6, 2016; San Diego, CA. Abstract 1150.

haematologica | 2018; 103(12)

ARTICLE

Platelet Biology & its Disorders

Inhibition of Btk by Btk-specific concentrations of ibrutinib and acalabrutinib delays but does not block platelet aggregation mediated by glycoprotein VI

Phillip L.R. Nicolson,1 Craig E. Hughes,2 Stephanie Watson,1 Sophie H. Nock,2 Alexander T. Hardy,1 Callum N. Watson,1 Samantha J. Montague,3 Hayley Clifford,4 Aarnoud P. Huissoon,4 Jean-Daniel Malcor,5 Mark R. Thomas,1 Alice Y. Pollitt,2 Michael G. Tomlinson,6 Guy Pratt7 and Steve P. Watson1,8

Institute of Cardiovascular Sciences, College of Medical and Dental Sciences, University of Birmingham, UK; 2Institute for Cardiovascular and Metabolic Research, Harborne Building, University of Reading, UK; 3ACRF Department of Cancer Biology and Therapeutics, John Curtin School of Medical Research, Australian National University, Canberra, ACT, 2601, Australia; 4Department of Immunology, Heartlands Hospital, Birmingham, UK; 5Department of Biochemistry, University of Cambridge, UK; 6 Department of Biosciences, College of Life and Environmental Sciences, University of Birmingham, UK; 7Department of Haematology, Queen Elizabeth Hospital, Birmingham, UK and 8Centre of Membrane Proteins and Receptors (COMPARE), Universities of Birmingham and Nottingham, Midlands, UK 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2097-2108

ABSTRACT

brutinib and acalabrutinib are irreversible inhibitors of Bruton tyrosine kinase used in the treatment of B-cell malignancies. They bind irreversibly to cysteine 481 of Bruton tyrosine kinase, blocking autophosphorylation on tyrosine 223 and phosphorylation of downstream substrates including phospholipase C-g2. In the present study, we demonstrate that concentrations of ibrutinib and acalabrutinib that block Bruton tyrosine kinase activity, as shown by loss of phosphorylation at tyrosine 223 and phospholipase C-g2, delay but do not block aggregation in response to a maximally-effective concentration of collagen-related peptide or collagen. In contrast, 10- to 20-fold higher concentrations of ibrutinib or acalabrutinib block platelet aggregation in response to glycoprotein VI agonists. Ex vivo studies on patients treated with ibrutinib, but not acalabrutinib, showed a reduction of platelet aggregation in response to collagen-related peptide indicating that the clinical dose of ibrutinib but not acalabrutinib is supramaximal for Bruton tyrosine kinase blockade. Unexpectedly, low concentrations of ibrutinib inhibited aggregation in response to collagen-related peptide in patients deficient in Bruton tyrosine kinase. The increased bleeding seen with ibrutinib over acalabrutinib is due to off-target actions of ibrutinib that occur because of unfavorable pharmacodynamics.

Correspondence: p.nicolson@bham.ac.uk or s.p.watson@bham.ac.uk Received: March 16 2018. Accepted: July 18, 2018. Pre-published: July 19 2018. doi:10.3324/haematol.2018.193391 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2097 ©2018 Ferrata Storti Foundation

Introduction The major physiological ligands that activate platelets in hemostasis and thrombosis signal through G protein-coupled and tyrosine kinase-linked receptors. The former includes receptors for thrombin (PAR1, PAR4), thromboxane A2 (TP) and ADP (P2Y1, P2Y12), and the latter receptors for collagen/fibrin (glycoprotein VI: GPVI), podoplanin (CLEC-2), von Willebrand factor (GPIb-IX-V) and fibrinogen (integrin αIIbβ3).1,2 GPVI is a receptor for collagen and fibrin which forms a complex with the Fc receptor g-chain (FcRg).2-4 GPVI triggers powerful platelet activation through Src, Syk and Tec family tyrosine kinases leading to activation of phospholipase C-g2 (PLCg2).5 GPVI is expressed exclusively on platelets and the platelet precursor cell, the megakaryocyte.6 Mice deficient in GPVI have a minor increase in tail bleeding times but fail to form occlusive thrombi in a FeCl3 injury arterial thrombosis assay.7 haematologica | 2018; 103(12)

2097

P.L.R. Nicolson et al.

Patients homozygous for an insertion that introduces a stop codon and prevents expression of the immunoglobulin receptor on the platelet surface have a relatively mild bleeding diathesis,8 although there are too few of such individuals to determine whether they are protected from thrombosis. Bruton tyrosine kinase (Btk) is a member of the Tec family of tyrosine kinases and mediates phosphorylation and activation of PLCg2 downstream of GPVI and the B-cell antigen receptor. The irreversible Btk inhibitor ibrutinib has been introduced into the clinic for treatment of B-cell malignancies but has been reported to increase rates of major hemorrhage in a subgroup of patients.9,10 The increase in bleeding has been attributed to a loss of platelet activation by GPVI11-13 and GPIb,11 with the inhibition of the two receptors having been shown to correlate.14 In contrast to ibrutinib-treated subjects, patients with Xlinked agammaglobulinemia (XLA) do not bleed excessively.15 XLA is caused by mutations in the BTK gene which result in a loss or reduction of Btk expression, or expression of a non-functional protein. A potential explanation for this difference in bleeding propensity is that ibrutinib blocks activation of platelets by both Btk and the closely related kinase Tec. Tec is expressed in human and mouse platelets, and has been shown to support PLCg2 activation in mouse platelets.16 Interestingly, major hemorrhage is not seen in patients treated with the structurally related Btk inhibitor, acalabrutinib, despite this also inhibiting Btk by covalent modification of C481.9,17 It has been postulated that this is due to its greater selectivity for Btk over Tec in comparison to ibrutinib.17,18 In the present study we compared the inhibitory effects of ibrutinib and acalabrutinib on platelet activation and protein phosphorylation by GPVI alongside ex vivo studies on patients prescribed the two inhibitors, as well as on XLA patients.

Methods Reagents

the protease inhibitors sodium orthovanadate (5 mM), leupeptin (10 mg/mL), AEBSF (200 mg/mL), aprotinin (10 mg/mL) and pepstatin (1 mg/mL). Platelet lysates were precleared, and detergentinsoluble debris was discarded. An aliquot was dissolved with SDS sample buffer for detection of total tyrosine phosphorylation. Lysates were incubated with either the indicated antibodies and protein A- or protein G-Sepharose. Lysates were separated by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDSPAGE), electro-transferred, and western blotted. Western blots were imaged using ECL autoradiography film. In order to analyze levels of phosphorylation, western blot films were scanned and band intensity measured using ImageJ 1.5 with values normalized to basal levels. Results were averaged and IC50 values were calculated from these data.

Other Details on the methods for blood sampling, platelet preparation, granule release, [Ca2+]i mobilization, measurement of platelet adhesion under flow, cell lines, plasmids, transfections and the luciferase assay can be found in the Online Supplementary Information.

Statistical analysis All data are presented as mean ± standard error of the mean (SEM) with statistical significance taken as P<0.05 unless otherwise stated. Statistical analyses, unless otherwise specified, were performed using one-way analysis of variance (ANOVA) with a Bonferroni post-test. Ex vivo platelet aggregation was determined by optical densities, which were compared using a one-way ANOVA with the Tukey multiple comparison test. Correlations of aggregation with tyrosine phosphorylation were assessed using the Pearson correlation coefficient. IC50 values were analyzed using the Welch t-test. All statistical analyses were performed using GraphPad Prism 7.

Ethical approval Ethical approval for collecting blood from patients and healthy volunteers was granted by the National Research Ethics Service (10/H1206/58) and Birmingham University Internal Ethical Review (ERN_11-0175), respectively. Work on HLA patients has ethical approval via the University of Birmingham HBRC 16-251 Amendment 1.

Details on the source of reagents and chemical analyses can be found in the Online Supplementary Information.

Results Light transmission aggregometry Aggregation was measured in siliconized glass vials at 37°C in a Model 700 aggregometer (ChronoLog, Havertown, PA, USA) with stirring at 1200 rpm. Platelets were warmed to 37°C for 5 min before the experiments. Platelets were pre-incubated with ibrutinib, acalabrutinib or dimethyl sulfoxide (DMSO) vehicle for 5 min prior to agonist addition unless otherwise stated. Results were averaged and the half maximal inhibitory concentration (IC50) values were calculated from these data.

Protein phosphorylation

Washed platelets were pre-treated with 9 mM eptifibatide to block integrin αIIbβ3 activation. Agonists were added while stirring at 1200 rpm in an aggregometer at 37°C for 180 s unless otherwise stated. The platelets were stimulated in the presence of ibrutinib (17 nM - 7 mM), acalabrutinib (50 nM – 200 mM) or vehicle (DMSO). For whole cell lysate experiments, activation was terminated with 5X SDS reducing sample buffer. For immunoprecipitation, 8x108/mL platelets were used and reactions were terminated by addition of 2X ice-cold Nonidet P-40 lysis buffer containing 2098

Inhibition of GPVI-induced platelet aggregation by high concentrations of ibrutinib is reversible Ibrutinib is 97% bound to plasma proteins and unbound levels reach approximately 0.5 mM in patients.11 At this concentration, ibrutinib has been shown to block GPVIinduced platelet aggregation.12,19 If this is due to inhibition of Btk and other Tec kinases then the inhibition should be irreversible and time-dependent (i.e. inhibition should increase with time). To test this, platelets were treated with a concentration of ibrutinib that causes complete inhibition of GPVI-mediated aggregation in washed platelets before washout of ibrutinib and stimulation with the GPVI-specific agonist collagen-related peptide (CRP). Platelets showed almost full recovery on washout demonstrating that the inhibitory effect is not due solely to covalent modification of Btk or Tec (Figure 1A,B). In support of this, incubation of washed platelets with a high concentration of ibrutinib (700 nM) for ≥30 s was sufficient to block aggregation in response to a high dose of CRP (Figure 1C). haematologica | 2018; 103(12)

Effects of Btk inhibitors on platelet activation

However, at a lower concentration, ibrutinib (70 nM) caused a time-dependent delay in aggregation in response to a high concentration of CRP, which was apparent at incubation times of ≥5 min (Figures 1D and 2Ai). This concentration of ibrutinib also caused a reduced response to a sub-maximal concentration of CRP (Online Supplementary Figure S2A). The time-dependent delay is consistent with an irreversible action and contrasts with the rapid onset of inhibition observed at the high concentration of ibrutinib. A similar set of observations was seen in washed platelets stimulated by collagen (Online Supplementary Figure S1A). Similar results were also seen in the presence of 0.3% bovine serum albumin, which was used in case the results were influenced by adsorption of ibrutinib to the surface of the aggregometer tube (Online Supplementary Figure S1B).

Platelet secretion and Ca2+ mobilization play key roles in platelet activation. Consistent with the results for aggregation, low (70 nM) and high (700 nM) concentrations of ibrutinib had, respectively, no effect or blocked ATP secretion in response to a high concentration of CRP (Figure 2Bi, 2Biii). Similarly, the peak Ca2+ concentration following administration of a high concentration of CRP was not altered in the presence of a low concentration of ibrutinib (70 nM) but was markedly reduced by a high concentration (700 nM). The dose-response curve for inhibition of aggregation was similar to that for loss of Ca2+ mobilization (Figure 2Bii, 2Biii) and was not affected by the presence of the cyclooxygenase inhibitor indomethacin (Online Supplementary Figure S1D). There was no statistical difference between the IC50 of ibrutinib for secretion,

Figure 1. Increasing ibrutinib incubation time has no effect on degree of inhibition of platelet aggregation and this inhibition is reversed by washing. (A) Representative traces of washed platelets at 4x108/mL stimulated with CRP (10 mg/mL for 180 s). Prior to addition of the agonist, platelets were pre-incubated with either ibrutinib or vehicle (DMSO) for 5 min. Alongside, washed platelets at 4x108/mL identically treated with either ibrutinib or vehicle were washed twice in Tyrode buffer and platelets resuspended to 4x108/mL; platelets were then stimulated with CRP (10 mg/mL for 180 s). The data shown are representative of three identical experiments. (B) Mean data for (A) (n=3) analyzed with one-way ANOVA. Results are shown as mean ± SEM. *P<0.05. (C) Washed platelets (4x108/mL) were incubated with ibrutinib or vehicle (DMSO) for 30 s - 60 min before being stimulated with CRP (10 mg/mL). The optical density (OD) of platelet suspensions was measured in a ChronoLog Model 700 aggregometer with stirring at 1200 rpm. Traces representative of three similar experiments are shown. (D) Delay in aggregation seen with ibrutinib-treated washed platelets (n=3) analyzed with one-way ANOVA. Results are shown as mean ± SEM. *P<0.05.

haematologica | 2018; 103(12)

2099

P.L.R. Nicolson et al.

aggregation or Ca2+ mobilization. Taken together these results show that ibrutinib has two distinct effects on platelet activation by CRP. At a low concentration (70 nM), ibrutinib delays but does not inhibit activation, whereas at a 10-fold higher concentration of ibrutinib (700 nM) activation is blocked. The latter action is reversible indicating that it is not mediated by covalent modification of Btk or other Tec kinases.

Low-dose ibrutinib blocks Btk but not Tec The concentration-response curve to ibrutinib on tyrosine phosphorylation was investigated in washed platelets in the same conditions as for the platelet function studies above. CRP induced robust tyrosine phosphorylation in whole cell lysates which was dose-dependently inhibited by ibrutinib. Correspondingly with the results for aggregation, this inhibitory effect of ibrutinib on global tyrosine phosphorylation was also reversible on washout (Figure

Aii

Bii

3Ai). Using phosphospecific antibodies, we were able to see that the inhibition of Src pY418 (which lies upstream of Btk) was also reversible but that autophosphorylation of Btk at pY223 and Btk substrates PLCg2 pY753 and pY121711,20 was irreversible (Figure 3Ai-ii). A detailed analysis of the dose response to ibrutinib on a wider range of proteins in the GPVI signaling cascade was investigated using further phosphospecific antibodies (Figure 3Bi). Autophosphorylation of Btk and downstream PLCg2 was reduced to basal levels by a low dose of ibrutinib (70 nM) (Figure 3Bii). In contrast, phosphorylation of Btk on Y551, which is mediated by Src family kinases,21 and proteins that lie upstream of Btk, namely Src Y418, Syk Y525/6, SLP-76 Y145 and LAT Y200, was not altered (Figure 3Biii-iv). Inhibition of phosphorylation of Src on its activation site, Y418, was observed at a 10-fold higher concentration of ibrutinib (Figure 3Biv) and was shown to correlate with inhibition of aggregation (Pearson correla-

Biii

Figure 2. Ibrutinib dose-dependently inhibits glycoprotein VI-mediated platelet aggregation, ATP secretion and Ca2+ mobilization. (A) (i) Representative traces showing the effect of increasing doses of in vitro ibrutinib incubated for 5 min with washed platelets at 4x108/mL. (ii) Ibrutinib dose-response curves in washed platelets (n=7). (B) Representative traces showing the effect of increasing doses of in vitro ibrutinib incubated with washed platelets at 4x108/mL for 5 min on (i) ATP secretion and (ii) Ca2+ mobilization in response to stimulation with CRP (10 mg/mL) for 180 s. (iii) Ibrutinib dose-response curves in washed platelets on ATP secretion (n=3) and Ca2+ mobilization (n=3). The dose-response curve for inhibition of washed platelet aggregation from (Aiii) is shown as a dotted line to enable comparison. Results are shown as mean Âą SEM. All experiments were stimulated with CRP (10 mg/mL). For comparison of IC50: ns = nonsignificant.

2100

haematologica | 2018; 103(12)

Effects of Btk inhibitors on platelet activation

Aii

Bii

Biii

Biv

Cii

Figure 3. Ibrutinib dose-dependently inhibits glycoprotein VI-mediated signaling. (A) Eptifibatide (9 mM)-treated washed human platelets (4×108/mL) were stimulated with CRP (10 mg/mL for 180 s) followed by lysis with 5X SDS reducing sample buffer. Prior to addition of agonist, platelets were pre-incubated with either ibrutinib or vehicle (DMSO). Some platelets underwent two further washing steps prior to addition of agonist. (i) Whole cell lysates were then separated by SDS-PAGE and western blotted with the stated antibodies for whole cell phosphorylation, kinases and proteins downstream of GPVI. Blots are representative of three experiments. (ii) Percentage tyrosine phosphorylation as compared to that of non-washed vehicle platelets was measured and is represented as the mean ± SEM of three identical experiments. (Bi) Washed platelets were treated as in (A) but with a wider range of ibrutinib doses. (ii - iv) The percentage of tyrosine phosphorylation as compared to that of vehicle-treated platelets was measured and is represented as the mean ± SEM of four identical experiments. The dose-response curve for inhibition of washed platelet aggregation from Figure 2Aii is shown as a dotted line to enable comparison. (C) Eptifibatide (9 mM)-treated washed human platelets (8×108/mL) were stimulated with CRP (10 mg/mL for 180 s) followed by lysis with 2X ice cold lysis sample buffer. Lysates were pre-cleared and Tec was immunoprecipitated before addition of SDS reducing sample buffer and separation by SDS-PAGE and western blotted with the anti-pY antibody 4G10. Membranes were stripped and then reprobed with the pan-Tec antibody. (i) The trace is representative of three identical experiments. (ii) The percentage of tyrosine phosphorylation as compared to that of vehicle-treated platelets was measured and is represented as the mean ± SEM of three identical experiments. *P<0.05. ns = non-significant.

haematologica | 2018; 103(12)

2101

P.L.R. Nicolson et al.

tion coefficient 0.959). Inhibition of whole cell phosphorylation and phosphorylation of Syk Y525/6, SLP-76 Y145, Btk Y551 and LAT Y200 was seen at a 7 mM dose of ibrutinib which is 10-fold higher than the maximal concentration in patients (Figure 3Bii, 3Biii). The IC50 values for each phosphorylation event, when they could be calculated, are included in Online Supplementary Table S1. Blockade of Btk pY223 and PLCg2 phosphorylation by 70 nM ibrutinib was also observed with lower concentrations of CRP or in the absence of the integrin αIIbβ3 blocker, eptifibatide (Online Supplementary Figures S2Aiv and S2B). There was no significant increase in phosphorylation of PLCg2 up to 180 s in response to CRP in the presence of 70 nM ibrutinib (Online Supplementary Figure S2Ci-iii). Due to the absence of phosphospecific antibodies for the Btk-related Tec family kinase, Tec, the effect of ibrutinib on Tec phosphorylation was investigated following immunoprecipitation and re-probing with the antiphosphotyrosine monoclonal antibody 4G10. The effect of ibrutinib was biphasic with partial blockade at 170 nM and full blockade observed at 7 mM (Figure 3C). These results demonstrate that a concentration of 70 nM ibrutinib is sufficient to block Btk at its autophosphorylation site and on PLCg2 on Y753, Y759 and Y1217 and

that this effect is irreversible. At this concentration of ibrutinib, aggregation in response to a high concentration of CRP is delayed but is not blocked. At a 10- to 20-fold higher concentration, ibrutinib reversibly blocks aggregation in parallel with reversible loss of phosphorylation of Src on Y418. Reversible inhibition of tyrosine phosphorylation of other proteins is seen at a 100-fold higher concentration than that required for blockade of Btk. Ibrutinib causes biphasic inhibition of phosphorylation of Tec, with inhibition occurring at 3- to 5-fold higher concentrations than those required to block phosphorylation of Btk on Y223, and full blockade at a 100-fold higher concentration. These results are consistent with loss of Tec autophosphorylation at 170 nM of ibrutinib and loss of phosphorylation on the activation site by higher concentrations.

Low-dose ibrutinib has no effect on platelet adhesion and aggregation in response to collagen under flow conditions The relevance of the observation that aggregation is delayed but not blocked in response to high concentrations of CRP and collagen was addressed using flow studies in which GPVI functions in conjunction with other tyrosine kinase-linked receptors that also signal via Btk,

Aii

Bii

Figure 4. A low dose of ibrutinib has no effect on platelet adhesion to collagen under flow. (A) Washed platelets (10x108/mL) were incubated with ibrutinib or vehicle (DMSO) for 5 min and stimulated with CRP (10 mg/mL). (i) Representative trace and (ii) mean data from three identical experiments show a characteristic delay associated with inhibition of Btk autophosphorylation. (B) Platelets were reconstituted with autologous red blood cells and platelet-poor plasma and flowed at arterial shear over collagen-coated microcapillaries for 3 min before being fixed and imaged. (i) Representative differential interference contrast images are shown. (ii) Platelet coverage as a percentage of values for vehicle-treated platelets was calculated and is shown as mean ± SEM from three identical experiments. *P<0.05, ns: non-significant.

2102

haematologica | 2018; 103(12)

Effects of Btk inhibitors on platelet activation namely GPIb and integrin αIIbβ3. To ensure that a known degree of Btk blockade was achieved, washed platelets were incubated with ibrutinib at a concentration sufficient to fully and irreversibly inhibit Btk kinase activity (70 nM). Inhibition of Btk autophosphorylation was confirmed by

a delay in aggregation in response to CRP (Figure 4A) and by measurement of phosphorylation (data not shown). Following incubation, platelets were reconstituted with autologous red blood cells and platelet-poor plasma and flowed over collagen at arterial shear rates. Adhesion of

Aii Bii

Aiii

Biii

Aiv

Biv

Figure 5. Patients with X-linked agammaglobulinemia, who lack Btk expression are more sensitive than healthy donors to ibrutinib inhibition of glycoprotein VI-mediated platelet aggregation. Ibrutinib, but not acalabrutinib, blocks glycoprotein VI-mediated platelet aggregation ex vivo. (A) Citrated blood was taken from XLA patients. (i) Whole cell lysates were then separated by SDS-PAGE and western blot with the polyclonal N-terminal Btk antibody. Platelet-rich plasma (PRP) from XLA patients was stimulated with CRP (10 mg/mL) for 180 s. (ii) A representative aggregation trace of XLA patients or healthy donor (HD). (iii) Ibrutinib dose-response curves in washed platelets of XLA patients (n=4). Healthy donor responses from Figure 2Aiii are shown as a dotted line for comparison. (iv) Whole cell lysates were then separated by SDS-PAGE and western blotted with the phosphospecific antibody to PLCg2 pY1217 (n=3). The aggregation curve for XLA patients is shown as a dotted line for comparison. (B) Patients taking ibrutinib 420 mg once daily, acalabrutinib 100 mg twice daily or a control chemotherapy regime of fludarabine (25 mg/m2 IV days 1-3), cyclophosphamide (250 mg/m2 IV days 1-3) and rituximab (375 mg/m2 IV day 1) (FCR) had citrated blood taken on day 28 of the treatment cycle (2-3 h after the dose of Btk inhibitor). PRP from this blood was then stimulated with CRP (10 mg/mL) for 180 s. (i) A representative trace. (ii) Mean and SEM from five, nine and three patients for FCR, ibrutinib and acalabrutinib respectively. (iii) Comparison of platelet counts in PRP from all groups of patients. Statistical analysis was performed using a one-way ANOVA with the Tukey multiple comparisons test, *P<0.05, ns=not significant. (iv) A representative western blot from eptifibatide (9 mM)-treated washed platelets (4×108/mL) from the same patients stimulated with CRP (10 mg/mL for 180 s) followed by lysis with 5X SDS sample buffer and probed with Btk pY223 and PLCg2 pY1217 phosphospecific antibody.

haematologica | 2018; 103(12)

2103

P.L.R. Nicolson et al.

ibrutinib-treated platelets was unchanged when compared to that of vehicle-treated platelets (Figure 4B).

Btk-specific concentrations of ibrutinib block GPVI mediated aggregation in patients with X-linked agammaglobulinemia The B-cell immunodeficiency XLA is caused by mutations in the BTK gene. Using knowledge of patients’ mutations (Online Supplementary Table S2), and an antibody to the N-terminus of Btk, we selected unrelated patients lacking Btk protein to test for off-target effects of ibrutinib (Figure 5Ai). Strikingly, the concentration-response curve for inhibition of CRP-induced aggregation by ibrutinib was shifted to the left in the XLA patients when compared to that of the healthy donors (Figure 5Aii-iii), whereas the curve for inhibition of PLCg2 phosphorylation was unchanged (Figure 5Aiv). Since the only known difference between XLA patients and controls is the absence of Btk, this demonstrates an off-target effect of ibrutinib that was unmasked in the absence of Btk protein. This off-target action occurred over a similar concentration range to that required for inhibition of Btk. The GPCR agonists ADP and PAR1 peptide stimulated robust aggregation in XLA patients, as previously demonstrated16 (data not shown). One possible explanation for the increased sensitivity of XLA patients to ibrutinib relative to controls is that Btk also functions as an adapter protein in the GPVI signaling pathway (as the only known difference in these two groups is Btk protein). To investigate this, we transfected Btk-deficient DT40 chicken B cells with GPVI and its signaling partner, FcRg, in the presence of wild-type (WT) or kinase-dead (KD) Btk. Importantly, these cells express PLCg2 but do not express other Tec family kinases.22,23 The

Acalabrutinib inhibits Btk Y223 phosphorylation and platelet aggregation, secretion and Ca2+ mobilization by glycoprotein VI Studies were extended to a second-generation Btk inhibitor, acalabrutinib, which, like ibrutinib, irreversibly binds to Btk at C481 and is highly plasma protein-bound. Acalabrutinib has a higher selectivity over other tyrosine kinases relative to ibrutinib, including Src, Syk and Tec,25 but a 5-fold lower IC50 for Btk.17 In patients the mean peak free drug concentration of acalabrutinib is 1.3 mM.17 Acalabrutinib has a similar, dose-dependent effect on platelet aggregation to that of ibrutinib. At a concentration of 2 mM in washed platelets, acalabrutinib induced a slight delay in aggregation (Online Supplementary Figure S3A) but had no effect on the overall magnitude of response (Figure 7Aii). The difference in the dose-dependency relative to ibrutinib is consistent with the lower IC50 of acalabrutinib for Btk. Similar to ibrutinib, inhibition of platelet aggregation by CRP occurs at acalabrutinib concentrations that are one order of magnitude higher than those which cause

2104

K430E mutant of Btk has been previously reported to lack kinase activity.24 Cells lacking Btk or GPVI were unresponsive to collagen. Cells transfected with WT or KD Btk reconstituted NFAT signaling to a similar degree (Figure 6A,B), demonstrating that Btk also functions as an adapter protein in the GPVI signaling pathway. A low dose of ibrutinib had no effect on cells transfected with WT or KD Btk whereas a high dose of ibrutinib blocked NFAT signaling in both WT and KD transfected cells (Figure 6C). Together these results demonstrate that Btk functions as an adapter protein, as well as a kinase, in XLA platelets and in transfected DT40 cells.

Figure 6. Kinase-dead Btk is sufficient for glycoprotein VI signaling. Btk-deficient DT40 cells were transfected with either wild-type (WT) or kinase-dead (KD) Btk with or without GPVI/FcRg. All cells were transfected with a NFATluciferase reporter plasmid. Cells were stimulated with collagen (10 mg/mL) in the presence of serum. (A) Luciferase activity was measured and is shown as mean ± SEM of five identical experiments. (B) Representative western blot showing equal Btk expression in WT and KD-transfected cells. (C) Cells were stimulated with collagen (10 mg/mL) in the presence or absence of ibrutinib (0.5 mM – 10 mM). Serum was excluded during stimulation to avoid plasma binding of the drugs. Luciferase activity between vehicle and drug-treated samples was measured and is shown as the mean ± SEM of three independent experiments. *P<0.05, **P<0.01.

haematologica | 2018; 103(12)

Effects of Btk inhibitors on platelet activation

Aii

Aiii

Aiv

Bii

Figure 7. Acalabrutinib dose-dependently inhibits glycoprotein VI-mediated signaling. (A) Eptifibatide (9 mM)-treated washed human platelets (4×108/mL) were stimulated with CRP (10 mg/mL for 180 s) followed by lysis with 5X SDS reducing sample buffer. Prior to addition of agonist, platelets were pre-incubated with either acalabrutinib or vehicle (DMSO). (i) Whole cell lysates were then separated by SDS-PAGE and western blot with the stated antibodies for whole cell phosphorylation, kinases and proteins downstream of GPVI. Blots are representative of three experiments. (ii - iv) The percentage of tyrosine phosphorylation as compared to that of vehicle-treated platelets was measured and is represented as the mean ± SEM of three identical experiments. The dose-response curve for inhibition of aggregation from Online Supplementary Figure S3D is shown as a dotted line to enable comparison. (B) Eptifibatide (9 mM)-treated washed human platelets (8×108/mL) were stimulated with CRP (10 mg/mL for 180 s) followed by lysis with 2X ice cold lysis sample buffer. Lysates were precleared and Tec was immunoprecipitated before addition of SDS reducing sample buffer and separation by SDS-PAGE and western blot with the anti-pY antibody 4G10. Membranes were stripped and reprobed with the pan-Tec antibody. The trace is representative of three identical experiments. (C) Btk-deficient DT40 cells were transfected with either wild-type (WT) or kinase-dead (KD) Btk with or without GPVI/FcRg. All cells were transfected with a NFAT-luciferase reporter plasmid. Cells were stimulated with collagen (10 mg/mL) in the presence or absence of acalabrutinib (0.5-10 mM). Serum was excluded during stimulation to avoid plasma binding of the drugs. Luciferase activity between vehicle and drug-treated samples was measured and is shown as the mean ± SEM of three independent experiments. ns = non-significant.

haematologica | 2018; 103(12)

2105

P.L.R. Nicolson et al.

a delay in aggregation; and the curves for inhibition of ATP secretion and Ca2+ mobilization lie slightly to the left of that for aggregation (Online Supplementary Figure S3AE). As with ibrutinib, acalabrutinib blocked tyrosine phosphorylation of Btk on Y223 and PLCg2 on Y759 and Y1217 at a concentration (2 mM) that caused a delay in onset but no reduction in aggregation (Figure 7Ai,ii). Higher concentrations of acalabrutinib (up to 200 mM) had no effect on phosphorylation of Src Y418, Syk Y525/6 and LAT Y200 but caused a small reduction in phosphorylation of Btk Y551 and SLP-76 Y145 (Figure 7Ai,iii-iv). Interestingly, acalabrutinib also caused a biphasic inhibition of Tec phosphorylation with partial inhibition observed at approximately 1 mM and full blockade at 200 mM (Figure 7B). The IC50 values for each phosphorylation event are included in Online Supplementary Table S1. Concentrations of acalabrutinib that blocked phosphorylation of Btk in platelets had no effect on NFAT activation by CRP in DT40 cells transfected with WT or KD Btk (Figure 7C). Allowing for the fact that acalabrutinib has a 5-fold lower potency for Btk, these results are in line with those for ibrutinib.

Glycoprotein VI-mediated platelet aggregation is blocked ex vivo in patients taking ibrutinib, but not acalabrutinib We investigated the effect of ibrutinib and acalabrutinib in patients with chronic lymphoid leukemia (CLL) taking ibrutinib 420 mg once daily, acalabrutinib 100 mg twice daily or a non-Btk targeting control chemotherapy regimen. GPVI-induced platelet aggregation was blocked in the platelet-rich plasma of patients taking ibrutinib but was not blocked in patients taking acalabrutinib or in the control group despite complete inhibition of autophosphorylation of Btk pY223 and its downstream substrate PLCg2 at pY1217 by both inhibitors (Figure 5Bi-iv). Platelet aggregation induced by the GPCR agonists, ADP and PAR1 peptide, was not altered in the patients taking either inhibitor (data not shown).

Discussion In this study we show that (i) irreversible blockade of Btk by ibrutinib and acalabrutinib delays but does not block the platelet aggregation induced by high concentrations of GPVI agonists; (ii) blockade of GPVI-mediated aggregation by ibrutinib and acalabrutinib occurs at a concentration one to two orders of magnitude higher than is required to block Btk due to an off-target action which is reversible; (iii) the ratio between inhibition of Btk kinase activity and platelet aggregation induced by GPVI is the same for ibrutinib and acalabrutinib; (iv) clinically relevant concentrations of ibrutinib but not acalabrutinib block activation of platelets by GPVI; (v) platelet adhesion and aggregation under flow conditions is maintained following inhibition of Btk; (vi) Btk supports platelet activation by GPVI by acting as an adapter protein and as a tyrosine kinase; and (vii) ibrutinib blocks platelet aggregation in XLA patients at concentrations that block Btk. These results show that platelets, in which Btk kinase function and downstream PLCg2 phosphorylation have been blocked, have a slight delay in aggregation in response to high concentrations of GPVI ligands, while platelet adhesion and aggregation under arterial flow con2106

ditions are unaltered. These observations, together with reports that patients with XLA or those treated with acalabrutinib do not experience major bleeding,15,17 provide powerful evidence that inhibition of Btk does not give rise to major bleeding. The major bleeding observed in patients treated with ibrutinib relative to acalabrutinib is due to the differential dosing regimens of the two Btk inhibitors, with the clinical dose of ibrutinib blocking activation of platelets by GPVI due to one or more off-target effects. The conclusion that inhibition of Btk does not give rise to major bleeding on treatment with ibrutinib contrasts with the conclusion of the studies by Levade et al.11 and Bye et al.13. Levade et al.11 demonstrated a close correlation between inhibition of autophosphorylation of Btk at Y223 and aggregation in GPVI-activated platelets. While we are unable to explain this in the light of the present observations, we note that Levade et al.11 also reported that phosphorylation of PLCg2 at Y753, which is mediated by Btk, was inhibited at a 10-fold lower concentration of ibrutinib as is seen in the present study. Bye et al.13 used a single, supramaximal concentration of ibrutinib which also blocked Src phosphorylation for their biochemical and flow-based assays. The determination of full concentration-response curves in the present study has highlighted the mismatch between inhibition of Btk and loss of platelet aggregation, and has provided evidence that the bleeding diathesis that is seen in some ibrutinib-treated patients is due to off-target effects. An unexpected observation in the present study was that platelets are able to aggregate in response to a high dose of CRP despite the absence of detectable PLCg2 phosphorylation. One explanation for this is that Btk also supports activation of PLCg2 as an adapter protein as shown by the observation that transfection of KD Btk restores GPVI signaling in DT40 cells. A similar result has been previously shown for Btk in B-cell receptor signaling.24,26 This is in keeping with previous studies showing that phosphorylation of PLCg2 at Y1217 is not required for its enzymatic activity in Ramos cells.27 We were surprised to find that platelets from XLA patients, who lack Btk protein, have increased susceptibility to ibrutinib relative to platelets from controls. The only known difference between the XLA patients and controls in the presence of ibrutinib is the absence of Btk protein, although this could also change the balance of activatory and inhibitory phosphorylation within the GPVI signaling cascade. Furthermore, the absence of Btk renders aggregation of these platelets critically dependent on PLCg2 phosphorylation in contrast to controls. The target for ibrutinib which gives rise to inhibition of aggregation in the XLA patients is not known. There are several kinases that are inhibited by ibrutinib over a similar range of concentrations to that for inhibition of Btk.18 Within this group only Csk is known to be expressed in platelets.28 We have shown that blockade of GPVI-mediated platelet aggregation by ibrutinib is reversible, which contrasts with the irreversible blockade of Btk and Tec.18 The reversibility provides evidence that blockade is not mediated by inhibition of Tec family kinases, as was previously postulated,9,11,17 because Tec also has a cysteine residue in its ATP binding domain analogous to C481 on Btk. This is further supported by the observation that ibrutinib-mediated blockade of NFAT signaling in DT40 cells, which lack Tec, follows a similar pattern as that for platelet aggregahaematologica | 2018; 103(12)

Effects of Btk inhibitors on platelet activation

tion; namely no effect at low doses with blockade at high doses. For ibrutinib, we have shown that inhibition of aggregation correlates strongly with loss of phosphorylation at Src Y418. However, this is not altered by acalabrutinib demonstrating an as yet unidentified off-target action. Bye et al.13 also showed that ibrutinib dose-dependently inhibits phosphorylation of Src Y418. However, in a different study they found that both low-dose ibrutinib and acalabrutinib potentiated Src Y418 phosphorylation.29 We were not able to replicate this latter finding. We have shown that, despite acalabrutinib’s more favorable selectivity to Btk over other Src, Syk and Tec kinases in in vitro kinase assays, the window between Btk inhibition and blockade of GPVI-induced aggregation in vitro is similar to that of ibrutinib. Despite this, acalabrutinib, but not ibrutinib, fails to block GPVI-mediated platelet activation ex vivo. We propose that this is because of the differential dosing and pharmacodynamics of the two Btk inhibitors. Acalabrutinib is used at a dose of 1.5 mg/kg twice daily17,30 and ibrutinib at a single daily dose of 6 mg/kg in CLL or 8 mg/kg in mantle cell lymphoma. Pharmacokinetic studies have shown that ibrutinib achieves Btk occupancy of >95% at doses of 2.5 mg/kg but that doses of 6 mg/kg are required to maintain this over 24 h.31 Acalabrutinib at 1.5 mg/kg twice daily also achieves full Btk occupancy over 24 h.17 The peak unbound plasma concentration of ibrutinib in patients is 0.5 mM11 and that of acalabrutinib 1.3 mM.17 The initial and terminal half-lives of ibrutinib are 2-3 h and 4-8 h, respectively.31 The half-life of acalabrutinib is 1 h.17 Despite the peak concentration of acalabrutinib being approximately 2-fold higher than the concentration of ibrutinib, the 5-fold lower potency of acalabrutinib as an inhibitor of Btk30 means that, in potency terms, it is dosed at a lower level consistent with the lack of inhibition of GPVI. This implies that ibrutinib could be used at a lower concentration to achieve Btk blockade. Indeed, there is retrospective clinical evidence that doses less than 6 mg/kg are as effective as 6 mg/kg for treating CLL32 and a prospective clinical trial using doses as low as 2.5 mg/kg is being undertaken.33 It is important to consider the incidence of minor and major bleeding in patients taking ibrutinib for CLL or at the higher dose for mantle cell lymphoma. In reported studies involving patients treated with ibrutinib for mantle cell lymphoma, minor and major bleeding was seen in 9-15% and 1-5% of patients, respectively.34-36 In the study with the largest cohort of patients with mantle cell lymphoma, the rate of major hemorrhages was 5%. This is comparable to the 4-8% major hemorrhage rate seen in patients taking ibrutinib for CLL.9 Thus, there is no increase in bleeding rates with higher doses of ibrutinib. This implies that the inhibitory effect of 420 mg ibrutinib on platelets is at a physiological maximum. During the writing of this manuscript, Bye et al. reported thrombus instability on collagen in a flow adhesion assay in blood treated in vitro with high doses of ibrutinib and ex vivo in patients treated with ibrutinib.29 This is consistent with our findings that Btk kinase function is not required for platelet adhesion to collagen under flow, but that offtarget effects of ibrutinib seen with higher doses mediate this inhibition. They also reported complete blockade of platelet aggregation in response to supramaximal concentrations of collagen in patients receiving ibrutinib or acalhaematologica | 2018; 103(12)

abrutinib29 in contrast to the findings of this study. Bye et al. used the Optimul 96-well microtiter assay to measure aggregation rather than the widely used light transmission aggregometry. We have shown that Optimul is a more sensitive assay than light transmission aggregometry.37,38 We suggest that the delay in onset of aggregation observed using light transmission aggregometry with concentrations of ibrutinib or acalabrutinib that just block Btk manifest as complete blockade in the Optimul assay. Bye et al. also concluded that the increased bleeding observed with ibrutinib is due to blockade of Src family kinases (SFK).29 We agree that bleeding caused by ibrutinib is due to off-target actions, and that acalabrutinib has a greater selectively to Btk over SFK relative to ibrutinib. However, our results show that a similar fold increase in the concentration of ibrutinib and acalabrutinib causes inhibition of platelet aggregation in response to CRP but without concomitant SFK blockade in the acalabrutinib-treated platelets. Thus, the off-target action of ibrutinib and acalabrutinib that inhibits aggregation cannot be explained solely by differential blockade of SFK. The results of our study explain the lack of major bleeding side effects experienced by patients taking acalabrutinib and suggest that the bleeding side effect of ibrutinib can potentially be abolished by reducing the dose. Furthermore, this study also shows that the bleeding caused by ibrutinib is not due to an irreversible action. This predicts that the GPVI blockade wears off over a period of 24 h as the drug is cleared.31 We hypothesize that, in the event of a major bleed, there may be no need to use expensive and potentially harmful platelet transfusions to correct the signaling deficit. Nevertheless, each clinical scenario should be judged on its own merits and individual clinicians’ discretion is crucial. In conclusion, the present study shows that inhibition of Btk kinase activity causes only partial inhibition of GPVI signaling in platelets and provides evidence that Btk supports GPVI signaling by functioning as an adapter protein as well as a kinase. The excessive bleeding induced by ibrutinib relative to acalabrutinib is likely to reflect a nonTec family kinase off-target inhibitory effect of ibrutinib, probably on Src. Acknowledgments This work was supported by British Haeart Foundation (BHF) Programme grants (RG/13/18/30563), a BHF clinical fellowship to PLRN (FS/17/20/32738), an AMS springboard grant to AYP (SBF002\1099) and BHF studentship to ATH, the University of Birmingham’s Institute of Translation Medicine and Institute of Cardiovascular Sciences; SPW holds a BHF Chair (CH03/003). We would like to thank Alex Bye and Jon Gibbins for their expertise on the Ca2+ mobilization assay. We would also like to thank Vicky Simms, Natalie Poulter and Steve Thomas for their help with the flow adhesion assay and Mark Crowther, Nick Pemberton, Salim Shafeek, Kate Arthur, Gaynor Pemberton, Lesley Candlin and Rebekah Hart at Worcestershire Royal Hospital, Shankara Paneesha, Alison Hardy and Melanie Kelly at Birmingham Heartlands Hospital and Tina McSkeane, Gillian Marshall and Michelle Harry at the Queen Elizabeth Hospital for provision of patients’ samples. Finally, we would like to thank Andrew Wilkinson and Robert Neely from the School of Chemistry at the University of Birmingham for their help with the chemical analysis of ibrutinib and acalabrutinib. 2107

P.L.R. Nicolson et al.

References 1. Li Z, Delaney MK, O'Brien KA, Du X. Signalling during platelet adhesion and activation. Arterioscler. Thromb Vasc Biol. 2010;30(12):2341–2349. 2. Alshehri OM, Hughes CE, Montague S, et al. Fibrin activates GPVI in human and mouse platelets. Blood. 2015;126(13):1601– 1608. 3. Nieswandt B. Platelet-collagen interaction: is GPVI the central receptor? Blood. 2003;102 (2):449–461. 4. Mammadova-Bach E, Ollivier V, Loyau S, et al. Platelet glycoprotein VI binds to polymerized fibrin and promotes thrombin generation. Blood. 2015;126(5):683–691. 5. Watson SP, Herbert JMJ, Pollitt AY. GPVI and CLEC-2 in hemostasis and vascular integrity. J Thromb Haemost. 2010;8(7):1456– 1467. 6. Jandrot-Perrus M, Busfield S, Lagrue AH, et al. Cloning, characterization, and functional studies of human and mouse glycoprotein VI: a platelet-specific collagen receptor from the immunoglobulin superfamily. Blood. 2000;96(1):1798–1807. 7. Bender M, Hagedorn I, Nieswandt B. Genetic and antibody-induced glycoprotein VI deficiency equally protects mice from mechanically and FeCl3-induced thrombosis. J Thromb Haemost. 2011;9(7):1423– 1426. 8. Nurden AT, Nurden P. Congenital platelet disorders and understanding of platelet function. Br J Haematol. 2013;165(2):165–178. 9. Shatzel JJ, Olson SR, Tao DL, et al. Ibrutinibassociated bleeding: pathogenesis, management, and risk reduction strategies. J Thromb Haemost. 2015;38(1):42–49. 10. Caron F, Leong DP, Hillis C, Fraser G, Siegal D. Current understanding of bleeding with ibrutinib use: a systematic review and metaanalysis. Blood Adv. 2017;1(12):772–778. 11. Levade M, David E, Garcia C, Laurent P, Payrastre B. Ibrutinib treatment affects collagen and von Willebrand factor-dependent platelet functions. Blood. 2014;124(26): 3991–3995. 12. Kamel S, Horton L, Ysebaert L, et al. Ibrutinib inhibits collagen-mediated but not ADP-mediated platelet aggregation. Leukemia. 2015;29(4):783–787. 13. Bye AP, Unsworth AJ, Vaiyapuri S, Stainer AR, Fry MJ, Gibbins JM. Ibrutinib inhibits platelet integrin αIIbβ3 outside-in signalling and thrombus stability but not adhesion to collagen. Arterioscler Thromb Vasc Biol. 2015;35(11):2326-2335. 14. Kazianka L, Drucker C, Skrabs C, et al. Ristocetin-induced platelet aggregation for monitoring of bleeding tendency in CLL treated with ibrutinib. Leukemia. 2017;31(5):1117–1122. 15. Quek LS, Bolen J, Watson SP. A role for Bruton’s tyrosine kinase (Btk) in platelet activation by collagen. Curr Biol. 1998;8(20): 1137-1140.

2108

16. Atkinson BT, Ellmeier W, Watson SP. Tec regulates platelet activation by GPVI in the absence of Btk. Blood. 2003;102(10):3592– 3599. 17. Byrd JC, Harrington B, O'Brien S, et al. Acalabrutinib (ACP-196) in relapsed chronic lymphocytic leukemia. N Engl J Med. 2016;374(4):323–332. 18. Honigberg L, 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 USA. 2010;107(29):13075–13080. 19. Rushworth SA, MacEwan DJ, Bowles KM. Ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(13): 1277–1279. 20. Watanabe D, Hashimoto S, Ishiai M, et al. Four tyrosine residues in phospholipase Cgamma 2, identified as Btk-dependent phosphorylation sites, are required for B cell antigen receptor-coupled calcium signalling. J Biol Chem. 2001;276(42):38595–38601. 21. Wahl MI, Fluckiger AC, Kato RM, et al. Phosphorylation of two regulatory tyrosine residues in the activation of Bruton’s tyrosine kinase via alternative receptors. Proc Natl Acad Sci USA. 1997;94(21):11526– 11533. 22. Tomlinson MG, Kurosaki T, Berson AE, Fujii GH, Bolen JB. Reconstitution of Btk signalling by the atypical tec family tyrosine kinases Bmx and Txk. J Biol Chem. 1999;274(19):13577–13585. 23. Takata M, Homma Y, Kurosaki T. Requirement of phospholipase C-g2 activation in surface immunoglobulin M-induced B cell apoptosis. J Exp Med. 1995;182(4): 907–914. 24. Tomlinson MG, Woods DB, McMahon M, et al. A conditional form of Bruton's tyrosine kinase is sufficient to activate multiple downstream signalling pathways via PLC gamma 2 in B cells. BMC Immunol. 2001;2(4):1–12. 25. Wu J, Zhang M, Liu D. Acalabrutinib (ACP196): a selective second- generation BTK inhibitor. J Hematol Oncol. 2016;9(1):1–4. 26. Salto K, Tolias KF, Abdelhafid S, et al. Btk regulates PtdIns-4,5-P2 synthesis: importance for calcium signalling and PI3K activity. Immunity. 2003;19:669–678. 27. Kim YJ, Sekiya F, Poulin B, Bae YS, Rhee SG. Mechanism of B-cell receptor-induced phosphorylation and activation of phospholipase C-g2. Mol Cell Biol. 2004;24(22):9986–9999. 28. Burkhart JM, Vaudel M, Gambaryan S, et al. The first comprehensive and quantitative analysis of human platelet protein composition allows the comparative analysis of structural and functional pathways. Blood. 2012;120(15):e73–e82. 29. Bye AP, Unsworth AJ, Desborough MJ, et al. Severe platelet dysfunction in NHL patients receiving ibrutinib is absent in patients receiving acalabrutinib. Blood Adv. 2017;1(26):2610–2623. 30. Covey T, Barf T, Gulrajani M, et al. Abstract

31.

32.

33.

34.

35.

36.

37.

38.

39. 40.

41.

42.

2596: ACP-196: a novel covalent Bruton's tyrosine kinase (Btk) inhibitor with improved selectivity and in vivo target coverage in chronic lymphocytic Leukemia (CLL) patients. Cancer Res. 2015;75(15 Suppl):2596–2596. Advani RH, Buggy JJ, Sharman JP, et al. Bruton tyrosine kinase inhibitor ibrutinib (PCI-32765) has significant activity in patients with relapsed/refractory B-cell malignancies. J Clin Oncol. 2012;31(1):88– 94. Banerjee R, Timlin C, Fitzpatrick D, et al. Comparable outcomes in chronic lymphocytic leukeemia patients treated with reduced dose ibrutinib: results from a multicenter study. Haematologica. 2016;101(s1): 56-57. Bose P, Gandhi VV, Keating MJ. Pharmacokinetic and pharmacodynamic evaluation of ibrutinib for the treatment of chronic lymphocytic leukeemia: rationale for lower doses. Expert Opin Drug Metab Toxicol. 2016;11:1–12. Dreyling, M, Jurczak W, Silva RS, et al. Ibrutinib versus temsirolimus in patients with relapsed or refractory mantle-cell lymphoma: an international, randomised, openlabel, phase 3 study. Lancet. 2016;387 (10020):770-779. Rule, S, Dreyling, M, Goy, A, et al. Outcomes in 370 patients with mantle cell lymphoma treated with ibrutinib: a pooled analysis from three open-label studies. Br J Haematol. 2017;179(3):430-438. Wang, ML, Rule S, Martin, P, et al. Targeting BTK with ibrutinib in relapsed or refractory mantle-cell lymphoma. N Eng J Med. 2013;369(6):507-516. Lordkipanidzé M, Lowe GC, Kirkby NS, et al. Characterization of multiple platelet activation pathways in patients with bleeding as a high-throughput screening option: use of 96-well Optimul assay. Blood. 2014;123 (8):e11-e22. Chan, MV, Leadbeater, PD, Watson, SP, Warner TD. Not all light transmission aggregation assays are created equal: qualitative differences between light tramsission and 96-well plate aggregometry. Platelets. 2018 May 1:1-4. [Epub ahead of print] Hughes CE, Pollitt AY, Mori J, et al. CLEC-2 activates Syk through dimerization. Blood. 2010;115(14):2947–2955. Grynkiewicz G, Poenie M, Tsien RY. A new generation of Ca2+ indicators with greatly improved fuorescence properties. J Biol Chem. 2001;260(6):3440–3450. Takata M, Kurosaki T. A role for Bruton's tyrosine kinase in B cell antigen receptormediated activation of phospholipase C-g2. J Exp Med. 1996;184(1):31–40. Hughes CE, Sinha U, Pandey A, et al. Critical role for an acidic amino acid region in platelet signalling by the HemITAM (Hemi-immunoreceptor Tyrosine-based Activation Motif) containing receptor CLEC-2 (C-type lectin receptor-2). J Biol Chem. 2013;288(7):5127–5135.

haematologica | 2018; 103(12)

ARTICLE

Stem Cell Transplantation

Diffuse alveolar hemorrhage is most often fatal and is affected by graft source, conditioning regimen toxicity, and engraftment kinetics

Fatma Keklik,1 Ezzideen Barjes Alrawi,1 Qing Cao,2 Nelli Bejanyan,1 Armin Rashidi,1 Aleksandr Lazaryan,1 Patrick Arndt,3 Erhan H. Dincer,3 Veronika Bachanova,1 Erica D. Warlick,1 Margaret L. MacMillan,4 Mukta Arora,1 Jeffrey Miller,1 Claudio G Brunstein,1 Daniel J. Weisdorf1 and Celalettin Ustun1

Division of Hematology-Oncology and Transplantation, Department of Medicine; Biostatistics and Bioinformatics; 3Division of Pulmonary, Allergy, Critical Care and Sleep Medicine, Department of Medicine and 4Division of Pediatric Blood and Marrow Transplantation, Department of Pediatrics, University of Minnesota, Minneapolis, MN, USA 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(12):2109-2115

ABSTRACT

iffuse alveolar hemorrhage after hematopoietic stem cell transplantation is a frequently fatal complication with no standard therapy. Although significant changes in supportive and intensive care measures for patients undergoing hematopoietic stem cell transplantation have been made over the past decades, the impact of these changes on the incidence and outcome of patients with diffuse alveolar hemorrhage has not been examined. We analyzed 1228 patients who underwent allogeneic hematopoietic stem cell transplantation between 2008-2015 at the University of Minnesota to study the incidence, risk factors, and outcomes of diffuse alveolar hemorrhage. Diffuse alveolar hemorrhage developed in 5% of allogeneic hematopoietic stem cell transplant recipients, at a median of 30 days (range +3 to +168 days) after transplantation. The incidence of diffuse alveolar hemorrhage was significantly greater in recipients of umbilical cord blood than peripheral blood or bone marrow grafts (HR: 2.08, 95% CI: 1.163.74; P=0.01). In multivariate analysis, delayed neutrophil engraftment or primary graft failure was a risk factor for diffuse alveolar hemorrhage following peripheral blood or bone marrow hematopoietic stem cell transplants (HR: 5.51, 95% CI: 1.26-24; P=0.02) and delayed platelet engraftment was associated with significantly increased diffuse alveolar hemorrhage in umbilical cord blood transplant recipients (HR: 6.96, 95% CI: 2.39-20.29; P<0.05). Myeloablative regimens including total body irradiation were also risk factors for diffuse alveolar hemorrhage (HR: 1.8, 95% CI: 1.03-3.13, P=0.05) in both peripheral blood or bone marrow and umbilical cord blood hematopoietic stem cell transplants (HR: 1.87, 95% CI: 0.95-3.71). Patients with diffuse alveolar hemorrhage had an inferior 6-month treatment-related mortality (HR: 6.09, 95% CI: 4.338.56, P<0.01) and 2-year overall survival (HR: 4.16, 95% CI: 3.06-5.64; P<0.01) using either graft source. The etiology of diffuse alveolar hemorrhage is multifactorial, involving lung injury influenced by high-dose total body irradiation, graft source, and delayed engraftment or graft failure. The survival of patients with diffuse alveolar hemorrhage after hematopoietic stem cell transplantation remains poor. Clinical interventions or experimental studies (e.g., cell expansion for umbilical cord blood transplants or thrombopoietin use) that modulate these risk factors may limit the incidence and improve the outcomes of diffuse alveolar hemorrhage. haematologica | 2018; 103(12)

Correspondence: celalettin_ustun@rush.edu

Received: January 23, 2018. Accepted: July 27, 2018. Pre-published: August 3, 2018. doi:10.3324/haematol.2018.189134 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/12/2109 ÂŠ2018 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2109

F. Keklik et al.

Introduction Pulmonary complications occur frequently in patients after hematopoietic stem cell transplantation (HCT). Diffuse alveolar hemorrhage (DAH) is a serious pulmonary complication with a high mortality rate after HCT.1,2 The incidence of DAH varies between 3% and 10% after allogeneic HCT.3,4 The clinical symptoms of DAH include cough, hypoxemia, fever, and rarely hemoptysis.5 Chest radiography shows non-specific bilateral areas of ground-glass attenuation and patchy areas of consolidation.6 Bronchoalveolar lavage findings are characteristic for the diagnosis of DAH with progressively bloodier lavage returns. The etiology of DAH remains unknown, but lung tissue injury, inflammation and cytokine release have all been implicated in the pathogenesis of DAH.5 Some may require that no microorganism is isolated from the bronchoalveolar lavage fluid; however, DAH syndromes after allogeneic HCT have been shown to have similar clinical characteristics and presentation with or without an associated infection.3,7 The reported outcomes of patients with DAH after allogeneic HCT have been dismal. Generally, patients with DAH require intensive care unit support and often mechanical ventilation because of severe hypoxemia. The mortality rate can reach 70-100% in patients with DAH because of multiorgan failure and/or sepsis.5,8-10 Recent changes in allogeneic HCT and supportive care practices include: (i) more frequent use of alternative donor allografts [e.g., umbilical cord blood (UCB) and haploidentical donors]; (ii) use of reduced intensity conditioning for older patients or those with comorbidities; (iii) advances in the use of antibiotics targeting fungal and viral infection; and (iv) improvements in intensive care unit support. We therefore evaluated the impact of these changes on the contemporary incidence, risks, and outcomes of DAH after allogeneic HCT.

Methods All patients who received peripheral blood stem cells (PB), bone marrow (BM) or UCB HCT between 2008 and 2015 at the University of Minnesota were included. The data were prospectively collected in our institutional Blood and Marrow Transplant Database on patients who had provided written informed consent to Institutional Research Board-approved studies, supplemented by individual medical record review as needed.

Definitions Patients received myeloablative or reduced intensity conditioning regimens (based on age ≥55 years or significant comorbidities, as previously described).11 The dose of total body irradiation (TBI) differed, being 1320 cGy in fractionated doses for myeloablative conditioning and 200 cGy for reduced intensity conditioning regimens. UCB HCT grafts were matched at four to six loci of six HLA-A, -B (antigen level), and -DRB1 (allele level) to the recipient and in patients receiving two UCB units were generally similarly matched to each other.12 HLA-matched was defined as 6/6 (or 45/6 + 5-6/6 for double UCB) and 8/8 for PB and BM graft sources with graft nucleated cell doses and CD34 content as previously described.13,14 Standard risk was defined as patients with leukemia, lymphoma or other malignancy in first or second complete remission, chronic myelogenous leukemia in first chronic phase, 2110

myelodysplastic syndrome without excess blasts, or nonmalignant diseases. All other patients were considered high risk. DAH was diagnosed by strict clinical criteria and alveolar lavage in all patients during fiberoptic bronchoscopy. The criteria included acute onset of hypoxemia with presence of diffuse pulmonary infiltrates on a chest X-ray or computed tomography scan and the presence of progressively bloodier return with each subsequent bronchoalveolar lavage.2,15 Aliquots of saline were successively instilled and aspirated and then visually examined to detect the presence of hemorrhage. The bronchoalveolar lavage fluid was submitted for cytological and microbiological examination. Cytological evaluation included specimen review using potassium hydroxide, Giemsa, and Papanicolaou stains. Appropriate microbiological and cytological examinations for bacteria, fungi, mycobacteria, Pneumocystis jiroveci, and viruses were also performed. For this study, we classified all patients as having DAH regardless of documented infection from bronchoalveolar lavage given our previous analysis demonstrating that infection-associated alveolar hemorrhage and DAH were related clinical syndromes with similar clinical presentation and risks.3 Supportive measures for the management of alveolar hemorrhage included correction of platelet and coagulation abnormalities, careful maintenance of fluid and electrolyte balance, and aggressive ventilator and oxygen support. All patients received prophylactic and empiric antimicrobial agents as clinically indicated. Corticosteroids, administered in most patients (>90%), consisted of a standard regimen of high-dose methylprednisolone, 500 mg twice a day for 3 days, followed by 250 mg twice a day for 3 days, 125 mg twice a day for 3 days, 60 mg twice a day for 3 days, and then 60 mg once a day tapered off over a 2-month period. Pediatric patients also received a similar dose and schedule of methylprednisolone with the dose of methylprednisolone adjusted per body surface area (i.e., starting dose 250 mg/m2/dose intravenously, twice a day for 3 days). Neutrophil engraftment was defined as the first of 3 consecutive days with an absolute neutrophil count ≥0.5×109/L. Event times for neutrophil and platelet recovery were measured from the date of transplantation and were censored for death or disease progression before day 21 without neutrophil recovery. Platelet engraftment was defined as the first day when the platelet count was >20×109/L and subsequently remained so without transfusions for 7 days. Neutrophil counts that never decreased below 0.5×109/L and platelet counts that never decreased below 20×109/L (without transfusion support) were defined as indicating engraftment at day +1 for either cell lineage. Primary graft failure was defined as a lack of neutrophil recovery at day 42 or less than 10% marrow reconstitution of donor origin, even with neutrophil recovery.

Statistical analysis The baseline characteristics of the patients and their transplants and information on post-transplantation complications and outcomes were prospectively collected by the Biostatistical Support Group at the University of Minnesota using standardized collection procedures. Demographic and transplant characteristics were summarized by standard descriptive statistical methods. The statistical comparison of categorical variables was performed using a chi-square test, while the Kruskal-Wallis (Wilcoxon) rank-sum test was used for comparisons of continuous variables between patients with and without DAH. A cumulative incidence estimator was used to calculate the probabilities of neutrophil engraftment, DAH and infection, reflecting non-event deaths as a competing risk. The cumulative incidence of treatment-related mortality was also calculated, reflecting relapse as a competing risk.16 Fine and Gray regression analysis was used to compare the differences between cumulative haematologica | 2018; 103(12)

Diffuse alveolar hemorrhage after HCT

incidence curves for the endpoints of neutrophil engraftment, treatment-related mortality, DAH and infection.17 The KaplanMeier method was used to estimate the probabilities of diseasefree survival and overall survival through 2 years after allogeneic HCT and the log-rank test was used for univariate comparisons.18 A Cox proportional hazard regression model was used to estimate differences between the adjusted survival curves,19 with DAH being treated as a time-dependent variable. Factors that were considered in regression models included gender (male versus female), age at transplant (0-35 years versus ≥35 years), recipient’s cytomegalovirus serostatus (positive versus negative), intensity of the transplant conditioning regimens (myeloablative versus reduced intensity), graft source (PB/BM versus UCB), TBI use (yes versus no), composite factor of TBI and conditioning intensity, diagnosis (malignant versus nonmalignant), neutrophil engraftment (treated as a time-dependent variable), platelet engraftment (treated as a time-dependent variable), disease risk (standard risk versus high risk) and greatest mismatch for HLA disparity considering the worst matched of the two UCB units (matched versus mismatched). Factors with a univariate P-value <0.15 or those with established potential clinical importance were included in the multivariate analysis. Prognostic factor models for all clinical outcomes were built using a backward selection method (P<0.05 was considered significant for remaining in the model). All statistical analyses were implemented using Statistical Analysis System statistical software version 9.3 (SAS Institute Inc., Cary, NC, USA). The cut-off significance level for all P values was 0.05.

Results A total of 1228 patients undergoing allogenic HCT were included in the study: 658 received PB/BM grafts and 570 received UCB grafts. There were significant differences between PB/BM recipients and UCB graft recipients regarding patient- and disease-characteristics (Online Supplementary Table S1). The median total nucleated cell count was higher and the median time-to-engraftment was shorter in PB/BM graft recipients than in UCB graft recipients. DAH was diagnosed in 59 patients (5%) (Table 1) at a median of 30 days (26 days for BM/PB and 33 days for UCB grafts) after HCT. The median time from HCT to neutrophil engraftment was 19 days (14 days for BM/PB and 22 days for UCB) while that for platelet engraftment was 63 days (20 days for BM/PB and 84 days for UCB grafts). Fifty-one percent of patients had DAH within the 2 weeks preceding or after neutrophil engraftment. Seventy-nine percent of patients with DAH had no platelet engraftment at the time of DAH. The patients with DAH had a median age of 32 years, 66% were male, 85% had a malignant disease, 42% had a history of smoking, 5.1% had a history of pre-HCT lung disease, 54% were seropositive for cytomegalovirus, 59% received myeloablative conditioning, 86% received TBI, 64% were HLA-matched and 20% had a sibling donor. At the time of DAH, the median platelet count was 24x109/L (range, 1.0-114), 34% had renal dysfunction (creatinine >1.3 mg/dL), 34% had abnormal liver function (alanine transaminase >35 U/L), 8% had an elevated international normalized ratio >1.5, 25% had a prolonged partial thromboplastin time (≥37 s, although in most cases it was <40 s), and no tested patient (n=37) had a low fibrinogen level <180 mg/dL. Forty-six percent of patients had fever, haematologica | 2018; 103(12)

27% had a documented systemic infection, and in 19% of cases an infectious organism was isolated from the bronchoalveolar lavage fluid along with the findings of DAH. In patients with DAH, the incidence of grade II-IV acute graft-versus-host disease (GvHD) by day +100 was 36%, with most cases of grade II-IV GvHD (69.5%) occurring Table 1. Patient and transplant characteristics.

No DAH n (%)

DAH n (%)

P value

Number Sex

1169 (95%) 59 (5%) Male 708 (61) 39 (66) 0.40 Female 461 (39) 20 (34) Age at HCT (years) 0-35 591 (51) 31 (52) 0.77 ≥35 578 (49) 28 (47) Malignant disease Yes 830 (71) 50 (85) 0.02 No 339 (29) 9 (15) Disease risk Standard 800 (68) 36 (61) 0.23 High 369 (32) 23 (39) Recipient CMV serology Positive 651 (56) 32 (55) 0.87 Negative 506 (44) 26 (45) Conditioning intensity MAC 534 (46) 35 (59) 0.04 RIC 635 (54) 24 (41) TBI Yes 865 (74) 51 (86) 0.03 No 304 (26) 8 (14) Conditioning intensity & TBI MAC TBI 339 (29) 28 (47) 0.01 MAC no TBI 195 (17) 7 (12) RIC 635 (54) 24 (40) Graft source BM/PB 639 (55) 19 (32) <0.01 UCB 530 (45) 40 (68) Graft source PB 342 (53) 10 (53) 0.97 BM 297 (47) 9 (47) Graft source & BM/PB-MAC 261 (49) 10 (29) 0.02 conditioning intensity UCB-MAC 273 (51) 25 (71) Graft cell dose infused TNC x108/kg, median (range) BM/PB 5.22 (0.01-35.50) 3.10 (0.02-12.57) 0.01 UCB 0.46 (0.03-99.01) 0.39 (0.21-2.04) 0.01 CD34 dose, median (range) 2.67 (0.01-64.46) 0.95 (0.02-10.14) <0.01 HLA matching M BM/PB 543 (46%) 13 (22%) <0.01 MM BM/PB 96 (8%) 6 (10%) UCB 530 (45%) 40 (68%) Post-transplant events Neutrophil engraftment (n) 1099 45 Median days (range) 14 (0-42) 19 (0-41) <0.01 Platelet engraftment (n) 984 22 Median days (range) 30 (0-175) 63 (0-168) <0.01 Engraftment day +1 Neutrophils, n (%) Yes 96 (8.2%) 1 (1.6%) 0.04 No 1073 (91.8%) 58 (98.4%) Platelets, n (%) Yes 18 (1.5%) 1 (1.6%) 0.61 No 1151 (98.5%) 58 (98.4%) Graft failure Yes 51 (4%) 11 (19%) <0.01 No 1118 (96%) 48 (81%) Acute GvHD grade II-IV (n) 383 23 Median days (range) 35 (11-174) 34 (14-148) 0.40 <day 100 (n=360) 360 (31) 21 (36) 0.13 BM: bone marrow; CMV: cytomegalovirus; DAH: diffuse alveolar hemorrhage; GvHD: graft-versushost disease; HCT: hematopoietic stem cell transplantation; M: matched; MAC: myeloablative conditioning; MM: mismatched; TNC: total nucleated cell count; PB: peripheral blood; RIC: reduced-intensity conditioning; TBI: total body irradiation; TRM: treatment-related mortality; UCB: umbilical cord blood. Significant differences are shown in bold.

2111

F. Keklik et al.

prior to DAH. Most patients (92%) received high-dose steroids for the treatment of DAH and seven patients (12%) additionally received an anti-tumor necrotizing factor: etanercept in six cases and infliximab in one. In three patients, anti-tumor necrotizing factor was mainly used for the treatment of concurrent severe gut GvHD. Seventy-five percent of the patients required mechanical ventilator support.

Risk factors DAH occurred more often in patients who received UCB grafts, TBI at a myeloablative dose, and HLA-mismatched donor grafts: it was also more common among those who had a malignant disease (Table 1). The median time to platelet engraftment (63 days versus 30 days) and to neutrophil engraftment (19 days versus 14 days) was significantly delayed in the group with DAH compared with the group that did not develop DAH (P<0.01 for each). Primary graft failure was significantly more frequent in the DAH group (Table 1). Only 1.6% of patients who never experienced severe thrombocytopenia had DAH compared with 8.2% of patients with severe thrombocytopenia (P=0.04). The median graft cell dose was significantly lower in the group that had DAH (3.1x108/kg) than in the group that did not develop DAH (5.22 x 108/kg) (P<0.01) among PB/BM HCT recipients, whereas among UCB HCT recipients it was lower in those with DAH than in those without (0.39x108/kg versus 0.46x108/kg, respectively; P<0.01) (Table 1). Sex, age, disease risk, recipient cytomegalovirus serology, and the incidence of acute GvHD were all similar between the groups with and without DAH. DAH was observed in 7% (40/570) of the UCB graft recipients (Table 2). These UCB graft recipients received myeloablative TBI more often than UCB graft recipients without DAH did (90% versus 76%; P=0.05), more often had double UCB grafts (90% versus 62%; P<0.01) and received fewer cells/kg recipient weight [0.39x108/kg (range, 0.21-2.04x108/kg) versus 0.46x1088/kg (range, 0.0399.01x107/kg; P<0.01]. Neutrophil and platelet engraftment failure (by the time of death or DAH) was strongly associated with DAH (30% versus 9%, and 70% versus 21%, respectively, in the groups with and without DAH) (Table 2). Multivariate analysis showed that UCB HCT recipients had a 2-fold higher incidence of DAH than PB/BM HCT recipients (HR: 2.08, 95% CI: 1.16-3.74; P=0.01). Delayed neutrophil engraftment or graft failure was a risk factor for DAH in PB/BM HCT recipients (HR: 5.51, 95% CI: 1.2624; P=0.02) whereas delayed platelet engraftment was associated with significantly increased DAH in UCB HCT recipients (HR: 6.96, 95% CI: 2.39-20.29; P<0.05) (Table 3A,B). Two different engraftment models were tested because neutrophil engraftment was strongly correlated with platelet engraftment. TBI at a myeloablative dose was also a risk factor for DAH (HR: 1.8, 95% CI: 1.033.13; P=0.05), an effect that was more pronounced in UCB HCT recipients (HR: 1.87, 95% CI: 0.95-3.71; P=0.08) than in BM/PB HCT recipients (HR: 1.63, 95% CI: 0.62-4.31; P=0.60) (Table 3A,B).

95% CI: 3.06-5.64; P<0.01) following HCT with either graft source (PB/BM or UCB). UCB was also a risk factor for poorer treatment-related mortality at 6 months (HR: 1.43, 95% CI: 1.08-1.9; P=0.01) and overall survival (HR: 1.22, 95% CI: 1.02-1.48; P=0.03). A total of 44 patients with DAH (74.5%) required intubation. Of the DAH patients requiring intubation, 54% and 66% died 30 and 60 days after intubation while only 13% and 26% of DAH patients not requiring intubation died by 30 and 60 days after the diagnosis of DAH (P=0.01). Among the intubated patients, UCB HCT recipients had a higher mortality at 6 months compared with the BM/PB graft recipients (84% versus 56%; P=0.05) (Online Supplementary Figure S1). Of seven patients who received an anti-tumor necrotizing factor drug, only one (14%) survived.

Table 2. Characteristics of patients receiving umbilical cord blood hematopoietic stem cell transplantation.

No DAH

DAH

P value

Treatment-related mortality and survival

Number N=530 N=40 Sex Male 317(60%) 25(63%) Female 213(40%) 15(38%) Age at HCT (years) Median (range) 28(0-73) 32(2-72) Malignant disease Yes 401(76%) 37(92%) No 129(24%) 3(8%) Disease risk Standard 381(72%) 25(63%) High 149(28%) 15(38%) Conditioning intensity MAC 273(52%) 25(63%) RIC 257(48%) 15(38%) TBI Yes 405(76%) 36(90%) No 125(24%) 4(10%) Conditioning intensity & TBI MAC TBI 177 (33%) 21 (53%) MAC no TBI 86 (18%) 4 (10%) RIC 272 (48%) 15 (37%) UCB units Single 199(38%) 4(10%) Double 331(62%) 36(90%) TNC x 108/kg N 530 40 Median (Range) 0.46 (0.03-99.01) 0.39 (0.21-2.04) TNC x 108/kg by unit number Single, median (range) 0.59 (0.03-14.28) 0.35 (0.21-0.37) Double, median (range) 0.43 (0.19-99.01) 0.41 (0.23-2.04) ANC engraftment Median days (range) 18 (0-42) 22 (0-41) Platelet engraftment Median days (range) 44 (0-175) 84 (48-168) Graft failure Yes 36 (7%) 9 (23%) No 494(93%) 31(78%)

DAH was associated with higher treatment-related mortality at 6 months (HR: 6.09, 95% CI: 4.33-8.56; P<0.01) and a lower overall survival at 2 years (HR: 4.16,

ANC: absolute neutrophil count; DAM: diffuse alveolar hemorrhage; HCT: hematopoietic stem cell transplant; MAC: myeloablative conditioning; RIC: reduced intensity conditioning; TBI: total body irradiation; TNC: total nucleated cell count; UCB: umbilical cord blood; Significant differences are shown in bold.

2112

0.74

0.17 <0.01

0.21

0.18

0.05

0.04

<0.01

0.01

0.03 0.22 0.24 <0.01 <0.01

haematologica | 2018; 103(12)

Diffuse alveolar hemorrhage after HCT

Discussion We observed that even in this recent era, the overall incidence of DAH was 5% for all patients, similar to the incidence recorded in earlier studies.5,20,21 We found that myeloablative TBI, UCB HCT, and delayed engraftment or graft failure were significant risk factors for DAH. Thrombocytopenia is important in hemorrhagic complications after allogeneic HCT;22,23 however, its importance in DAH is controversial.15,24,25 The relation of platelet recovery with DAH is not direct. In our study, both the severity and duration of thrombocytopenia were significantly associated with DAH. However, the risks of DAH cannot be fully explained solely by low platelet counts given that the median platelet count was >20x109/L in our patients with DAH. Robbins et al. also showed that platelet transfusions did not prevent the development and/or progression of DAH.15 Patients with DAH more often had malignant disease and myeloablative conditioning, but reduced intensity conditioning was not associated with DAH. We also found that myeloablative conditioning containing TBI was a risk factor for DAH. These findings strongly suggest that direct lung injury at the time of conditioning by myeloablative dose TBI predisposes to DAH after HCT. Others have suggested that irradiation induces cellular damage and plays an important direct

role in lung injury26 and the association with high-dose TBI during conditioning contributes to the risk of DAH.15,26-28 Moreover, TBI-induced lung injury may prolong or deepen thrombocytopenia given that in human and some animal studies the lung can contribute to platelet production.29-31 TBI may also damage the vascular and rheological microenvironments of the pulmonary capillaries which may be more hemostatic than the microenvironments of pulmonary epithelial tissues.32 A longer duration of thrombocytopenia was associated with increased DAH in our study. It is well-known that UCB HCT is associated with delayed engraftment/graft failure compared with other related or unrelated grafts.12,15,33-35 Moreover, UCB HCT recipients also received more myeloablative conditioning with TBI, another risk factor for DAH. Overall, we found that UCB grafts were associated with more DAH than PB/BM grafts. In multivariate analysis, UCB HCT was confirmed as an independent risk factor for DAH. Delayed neutrophil engraftment was another risk factor for DAH, particularly for patients receiving PB/BM grafts. It is known that a higher number of stem cells/total nucleated cells expedites engraftment.36 Patients with DAH received fewer total nucleated cells and more often had delayed engraftment than patients without DAH in both the UCB and PB/BM HCT recipients. Among the patients undergoing single unit UCB HCT, the association

Table 3A. Multivariate analysis of risks for diffuse alveolar hemorrhage: neutrophil engraftment.

All HCT Variables Conditioning RIC MAC no TBI MAC TBI Graft source BM/PB UCB ANC engraftment* Yes No

UCB HCT P value

HR (95% CI)

0.05 1.00 0.83 (0.36-1.93) 1.80 (1.03-3.13) 0.01 1.00 2.08 (1.16-3.74)

P value

RR (99% CI)

0.08 1.00 0.69 (0.23-2.04) 1.87 (0.95-3.71) ----

0.07 1.00 2.16 (0.93-5.01)

PB/BM HCT P value

HR (95% CI)

0.60 1.00 1.08 (0.28-4.16) 1.63 (0.62-4.31) ----

0.41 1.00 1.57 (0.53-4.62)

0.02 1.00 5.51 (1.26-24.00)

*As a time-dependent variable.

Table 3B. Multivariate analysis for diffuse alveolar hemorrhage: platelet engraftment.

All HCT Variables Conditioning RIC MAC No TBI MAC TBI Graft source BM/PB UCB Platelet engraftment* Yes No

HR (95% CI)

UCB HCT P value

HR (95% CI)

0.05 1.00 0.78 (0.34-1.79) 1.76 (1.02-3.04) 0.07 1.00 1.65 (0.96-2.83)

RR (99% CI)

0.08 1.00 0.65 (0.22-1.91) 1.78 (0.91-3.48) ----

<0.01 1.00 4.44 (2.10-9.42)

PB/BM HCT P value

0.60 1.00 1.10 (0.32-3.80) 1.64 (0.62-4.32) ----

<0.01 1.00 6.96 (2.39-20.29)

P value

0.15 1.00 2.26 (0.74-6.90)

*As a time-dependent variable. ANC: absolute neutrophil count; BM: bone marrow; HCT: hematopoietic stem cell transplantation; HR: hazard ratio; MAC: myeloablative conditioning; PB: peripheral blood; RIC: reduced intensity conditioning; TBI: total body irradiation; UCB: umbilical cord blood.

haematologica | 2018; 103(12)

2113

F. Keklik et al.

between lower infused cell dose and DAH was even more evident. Most cases of DAH occurred within 2 weeks of neutrophil engraftment. This suggests that a sudden neutrophil influx may contribute to lung injury.1,7,15,37-39 This phenomenon can even occur in a periengraftment period when patients still have neutropenia.15,38,40 Older age (>40 years),3,5,15 severe acute GvHD,3,7,41 and compromised renal function15,41 have each been reported as risk factors for DAH. In our study, neither age nor acute GvHD was recognized as a risk factor. Kidney dysfunction was observed in one-third of the patients with DAH. Clinically apparent coagulopathy, as determined by standard tests, was uncommon in our DAH patients, as in earlier studies.42 Pretransplant respiratory infections have also been described as risk factors for DAH after HCT.10 Although there is no study specifically evaluating the association between smoking and pulmonary complications among HCT recipients, tobacco use and prior lung disease have not been reported as risk factors for DAH.43 Likewise, no correlation was found between smoking status, bronchiolitis index determined by bronchoscopy, or inflammatory cell fractions, and the likelihood of developing DAH in autologous HCT recipients.44 We observed that almost half of the patients with DAH had a history of smoking but were rarely diagnosed with a specific lung

References 1. Wah TM, Moss HA, Robertson RJ, Barnard DL. Pulmonary complications following bone marrow transplantation. Br J Radiol. 2003;76(906):373-379. 2. Yen KT, Lee AS, Krowka MJ, Burger CD. Pulmonary complications in bone marrow transplantation: a practical approach to diagnosis and treatment. Clin Chest Med. 2004;25(1):189-201. 3. Majhail NS, Parks K, Defor TE, Weisdorf DJ. Diffuse alveolar hemorrhage and infection-associated alveolar hemorrhage following hematopoietic stem cell transplantation: related and high-risk clinical syndromes. Biol Blood Marrow Transplant. 2006;12(10):1038-1046. 4. Gupta S, Jain A, Warneke CL, et al. Outcome of alveolar hemorrhage in hematopoietic stem cell transplant recipients. Bone Marrow Transplant. 2007;40(1):71-78. 5. Afessa B, Tefferi A, Litzow MR, Peters SG. Outcome of diffuse alveolar hemorrhage in hematopoietic stem cell transplant recipients. Am J Respir Crit Care Med. 2002;166(10):1364-1368. 6. Franquet T, Muller NL, Lee KS, Gimenez A, Flint JD. High-resolution CT and pathologic findings of noninfectious pulmonary complications after hematopoietic stem cell transplantation. AJR Am J Roentgenol. 2005;184(2):629-637. 7. Roychowdhury M, Pambuccian SE, Aslan DL, et al. Pulmonary complications after bone marrow transplantation: an autopsy study from a large transplantation center. Arch Pathol Lab Med. 2005;129(3):366-371. 8. Baker MS, Diab KJ, Carlos WG, Mathur P.

2114

10.

11.

12.

13.

14.

15.

16.

disease prior to their allogeneic HCT. The therapy of DAH remains empirical and thus inadequate, due in part to the unknown pathogenesis of the condition. Because the immune response and inflammation are suggested to contribute to the pathogenesis of DAH, steroid treatment and mechanical ventilator support for acute respiratory failure are used commonly, although often unsuccessfully, for therapy.21,25,38,45 Etanercept or other anti-cytokine agents,46 drugs targeting coagulopathy such as aminocaproic acid and recombinant factor VIIa,47-49 and prophylactic use of an interleukin-1 receptor antagonist to prevent GvHD50 have also been used. All led to poor outcomes in patients with DAH and the mortality rates for this syndrome remain very high3,9,37,38,51-53 due to respiratory failure, sepsis and multi-organ failure.38,54 Our study confirms that the pathogenesis of DAH is complex, affected by conditioning regimen, graft source, and engraftment kinetics and that the outcome remains poor, particularly for patients requiring intubation/mechanical ventilation. Improved management of DAH awaits better understanding of the complex relationship of these multiple risk factors and the definition of the best strategy to expedite engraftment and limit lung injury. Formal testing in comparative trials of therapeutic strategies is needed to validate approaches and limit the high mortality of this devastating syndrome.

Intrapulmonary recombinant factor VII as an effective treatment for diffuse alveolar hemorrhage: a case series. J Bronchol Interv Pulmonol. 2016;23(3):255-258. Agusti C, Ramirez J, Picado C, et al. Diffuse alveolar hemorrhage in allogeneic bone marrow transplantation. A postmortem study. Am J Respir Crit Care Med. 1995;151(4):1006-1010. Diab M, ZazaDitYafawi J, Soubani AO. Major pulmonary complications after hematopoietic stem cell transplant. Exp Clin Transplant. 2016;14(3):259-270. Ustun C, Courville EL, DeFor T, et al. Myeloablative, but not reduced-intensity, conditioning overcomes the negative effect of flow-cytometric evidence of leukemia in acute myeloid leukemia. Biol Blood Marrow Transplant. 2016;22(4):669-675. Brunstein CG, Gutman JA, Weisdorf DJ, et al. Allogeneic hematopoietic cell transplantation for hematologic malignancy: relative risks and benefits of double umbilical cord blood. Blood. 2010;116(22):4693-4699. Barker JN, Weisdorf DJ, DeFor TE, Blazar BR, Miller JS, Wagner JE. Rapid and complete donor chimerism in adult recipients of unrelated donor umbilical cord blood transplantation after reduced-intensity conditioning. Blood. 2003;102(5):1915-1919. Brunstein CG, Barker JN, Weisdorf DJ, et al. Umbilical cord blood transplantation after nonmyeloablative conditioning: impact on transplantation outcomes in 110 adults with hematologic disease. Blood. 2007;110(8):3064-3070. Robbins RA, Linder J, Stahl MG, et al. Diffuse alveolar hemorrhage in autologous bone marrow transplant recipients. Am J Med. 1989;87(5):511-518. Lin DY. Non-parametric inference for cumulative incidence functions in compet-

17.

18. 19. 20.

21.

22.

23.

24.

25.

ing risks studies. Stat Med. 1997;16(8):901910. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496509. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53(282):457-481. Cox DR. Regression models and lifetables. J Roy Stat Soc Ser B. 1972;34(2): 187-220. Feinstein MB, Mokhtari M, Ferreiro R, Stover DE, Jakubowski A. Fiberoptic bronchoscopy in allogeneic bone marrow transplantation: findings in the era of serum cytomegalovirus antigen surveillance. Chest. 2001;120(4):1094-1100. Heggen J, West C, Olson E, et al. Diffuse alveolar hemorrhage in pediatric hematopoietic cell transplant patients. Pediatrics. 2002;109(5):965-971. Lunde LE, Dasaraju S, Cao Q, et al. Hemorrhagic cystitis after allogeneic hematopoietic cell transplantation: risk factors, graft source and survival. Bone Marrow Transplant. 2015;50(11):14321437. Nevo S, Enger C, Hartley E, et al. Acute bleeding and thrombocytopenia after bone marrow transplantation. Bone Marrow Transplant. 2001;27(1):65-72. Jules-Elysee K, Stover DE, Yahalom J, White DA, Gulati SC. Pulmonary complications in lymphoma patients treated with high-dose therapy autologous bone marrow transplantation. Am Rev Respir Dis. 1992;146(2):485-491. Ben-Abraham R, Paret G, Cohen R, et al. Diffuse alveolar hemorrhage following allogeneic bone marrow transplantation in children. Chest. 2003;124(2):660-664.

haematologica | 2018; 103(12)

Diffuse alveolar hemorrhage after HCT

26. Nusair S, Breuer R, Shapira MY, Berkman N, Or R. Low incidence of pulmonary complications following nonmyeloablative stem cell transplantation. Eur Respir J. 2004;23(3):440-445. 27. Shankar G, Scott Bryson J, Darrell Jennings C, Kaplan AM, Cohen DA. Idiopathic pneumonia syndrome after allogeneic bone marrow transplantation in mice. Role of pretransplant radiation conditioning. Am J Respir Cell Mol Biol. 1999;20(6):1116-1124. 28. Escuissato DL, Warszawiak D, Marchiori E. Differential diagnosis of diffuse alveolar haemorrhage in immunocompromised patients. Curr Opin Infect Dis. 2015;28(4):337-342. 29. Lefrancais E, Ortiz-Munoz G, Caudrillier A, et al. The lung is a site of platelet biogenesis and a reservoir for haematopoietic progenitors. Nature. 2017;544(7648):105-+. 30. Zucker-Franklin D, Philipp CS. Platelet production in the pulmonary capillary bed: new ultrastructural evidence for an old concept. Am J Pathol. 2000;157(1):69-74. 31. Kaufman RM, Airo R, Pollack S, Crosby WH. Circulating megakaryocytes and platelet release in the lung. Blood. 1965;26(6):720-731. 32. Kroll MH, Afshar-Kharghan V. Platelets in pulmonary vascular physiology and pathology. Pulm Circ. 2012;2(3):291-308. 33. Weisdorf D, Eapen M, Ruggeri A, et al. Alternative donor transplantation for older patients with acute myeloid leukemia in first complete remission: a center for international blood and marrow transplant research-eurocord analysis. Biol Blood Marrow Transplant. 2014;20(6):816-822. 34. Brunstein CG, Fuchs EJ, Carter SL, et al. Alternative donor transplantation after reduced intensity conditioning: results of parallel phase 2 trials using partially HLAmismatched related bone marrow or unrelated double umbilical cord blood grafts. Blood. 2011;118(2):282-288. 35. Rodrigues CA, Rocha V, Dreger P, et al. Alternative donor hematopoietic stem cell transplantation for mature lymphoid malignancies after reduced-intensity conditioning regimen: similar outcomes with umbilical cord blood and unrelated donor peripheral blood. Haematologica. 2014;99(2):370377.

haematologica | 2018; 103(12)

36. Jillella AP, Ustun C. What is the optimum number of CD34+ peripheral blood stem cells for an autologous transplant? Stem Cells Dev. 2004;13(6):598-606. 37. Weisdorf DJ. Diffuse alveolar hemorrhage: an evolving problem? Leukemia. 2003;17(6):1049-1050. 38. Metcalf JP, Rennard SI, Reed EC, et al. Corticosteroids as adjunctive therapy for diffuse alveolar hemorrhage associated with bone marrow transplantation. University of Nebraska Medical Center Bone Marrow Transplant Group. Am J Med. 1994;96(4):327-334. 39. Chan CK, Hyland RH, Hutcheon MA. Pulmonary complications following bone marrow transplantation. Clin Chest Med. 1990;11(2):323-332. 40. Capizzi SA, Kumar S, Huneke NE, et al. Peri-engraftment respiratory distress syndrome during autologous hematopoietic stem cell transplantation. Bone Marrow Transplant. 2001;27(12):1299-1303. 41. Pena E, Souza CA, Escuissato DL, et al. Noninfectious pulmonary complications after hematopoietic stem cell transplantation: practical approach to imaging diagnosis. Radiographics. 2014;34(3):663-683. 42. Wanko SO, Broadwater G, Folz RJ, Chao NJ. Diffuse alveolar hemorrhage: retrospective review of clinical outcome in allogeneic transplant recipients treated with aminocaproic acid. Biol Blood Marrow Transplant. 2006;12(9):949-953. 43. Ho VT, Weller E, Lee SJ, Alyea EP, Antin JH, Soiffer RJ. Prognostic factors for early severe pulmonary complications after hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2001;7(4):223-229. 44. Sisson JH, Thompson AB, Anderson JR, et al. Airway inflammation predicts diffuse alveolar hemorrhage during bone marrow transplantation in patients with Hodgkin disease. Am Rev Respir Dis. 1992;146(2):439-443. 45. Graf L, Stern M. Acute phase after haematopoietic stem cell transplantation: bleeding and thrombotic complications. Hamostaseologie. 2012;32(1):56-62. 46. Yanik GA, Horowitz MM, Weisdorf DJ, Logan BR, Ho VT, Soiffer RJ, et al. Randomized, double-blind, placebo-con-

47.

48.

49.

50.

51.

52.

53.

54.

trolled trial of soluble tumor necrosis factor receptor: enbrel (etanercept) for the treatment of idiopathic pneumonia syndrome after allogeneic stem cell transplantation: blood and marrow transplant clinical trials network protocol. Biol Blood Marrow Transplant. 2014;20(6):858-864. Rathi NK, Tanner AR, Dinh A, et al. Low-, medium- and high-dose steroids with or without aminocaproic acid in adult hematopoietic SCT patients with diffuse alveolar hemorrhage. Bone Marrow Transplant. 2015;50(3):420-426. Heslet L, Nielsen JD, Levi M, Sengelov H, Johansson PI. Successful pulmonary administration of activated recombinant factor VII in diffuse alveolar hemorrhage. Crit Care. 2006;10(6):R177. Estella A, Jareno A, Perez-Bello Fontaina L. Intrapulmonary administration of recombinant activated factor VII in diffuse alveolar haemorrhage: a report of two case stories. Cases J. 2008;1(1):150. Antin JH, Weisdorf D, Neuberg D, et al. Interleukin-1 blockade does not prevent acute graft-versus-host disease: results of a randomized, double-blind, placebo-controlled trial of interleukin-1 receptor antagonist in allogeneic bone marrow transplantation. Blood. 2002;100(10):3479-3482. Raptis A, Mavroudis D, Suffredini A, Molldrem J, Rhee FV, Childs R, et al. Highdose corticosteroid therapy for diffuse alveolar hemorrhage in allogeneic bone marrow stem cell transplant recipients. Bone Marrow Transplant. 1999;24(8):879883. Srivastava A, Gottlieb D, Bradstock KF. Diffuse alveolar haemorrhage associated with microangiopathy after allogeneic bone marrow transplantation. Bone Marrow Transplant. 1995;15(6):863-867. Huaringa AJ, Leyva FJ, Giralt SA, Blanco J, Signes-Costa J, Velarde H, et al. Outcome of bone marrow transplantation patients requiring mechanical ventilation. Crit Care Med. 2000;28(4):1014-1017. Lewis ID, DeFor T, Weisdorf DJ. Increasing incidence of diffuse alveolar hemorrhage following allogeneic bone marrow transplantation: cryptic etiology and uncertain therapy. Bone Marrow Transplant. 2000;26(5):539-543.

2115